diff --git "a/checkpoint-92750/trainer_state.json" "b/checkpoint-92750/trainer_state.json"
--- "a/checkpoint-92750/trainer_state.json"
+++ "b/checkpoint-92750/trainer_state.json"
@@ -1,5 +1,5 @@
 {
-  "best_metric": 3.299715518951416,
+  "best_metric": 3.300025701522827,
   "best_model_checkpoint": "/scratch/cl5625/exceptions/models/100M_low_2000_634/checkpoint-90000",
   "epoch": 10.0,
   "eval_steps": 1000,
@@ -10,13815 +10,13815 @@
   "log_history": [
     {
       "epoch": 0.005390835579514825,
-      "grad_norm": 1.3841798305511475,
+      "grad_norm": 1.4145963191986084,
       "learning_rate": 0.0003,
-      "loss": 8.6258,
+      "loss": 8.6259,
       "step": 50
     },
     {
       "epoch": 0.01078167115902965,
-      "grad_norm": 2.7142674922943115,
+      "grad_norm": 3.1651010513305664,
       "learning_rate": 0.0006,
-      "loss": 6.9518,
+      "loss": 6.9317,
       "step": 100
     },
     {
       "epoch": 0.016172506738544475,
-      "grad_norm": 1.7501500844955444,
+      "grad_norm": 2.394423723220825,
       "learning_rate": 0.0005996762007555315,
-      "loss": 6.4766,
+      "loss": 6.4779,
       "step": 150
     },
     {
       "epoch": 0.0215633423180593,
-      "grad_norm": 2.089492082595825,
+      "grad_norm": 1.727589726448059,
       "learning_rate": 0.000599352401511063,
-      "loss": 6.2242,
+      "loss": 6.2178,
       "step": 200
     },
     {
       "epoch": 0.026954177897574125,
-      "grad_norm": 1.0795507431030273,
+      "grad_norm": 1.7129422426223755,
       "learning_rate": 0.0005990286022665946,
-      "loss": 6.0814,
+      "loss": 6.0805,
       "step": 250
     },
     {
       "epoch": 0.03234501347708895,
-      "grad_norm": 2.468144655227661,
+      "grad_norm": 1.352902889251709,
       "learning_rate": 0.0005987048030221263,
-      "loss": 5.9861,
+      "loss": 5.9783,
       "step": 300
     },
     {
       "epoch": 0.03773584905660377,
-      "grad_norm": 1.4130438566207886,
+      "grad_norm": 2.4004905223846436,
       "learning_rate": 0.0005983810037776578,
-      "loss": 5.8758,
+      "loss": 5.8715,
       "step": 350
     },
     {
       "epoch": 0.0431266846361186,
-      "grad_norm": 1.5332931280136108,
+      "grad_norm": 1.5558369159698486,
       "learning_rate": 0.0005980572045331894,
-      "loss": 5.7848,
+      "loss": 5.7796,
       "step": 400
     },
     {
       "epoch": 0.04851752021563342,
-      "grad_norm": 1.451593279838562,
+      "grad_norm": 1.405604362487793,
       "learning_rate": 0.0005977334052887209,
-      "loss": 5.713,
+      "loss": 5.7133,
       "step": 450
     },
     {
       "epoch": 0.05390835579514825,
-      "grad_norm": 1.3739091157913208,
+      "grad_norm": 1.7021785974502563,
       "learning_rate": 0.0005974096060442526,
-      "loss": 5.6508,
+      "loss": 5.6534,
       "step": 500
     },
     {
       "epoch": 0.05929919137466307,
-      "grad_norm": 1.2970751523971558,
+      "grad_norm": 1.6914323568344116,
       "learning_rate": 0.0005970858067997841,
-      "loss": 5.5723,
+      "loss": 5.5752,
       "step": 550
     },
     {
       "epoch": 0.0646900269541779,
-      "grad_norm": 1.2038410902023315,
+      "grad_norm": 1.8869633674621582,
       "learning_rate": 0.0005967620075553157,
-      "loss": 5.5093,
+      "loss": 5.5078,
       "step": 600
     },
     {
       "epoch": 0.07008086253369272,
-      "grad_norm": 1.2364015579223633,
+      "grad_norm": 0.8667061924934387,
       "learning_rate": 0.0005964382083108472,
-      "loss": 5.4167,
+      "loss": 5.418,
       "step": 650
     },
     {
       "epoch": 0.07547169811320754,
-      "grad_norm": 1.1776434183120728,
+      "grad_norm": 1.124100685119629,
       "learning_rate": 0.0005961144090663788,
-      "loss": 5.3687,
+      "loss": 5.3656,
       "step": 700
     },
     {
       "epoch": 0.08086253369272237,
-      "grad_norm": 1.2404123544692993,
+      "grad_norm": 1.3503167629241943,
       "learning_rate": 0.0005957906098219104,
-      "loss": 5.2965,
+      "loss": 5.2977,
       "step": 750
     },
     {
       "epoch": 0.0862533692722372,
-      "grad_norm": 0.8565037250518799,
+      "grad_norm": 1.0930562019348145,
       "learning_rate": 0.0005954668105774419,
-      "loss": 5.2536,
+      "loss": 5.2523,
       "step": 800
     },
     {
       "epoch": 0.09164420485175202,
-      "grad_norm": 0.8625679612159729,
+      "grad_norm": 0.9786685705184937,
       "learning_rate": 0.0005951430113329735,
-      "loss": 5.1917,
+      "loss": 5.1895,
       "step": 850
     },
     {
       "epoch": 0.09703504043126684,
-      "grad_norm": 1.4676177501678467,
+      "grad_norm": 1.2884083986282349,
       "learning_rate": 0.0005948192120885051,
-      "loss": 5.1705,
+      "loss": 5.168,
       "step": 900
     },
     {
       "epoch": 0.10242587601078167,
-      "grad_norm": 1.3620243072509766,
+      "grad_norm": 0.9569465517997742,
       "learning_rate": 0.0005944954128440366,
-      "loss": 5.129,
+      "loss": 5.1311,
       "step": 950
     },
     {
       "epoch": 0.1078167115902965,
-      "grad_norm": 1.802571177482605,
+      "grad_norm": 1.0159235000610352,
       "learning_rate": 0.0005941716135995682,
-      "loss": 5.0781,
+      "loss": 5.0746,
       "step": 1000
     },
     {
       "epoch": 0.1078167115902965,
-      "eval_accuracy": 0.22607515020446844,
-      "eval_loss": 5.030185699462891,
-      "eval_runtime": 184.9577,
-      "eval_samples_per_second": 97.379,
-      "eval_steps_per_second": 6.088,
+      "eval_accuracy": 0.22805328468001887,
+      "eval_loss": 5.009091377258301,
+      "eval_runtime": 181.7061,
+      "eval_samples_per_second": 99.122,
+      "eval_steps_per_second": 6.197,
       "step": 1000
     },
     {
       "epoch": 0.11320754716981132,
-      "grad_norm": 1.0640990734100342,
+      "grad_norm": 1.0572841167449951,
       "learning_rate": 0.0005938478143550997,
-      "loss": 5.0384,
+      "loss": 5.0385,
       "step": 1050
     },
     {
       "epoch": 0.11859838274932614,
-      "grad_norm": 0.9512579441070557,
+      "grad_norm": 1.0152630805969238,
       "learning_rate": 0.0005935240151106314,
-      "loss": 5.01,
+      "loss": 5.0085,
       "step": 1100
     },
     {
       "epoch": 0.12398921832884097,
-      "grad_norm": 1.4336494207382202,
+      "grad_norm": 1.1099059581756592,
       "learning_rate": 0.0005932002158661629,
-      "loss": 4.986,
+      "loss": 4.9873,
       "step": 1150
     },
     {
       "epoch": 0.1293800539083558,
-      "grad_norm": 0.986595094203949,
+      "grad_norm": 0.9290792346000671,
       "learning_rate": 0.0005928764166216945,
-      "loss": 4.9237,
+      "loss": 4.9231,
       "step": 1200
     },
     {
       "epoch": 0.1347708894878706,
-      "grad_norm": 0.9924365878105164,
+      "grad_norm": 0.9619135856628418,
       "learning_rate": 0.000592552617377226,
-      "loss": 4.8993,
+      "loss": 4.901,
       "step": 1250
     },
     {
       "epoch": 0.14016172506738545,
-      "grad_norm": 1.405576229095459,
+      "grad_norm": 1.0886056423187256,
       "learning_rate": 0.0005922288181327577,
-      "loss": 4.8778,
+      "loss": 4.8781,
       "step": 1300
     },
     {
       "epoch": 0.14555256064690028,
-      "grad_norm": 0.7532362341880798,
+      "grad_norm": 0.8178435564041138,
       "learning_rate": 0.0005919050188882893,
-      "loss": 4.8412,
+      "loss": 4.8423,
       "step": 1350
     },
     {
       "epoch": 0.1509433962264151,
-      "grad_norm": 0.9290658235549927,
+      "grad_norm": 0.95741206407547,
       "learning_rate": 0.0005915812196438207,
-      "loss": 4.8361,
+      "loss": 4.8349,
       "step": 1400
     },
     {
       "epoch": 0.15633423180592992,
-      "grad_norm": 1.0212785005569458,
+      "grad_norm": 1.099839448928833,
       "learning_rate": 0.0005912574203993524,
-      "loss": 4.8431,
+      "loss": 4.8422,
       "step": 1450
     },
     {
       "epoch": 0.16172506738544473,
-      "grad_norm": 1.3435090780258179,
+      "grad_norm": 0.8492836952209473,
       "learning_rate": 0.0005909336211548839,
-      "loss": 4.801,
+      "loss": 4.8014,
       "step": 1500
     },
     {
       "epoch": 0.16711590296495957,
-      "grad_norm": 1.0386897325515747,
+      "grad_norm": 0.9975847601890564,
       "learning_rate": 0.0005906098219104155,
-      "loss": 4.7558,
+      "loss": 4.7623,
       "step": 1550
     },
     {
       "epoch": 0.1725067385444744,
-      "grad_norm": 0.8441452383995056,
+      "grad_norm": 0.9260163307189941,
       "learning_rate": 0.000590286022665947,
-      "loss": 4.749,
+      "loss": 4.7475,
       "step": 1600
     },
     {
       "epoch": 0.1778975741239892,
-      "grad_norm": 0.9853036999702454,
+      "grad_norm": 0.8640759587287903,
       "learning_rate": 0.0005899622234214787,
-      "loss": 4.7201,
+      "loss": 4.7228,
       "step": 1650
     },
     {
       "epoch": 0.18328840970350405,
-      "grad_norm": 0.9956970810890198,
+      "grad_norm": 1.009647011756897,
       "learning_rate": 0.0005896384241770102,
-      "loss": 4.6933,
+      "loss": 4.6961,
       "step": 1700
     },
     {
       "epoch": 0.18867924528301888,
-      "grad_norm": 0.8019234538078308,
+      "grad_norm": 0.8946645259857178,
       "learning_rate": 0.0005893146249325418,
-      "loss": 4.7021,
+      "loss": 4.705,
       "step": 1750
     },
     {
       "epoch": 0.1940700808625337,
-      "grad_norm": 0.8837598562240601,
+      "grad_norm": 1.1306875944137573,
       "learning_rate": 0.0005889908256880733,
-      "loss": 4.6704,
+      "loss": 4.6712,
       "step": 1800
     },
     {
       "epoch": 0.19946091644204852,
-      "grad_norm": 1.046995997428894,
+      "grad_norm": 1.0593204498291016,
       "learning_rate": 0.0005886670264436049,
-      "loss": 4.6374,
+      "loss": 4.6409,
       "step": 1850
     },
     {
       "epoch": 0.20485175202156333,
-      "grad_norm": 1.0166348218917847,
+      "grad_norm": 0.8434944748878479,
       "learning_rate": 0.0005883432271991365,
-      "loss": 4.6279,
+      "loss": 4.6297,
       "step": 1900
     },
     {
       "epoch": 0.21024258760107817,
-      "grad_norm": 0.8063094615936279,
+      "grad_norm": 0.8913785815238953,
       "learning_rate": 0.0005880194279546681,
-      "loss": 4.6256,
+      "loss": 4.6276,
       "step": 1950
     },
     {
       "epoch": 0.215633423180593,
-      "grad_norm": 0.9398548603057861,
+      "grad_norm": 0.7697395086288452,
       "learning_rate": 0.0005876956287101996,
-      "loss": 4.5763,
+      "loss": 4.5779,
       "step": 2000
     },
     {
       "epoch": 0.215633423180593,
-      "eval_accuracy": 0.2714252357849155,
-      "eval_loss": 4.502427577972412,
-      "eval_runtime": 183.5351,
-      "eval_samples_per_second": 98.134,
-      "eval_steps_per_second": 6.135,
+      "eval_accuracy": 0.2714229540742714,
+      "eval_loss": 4.502364158630371,
+      "eval_runtime": 181.696,
+      "eval_samples_per_second": 99.127,
+      "eval_steps_per_second": 6.197,
       "step": 2000
     },
     {
       "epoch": 0.2210242587601078,
-      "grad_norm": 0.980785608291626,
+      "grad_norm": 0.9948697686195374,
       "learning_rate": 0.0005873718294657312,
-      "loss": 4.5575,
+      "loss": 4.5556,
       "step": 2050
     },
     {
       "epoch": 0.22641509433962265,
-      "grad_norm": 0.8288772106170654,
+      "grad_norm": 0.8819608688354492,
       "learning_rate": 0.0005870480302212628,
-      "loss": 4.5462,
+      "loss": 4.5434,
       "step": 2100
     },
     {
       "epoch": 0.23180592991913745,
-      "grad_norm": 0.9006871581077576,
+      "grad_norm": 0.8973624110221863,
       "learning_rate": 0.0005867242309767943,
-      "loss": 4.5289,
+      "loss": 4.5261,
       "step": 2150
     },
     {
       "epoch": 0.2371967654986523,
-      "grad_norm": 0.7568821310997009,
+      "grad_norm": 0.8225829601287842,
       "learning_rate": 0.0005864004317323259,
-      "loss": 4.4877,
+      "loss": 4.4894,
       "step": 2200
     },
     {
       "epoch": 0.24258760107816713,
-      "grad_norm": 0.8699207305908203,
+      "grad_norm": 0.7101904153823853,
       "learning_rate": 0.0005860766324878575,
-      "loss": 4.5022,
+      "loss": 4.5017,
       "step": 2250
     },
     {
       "epoch": 0.24797843665768193,
-      "grad_norm": 0.9681375026702881,
+      "grad_norm": 0.8717466592788696,
       "learning_rate": 0.000585752833243389,
-      "loss": 4.4781,
+      "loss": 4.4742,
       "step": 2300
     },
     {
       "epoch": 0.25336927223719674,
-      "grad_norm": 1.0356428623199463,
+      "grad_norm": 0.9836375117301941,
       "learning_rate": 0.0005854290339989206,
-      "loss": 4.4547,
+      "loss": 4.4495,
       "step": 2350
     },
     {
       "epoch": 0.2587601078167116,
-      "grad_norm": 1.0217968225479126,
+      "grad_norm": 0.9477341175079346,
       "learning_rate": 0.0005851052347544521,
-      "loss": 4.4496,
+      "loss": 4.4442,
       "step": 2400
     },
     {
       "epoch": 0.2641509433962264,
-      "grad_norm": 0.9390143752098083,
+      "grad_norm": 0.7532225847244263,
       "learning_rate": 0.0005847814355099838,
-      "loss": 4.4324,
+      "loss": 4.4306,
       "step": 2450
     },
     {
       "epoch": 0.2695417789757412,
-      "grad_norm": 0.8502791523933411,
+      "grad_norm": 0.6514259576797485,
       "learning_rate": 0.0005844576362655154,
-      "loss": 4.4238,
+      "loss": 4.4197,
       "step": 2500
     },
     {
       "epoch": 0.2749326145552561,
-      "grad_norm": 0.9871562719345093,
+      "grad_norm": 0.9490562081336975,
       "learning_rate": 0.0005841338370210469,
-      "loss": 4.3975,
+      "loss": 4.3958,
       "step": 2550
     },
     {
       "epoch": 0.2803234501347709,
-      "grad_norm": 0.9953110814094543,
+      "grad_norm": 0.856418788433075,
       "learning_rate": 0.0005838100377765785,
-      "loss": 4.3725,
+      "loss": 4.3722,
       "step": 2600
     },
     {
       "epoch": 0.2857142857142857,
-      "grad_norm": 0.8370280861854553,
+      "grad_norm": 0.8336089849472046,
       "learning_rate": 0.0005834862385321101,
-      "loss": 4.3714,
+      "loss": 4.3709,
       "step": 2650
     },
     {
       "epoch": 0.29110512129380056,
-      "grad_norm": 0.8906695246696472,
+      "grad_norm": 0.8533580899238586,
       "learning_rate": 0.0005831624392876417,
-      "loss": 4.3636,
+      "loss": 4.3617,
       "step": 2700
     },
     {
       "epoch": 0.29649595687331537,
-      "grad_norm": 0.8926903009414673,
+      "grad_norm": 1.0777424573898315,
       "learning_rate": 0.0005828386400431731,
-      "loss": 4.3712,
+      "loss": 4.3696,
       "step": 2750
     },
     {
       "epoch": 0.3018867924528302,
-      "grad_norm": 0.6380210518836975,
+      "grad_norm": 0.7660175561904907,
       "learning_rate": 0.0005825148407987048,
-      "loss": 4.3652,
+      "loss": 4.3629,
       "step": 2800
     },
     {
       "epoch": 0.30727762803234504,
-      "grad_norm": 0.9012649655342102,
+      "grad_norm": 0.8766116499900818,
       "learning_rate": 0.0005821910415542363,
-      "loss": 4.3264,
+      "loss": 4.3306,
       "step": 2850
     },
     {
       "epoch": 0.31266846361185985,
-      "grad_norm": 0.8436964750289917,
+      "grad_norm": 0.9792775511741638,
       "learning_rate": 0.0005818672423097679,
-      "loss": 4.3127,
+      "loss": 4.3132,
       "step": 2900
     },
     {
       "epoch": 0.31805929919137466,
-      "grad_norm": 0.8248408436775208,
+      "grad_norm": 0.8375799059867859,
       "learning_rate": 0.0005815434430652994,
-      "loss": 4.3424,
+      "loss": 4.3457,
       "step": 2950
     },
     {
       "epoch": 0.32345013477088946,
-      "grad_norm": 0.7463982105255127,
+      "grad_norm": 0.7174794673919678,
       "learning_rate": 0.0005812196438208311,
-      "loss": 4.3107,
+      "loss": 4.3057,
       "step": 3000
     },
     {
       "epoch": 0.32345013477088946,
-      "eval_accuracy": 0.2991547565898248,
-      "eval_loss": 4.231179237365723,
-      "eval_runtime": 183.5046,
-      "eval_samples_per_second": 98.15,
-      "eval_steps_per_second": 6.136,
+      "eval_accuracy": 0.29961283716485065,
+      "eval_loss": 4.228518486022949,
+      "eval_runtime": 181.6163,
+      "eval_samples_per_second": 99.171,
+      "eval_steps_per_second": 6.2,
       "step": 3000
     },
     {
       "epoch": 0.3288409703504043,
-      "grad_norm": 0.7356308102607727,
+      "grad_norm": 0.6986973285675049,
       "learning_rate": 0.0005808958445763626,
-      "loss": 4.3028,
+      "loss": 4.2999,
       "step": 3050
     },
     {
       "epoch": 0.33423180592991913,
-      "grad_norm": 0.7332613468170166,
+      "grad_norm": 0.7411074042320251,
       "learning_rate": 0.0005805720453318942,
-      "loss": 4.281,
+      "loss": 4.2815,
       "step": 3100
     },
     {
       "epoch": 0.33962264150943394,
-      "grad_norm": 0.9393066763877869,
+      "grad_norm": 0.8088811635971069,
       "learning_rate": 0.0005802482460874257,
-      "loss": 4.2707,
+      "loss": 4.2689,
       "step": 3150
     },
     {
       "epoch": 0.3450134770889488,
-      "grad_norm": 0.7228361964225769,
+      "grad_norm": 0.80063396692276,
       "learning_rate": 0.0005799244468429573,
-      "loss": 4.2769,
+      "loss": 4.2754,
       "step": 3200
     },
     {
       "epoch": 0.3504043126684636,
-      "grad_norm": 0.7841880917549133,
+      "grad_norm": 0.8073013424873352,
       "learning_rate": 0.0005796006475984889,
-      "loss": 4.2499,
+      "loss": 4.2485,
       "step": 3250
     },
     {
       "epoch": 0.3557951482479784,
-      "grad_norm": 0.7926186919212341,
+      "grad_norm": 0.7534303665161133,
       "learning_rate": 0.0005792768483540205,
-      "loss": 4.2548,
+      "loss": 4.2577,
       "step": 3300
     },
     {
       "epoch": 0.3611859838274933,
-      "grad_norm": 0.8690990805625916,
+      "grad_norm": 0.9090972542762756,
       "learning_rate": 0.000578953049109552,
-      "loss": 4.254,
+      "loss": 4.2544,
       "step": 3350
     },
     {
       "epoch": 0.3665768194070081,
-      "grad_norm": 0.7845708727836609,
+      "grad_norm": 0.741401731967926,
       "learning_rate": 0.0005786292498650836,
-      "loss": 4.2325,
+      "loss": 4.231,
       "step": 3400
     },
     {
       "epoch": 0.3719676549865229,
-      "grad_norm": 0.8615543246269226,
+      "grad_norm": 0.9652010202407837,
       "learning_rate": 0.0005783054506206152,
-      "loss": 4.2247,
+      "loss": 4.223,
       "step": 3450
     },
     {
       "epoch": 0.37735849056603776,
-      "grad_norm": 0.7429792284965515,
+      "grad_norm": 0.6787908673286438,
       "learning_rate": 0.0005779816513761467,
-      "loss": 4.216,
+      "loss": 4.2163,
       "step": 3500
     },
     {
       "epoch": 0.38274932614555257,
-      "grad_norm": 0.7726993560791016,
+      "grad_norm": 0.63587486743927,
       "learning_rate": 0.0005776578521316782,
-      "loss": 4.1962,
+      "loss": 4.1965,
       "step": 3550
     },
     {
       "epoch": 0.3881401617250674,
-      "grad_norm": 0.7264295220375061,
+      "grad_norm": 0.8312519788742065,
       "learning_rate": 0.0005773340528872099,
-      "loss": 4.2044,
+      "loss": 4.2034,
       "step": 3600
     },
     {
       "epoch": 0.3935309973045822,
-      "grad_norm": 0.7918074131011963,
+      "grad_norm": 0.82889324426651,
       "learning_rate": 0.0005770102536427414,
-      "loss": 4.219,
+      "loss": 4.2196,
       "step": 3650
     },
     {
       "epoch": 0.39892183288409705,
-      "grad_norm": 0.7423751950263977,
+      "grad_norm": 0.6806962490081787,
       "learning_rate": 0.000576686454398273,
-      "loss": 4.2045,
+      "loss": 4.2082,
       "step": 3700
     },
     {
       "epoch": 0.40431266846361186,
-      "grad_norm": 0.6395682096481323,
+      "grad_norm": 0.7568163871765137,
       "learning_rate": 0.0005763626551538045,
-      "loss": 4.1897,
+      "loss": 4.1913,
       "step": 3750
     },
     {
       "epoch": 0.40970350404312667,
-      "grad_norm": 0.6632921695709229,
+      "grad_norm": 0.6309810876846313,
       "learning_rate": 0.0005760388559093362,
-      "loss": 4.182,
+      "loss": 4.1834,
       "step": 3800
     },
     {
       "epoch": 0.41509433962264153,
-      "grad_norm": 0.6251309514045715,
+      "grad_norm": 0.6354190707206726,
       "learning_rate": 0.0005757150566648678,
-      "loss": 4.1772,
+      "loss": 4.1777,
       "step": 3850
     },
     {
       "epoch": 0.42048517520215634,
-      "grad_norm": 0.8533205986022949,
+      "grad_norm": 0.8854219317436218,
       "learning_rate": 0.0005753912574203993,
-      "loss": 4.1699,
+      "loss": 4.1682,
       "step": 3900
     },
     {
       "epoch": 0.42587601078167114,
-      "grad_norm": 0.707554817199707,
+      "grad_norm": 0.7516558766365051,
       "learning_rate": 0.0005750674581759309,
-      "loss": 4.1668,
+      "loss": 4.1666,
       "step": 3950
     },
     {
       "epoch": 0.431266846361186,
-      "grad_norm": 0.7947620153427124,
+      "grad_norm": 0.8482111692428589,
       "learning_rate": 0.0005747436589314624,
-      "loss": 4.1524,
+      "loss": 4.154,
       "step": 4000
     },
     {
       "epoch": 0.431266846361186,
-      "eval_accuracy": 0.3127696153503061,
-      "eval_loss": 4.084148406982422,
-      "eval_runtime": 183.63,
-      "eval_samples_per_second": 98.083,
-      "eval_steps_per_second": 6.132,
+      "eval_accuracy": 0.3124987437009846,
+      "eval_loss": 4.088163375854492,
+      "eval_runtime": 181.3707,
+      "eval_samples_per_second": 99.305,
+      "eval_steps_per_second": 6.208,
       "step": 4000
     },
     {
       "epoch": 0.4366576819407008,
-      "grad_norm": 0.5997743606567383,
+      "grad_norm": 0.7016038298606873,
       "learning_rate": 0.0005744198596869941,
-      "loss": 4.1636,
+      "loss": 4.1634,
       "step": 4050
     },
     {
       "epoch": 0.4420485175202156,
-      "grad_norm": 0.7036544680595398,
+      "grad_norm": 0.5946022868156433,
       "learning_rate": 0.0005740960604425255,
-      "loss": 4.156,
+      "loss": 4.1587,
       "step": 4100
     },
     {
       "epoch": 0.4474393530997305,
-      "grad_norm": 0.7143358588218689,
+      "grad_norm": 0.6194251775741577,
       "learning_rate": 0.0005737722611980572,
-      "loss": 4.1384,
+      "loss": 4.1437,
       "step": 4150
     },
     {
       "epoch": 0.4528301886792453,
-      "grad_norm": 0.6587009429931641,
+      "grad_norm": 0.6552414894104004,
       "learning_rate": 0.0005734484619535887,
-      "loss": 4.113,
+      "loss": 4.1145,
       "step": 4200
     },
     {
       "epoch": 0.4582210242587601,
-      "grad_norm": 0.704142153263092,
+      "grad_norm": 0.6080745458602905,
       "learning_rate": 0.0005731246627091203,
-      "loss": 4.1487,
+      "loss": 4.1491,
       "step": 4250
     },
     {
       "epoch": 0.4636118598382749,
-      "grad_norm": 0.5820618867874146,
+      "grad_norm": 0.6025022864341736,
       "learning_rate": 0.0005728008634646518,
-      "loss": 4.1219,
+      "loss": 4.1221,
       "step": 4300
     },
     {
       "epoch": 0.46900269541778977,
-      "grad_norm": 0.6929563879966736,
+      "grad_norm": 0.7329853177070618,
       "learning_rate": 0.0005724770642201835,
-      "loss": 4.1263,
+      "loss": 4.1343,
       "step": 4350
     },
     {
       "epoch": 0.4743935309973046,
-      "grad_norm": 0.7341722846031189,
+      "grad_norm": 0.6319265365600586,
       "learning_rate": 0.000572153264975715,
-      "loss": 4.1214,
+      "loss": 4.1239,
       "step": 4400
     },
     {
       "epoch": 0.4797843665768194,
-      "grad_norm": 0.5978211164474487,
+      "grad_norm": 0.5935447216033936,
       "learning_rate": 0.0005718294657312466,
-      "loss": 4.0946,
+      "loss": 4.0961,
       "step": 4450
     },
     {
       "epoch": 0.48517520215633425,
-      "grad_norm": 0.6326689124107361,
+      "grad_norm": 0.6037242412567139,
       "learning_rate": 0.0005715056664867781,
-      "loss": 4.0929,
+      "loss": 4.0931,
       "step": 4500
     },
     {
       "epoch": 0.49056603773584906,
-      "grad_norm": 0.6336254477500916,
+      "grad_norm": 0.7396536469459534,
       "learning_rate": 0.0005711818672423097,
-      "loss": 4.0867,
+      "loss": 4.0852,
       "step": 4550
     },
     {
       "epoch": 0.49595687331536387,
-      "grad_norm": 0.6374770998954773,
+      "grad_norm": 0.6765477061271667,
       "learning_rate": 0.0005708580679978413,
-      "loss": 4.0936,
+      "loss": 4.0949,
       "step": 4600
     },
     {
       "epoch": 0.5013477088948787,
-      "grad_norm": 0.7158097624778748,
+      "grad_norm": 0.7021300196647644,
       "learning_rate": 0.0005705342687533729,
-      "loss": 4.0868,
+      "loss": 4.087,
       "step": 4650
     },
     {
       "epoch": 0.5067385444743935,
-      "grad_norm": 0.6559752225875854,
+      "grad_norm": 0.6416023969650269,
       "learning_rate": 0.0005702104695089044,
-      "loss": 4.0694,
+      "loss": 4.069,
       "step": 4700
     },
     {
       "epoch": 0.5121293800539084,
-      "grad_norm": 0.6569370627403259,
+      "grad_norm": 0.6855347752571106,
       "learning_rate": 0.000569886670264436,
-      "loss": 4.0812,
+      "loss": 4.0833,
       "step": 4750
     },
     {
       "epoch": 0.5175202156334232,
-      "grad_norm": 0.6718046069145203,
+      "grad_norm": 0.5605803728103638,
       "learning_rate": 0.0005695628710199675,
-      "loss": 4.0682,
+      "loss": 4.071,
       "step": 4800
     },
     {
       "epoch": 0.522911051212938,
-      "grad_norm": 0.6529590487480164,
+      "grad_norm": 0.6609882116317749,
       "learning_rate": 0.0005692390717754991,
-      "loss": 4.0635,
+      "loss": 4.0662,
       "step": 4850
     },
     {
       "epoch": 0.5283018867924528,
-      "grad_norm": 0.6065393686294556,
+      "grad_norm": 0.7520987391471863,
       "learning_rate": 0.0005689152725310306,
-      "loss": 4.0566,
+      "loss": 4.0593,
       "step": 4900
     },
     {
       "epoch": 0.5336927223719676,
-      "grad_norm": 0.711345911026001,
+      "grad_norm": 0.7330611348152161,
       "learning_rate": 0.0005685914732865623,
-      "loss": 4.0593,
+      "loss": 4.0614,
       "step": 4950
     },
     {
       "epoch": 0.5390835579514824,
-      "grad_norm": 0.6914499998092651,
+      "grad_norm": 0.5883002877235413,
       "learning_rate": 0.0005682676740420939,
-      "loss": 4.0724,
+      "loss": 4.0678,
       "step": 5000
     },
     {
       "epoch": 0.5390835579514824,
-      "eval_accuracy": 0.32181616344262765,
-      "eval_loss": 3.98882794380188,
-      "eval_runtime": 183.5537,
-      "eval_samples_per_second": 98.124,
-      "eval_steps_per_second": 6.134,
+      "eval_accuracy": 0.32201423765711346,
+      "eval_loss": 3.987787961959839,
+      "eval_runtime": 181.5162,
+      "eval_samples_per_second": 99.225,
+      "eval_steps_per_second": 6.203,
       "step": 5000
     },
     {
       "epoch": 0.5444743935309974,
-      "grad_norm": 0.6853660345077515,
+      "grad_norm": 0.7446970343589783,
       "learning_rate": 0.0005679438747976254,
-      "loss": 4.0452,
+      "loss": 4.0443,
       "step": 5050
     },
     {
       "epoch": 0.5498652291105122,
-      "grad_norm": 0.6142929792404175,
+      "grad_norm": 0.5801444053649902,
       "learning_rate": 0.000567620075553157,
-      "loss": 4.0442,
+      "loss": 4.0443,
       "step": 5100
     },
     {
       "epoch": 0.555256064690027,
-      "grad_norm": 0.6437095999717712,
+      "grad_norm": 0.6403396129608154,
       "learning_rate": 0.0005672962763086886,
-      "loss": 4.0475,
+      "loss": 4.048,
       "step": 5150
     },
     {
       "epoch": 0.5606469002695418,
-      "grad_norm": 0.7355647683143616,
+      "grad_norm": 0.6608542799949646,
       "learning_rate": 0.0005669724770642202,
-      "loss": 4.0336,
+      "loss": 4.0321,
       "step": 5200
     },
     {
       "epoch": 0.5660377358490566,
-      "grad_norm": 0.6523656249046326,
+      "grad_norm": 0.6073299050331116,
       "learning_rate": 0.0005666486778197517,
-      "loss": 4.0297,
+      "loss": 4.0307,
       "step": 5250
     },
     {
       "epoch": 0.5714285714285714,
-      "grad_norm": 0.6674039959907532,
+      "grad_norm": 0.6488654017448425,
       "learning_rate": 0.0005663248785752833,
-      "loss": 4.0411,
+      "loss": 4.0401,
       "step": 5300
     },
     {
       "epoch": 0.5768194070080862,
-      "grad_norm": 0.599259078502655,
+      "grad_norm": 0.6238702535629272,
       "learning_rate": 0.0005660010793308148,
       "loss": 4.0249,
       "step": 5350
     },
     {
       "epoch": 0.5822102425876011,
-      "grad_norm": 0.6059934496879578,
+      "grad_norm": 0.6104423999786377,
       "learning_rate": 0.0005656772800863465,
-      "loss": 4.0416,
+      "loss": 4.0438,
       "step": 5400
     },
     {
       "epoch": 0.5876010781671159,
-      "grad_norm": 0.5647556185722351,
+      "grad_norm": 0.5740390419960022,
       "learning_rate": 0.0005653534808418779,
-      "loss": 4.0184,
+      "loss": 4.0193,
       "step": 5450
     },
     {
       "epoch": 0.5929919137466307,
-      "grad_norm": 0.6619639992713928,
+      "grad_norm": 0.6076253056526184,
       "learning_rate": 0.0005650296815974096,
-      "loss": 4.0231,
+      "loss": 4.023,
       "step": 5500
     },
     {
       "epoch": 0.5983827493261455,
-      "grad_norm": 0.6255214810371399,
+      "grad_norm": 0.5750396251678467,
       "learning_rate": 0.0005647058823529411,
-      "loss": 4.0157,
+      "loss": 4.0152,
       "step": 5550
     },
     {
       "epoch": 0.6037735849056604,
-      "grad_norm": 0.5285488963127136,
+      "grad_norm": 0.5481240153312683,
       "learning_rate": 0.0005643820831084727,
-      "loss": 4.005,
+      "loss": 4.0053,
       "step": 5600
     },
     {
       "epoch": 0.6091644204851752,
-      "grad_norm": 0.8316670656204224,
+      "grad_norm": 0.7279238104820251,
       "learning_rate": 0.0005640582838640042,
-      "loss": 4.0181,
+      "loss": 4.0191,
       "step": 5650
     },
     {
       "epoch": 0.6145552560646901,
-      "grad_norm": 0.7264182567596436,
+      "grad_norm": 0.7080326676368713,
       "learning_rate": 0.0005637344846195358,
-      "loss": 4.0123,
+      "loss": 4.0134,
       "step": 5700
     },
     {
       "epoch": 0.6199460916442049,
-      "grad_norm": 0.6089359521865845,
+      "grad_norm": 0.588047206401825,
       "learning_rate": 0.0005634106853750674,
-      "loss": 4.0165,
+      "loss": 4.0209,
       "step": 5750
     },
     {
       "epoch": 0.6253369272237197,
-      "grad_norm": 0.769210696220398,
+      "grad_norm": 0.7702767252922058,
       "learning_rate": 0.000563086886130599,
-      "loss": 3.9894,
+      "loss": 3.9901,
       "step": 5800
     },
     {
       "epoch": 0.6307277628032345,
-      "grad_norm": 0.7284867763519287,
+      "grad_norm": 0.7166429162025452,
       "learning_rate": 0.0005627630868861305,
-      "loss": 3.9899,
+      "loss": 3.9905,
       "step": 5850
     },
     {
       "epoch": 0.6361185983827493,
-      "grad_norm": 0.604616105556488,
+      "grad_norm": 0.5853614807128906,
       "learning_rate": 0.0005624392876416621,
-      "loss": 3.9758,
+      "loss": 3.9772,
       "step": 5900
     },
     {
       "epoch": 0.6415094339622641,
-      "grad_norm": 0.685613751411438,
+      "grad_norm": 0.6194868087768555,
       "learning_rate": 0.0005621154883971937,
-      "loss": 3.9787,
+      "loss": 3.9817,
       "step": 5950
     },
     {
       "epoch": 0.6469002695417789,
-      "grad_norm": 0.685235321521759,
+      "grad_norm": 0.6746755242347717,
       "learning_rate": 0.0005617916891527253,
       "loss": 3.9609,
       "step": 6000
     },
     {
       "epoch": 0.6469002695417789,
-      "eval_accuracy": 0.32830632639045004,
-      "eval_loss": 3.917071580886841,
-      "eval_runtime": 183.57,
-      "eval_samples_per_second": 98.115,
-      "eval_steps_per_second": 6.134,
+      "eval_accuracy": 0.3282789458627208,
+      "eval_loss": 3.917991876602173,
+      "eval_runtime": 181.6023,
+      "eval_samples_per_second": 99.178,
+      "eval_steps_per_second": 6.2,
       "step": 6000
     },
     {
       "epoch": 0.6522911051212938,
-      "grad_norm": 0.6348211169242859,
+      "grad_norm": 0.6664725542068481,
       "learning_rate": 0.0005614678899082568,
-      "loss": 3.9973,
+      "loss": 3.995,
       "step": 6050
     },
     {
       "epoch": 0.6576819407008087,
-      "grad_norm": 0.5842711925506592,
+      "grad_norm": 0.5693532228469849,
       "learning_rate": 0.0005611440906637884,
       "loss": 3.9818,
       "step": 6100
     },
     {
       "epoch": 0.6630727762803235,
-      "grad_norm": 0.6440193057060242,
+      "grad_norm": 0.8059385418891907,
       "learning_rate": 0.00056082029141932,
-      "loss": 3.9752,
+      "loss": 3.9733,
       "step": 6150
     },
     {
       "epoch": 0.6684636118598383,
-      "grad_norm": 0.6132529377937317,
+      "grad_norm": 0.6112721562385559,
       "learning_rate": 0.0005604964921748515,
-      "loss": 3.9854,
+      "loss": 3.9865,
       "step": 6200
     },
     {
       "epoch": 0.6738544474393531,
-      "grad_norm": 0.6459117531776428,
+      "grad_norm": 0.6212737560272217,
       "learning_rate": 0.000560172692930383,
-      "loss": 3.9462,
+      "loss": 3.9479,
       "step": 6250
     },
     {
       "epoch": 0.6792452830188679,
-      "grad_norm": 0.6339119672775269,
+      "grad_norm": 0.5734388828277588,
       "learning_rate": 0.0005598488936859147,
-      "loss": 3.9565,
+      "loss": 3.9591,
       "step": 6300
     },
     {
       "epoch": 0.6846361185983828,
-      "grad_norm": 0.6577672958374023,
+      "grad_norm": 0.7219184041023254,
       "learning_rate": 0.0005595250944414463,
-      "loss": 3.9497,
+      "loss": 3.9504,
       "step": 6350
     },
     {
       "epoch": 0.6900269541778976,
-      "grad_norm": 0.6435871720314026,
+      "grad_norm": 0.6439206004142761,
       "learning_rate": 0.0005592012951969778,
-      "loss": 3.9424,
+      "loss": 3.9434,
       "step": 6400
     },
     {
       "epoch": 0.6954177897574124,
-      "grad_norm": 0.7063788771629333,
+      "grad_norm": 0.6976845860481262,
       "learning_rate": 0.0005588774959525094,
-      "loss": 3.9694,
+      "loss": 3.9712,
       "step": 6450
     },
     {
       "epoch": 0.7008086253369272,
-      "grad_norm": 0.5613532662391663,
+      "grad_norm": 0.5976806282997131,
       "learning_rate": 0.000558553696708041,
-      "loss": 3.9543,
+      "loss": 3.9546,
       "step": 6500
     },
     {
       "epoch": 0.706199460916442,
-      "grad_norm": 0.559969425201416,
+      "grad_norm": 0.5768620371818542,
       "learning_rate": 0.0005582298974635726,
-      "loss": 3.9512,
+      "loss": 3.9514,
       "step": 6550
     },
     {
       "epoch": 0.7115902964959568,
-      "grad_norm": 0.58429354429245,
+      "grad_norm": 0.6398442983627319,
       "learning_rate": 0.0005579060982191041,
-      "loss": 3.9474,
+      "loss": 3.949,
       "step": 6600
     },
     {
       "epoch": 0.7169811320754716,
-      "grad_norm": Infinity,
-      "learning_rate": 0.0005575887749595251,
-      "loss": 3.9414,
+      "grad_norm": 0.6361405849456787,
+      "learning_rate": 0.0005575822989746357,
+      "loss": 3.9437,
       "step": 6650
     },
     {
       "epoch": 0.7223719676549866,
-      "grad_norm": 0.589440107345581,
-      "learning_rate": 0.0005572649757150566,
-      "loss": 3.9341,
+      "grad_norm": 0.5549120903015137,
+      "learning_rate": 0.0005572584997301672,
+      "loss": 3.9364,
       "step": 6700
     },
     {
       "epoch": 0.7277628032345014,
-      "grad_norm": 0.7343171834945679,
-      "learning_rate": 0.0005569411764705882,
-      "loss": 3.9383,
+      "grad_norm": 0.8019936084747314,
+      "learning_rate": 0.0005569347004856989,
+      "loss": 3.9399,
       "step": 6750
     },
     {
       "epoch": 0.7331536388140162,
-      "grad_norm": 0.6029019355773926,
-      "learning_rate": 0.0005566173772261198,
-      "loss": 3.9469,
+      "grad_norm": 0.6074087023735046,
+      "learning_rate": 0.0005566109012412303,
+      "loss": 3.9487,
       "step": 6800
     },
     {
       "epoch": 0.738544474393531,
-      "grad_norm": 0.5020064115524292,
-      "learning_rate": 0.0005562935779816513,
-      "loss": 3.931,
+      "grad_norm": 0.5666708946228027,
+      "learning_rate": 0.000556287101996762,
+      "loss": 3.9318,
       "step": 6850
     },
     {
       "epoch": 0.7439353099730458,
-      "grad_norm": 0.6130984425544739,
-      "learning_rate": 0.0005559697787371828,
-      "loss": 3.9264,
+      "grad_norm": 0.6135334968566895,
+      "learning_rate": 0.0005559633027522935,
+      "loss": 3.9286,
       "step": 6900
     },
     {
       "epoch": 0.7493261455525606,
-      "grad_norm": 0.6600217819213867,
-      "learning_rate": 0.0005556459794927145,
-      "loss": 3.9137,
+      "grad_norm": 0.6896835565567017,
+      "learning_rate": 0.0005556395035078251,
+      "loss": 3.9139,
       "step": 6950
     },
     {
       "epoch": 0.7547169811320755,
-      "grad_norm": 0.5343778133392334,
-      "learning_rate": 0.000555322180248246,
-      "loss": 3.9274,
+      "grad_norm": 0.536331057548523,
+      "learning_rate": 0.0005553157042633566,
+      "loss": 3.9291,
       "step": 7000
     },
     {
       "epoch": 0.7547169811320755,
-      "eval_accuracy": 0.33360293736563035,
-      "eval_loss": 3.858804941177368,
-      "eval_runtime": 183.535,
-      "eval_samples_per_second": 98.134,
-      "eval_steps_per_second": 6.135,
+      "eval_accuracy": 0.33301186565591956,
+      "eval_loss": 3.8630003929138184,
+      "eval_runtime": 181.6998,
+      "eval_samples_per_second": 99.125,
+      "eval_steps_per_second": 6.197,
       "step": 7000
     },
     {
       "epoch": 0.7601078167115903,
-      "grad_norm": 0.6115443110466003,
-      "learning_rate": 0.0005549983810037776,
-      "loss": 3.931,
+      "grad_norm": 0.5458576083183289,
+      "learning_rate": 0.0005549919050188882,
+      "loss": 3.9324,
       "step": 7050
     },
     {
       "epoch": 0.7654986522911051,
-      "grad_norm": 0.5609748959541321,
-      "learning_rate": 0.0005546745817593091,
-      "loss": 3.9311,
+      "grad_norm": 0.5892027020454407,
+      "learning_rate": 0.0005546681057744198,
+      "loss": 3.9325,
       "step": 7100
     },
     {
       "epoch": 0.77088948787062,
-      "grad_norm": 0.6355463266372681,
-      "learning_rate": 0.0005543507825148408,
-      "loss": 3.9063,
+      "grad_norm": 0.6889460682868958,
+      "learning_rate": 0.0005543443065299514,
+      "loss": 3.9066,
       "step": 7150
     },
     {
       "epoch": 0.7762803234501348,
-      "grad_norm": 0.6557473540306091,
-      "learning_rate": 0.0005540269832703723,
-      "loss": 3.9159,
+      "grad_norm": 0.6623647212982178,
+      "learning_rate": 0.000554020507285483,
+      "loss": 3.9168,
       "step": 7200
     },
     {
       "epoch": 0.7816711590296496,
-      "grad_norm": 0.6337311863899231,
-      "learning_rate": 0.0005537031840259039,
-      "loss": 3.9184,
+      "grad_norm": 0.6174417734146118,
+      "learning_rate": 0.0005536967080410145,
+      "loss": 3.9172,
       "step": 7250
     },
     {
       "epoch": 0.7870619946091644,
-      "grad_norm": 0.5891764760017395,
-      "learning_rate": 0.0005533793847814354,
-      "loss": 3.9032,
+      "grad_norm": 0.5966586470603943,
+      "learning_rate": 0.0005533729087965462,
+      "loss": 3.9069,
       "step": 7300
     },
     {
       "epoch": 0.7924528301886793,
-      "grad_norm": 0.6272363066673279,
-      "learning_rate": 0.000553055585536967,
-      "loss": 3.9148,
+      "grad_norm": 0.5991404056549072,
+      "learning_rate": 0.0005530491095520777,
+      "loss": 3.9153,
       "step": 7350
     },
     {
       "epoch": 0.7978436657681941,
-      "grad_norm": 0.6207209229469299,
-      "learning_rate": 0.0005527317862924987,
+      "grad_norm": 0.6612049341201782,
+      "learning_rate": 0.0005527253103076093,
       "loss": 3.8796,
       "step": 7400
     },
     {
       "epoch": 0.8032345013477089,
-      "grad_norm": 0.5580083131790161,
-      "learning_rate": 0.0005524079870480301,
-      "loss": 3.8865,
+      "grad_norm": 0.5842414498329163,
+      "learning_rate": 0.0005524015110631408,
+      "loss": 3.89,
       "step": 7450
     },
     {
       "epoch": 0.8086253369272237,
-      "grad_norm": 0.7376517057418823,
-      "learning_rate": 0.0005520841878035618,
-      "loss": 3.9018,
+      "grad_norm": 0.7356441617012024,
+      "learning_rate": 0.0005520777118186724,
+      "loss": 3.9037,
       "step": 7500
     },
     {
       "epoch": 0.8140161725067385,
-      "grad_norm": 0.6539864540100098,
-      "learning_rate": 0.0005517603885590933,
-      "loss": 3.8971,
+      "grad_norm": 0.5401440858840942,
+      "learning_rate": 0.0005517539125742039,
+      "loss": 3.8968,
       "step": 7550
     },
     {
       "epoch": 0.8194070080862533,
-      "grad_norm": 0.640784502029419,
-      "learning_rate": 0.0005514365893146249,
-      "loss": 3.9161,
+      "grad_norm": 0.6095986366271973,
+      "learning_rate": 0.0005514301133297355,
+      "loss": 3.9171,
       "step": 7600
     },
     {
       "epoch": 0.8247978436657682,
-      "grad_norm": 0.5456549525260925,
-      "learning_rate": 0.0005511127900701564,
-      "loss": 3.8959,
+      "grad_norm": 0.5622900724411011,
+      "learning_rate": 0.0005511063140852671,
+      "loss": 3.8997,
       "step": 7650
     },
     {
       "epoch": 0.8301886792452831,
-      "grad_norm": 0.5484740734100342,
-      "learning_rate": 0.000550788990825688,
-      "loss": 3.871,
+      "grad_norm": 0.6393899917602539,
+      "learning_rate": 0.0005507825148407987,
+      "loss": 3.8719,
       "step": 7700
     },
     {
       "epoch": 0.8355795148247979,
-      "grad_norm": 0.605842649936676,
-      "learning_rate": 0.0005504651915812196,
-      "loss": 3.8834,
+      "grad_norm": 0.6330990791320801,
+      "learning_rate": 0.0005504587155963302,
+      "loss": 3.8863,
       "step": 7750
     },
     {
       "epoch": 0.8409703504043127,
-      "grad_norm": 0.5898163914680481,
-      "learning_rate": 0.0005501413923367512,
-      "loss": 3.8977,
+      "grad_norm": 0.555696427822113,
+      "learning_rate": 0.0005501349163518618,
+      "loss": 3.8994,
       "step": 7800
     },
     {
       "epoch": 0.8463611859838275,
-      "grad_norm": 0.7656612992286682,
+      "grad_norm": Infinity,
       "learning_rate": 0.0005498175930922827,
       "loss": 3.8864,
       "step": 7850
     },
     {
       "epoch": 0.8517520215633423,
-      "grad_norm": 0.8152493834495544,
+      "grad_norm": 0.6537197232246399,
       "learning_rate": 0.0005494937938478143,
-      "loss": 3.8847,
+      "loss": 3.8855,
       "step": 7900
     },
     {
       "epoch": 0.8571428571428571,
-      "grad_norm": 0.5602722764015198,
+      "grad_norm": 0.5033941268920898,
       "learning_rate": 0.0005491699946033459,
-      "loss": 3.8864,
+      "loss": 3.8855,
       "step": 7950
     },
     {
       "epoch": 0.862533692722372,
-      "grad_norm": 0.5723937153816223,
+      "grad_norm": 0.5997176766395569,
       "learning_rate": 0.0005488461953588775,
-      "loss": 3.8618,
+      "loss": 3.8673,
       "step": 8000
     },
     {
       "epoch": 0.862533692722372,
-      "eval_accuracy": 0.33786419497282644,
-      "eval_loss": 3.8131983280181885,
-      "eval_runtime": 183.2819,
-      "eval_samples_per_second": 98.269,
-      "eval_steps_per_second": 6.144,
+      "eval_accuracy": 0.33755735921763835,
+      "eval_loss": 3.8170533180236816,
+      "eval_runtime": 181.3255,
+      "eval_samples_per_second": 99.33,
+      "eval_steps_per_second": 6.21,
       "step": 8000
     },
     {
       "epoch": 0.8679245283018868,
-      "grad_norm": 0.5539727210998535,
+      "grad_norm": 0.618726909160614,
       "learning_rate": 0.000548522396114409,
-      "loss": 3.879,
+      "loss": 3.8841,
       "step": 8050
     },
     {
       "epoch": 0.8733153638814016,
-      "grad_norm": 0.7839546203613281,
+      "grad_norm": 0.7399865984916687,
       "learning_rate": 0.0005481985968699406,
-      "loss": 3.8589,
+      "loss": 3.861,
       "step": 8100
     },
     {
       "epoch": 0.8787061994609164,
-      "grad_norm": 0.5483460426330566,
+      "grad_norm": 0.5887700915336609,
       "learning_rate": 0.0005478747976254721,
-      "loss": 3.8715,
+      "loss": 3.8717,
       "step": 8150
     },
     {
       "epoch": 0.8840970350404312,
-      "grad_norm": 0.5753620266914368,
+      "grad_norm": 0.5937870144844055,
       "learning_rate": 0.0005475509983810037,
-      "loss": 3.8617,
+      "loss": 3.864,
       "step": 8200
     },
     {
       "epoch": 0.889487870619946,
-      "grad_norm": 0.5939356088638306,
+      "grad_norm": 0.5936031937599182,
       "learning_rate": 0.0005472271991365352,
-      "loss": 3.8748,
+      "loss": 3.8783,
       "step": 8250
     },
     {
       "epoch": 0.894878706199461,
-      "grad_norm": 0.5494759678840637,
+      "grad_norm": 0.5343406200408936,
       "learning_rate": 0.0005469033998920669,
-      "loss": 3.8485,
+      "loss": 3.8528,
       "step": 8300
     },
     {
       "epoch": 0.9002695417789758,
-      "grad_norm": 0.5566021800041199,
+      "grad_norm": 0.5612528324127197,
       "learning_rate": 0.0005465796006475984,
-      "loss": 3.8652,
+      "loss": 3.8672,
       "step": 8350
     },
     {
       "epoch": 0.9056603773584906,
-      "grad_norm": 0.5274083018302917,
+      "grad_norm": 0.5439274311065674,
       "learning_rate": 0.00054625580140313,
-      "loss": 3.866,
+      "loss": 3.8675,
       "step": 8400
     },
     {
       "epoch": 0.9110512129380054,
-      "grad_norm": 0.5682411193847656,
+      "grad_norm": 0.6031448245048523,
       "learning_rate": 0.0005459320021586615,
-      "loss": 3.8363,
+      "loss": 3.8391,
       "step": 8450
     },
     {
       "epoch": 0.9164420485175202,
-      "grad_norm": 0.5468075275421143,
+      "grad_norm": 0.5453014969825745,
       "learning_rate": 0.0005456082029141932,
-      "loss": 3.8486,
+      "loss": 3.8505,
       "step": 8500
     },
     {
       "epoch": 0.921832884097035,
-      "grad_norm": 0.5886979699134827,
+      "grad_norm": 0.5259692668914795,
       "learning_rate": 0.0005452844036697248,
-      "loss": 3.8496,
+      "loss": 3.8519,
       "step": 8550
     },
     {
       "epoch": 0.9272237196765498,
-      "grad_norm": 0.566277027130127,
+      "grad_norm": 0.6086834669113159,
       "learning_rate": 0.0005449606044252563,
-      "loss": 3.8782,
+      "loss": 3.8812,
       "step": 8600
     },
     {
       "epoch": 0.9326145552560647,
-      "grad_norm": 0.6201555728912354,
+      "grad_norm": 0.5958359837532043,
       "learning_rate": 0.0005446368051807879,
-      "loss": 3.8507,
+      "loss": 3.8548,
       "step": 8650
     },
     {
       "epoch": 0.9380053908355795,
-      "grad_norm": 0.5942217111587524,
+      "grad_norm": 0.5887793898582458,
       "learning_rate": 0.0005443130059363194,
-      "loss": 3.8613,
+      "loss": 3.8635,
       "step": 8700
     },
     {
       "epoch": 0.9433962264150944,
-      "grad_norm": 0.5227022767066956,
+      "grad_norm": 0.5301551818847656,
       "learning_rate": 0.0005439892066918511,
-      "loss": 3.8428,
+      "loss": 3.8457,
       "step": 8750
     },
     {
       "epoch": 0.9487870619946092,
-      "grad_norm": 0.6055588126182556,
+      "grad_norm": 0.6073969602584839,
       "learning_rate": 0.0005436654074473825,
-      "loss": 3.8415,
+      "loss": 3.8448,
       "step": 8800
     },
     {
       "epoch": 0.954177897574124,
-      "grad_norm": 0.5056185722351074,
+      "grad_norm": 0.530245840549469,
       "learning_rate": 0.0005433416082029142,
-      "loss": 3.8431,
+      "loss": 3.8465,
       "step": 8850
     },
     {
       "epoch": 0.9595687331536388,
-      "grad_norm": 0.6033360958099365,
+      "grad_norm": 0.5945138931274414,
       "learning_rate": 0.0005430178089584457,
-      "loss": 3.8452,
+      "loss": 3.8472,
       "step": 8900
     },
     {
       "epoch": 0.9649595687331537,
-      "grad_norm": 0.6905035972595215,
+      "grad_norm": 0.9510626792907715,
       "learning_rate": 0.0005426940097139773,
-      "loss": 3.8455,
+      "loss": 3.8487,
       "step": 8950
     },
     {
       "epoch": 0.9703504043126685,
-      "grad_norm": 0.6233043670654297,
-      "learning_rate": 0.0005423766864543982,
-      "loss": 3.8625,
+      "grad_norm": 0.5848538875579834,
+      "learning_rate": 0.0005423702104695088,
+      "loss": 3.8645,
       "step": 9000
     },
     {
       "epoch": 0.9703504043126685,
-      "eval_accuracy": 0.34123297775951444,
-      "eval_loss": 3.775101900100708,
-      "eval_runtime": 183.702,
-      "eval_samples_per_second": 98.045,
-      "eval_steps_per_second": 6.129,
+      "eval_accuracy": 0.34104250924717566,
+      "eval_loss": 3.7781906127929688,
+      "eval_runtime": 181.7189,
+      "eval_samples_per_second": 99.115,
+      "eval_steps_per_second": 6.196,
       "step": 9000
     },
     {
       "epoch": 0.9757412398921833,
-      "grad_norm": 0.6275179982185364,
-      "learning_rate": 0.0005420528872099298,
-      "loss": 3.839,
+      "grad_norm": 0.7085767984390259,
+      "learning_rate": 0.0005420464112250404,
+      "loss": 3.8406,
       "step": 9050
     },
     {
       "epoch": 0.9811320754716981,
-      "grad_norm": 0.5922442674636841,
-      "learning_rate": 0.0005417290879654613,
-      "loss": 3.8288,
+      "grad_norm": 0.6005401015281677,
+      "learning_rate": 0.000541722611980572,
+      "loss": 3.8315,
       "step": 9100
     },
     {
       "epoch": 0.9865229110512129,
-      "grad_norm": 0.5687377452850342,
-      "learning_rate": 0.000541405288720993,
-      "loss": 3.8416,
+      "grad_norm": 0.5444480776786804,
+      "learning_rate": 0.0005413988127361036,
+      "loss": 3.8438,
       "step": 9150
     },
     {
       "epoch": 0.9919137466307277,
-      "grad_norm": 0.5579700469970703,
-      "learning_rate": 0.0005410814894765245,
-      "loss": 3.8436,
+      "grad_norm": 0.5543394684791565,
+      "learning_rate": 0.0005410750134916351,
+      "loss": 3.8471,
       "step": 9200
     },
     {
       "epoch": 0.9973045822102425,
-      "grad_norm": 0.5312648415565491,
-      "learning_rate": 0.0005407576902320561,
-      "loss": 3.8355,
+      "grad_norm": 0.530593991279602,
+      "learning_rate": 0.0005407512142471667,
+      "loss": 3.8383,
       "step": 9250
     },
     {
       "epoch": 1.0026954177897573,
-      "grad_norm": 0.5382205247879028,
-      "learning_rate": 0.0005404338909875876,
-      "loss": 3.8133,
+      "grad_norm": 0.6533575057983398,
+      "learning_rate": 0.0005404274150026983,
+      "loss": 3.8157,
       "step": 9300
     },
     {
       "epoch": 1.0080862533692723,
-      "grad_norm": 0.6140266060829163,
-      "learning_rate": 0.0005401100917431192,
-      "loss": 3.7783,
+      "grad_norm": 0.5836777091026306,
+      "learning_rate": 0.0005401036157582299,
+      "loss": 3.7826,
       "step": 9350
     },
     {
       "epoch": 1.013477088948787,
-      "grad_norm": 0.5851580500602722,
-      "learning_rate": 0.0005397862924986508,
-      "loss": 3.7885,
+      "grad_norm": 0.5416027307510376,
+      "learning_rate": 0.0005397798165137614,
+      "loss": 3.7923,
       "step": 9400
     },
     {
       "epoch": 1.0188679245283019,
-      "grad_norm": 0.5980308651924133,
-      "learning_rate": 0.0005394624932541824,
-      "loss": 3.7783,
+      "grad_norm": 0.5336276888847351,
+      "learning_rate": 0.000539456017269293,
+      "loss": 3.7841,
       "step": 9450
     },
     {
       "epoch": 1.0242587601078168,
-      "grad_norm": 0.5626499652862549,
-      "learning_rate": 0.0005391386940097139,
-      "loss": 3.7758,
+      "grad_norm": 0.5434985160827637,
+      "learning_rate": 0.0005391322180248245,
+      "loss": 3.7778,
       "step": 9500
     },
     {
       "epoch": 1.0296495956873315,
-      "grad_norm": 0.5909863114356995,
-      "learning_rate": 0.0005388148947652455,
-      "loss": 3.7869,
+      "grad_norm": 0.5847581624984741,
+      "learning_rate": 0.0005388084187803561,
+      "loss": 3.7874,
       "step": 9550
     },
     {
       "epoch": 1.0350404312668464,
-      "grad_norm": 0.5526650547981262,
-      "learning_rate": 0.000538491095520777,
-      "loss": 3.7655,
+      "grad_norm": 0.5592184066772461,
+      "learning_rate": 0.0005384846195358876,
+      "loss": 3.7689,
       "step": 9600
     },
     {
       "epoch": 1.0404312668463611,
-      "grad_norm": 0.5704436898231506,
-      "learning_rate": 0.0005381672962763086,
-      "loss": 3.7576,
+      "grad_norm": 0.5576586127281189,
+      "learning_rate": 0.0005381608202914193,
+      "loss": 3.7629,
       "step": 9650
     },
     {
       "epoch": 1.045822102425876,
-      "grad_norm": 0.5485429167747498,
-      "learning_rate": 0.0005378434970318403,
-      "loss": 3.7695,
+      "grad_norm": 0.5352767109870911,
+      "learning_rate": 0.0005378370210469509,
+      "loss": 3.7724,
       "step": 9700
     },
     {
       "epoch": 1.0512129380053907,
-      "grad_norm": 0.5683411955833435,
-      "learning_rate": 0.0005375196977873718,
-      "loss": 3.7622,
+      "grad_norm": 0.5537406206130981,
+      "learning_rate": 0.0005375132218024824,
+      "loss": 3.7647,
       "step": 9750
     },
     {
       "epoch": 1.0566037735849056,
-      "grad_norm": 0.5670942068099976,
-      "learning_rate": 0.0005371958985429034,
-      "loss": 3.7672,
+      "grad_norm": 0.6254362463951111,
+      "learning_rate": 0.000537189422558014,
+      "loss": 3.771,
       "step": 9800
     },
     {
       "epoch": 1.0619946091644206,
-      "grad_norm": 0.6690695881843567,
-      "learning_rate": 0.0005368720992984349,
-      "loss": 3.7521,
+      "grad_norm": 0.5927165150642395,
+      "learning_rate": 0.0005368656233135455,
+      "loss": 3.7559,
       "step": 9850
     },
     {
       "epoch": 1.0673854447439353,
-      "grad_norm": 0.6253776550292969,
-      "learning_rate": 0.0005365483000539665,
-      "loss": 3.762,
+      "grad_norm": 0.5955566763877869,
+      "learning_rate": 0.0005365418240690772,
+      "loss": 3.7656,
       "step": 9900
     },
     {
       "epoch": 1.0727762803234502,
-      "grad_norm": 0.5629079937934875,
-      "learning_rate": 0.0005362245008094981,
-      "loss": 3.7594,
+      "grad_norm": 0.5806096792221069,
+      "learning_rate": 0.0005362180248246087,
+      "loss": 3.7619,
       "step": 9950
     },
     {
       "epoch": 1.0781671159029649,
-      "grad_norm": 0.5760037302970886,
-      "learning_rate": 0.0005359007015650297,
-      "loss": 3.7656,
+      "grad_norm": 0.6050166487693787,
+      "learning_rate": 0.0005358942255801403,
+      "loss": 3.767,
       "step": 10000
     },
     {
       "epoch": 1.0781671159029649,
-      "eval_accuracy": 0.34546098758303934,
-      "eval_loss": 3.7440903186798096,
-      "eval_runtime": 183.3277,
-      "eval_samples_per_second": 98.245,
-      "eval_steps_per_second": 6.142,
+      "eval_accuracy": 0.3450713583273366,
+      "eval_loss": 3.746429681777954,
+      "eval_runtime": 181.6002,
+      "eval_samples_per_second": 99.179,
+      "eval_steps_per_second": 6.2,
       "step": 10000
     },
     {
       "epoch": 1.0835579514824798,
-      "grad_norm": 0.5451662540435791,
-      "learning_rate": 0.0005355769023205612,
-      "loss": 3.755,
+      "grad_norm": 0.5626857876777649,
+      "learning_rate": 0.0005355704263356718,
+      "loss": 3.7569,
       "step": 10050
     },
     {
       "epoch": 1.0889487870619945,
-      "grad_norm": 0.5988211035728455,
-      "learning_rate": 0.0005352531030760928,
-      "loss": 3.7454,
+      "grad_norm": 0.5822886824607849,
+      "learning_rate": 0.0005352466270912035,
+      "loss": 3.7466,
       "step": 10100
     },
     {
       "epoch": 1.0943396226415094,
-      "grad_norm": 0.6496660709381104,
-      "learning_rate": 0.0005349293038316244,
-      "loss": 3.777,
+      "grad_norm": 0.5978696942329407,
+      "learning_rate": 0.000534922827846735,
+      "loss": 3.7817,
       "step": 10150
     },
     {
       "epoch": 1.0997304582210243,
-      "grad_norm": 0.5746974945068359,
+      "grad_norm": 0.5884588956832886,
       "learning_rate": 0.0005346055045871559,
-      "loss": 3.7792,
+      "loss": 3.7812,
       "step": 10200
     },
     {
       "epoch": 1.105121293800539,
-      "grad_norm": 0.5537252426147461,
+      "grad_norm": 0.562393307685852,
       "learning_rate": 0.0005342817053426874,
-      "loss": 3.7429,
+      "loss": 3.7455,
       "step": 10250
     },
     {
       "epoch": 1.110512129380054,
-      "grad_norm": 0.6647258400917053,
+      "grad_norm": 0.6107234954833984,
       "learning_rate": 0.0005339579060982191,
-      "loss": 3.7704,
+      "loss": 3.7743,
       "step": 10300
     },
     {
       "epoch": 1.1159029649595686,
-      "grad_norm": 0.5465642809867859,
+      "grad_norm": 0.4961622357368469,
       "learning_rate": 0.0005336341068537506,
-      "loss": 3.7703,
+      "loss": 3.7743,
       "step": 10350
     },
     {
       "epoch": 1.1212938005390836,
-      "grad_norm": 0.5680330991744995,
+      "grad_norm": 0.5662735104560852,
       "learning_rate": 0.0005333103076092822,
-      "loss": 3.7573,
+      "loss": 3.7591,
       "step": 10400
     },
     {
       "epoch": 1.1266846361185983,
-      "grad_norm": 0.5591064095497131,
+      "grad_norm": 0.5381325483322144,
       "learning_rate": 0.0005329865083648137,
-      "loss": 3.7225,
+      "loss": 3.725,
       "step": 10450
     },
     {
       "epoch": 1.1320754716981132,
-      "grad_norm": 0.5781257152557373,
+      "grad_norm": 0.5634454488754272,
       "learning_rate": 0.0005326627091203454,
-      "loss": 3.7439,
+      "loss": 3.7486,
       "step": 10500
     },
     {
       "epoch": 1.137466307277628,
-      "grad_norm": 0.5286954045295715,
+      "grad_norm": 0.5432537198066711,
       "learning_rate": 0.0005323389098758769,
-      "loss": 3.7463,
+      "loss": 3.7513,
       "step": 10550
     },
     {
       "epoch": 1.1428571428571428,
-      "grad_norm": 0.5636879205703735,
+      "grad_norm": 0.5721794366836548,
       "learning_rate": 0.0005320151106314085,
-      "loss": 3.7536,
+      "loss": 3.7555,
       "step": 10600
     },
     {
       "epoch": 1.1482479784366577,
-      "grad_norm": 0.5693058371543884,
+      "grad_norm": 0.5426144003868103,
       "learning_rate": 0.00053169131138694,
-      "loss": 3.7615,
+      "loss": 3.7636,
       "step": 10650
     },
     {
       "epoch": 1.1536388140161726,
-      "grad_norm": 0.62501060962677,
+      "grad_norm": 0.6116250157356262,
       "learning_rate": 0.0005313675121424716,
-      "loss": 3.7439,
+      "loss": 3.7487,
       "step": 10700
     },
     {
       "epoch": 1.1590296495956873,
-      "grad_norm": 0.5845157504081726,
+      "grad_norm": 0.5428005456924438,
       "learning_rate": 0.0005310437128980032,
-      "loss": 3.7449,
+      "loss": 3.7481,
       "step": 10750
     },
     {
       "epoch": 1.1644204851752022,
-      "grad_norm": 0.5758997201919556,
+      "grad_norm": 0.5519539713859558,
       "learning_rate": 0.0005307199136535348,
-      "loss": 3.7219,
+      "loss": 3.7245,
       "step": 10800
     },
     {
       "epoch": 1.169811320754717,
-      "grad_norm": 0.6027551889419556,
+      "grad_norm": 0.5696902871131897,
       "learning_rate": 0.0005303961144090663,
-      "loss": 3.7569,
+      "loss": 3.7621,
       "step": 10850
     },
     {
       "epoch": 1.1752021563342319,
-      "grad_norm": 0.5498311519622803,
+      "grad_norm": 0.5773632526397705,
       "learning_rate": 0.0005300723151645979,
-      "loss": 3.7575,
+      "loss": 3.762,
       "step": 10900
     },
     {
       "epoch": 1.1805929919137466,
-      "grad_norm": 0.5848629474639893,
+      "grad_norm": 0.6049562692642212,
       "learning_rate": 0.0005297485159201295,
-      "loss": 3.7171,
+      "loss": 3.7212,
       "step": 10950
     },
     {
       "epoch": 1.1859838274932615,
-      "grad_norm": 0.5120974779129028,
+      "grad_norm": 0.5656830668449402,
       "learning_rate": 0.000529424716675661,
-      "loss": 3.7343,
+      "loss": 3.7405,
       "step": 11000
     },
     {
       "epoch": 1.1859838274932615,
-      "eval_accuracy": 0.3476425202645785,
-      "eval_loss": 3.7190380096435547,
-      "eval_runtime": 183.3061,
-      "eval_samples_per_second": 98.256,
-      "eval_steps_per_second": 6.143,
+      "eval_accuracy": 0.34732123367531104,
+      "eval_loss": 3.7205822467803955,
+      "eval_runtime": 181.0713,
+      "eval_samples_per_second": 99.469,
+      "eval_steps_per_second": 6.219,
       "step": 11000
     },
     {
       "epoch": 1.1913746630727764,
-      "grad_norm": 0.5583203434944153,
-      "learning_rate": 0.000529107393416082,
-      "loss": 3.7451,
+      "grad_norm": 0.5404970049858093,
+      "learning_rate": 0.0005291009174311926,
+      "loss": 3.7514,
       "step": 11050
     },
     {
       "epoch": 1.196765498652291,
-      "grad_norm": 0.5904495716094971,
-      "learning_rate": 0.0005287835941716135,
-      "loss": 3.74,
+      "grad_norm": 0.5717231631278992,
+      "learning_rate": 0.0005287771181867242,
+      "loss": 3.7434,
       "step": 11100
     },
     {
       "epoch": 1.202156334231806,
-      "grad_norm": 0.5562983751296997,
-      "learning_rate": 0.0005284597949271452,
-      "loss": 3.7345,
+      "grad_norm": 0.5889883637428284,
+      "learning_rate": 0.0005284533189422558,
+      "loss": 3.7371,
       "step": 11150
     },
     {
       "epoch": 1.2075471698113207,
-      "grad_norm": 0.676979660987854,
-      "learning_rate": 0.0005281359956826767,
-      "loss": 3.7461,
+      "grad_norm": 0.5684009790420532,
+      "learning_rate": 0.0005281295196977873,
+      "loss": 3.7501,
       "step": 11200
     },
     {
       "epoch": 1.2129380053908356,
-      "grad_norm": 0.507462739944458,
-      "learning_rate": 0.0005278121964382083,
-      "loss": 3.756,
+      "grad_norm": 0.5395342111587524,
+      "learning_rate": 0.0005278057204533189,
+      "loss": 3.7609,
       "step": 11250
     },
     {
       "epoch": 1.2183288409703503,
-      "grad_norm": 0.57256019115448,
-      "learning_rate": 0.0005274883971937398,
-      "loss": 3.7256,
+      "grad_norm": 0.630828320980072,
+      "learning_rate": 0.0005274819212088505,
+      "loss": 3.7282,
       "step": 11300
     },
     {
       "epoch": 1.2237196765498652,
-      "grad_norm": 0.5681268572807312,
-      "learning_rate": 0.0005271645979492714,
-      "loss": 3.747,
+      "grad_norm": 0.5785669088363647,
+      "learning_rate": 0.0005271581219643821,
+      "loss": 3.7509,
       "step": 11350
     },
     {
       "epoch": 1.2291105121293802,
-      "grad_norm": 0.5520728230476379,
-      "learning_rate": 0.000526840798704803,
-      "loss": 3.7469,
+      "grad_norm": 0.559070348739624,
+      "learning_rate": 0.0005268343227199136,
+      "loss": 3.7508,
       "step": 11400
     },
     {
       "epoch": 1.2345013477088949,
-      "grad_norm": 0.5298144817352295,
-      "learning_rate": 0.0005265169994603346,
-      "loss": 3.7509,
+      "grad_norm": 0.520759642124176,
+      "learning_rate": 0.0005265105234754452,
+      "loss": 3.7558,
       "step": 11450
     },
     {
       "epoch": 1.2398921832884098,
-      "grad_norm": 0.5538373589515686,
-      "learning_rate": 0.0005261932002158661,
-      "loss": 3.7152,
+      "grad_norm": 0.5443264842033386,
+      "learning_rate": 0.0005261867242309767,
+      "loss": 3.7174,
       "step": 11500
     },
     {
       "epoch": 1.2452830188679245,
-      "grad_norm": 0.601010262966156,
-      "learning_rate": 0.0005258694009713977,
-      "loss": 3.7366,
+      "grad_norm": 0.5318492650985718,
+      "learning_rate": 0.0005258629249865083,
+      "loss": 3.7399,
       "step": 11550
     },
     {
       "epoch": 1.2506738544474394,
-      "grad_norm": 0.5493463277816772,
-      "learning_rate": 0.0005255456017269292,
-      "loss": 3.7331,
+      "grad_norm": 0.5049984455108643,
+      "learning_rate": 0.0005255391257420398,
+      "loss": 3.7374,
       "step": 11600
     },
     {
       "epoch": 1.256064690026954,
-      "grad_norm": 0.5921666622161865,
-      "learning_rate": 0.0005252218024824608,
-      "loss": 3.732,
+      "grad_norm": 0.6127288937568665,
+      "learning_rate": 0.0005252153264975715,
+      "loss": 3.7362,
       "step": 11650
     },
     {
       "epoch": 1.261455525606469,
-      "grad_norm": 0.5424702763557434,
-      "learning_rate": 0.0005248980032379924,
-      "loss": 3.7297,
+      "grad_norm": 0.5741930603981018,
+      "learning_rate": 0.000524891527253103,
+      "loss": 3.7368,
       "step": 11700
     },
     {
       "epoch": 1.266846361185984,
-      "grad_norm": 0.6088834404945374,
-      "learning_rate": 0.000524574203993524,
-      "loss": 3.7384,
+      "grad_norm": 0.643710732460022,
+      "learning_rate": 0.0005245677280086346,
+      "loss": 3.7427,
       "step": 11750
     },
     {
       "epoch": 1.2722371967654986,
-      "grad_norm": 0.5311514139175415,
-      "learning_rate": 0.0005242504047490555,
-      "loss": 3.7268,
+      "grad_norm": 0.5314951539039612,
+      "learning_rate": 0.0005242439287641661,
+      "loss": 3.7329,
       "step": 11800
     },
     {
       "epoch": 1.2776280323450135,
-      "grad_norm": 0.5391415953636169,
-      "learning_rate": 0.0005239266055045871,
-      "loss": 3.7088,
+      "grad_norm": 0.5321791768074036,
+      "learning_rate": 0.0005239201295196978,
+      "loss": 3.7142,
       "step": 11850
     },
     {
       "epoch": 1.2830188679245282,
-      "grad_norm": 0.5253583788871765,
-      "learning_rate": 0.0005236028062601186,
-      "loss": 3.7158,
+      "grad_norm": 0.5225123763084412,
+      "learning_rate": 0.0005235963302752293,
+      "loss": 3.7185,
       "step": 11900
     },
     {
       "epoch": 1.2884097035040432,
-      "grad_norm": 0.5387663841247559,
-      "learning_rate": 0.0005232790070156503,
-      "loss": 3.7243,
+      "grad_norm": 0.5190525650978088,
+      "learning_rate": 0.0005232725310307609,
+      "loss": 3.7293,
       "step": 11950
     },
     {
       "epoch": 1.2938005390835579,
-      "grad_norm": 0.5523077845573425,
-      "learning_rate": 0.0005229552077711818,
-      "loss": 3.7195,
+      "grad_norm": 0.5273523926734924,
+      "learning_rate": 0.0005229487317862924,
+      "loss": 3.7228,
       "step": 12000
     },
     {
       "epoch": 1.2938005390835579,
-      "eval_accuracy": 0.34966281206060096,
-      "eval_loss": 3.695082664489746,
-      "eval_runtime": 183.4346,
-      "eval_samples_per_second": 98.188,
-      "eval_steps_per_second": 6.138,
+      "eval_accuracy": 0.3490053534364355,
+      "eval_loss": 3.698843002319336,
+      "eval_runtime": 181.6527,
+      "eval_samples_per_second": 99.151,
+      "eval_steps_per_second": 6.199,
       "step": 12000
     },
     {
       "epoch": 1.2991913746630728,
-      "grad_norm": 0.5675397515296936,
-      "learning_rate": 0.0005226314085267134,
-      "loss": 3.701,
+      "grad_norm": 0.5660483241081238,
+      "learning_rate": 0.000522624932541824,
+      "loss": 3.7056,
       "step": 12050
     },
     {
       "epoch": 1.3045822102425877,
-      "grad_norm": 0.6067949533462524,
-      "learning_rate": 0.000522307609282245,
-      "loss": 3.7387,
+      "grad_norm": 0.6258788704872131,
+      "learning_rate": 0.0005223011332973557,
+      "loss": 3.7426,
       "step": 12100
     },
     {
       "epoch": 1.3099730458221024,
-      "grad_norm": 0.5414419174194336,
-      "learning_rate": 0.0005219838100377766,
-      "loss": 3.7205,
+      "grad_norm": 0.6055550575256348,
+      "learning_rate": 0.0005219773340528872,
+      "loss": 3.7245,
       "step": 12150
     },
     {
       "epoch": 1.3153638814016173,
-      "grad_norm": 0.6006377339363098,
-      "learning_rate": 0.000521660010793308,
-      "loss": 3.7045,
+      "grad_norm": 0.5716223120689392,
+      "learning_rate": 0.0005216535348084188,
+      "loss": 3.7106,
       "step": 12200
     },
     {
       "epoch": 1.320754716981132,
-      "grad_norm": 0.6085687875747681,
-      "learning_rate": 0.0005213362115488396,
-      "loss": 3.7319,
+      "grad_norm": 0.6301817893981934,
+      "learning_rate": 0.0005213297355639503,
+      "loss": 3.7365,
       "step": 12250
     },
     {
       "epoch": 1.326145552560647,
-      "grad_norm": 0.5864773392677307,
-      "learning_rate": 0.0005210124123043713,
-      "loss": 3.6989,
+      "grad_norm": 0.5893067717552185,
+      "learning_rate": 0.0005210059363194819,
+      "loss": 3.7027,
       "step": 12300
     },
     {
       "epoch": 1.3315363881401616,
-      "grad_norm": 0.5509727001190186,
-      "learning_rate": 0.0005206886130599028,
-      "loss": 3.725,
+      "grad_norm": 0.5365200042724609,
+      "learning_rate": 0.0005206821370750134,
+      "loss": 3.7292,
       "step": 12350
     },
     {
       "epoch": 1.3369272237196765,
-      "grad_norm": 0.5826424360275269,
-      "learning_rate": 0.0005203648138154344,
-      "loss": 3.6955,
+      "grad_norm": 0.5425746440887451,
+      "learning_rate": 0.000520358337830545,
+      "loss": 3.6994,
       "step": 12400
     },
     {
       "epoch": 1.3423180592991915,
-      "grad_norm": 0.5667881369590759,
-      "learning_rate": 0.0005200410145709659,
-      "loss": 3.7183,
+      "grad_norm": 0.5802893042564392,
+      "learning_rate": 0.0005200345385860766,
+      "loss": 3.719,
       "step": 12450
     },
     {
       "epoch": 1.3477088948787062,
-      "grad_norm": 0.6350487470626831,
-      "learning_rate": 0.0005197172153264976,
-      "loss": 3.7223,
+      "grad_norm": 0.6360095739364624,
+      "learning_rate": 0.0005197107393416082,
+      "loss": 3.7263,
       "step": 12500
     },
     {
       "epoch": 1.353099730458221,
-      "grad_norm": 0.565371572971344,
-      "learning_rate": 0.0005193934160820291,
-      "loss": 3.7318,
+      "grad_norm": 0.5260937213897705,
+      "learning_rate": 0.0005193869400971397,
+      "loss": 3.7368,
       "step": 12550
     },
     {
       "epoch": 1.3584905660377358,
-      "grad_norm": 0.5364895462989807,
-      "learning_rate": 0.0005190696168375607,
-      "loss": 3.7087,
+      "grad_norm": 0.5390405654907227,
+      "learning_rate": 0.0005190631408526713,
+      "loss": 3.7125,
       "step": 12600
     },
     {
       "epoch": 1.3638814016172507,
-      "grad_norm": 0.6211722493171692,
-      "learning_rate": 0.0005187458175930922,
-      "loss": 3.7054,
+      "grad_norm": 0.5904570817947388,
+      "learning_rate": 0.0005187393416082029,
+      "loss": 3.7101,
       "step": 12650
     },
     {
       "epoch": 1.3692722371967654,
-      "grad_norm": 0.5237928628921509,
-      "learning_rate": 0.0005184220183486238,
-      "loss": 3.7025,
+      "grad_norm": 0.5123733282089233,
+      "learning_rate": 0.0005184155423637345,
+      "loss": 3.708,
       "step": 12700
     },
     {
       "epoch": 1.3746630727762803,
-      "grad_norm": 0.5414772033691406,
+      "grad_norm": 0.5602011680603027,
       "learning_rate": 0.0005180982191041554,
-      "loss": 3.7171,
+      "loss": 3.723,
       "step": 12750
     },
     {
       "epoch": 1.3800539083557952,
-      "grad_norm": 0.5802706480026245,
+      "grad_norm": 0.5690551400184631,
       "learning_rate": 0.000517774419859687,
-      "loss": 3.7097,
+      "loss": 3.7154,
       "step": 12800
     },
     {
       "epoch": 1.38544474393531,
-      "grad_norm": 0.5995144248008728,
+      "grad_norm": 0.5875492691993713,
       "learning_rate": 0.0005174506206152185,
-      "loss": 3.7307,
+      "loss": 3.7351,
       "step": 12850
     },
     {
       "epoch": 1.3908355795148248,
-      "grad_norm": 0.5822244882583618,
+      "grad_norm": 0.5407969355583191,
       "learning_rate": 0.0005171268213707501,
-      "loss": 3.7303,
+      "loss": 3.7336,
       "step": 12900
     },
     {
       "epoch": 1.3962264150943398,
-      "grad_norm": 0.5466347336769104,
+      "grad_norm": 0.5780190825462341,
       "learning_rate": 0.0005168030221262816,
-      "loss": 3.7282,
+      "loss": 3.7331,
       "step": 12950
     },
     {
       "epoch": 1.4016172506738545,
-      "grad_norm": 0.5338788032531738,
+      "grad_norm": 0.5492972731590271,
       "learning_rate": 0.0005164792228818132,
-      "loss": 3.7112,
+      "loss": 3.7158,
       "step": 13000
     },
     {
       "epoch": 1.4016172506738545,
-      "eval_accuracy": 0.35203926802287927,
-      "eval_loss": 3.6746132373809814,
-      "eval_runtime": 183.6347,
-      "eval_samples_per_second": 98.081,
-      "eval_steps_per_second": 6.132,
+      "eval_accuracy": 0.3514064736042477,
+      "eval_loss": 3.6766467094421387,
+      "eval_runtime": 181.6146,
+      "eval_samples_per_second": 99.172,
+      "eval_steps_per_second": 6.2,
       "step": 13000
     },
     {
       "epoch": 1.4070080862533692,
-      "grad_norm": 0.6075963973999023,
+      "grad_norm": 0.6273307204246521,
       "learning_rate": 0.0005161554236373448,
-      "loss": 3.692,
+      "loss": 3.6963,
       "step": 13050
     },
     {
       "epoch": 1.412398921832884,
-      "grad_norm": 0.5783334374427795,
-      "learning_rate": 0.0005158381003777657,
-      "loss": 3.7088,
+      "grad_norm": 0.6057804822921753,
+      "learning_rate": 0.0005158316243928764,
+      "loss": 3.7129,
       "step": 13100
     },
     {
       "epoch": 1.417789757412399,
-      "grad_norm": 0.4899292588233948,
-      "learning_rate": 0.0005155143011332973,
-      "loss": 3.7093,
+      "grad_norm": 0.5344161987304688,
+      "learning_rate": 0.0005155078251484079,
+      "loss": 3.7164,
       "step": 13150
     },
     {
       "epoch": 1.4231805929919137,
-      "grad_norm": 0.5416874885559082,
-      "learning_rate": 0.0005151905018888289,
-      "loss": 3.7333,
+      "grad_norm": 0.5618059635162354,
+      "learning_rate": 0.0005151840259039395,
+      "loss": 3.7404,
       "step": 13200
     },
     {
       "epoch": 1.4285714285714286,
-      "grad_norm": 0.5604920983314514,
-      "learning_rate": 0.0005148667026443604,
-      "loss": 3.6912,
+      "grad_norm": 0.5437524914741516,
+      "learning_rate": 0.000514860226659471,
+      "loss": 3.6961,
       "step": 13250
     },
     {
       "epoch": 1.4339622641509435,
-      "grad_norm": 0.5507873892784119,
-      "learning_rate": 0.000514542903399892,
-      "loss": 3.698,
+      "grad_norm": 0.5152619481086731,
+      "learning_rate": 0.0005145364274150027,
+      "loss": 3.7015,
       "step": 13300
     },
     {
       "epoch": 1.4393530997304582,
-      "grad_norm": 0.5818184614181519,
-      "learning_rate": 0.0005142191041554237,
-      "loss": 3.6776,
+      "grad_norm": 0.5904302597045898,
+      "learning_rate": 0.0005142126281705343,
+      "loss": 3.6824,
       "step": 13350
     },
     {
       "epoch": 1.444743935309973,
-      "grad_norm": 0.6013184785842896,
-      "learning_rate": 0.0005138953049109552,
-      "loss": 3.7165,
+      "grad_norm": 0.5918348431587219,
+      "learning_rate": 0.0005138888289260658,
+      "loss": 3.7208,
       "step": 13400
     },
     {
       "epoch": 1.4501347708894878,
-      "grad_norm": 0.6403388381004333,
-      "learning_rate": 0.0005135715056664868,
-      "loss": 3.7128,
+      "grad_norm": 0.5408966541290283,
+      "learning_rate": 0.0005135650296815974,
+      "loss": 3.7172,
       "step": 13450
     },
     {
       "epoch": 1.4555256064690028,
-      "grad_norm": 0.595710813999176,
-      "learning_rate": 0.0005132477064220183,
-      "loss": 3.6982,
+      "grad_norm": 0.6010380387306213,
+      "learning_rate": 0.0005132412304371289,
+      "loss": 3.702,
       "step": 13500
     },
     {
       "epoch": 1.4609164420485174,
-      "grad_norm": 0.6105093955993652,
-      "learning_rate": 0.0005129239071775499,
-      "loss": 3.7149,
+      "grad_norm": 0.5562928318977356,
+      "learning_rate": 0.0005129174311926605,
+      "loss": 3.7201,
       "step": 13550
     },
     {
       "epoch": 1.4663072776280324,
-      "grad_norm": 0.566348135471344,
-      "learning_rate": 0.0005126001079330814,
-      "loss": 3.6997,
+      "grad_norm": 0.5022876262664795,
+      "learning_rate": 0.000512593631948192,
+      "loss": 3.7046,
       "step": 13600
     },
     {
       "epoch": 1.4716981132075473,
-      "grad_norm": 0.5651235580444336,
-      "learning_rate": 0.000512276308688613,
-      "loss": 3.7034,
+      "grad_norm": 0.5742781162261963,
+      "learning_rate": 0.0005122698327037237,
+      "loss": 3.7083,
       "step": 13650
     },
     {
       "epoch": 1.477088948787062,
-      "grad_norm": 0.5697197318077087,
-      "learning_rate": 0.0005119525094441446,
-      "loss": 3.7153,
+      "grad_norm": 0.5881465077400208,
+      "learning_rate": 0.0005119460334592552,
+      "loss": 3.7193,
       "step": 13700
     },
     {
       "epoch": 1.482479784366577,
-      "grad_norm": 0.6158642172813416,
-      "learning_rate": 0.0005116287101996762,
-      "loss": 3.6978,
+      "grad_norm": 0.5863586664199829,
+      "learning_rate": 0.0005116222342147868,
+      "loss": 3.7016,
       "step": 13750
     },
     {
       "epoch": 1.4878706199460916,
-      "grad_norm": 0.5030224323272705,
-      "learning_rate": 0.0005113049109552077,
-      "loss": 3.6963,
+      "grad_norm": 0.5511084198951721,
+      "learning_rate": 0.0005112984349703183,
+      "loss": 3.6999,
       "step": 13800
     },
     {
       "epoch": 1.4932614555256065,
-      "grad_norm": 0.5537750720977783,
-      "learning_rate": 0.0005109811117107393,
-      "loss": 3.6907,
+      "grad_norm": 0.5578559041023254,
+      "learning_rate": 0.00051097463572585,
+      "loss": 3.6964,
       "step": 13850
     },
     {
       "epoch": 1.4986522911051212,
-      "grad_norm": 0.6205897927284241,
-      "learning_rate": 0.0005106573124662708,
-      "loss": 3.6805,
+      "grad_norm": 0.5994284749031067,
+      "learning_rate": 0.0005106508364813815,
+      "loss": 3.6854,
       "step": 13900
     },
     {
       "epoch": 1.5040431266846361,
-      "grad_norm": 0.5059426426887512,
-      "learning_rate": 0.0005103335132218025,
-      "loss": 3.7003,
+      "grad_norm": 0.5142702460289001,
+      "learning_rate": 0.0005103270372369131,
+      "loss": 3.7046,
       "step": 13950
     },
     {
       "epoch": 1.509433962264151,
-      "grad_norm": 0.5683820843696594,
-      "learning_rate": 0.000510009713977334,
-      "loss": 3.6911,
+      "grad_norm": 0.5755314230918884,
+      "learning_rate": 0.0005100032379924446,
+      "loss": 3.6962,
       "step": 14000
     },
     {
       "epoch": 1.509433962264151,
-      "eval_accuracy": 0.35447570037923115,
-      "eval_loss": 3.6494083404541016,
-      "eval_runtime": 183.437,
-      "eval_samples_per_second": 98.186,
-      "eval_steps_per_second": 6.138,
+      "eval_accuracy": 0.3536844900501661,
+      "eval_loss": 3.6539077758789062,
+      "eval_runtime": 185.2361,
+      "eval_samples_per_second": 97.233,
+      "eval_steps_per_second": 6.079,
       "step": 14000
     },
     {
       "epoch": 1.5148247978436657,
-      "grad_norm": 0.5536700487136841,
-      "learning_rate": 0.0005096859147328656,
-      "loss": 3.6698,
+      "grad_norm": 0.5519759058952332,
+      "learning_rate": 0.0005096794387479762,
+      "loss": 3.6721,
       "step": 14050
     },
     {
       "epoch": 1.5202156334231804,
-      "grad_norm": 0.5463404655456543,
-      "learning_rate": 0.0005093621154883971,
-      "loss": 3.7006,
+      "grad_norm": 0.5108723640441895,
+      "learning_rate": 0.0005093556395035078,
+      "loss": 3.7055,
       "step": 14100
     },
     {
       "epoch": 1.5256064690026954,
-      "grad_norm": 0.5179774761199951,
-      "learning_rate": 0.0005090383162439288,
-      "loss": 3.6828,
+      "grad_norm": 0.5432536005973816,
+      "learning_rate": 0.0005090318402590394,
+      "loss": 3.6854,
       "step": 14150
     },
     {
       "epoch": 1.5309973045822103,
-      "grad_norm": 0.5397458076477051,
-      "learning_rate": 0.0005087145169994602,
-      "loss": 3.6819,
+      "grad_norm": 0.48810628056526184,
+      "learning_rate": 0.0005087080410145709,
+      "loss": 3.6845,
       "step": 14200
     },
     {
       "epoch": 1.536388140161725,
-      "grad_norm": 0.5047475695610046,
-      "learning_rate": 0.0005083907177549918,
-      "loss": 3.6879,
+      "grad_norm": 0.5068284273147583,
+      "learning_rate": 0.0005083842417701025,
+      "loss": 3.6939,
       "step": 14250
     },
     {
       "epoch": 1.54177897574124,
-      "grad_norm": 0.5295481085777283,
-      "learning_rate": 0.0005080669185105234,
-      "loss": 3.6971,
+      "grad_norm": 0.5083636045455933,
+      "learning_rate": 0.000508060442525634,
+      "loss": 3.6993,
       "step": 14300
     },
     {
       "epoch": 1.5471698113207548,
-      "grad_norm": 0.49893248081207275,
-      "learning_rate": 0.000507743119266055,
-      "loss": 3.7041,
+      "grad_norm": 0.5182152390480042,
+      "learning_rate": 0.0005077366432811656,
+      "loss": 3.7094,
       "step": 14350
     },
     {
       "epoch": 1.5525606469002695,
-      "grad_norm": 0.6531094312667847,
-      "learning_rate": 0.0005074193200215865,
-      "loss": 3.6874,
+      "grad_norm": 0.5970607995986938,
+      "learning_rate": 0.0005074128440366971,
+      "loss": 3.6904,
       "step": 14400
     },
     {
       "epoch": 1.5579514824797842,
-      "grad_norm": 0.5354477763175964,
-      "learning_rate": 0.0005070955207771181,
-      "loss": 3.7074,
+      "grad_norm": 0.5772601366043091,
+      "learning_rate": 0.0005070890447922288,
+      "loss": 3.7105,
       "step": 14450
     },
     {
       "epoch": 1.5633423180592994,
-      "grad_norm": 0.7448453903198242,
-      "learning_rate": 0.0005067717215326498,
-      "loss": 3.6686,
+      "grad_norm": 0.5723680853843689,
+      "learning_rate": 0.0005067652455477604,
+      "loss": 3.6743,
       "step": 14500
     },
     {
       "epoch": 1.568733153638814,
-      "grad_norm": 0.5273525714874268,
-      "learning_rate": 0.0005064479222881813,
-      "loss": 3.6823,
+      "grad_norm": 0.5193287134170532,
+      "learning_rate": 0.0005064414463032919,
+      "loss": 3.6868,
       "step": 14550
     },
     {
       "epoch": 1.5741239892183287,
-      "grad_norm": 0.5093368887901306,
-      "learning_rate": 0.0005061241230437129,
-      "loss": 3.6727,
+      "grad_norm": 0.5231333374977112,
+      "learning_rate": 0.0005061176470588235,
+      "loss": 3.678,
       "step": 14600
     },
     {
       "epoch": 1.5795148247978437,
-      "grad_norm": 0.5493596196174622,
-      "learning_rate": 0.0005058003237992444,
-      "loss": 3.6886,
+      "grad_norm": 0.5787746906280518,
+      "learning_rate": 0.0005057938478143551,
+      "loss": 3.6933,
       "step": 14650
     },
     {
       "epoch": 1.5849056603773586,
-      "grad_norm": 0.5913286209106445,
-      "learning_rate": 0.000505476524554776,
-      "loss": 3.6968,
+      "grad_norm": 0.5243505239486694,
+      "learning_rate": 0.0005054700485698867,
+      "loss": 3.6995,
       "step": 14700
     },
     {
       "epoch": 1.5902964959568733,
-      "grad_norm": 0.5922813415527344,
-      "learning_rate": 0.0005051527253103076,
-      "loss": 3.6575,
+      "grad_norm": 0.5743887424468994,
+      "learning_rate": 0.0005051462493254182,
+      "loss": 3.664,
       "step": 14750
     },
     {
       "epoch": 1.595687331536388,
-      "grad_norm": 0.562714695930481,
-      "learning_rate": 0.0005048289260658392,
-      "loss": 3.6826,
+      "grad_norm": 0.5387320518493652,
+      "learning_rate": 0.0005048224500809498,
+      "loss": 3.6878,
       "step": 14800
     },
     {
       "epoch": 1.6010781671159031,
-      "grad_norm": 0.5367226600646973,
-      "learning_rate": 0.0005045051268213707,
-      "loss": 3.6719,
+      "grad_norm": 0.5582491755485535,
+      "learning_rate": 0.0005044986508364813,
+      "loss": 3.6774,
       "step": 14850
     },
     {
       "epoch": 1.6064690026954178,
-      "grad_norm": 0.5268148183822632,
+      "grad_norm": 0.506304919719696,
       "learning_rate": 0.0005041813275769023,
-      "loss": 3.6715,
+      "loss": 3.6756,
       "step": 14900
     },
     {
       "epoch": 1.6118598382749325,
-      "grad_norm": 0.5168265104293823,
+      "grad_norm": 0.5047406554222107,
       "learning_rate": 0.0005038575283324338,
-      "loss": 3.6692,
+      "loss": 3.6757,
       "step": 14950
     },
     {
       "epoch": 1.6172506738544474,
-      "grad_norm": 0.5893277525901794,
+      "grad_norm": 0.6384860277175903,
       "learning_rate": 0.0005035337290879654,
-      "loss": 3.6767,
+      "loss": 3.6818,
       "step": 15000
     },
     {
       "epoch": 1.6172506738544474,
-      "eval_accuracy": 0.356096910118311,
-      "eval_loss": 3.634113311767578,
-      "eval_runtime": 183.2068,
-      "eval_samples_per_second": 98.31,
-      "eval_steps_per_second": 6.146,
+      "eval_accuracy": 0.35535035612613775,
+      "eval_loss": 3.637713670730591,
+      "eval_runtime": 183.9749,
+      "eval_samples_per_second": 97.899,
+      "eval_steps_per_second": 6.12,
       "step": 15000
     },
     {
       "epoch": 1.6226415094339623,
-      "grad_norm": 0.5801236033439636,
+      "grad_norm": 0.591760516166687,
       "learning_rate": 0.000503209929843497,
-      "loss": 3.6864,
+      "loss": 3.692,
       "step": 15050
     },
     {
       "epoch": 1.628032345013477,
-      "grad_norm": 0.5008922815322876,
+      "grad_norm": 0.5207642316818237,
       "learning_rate": 0.0005028926065839179,
-      "loss": 3.6614,
+      "loss": 3.6668,
       "step": 15100
     },
     {
       "epoch": 1.633423180592992,
-      "grad_norm": 0.5347914099693298,
+      "grad_norm": 0.5464783906936646,
       "learning_rate": 0.0005025688073394495,
-      "loss": 3.6811,
+      "loss": 3.6845,
       "step": 15150
     },
     {
       "epoch": 1.6388140161725069,
-      "grad_norm": 0.6196523904800415,
+      "grad_norm": 0.5907894968986511,
       "learning_rate": 0.0005022450080949811,
-      "loss": 3.6564,
+      "loss": 3.6616,
       "step": 15200
     },
     {
       "epoch": 1.6442048517520216,
-      "grad_norm": 0.5722506046295166,
+      "grad_norm": 0.5714707374572754,
       "learning_rate": 0.0005019212088505126,
-      "loss": 3.6983,
+      "loss": 3.7021,
       "step": 15250
     },
     {
       "epoch": 1.6495956873315363,
-      "grad_norm": 0.5670409202575684,
+      "grad_norm": 0.5418184399604797,
       "learning_rate": 0.0005015974096060442,
-      "loss": 3.6822,
+      "loss": 3.6873,
       "step": 15300
     },
     {
       "epoch": 1.6549865229110512,
-      "grad_norm": 0.5775005221366882,
+      "grad_norm": 0.5553717613220215,
       "learning_rate": 0.0005012736103615758,
-      "loss": 3.676,
+      "loss": 3.6804,
       "step": 15350
     },
     {
       "epoch": 1.6603773584905661,
-      "grad_norm": 0.5482152700424194,
+      "grad_norm": 0.5480347871780396,
       "learning_rate": 0.0005009498111171074,
-      "loss": 3.6664,
+      "loss": 3.669,
       "step": 15400
     },
     {
       "epoch": 1.6657681940700808,
-      "grad_norm": 0.5493103861808777,
+      "grad_norm": 0.5169306993484497,
       "learning_rate": 0.0005006260118726389,
-      "loss": 3.6771,
+      "loss": 3.6834,
       "step": 15450
     },
     {
       "epoch": 1.6711590296495957,
-      "grad_norm": 0.5443813800811768,
+      "grad_norm": 0.5624404549598694,
       "learning_rate": 0.0005003022126281705,
-      "loss": 3.6884,
+      "loss": 3.6945,
       "step": 15500
     },
     {
       "epoch": 1.6765498652291106,
-      "grad_norm": 0.5360202193260193,
+      "grad_norm": 0.5373254418373108,
       "learning_rate": 0.000499978413383702,
-      "loss": 3.6633,
+      "loss": 3.667,
       "step": 15550
     },
     {
       "epoch": 1.6819407008086253,
-      "grad_norm": 0.5757080316543579,
+      "grad_norm": 0.5668867230415344,
       "learning_rate": 0.0004996546141392336,
-      "loss": 3.6626,
+      "loss": 3.6653,
       "step": 15600
     },
     {
       "epoch": 1.68733153638814,
-      "grad_norm": 0.5827114582061768,
+      "grad_norm": 0.5291263461112976,
       "learning_rate": 0.0004993308148947651,
-      "loss": 3.657,
+      "loss": 3.6627,
       "step": 15650
     },
     {
       "epoch": 1.692722371967655,
-      "grad_norm": 0.5391814708709717,
+      "grad_norm": 0.5443230271339417,
       "learning_rate": 0.0004990070156502968,
-      "loss": 3.673,
+      "loss": 3.6765,
       "step": 15700
     },
     {
       "epoch": 1.6981132075471699,
-      "grad_norm": 0.6123207211494446,
+      "grad_norm": 0.5464130640029907,
       "learning_rate": 0.0004986832164058284,
-      "loss": 3.6493,
+      "loss": 3.6531,
       "step": 15750
     },
     {
       "epoch": 1.7035040431266846,
-      "grad_norm": 0.5068091154098511,
+      "grad_norm": 0.4945217967033386,
       "learning_rate": 0.0004983594171613599,
-      "loss": 3.6679,
+      "loss": 3.6735,
       "step": 15800
     },
     {
       "epoch": 1.7088948787061995,
-      "grad_norm": 0.5883404016494751,
+      "grad_norm": 0.5244830250740051,
       "learning_rate": 0.0004980356179168915,
-      "loss": 3.6597,
+      "loss": 3.6627,
       "step": 15850
     },
     {
       "epoch": 1.7142857142857144,
-      "grad_norm": 0.5995203256607056,
+      "grad_norm": 0.6120002865791321,
       "learning_rate": 0.000497711818672423,
-      "loss": 3.6626,
+      "loss": 3.6649,
       "step": 15900
     },
     {
       "epoch": 1.719676549865229,
-      "grad_norm": 0.5240509510040283,
+      "grad_norm": 0.50753253698349,
       "learning_rate": 0.0004973880194279547,
-      "loss": 3.6495,
+      "loss": 3.6548,
       "step": 15950
     },
     {
       "epoch": 1.7250673854447438,
-      "grad_norm": 0.5879561305046082,
+      "grad_norm": 0.5200812816619873,
       "learning_rate": 0.0004970642201834862,
-      "loss": 3.6497,
+      "loss": 3.6543,
       "step": 16000
     },
     {
       "epoch": 1.7250673854447438,
-      "eval_accuracy": 0.3578040642916522,
-      "eval_loss": 3.616135597229004,
-      "eval_runtime": 183.5062,
-      "eval_samples_per_second": 98.149,
-      "eval_steps_per_second": 6.136,
+      "eval_accuracy": 0.3569923185667902,
+      "eval_loss": 3.622196912765503,
+      "eval_runtime": 183.5449,
+      "eval_samples_per_second": 98.129,
+      "eval_steps_per_second": 6.135,
       "step": 16000
     },
     {
       "epoch": 1.7304582210242587,
-      "grad_norm": 0.5466377139091492,
+      "grad_norm": 0.5553668141365051,
       "learning_rate": 0.0004967404209390178,
-      "loss": 3.6711,
+      "loss": 3.6755,
       "step": 16050
     },
     {
       "epoch": 1.7358490566037736,
-      "grad_norm": 0.5741949081420898,
+      "grad_norm": 0.5911309123039246,
       "learning_rate": 0.0004964166216945493,
-      "loss": 3.6722,
+      "loss": 3.6767,
       "step": 16100
     },
     {
       "epoch": 1.7412398921832883,
-      "grad_norm": 0.6283635497093201,
+      "grad_norm": 0.6108778715133667,
       "learning_rate": 0.000496092822450081,
-      "loss": 3.6545,
+      "loss": 3.6579,
       "step": 16150
     },
     {
       "epoch": 1.7466307277628033,
-      "grad_norm": 0.5305262207984924,
+      "grad_norm": 0.4974657893180847,
       "learning_rate": 0.0004957690232056125,
-      "loss": 3.6599,
+      "loss": 3.6648,
       "step": 16200
     },
     {
       "epoch": 1.7520215633423182,
-      "grad_norm": 0.5391592979431152,
+      "grad_norm": 0.5382362008094788,
       "learning_rate": 0.0004954452239611441,
-      "loss": 3.6603,
+      "loss": 3.6634,
       "step": 16250
     },
     {
       "epoch": 1.7574123989218329,
-      "grad_norm": 0.5375102162361145,
+      "grad_norm": 0.5095261931419373,
       "learning_rate": 0.0004951214247166756,
-      "loss": 3.6588,
+      "loss": 3.6637,
       "step": 16300
     },
     {
       "epoch": 1.7628032345013476,
-      "grad_norm": 0.5385131239891052,
+      "grad_norm": 0.5040801763534546,
       "learning_rate": 0.0004947976254722072,
-      "loss": 3.6598,
+      "loss": 3.6649,
       "step": 16350
     },
     {
       "epoch": 1.7681940700808625,
-      "grad_norm": 0.581318199634552,
+      "grad_norm": 0.5789003372192383,
       "learning_rate": 0.0004944738262277387,
-      "loss": 3.664,
+      "loss": 3.6668,
       "step": 16400
     },
     {
       "epoch": 1.7735849056603774,
-      "grad_norm": 0.5098201632499695,
+      "grad_norm": 0.5566280484199524,
       "learning_rate": 0.0004941500269832703,
-      "loss": 3.666,
+      "loss": 3.669,
       "step": 16450
     },
     {
       "epoch": 1.778975741239892,
-      "grad_norm": 0.5255228877067566,
+      "grad_norm": 0.5179877877235413,
       "learning_rate": 0.0004938262277388019,
-      "loss": 3.6614,
+      "loss": 3.6653,
       "step": 16500
     },
     {
       "epoch": 1.784366576819407,
-      "grad_norm": 0.5504864454269409,
+      "grad_norm": 0.5120202898979187,
       "learning_rate": 0.0004935024284943335,
-      "loss": 3.641,
+      "loss": 3.6457,
       "step": 16550
     },
     {
       "epoch": 1.789757412398922,
-      "grad_norm": 0.510482132434845,
+      "grad_norm": 0.5303856730461121,
       "learning_rate": 0.000493178629249865,
-      "loss": 3.6548,
+      "loss": 3.6579,
       "step": 16600
     },
     {
       "epoch": 1.7951482479784366,
-      "grad_norm": 0.5756171941757202,
+      "grad_norm": 0.579076886177063,
       "learning_rate": 0.0004928548300053966,
-      "loss": 3.6613,
+      "loss": 3.6652,
       "step": 16650
     },
     {
       "epoch": 1.8005390835579513,
-      "grad_norm": 0.5339798927307129,
+      "grad_norm": 0.5309374332427979,
       "learning_rate": 0.0004925310307609282,
-      "loss": 3.653,
+      "loss": 3.6564,
       "step": 16700
     },
     {
       "epoch": 1.8059299191374663,
-      "grad_norm": 0.588362991809845,
+      "grad_norm": 0.5917450785636902,
       "learning_rate": 0.0004922072315164598,
-      "loss": 3.6417,
+      "loss": 3.6481,
       "step": 16750
     },
     {
       "epoch": 1.8113207547169812,
-      "grad_norm": 0.5696936249732971,
+      "grad_norm": 0.5440935492515564,
       "learning_rate": 0.0004918834322719913,
-      "loss": 3.6488,
+      "loss": 3.6522,
       "step": 16800
     },
     {
       "epoch": 1.8167115902964959,
-      "grad_norm": 0.5738952159881592,
+      "grad_norm": 0.5752487182617188,
       "learning_rate": 0.0004915596330275229,
-      "loss": 3.6448,
+      "loss": 3.6494,
       "step": 16850
     },
     {
       "epoch": 1.8221024258760108,
-      "grad_norm": 0.5368397235870361,
+      "grad_norm": 0.5438686609268188,
       "learning_rate": 0.0004912358337830544,
-      "loss": 3.6527,
+      "loss": 3.656,
       "step": 16900
     },
     {
       "epoch": 1.8274932614555257,
-      "grad_norm": 0.5771840214729309,
+      "grad_norm": 0.5319474339485168,
       "learning_rate": 0.000490912034538586,
-      "loss": 3.6372,
+      "loss": 3.642,
       "step": 16950
     },
     {
       "epoch": 1.8328840970350404,
-      "grad_norm": 0.5940222144126892,
+      "grad_norm": 0.5695987343788147,
       "learning_rate": 0.0004905882352941175,
-      "loss": 3.6291,
+      "loss": 3.6349,
       "step": 17000
     },
     {
       "epoch": 1.8328840970350404,
-      "eval_accuracy": 0.359461020830823,
-      "eval_loss": 3.600602626800537,
-      "eval_runtime": 183.2755,
-      "eval_samples_per_second": 98.273,
-      "eval_steps_per_second": 6.144,
+      "eval_accuracy": 0.3584435951893282,
+      "eval_loss": 3.60434627532959,
+      "eval_runtime": 184.7306,
+      "eval_samples_per_second": 97.499,
+      "eval_steps_per_second": 6.095,
       "step": 17000
     },
     {
       "epoch": 1.838274932614555,
-      "grad_norm": 0.49475905299186707,
+      "grad_norm": 0.5294819474220276,
       "learning_rate": 0.0004902644360496492,
-      "loss": 3.6373,
+      "loss": 3.6421,
       "step": 17050
     },
     {
       "epoch": 1.8436657681940702,
-      "grad_norm": 0.5278207063674927,
+      "grad_norm": 0.5615407228469849,
       "learning_rate": 0.0004899406368051808,
-      "loss": 3.6397,
+      "loss": 3.6443,
       "step": 17100
     },
     {
       "epoch": 1.849056603773585,
-      "grad_norm": 0.5330104827880859,
-      "learning_rate": 0.0004896233135456017,
-      "loss": 3.6291,
+      "grad_norm": 0.5237312912940979,
+      "learning_rate": 0.0004896168375607123,
+      "loss": 3.6317,
       "step": 17150
     },
     {
       "epoch": 1.8544474393530996,
-      "grad_norm": 0.5108185410499573,
-      "learning_rate": 0.0004892995143011333,
-      "loss": 3.6578,
+      "grad_norm": 0.5544662475585938,
+      "learning_rate": 0.0004892930383162439,
+      "loss": 3.6602,
       "step": 17200
     },
     {
       "epoch": 1.8598382749326146,
-      "grad_norm": 0.5817229747772217,
-      "learning_rate": 0.0004889757150566648,
-      "loss": 3.6642,
+      "grad_norm": 0.5750020742416382,
+      "learning_rate": 0.0004889692390717754,
+      "loss": 3.6694,
       "step": 17250
     },
     {
       "epoch": 1.8652291105121295,
-      "grad_norm": 0.5322571992874146,
-      "learning_rate": 0.0004886519158121964,
-      "loss": 3.648,
+      "grad_norm": 0.5347704887390137,
+      "learning_rate": 0.0004886454398273071,
+      "loss": 3.6526,
       "step": 17300
     },
     {
       "epoch": 1.8706199460916442,
-      "grad_norm": 0.5755774974822998,
-      "learning_rate": 0.000488328116567728,
-      "loss": 3.636,
+      "grad_norm": 0.5284920930862427,
+      "learning_rate": 0.0004883216405828386,
+      "loss": 3.6427,
       "step": 17350
     },
     {
       "epoch": 1.8760107816711589,
-      "grad_norm": 0.5633023977279663,
-      "learning_rate": 0.0004880043173232595,
-      "loss": 3.6531,
+      "grad_norm": 0.5603639483451843,
+      "learning_rate": 0.00048799784133837017,
+      "loss": 3.6554,
       "step": 17400
     },
     {
       "epoch": 1.881401617250674,
-      "grad_norm": 0.5242737531661987,
-      "learning_rate": 0.0004876805180787911,
-      "loss": 3.6244,
+      "grad_norm": 0.547793984413147,
+      "learning_rate": 0.0004876740420939017,
+      "loss": 3.6277,
       "step": 17450
     },
     {
       "epoch": 1.8867924528301887,
-      "grad_norm": 0.5220695734024048,
-      "learning_rate": 0.0004873567188343227,
-      "loss": 3.6473,
+      "grad_norm": 0.5595480799674988,
+      "learning_rate": 0.0004873502428494333,
+      "loss": 3.6493,
       "step": 17500
     },
     {
       "epoch": 1.8921832884097034,
-      "grad_norm": 0.4793733060359955,
+      "grad_norm": 0.5159085392951965,
       "learning_rate": 0.0004870329195898542,
-      "loss": 3.6514,
+      "loss": 3.6534,
       "step": 17550
     },
     {
       "epoch": 1.8975741239892183,
-      "grad_norm": 0.6032518148422241,
+      "grad_norm": 0.5964058637619019,
       "learning_rate": 0.00048670912034538583,
-      "loss": 3.6398,
+      "loss": 3.6436,
       "step": 17600
     },
     {
       "epoch": 1.9029649595687332,
-      "grad_norm": 0.562566876411438,
+      "grad_norm": 0.5292296409606934,
       "learning_rate": 0.0004863853211009174,
-      "loss": 3.6368,
+      "loss": 3.6417,
       "step": 17650
     },
     {
       "epoch": 1.908355795148248,
-      "grad_norm": 0.5620681643486023,
+      "grad_norm": 0.5534155368804932,
       "learning_rate": 0.000486061521856449,
-      "loss": 3.6535,
+      "loss": 3.6574,
       "step": 17700
     },
     {
       "epoch": 1.9137466307277629,
-      "grad_norm": 0.548251748085022,
+      "grad_norm": 0.545872151851654,
       "learning_rate": 0.00048573772261198054,
-      "loss": 3.6267,
+      "loss": 3.6317,
       "step": 17750
     },
     {
       "epoch": 1.9191374663072778,
-      "grad_norm": 0.5563006401062012,
+      "grad_norm": 0.5668688416481018,
       "learning_rate": 0.00048541392336751214,
-      "loss": 3.632,
+      "loss": 3.6374,
       "step": 17800
     },
     {
       "epoch": 1.9245283018867925,
-      "grad_norm": 0.5012133121490479,
+      "grad_norm": 0.4825321435928345,
       "learning_rate": 0.0004850901241230437,
-      "loss": 3.6225,
+      "loss": 3.6252,
       "step": 17850
     },
     {
       "epoch": 1.9299191374663072,
-      "grad_norm": 0.5342919826507568,
+      "grad_norm": 0.5110409259796143,
       "learning_rate": 0.0004847663248785753,
-      "loss": 3.6266,
+      "loss": 3.6309,
       "step": 17900
     },
     {
       "epoch": 1.935309973045822,
-      "grad_norm": 0.5178313851356506,
+      "grad_norm": 0.5401217937469482,
       "learning_rate": 0.0004844425256341068,
-      "loss": 3.6355,
+      "loss": 3.6392,
       "step": 17950
     },
     {
       "epoch": 1.940700808625337,
-      "grad_norm": 0.5467429161071777,
+      "grad_norm": 0.572462260723114,
       "learning_rate": 0.00048411872638963834,
-      "loss": 3.643,
+      "loss": 3.6464,
       "step": 18000
     },
     {
       "epoch": 1.940700808625337,
-      "eval_accuracy": 0.36094337217927597,
-      "eval_loss": 3.584843873977661,
-      "eval_runtime": 183.5851,
-      "eval_samples_per_second": 98.107,
-      "eval_steps_per_second": 6.133,
+      "eval_accuracy": 0.3601135900750368,
+      "eval_loss": 3.5912423133850098,
+      "eval_runtime": 184.7034,
+      "eval_samples_per_second": 97.513,
+      "eval_steps_per_second": 6.096,
       "step": 18000
     },
     {
       "epoch": 1.9460916442048517,
-      "grad_norm": 0.5631021857261658,
+      "grad_norm": 0.5204117298126221,
       "learning_rate": 0.00048379492714516995,
-      "loss": 3.6179,
+      "loss": 3.6226,
       "step": 18050
     },
     {
       "epoch": 1.9514824797843666,
-      "grad_norm": 0.541903018951416,
+      "grad_norm": 0.5143402218818665,
       "learning_rate": 0.0004834711279007015,
-      "loss": 3.6427,
+      "loss": 3.6487,
       "step": 18100
     },
     {
       "epoch": 1.9568733153638815,
-      "grad_norm": 0.5587621927261353,
+      "grad_norm": 0.5703872442245483,
       "learning_rate": 0.0004831473286562331,
-      "loss": 3.6403,
+      "loss": 3.6425,
       "step": 18150
     },
     {
       "epoch": 1.9622641509433962,
-      "grad_norm": 1.227616548538208,
+      "grad_norm": 0.7369431257247925,
       "learning_rate": 0.00048282352941176465,
-      "loss": 3.6295,
+      "loss": 3.6339,
       "step": 18200
     },
     {
       "epoch": 1.967654986522911,
-      "grad_norm": 0.5264652371406555,
+      "grad_norm": 0.5618401169776917,
       "learning_rate": 0.00048249973016729626,
-      "loss": 3.645,
+      "loss": 3.6473,
       "step": 18250
     },
     {
       "epoch": 1.9730458221024259,
-      "grad_norm": 0.5398224592208862,
+      "grad_norm": 0.5731032490730286,
       "learning_rate": 0.0004821759309228278,
-      "loss": 3.6412,
+      "loss": 3.645,
       "step": 18300
     },
     {
       "epoch": 1.9784366576819408,
-      "grad_norm": 0.5698915123939514,
+      "grad_norm": 0.5541403293609619,
       "learning_rate": 0.00048185213167835936,
-      "loss": 3.6253,
+      "loss": 3.629,
       "step": 18350
     },
     {
       "epoch": 1.9838274932614555,
-      "grad_norm": 0.5805723071098328,
+      "grad_norm": 0.5562690496444702,
       "learning_rate": 0.00048152833243389096,
-      "loss": 3.6555,
+      "loss": 3.6606,
       "step": 18400
     },
     {
       "epoch": 1.9892183288409704,
-      "grad_norm": 0.5575518012046814,
+      "grad_norm": 0.5553107261657715,
       "learning_rate": 0.0004812045331894225,
-      "loss": 3.611,
+      "loss": 3.6153,
       "step": 18450
     },
     {
       "epoch": 1.9946091644204853,
-      "grad_norm": 0.5759623050689697,
+      "grad_norm": 0.5449579954147339,
       "learning_rate": 0.0004808807339449541,
-      "loss": 3.617,
+      "loss": 3.6234,
       "step": 18500
     },
     {
       "epoch": 2.0,
-      "grad_norm": 1.1932631731033325,
+      "grad_norm": 1.0665850639343262,
       "learning_rate": 0.0004805569347004856,
-      "loss": 3.6358,
+      "loss": 3.6395,
       "step": 18550
     },
     {
       "epoch": 2.0053908355795147,
-      "grad_norm": 0.5486853122711182,
+      "grad_norm": 0.5312926769256592,
       "learning_rate": 0.0004802331354560173,
-      "loss": 3.5442,
+      "loss": 3.5485,
       "step": 18600
     },
     {
       "epoch": 2.01078167115903,
-      "grad_norm": 0.5466881394386292,
+      "grad_norm": 0.5661903619766235,
       "learning_rate": 0.00047990933621154877,
-      "loss": 3.5451,
+      "loss": 3.5506,
       "step": 18650
     },
     {
       "epoch": 2.0161725067385445,
-      "grad_norm": 0.5424332618713379,
+      "grad_norm": 0.5767152905464172,
       "learning_rate": 0.0004795855369670804,
-      "loss": 3.5382,
+      "loss": 3.5433,
       "step": 18700
     },
     {
       "epoch": 2.0215633423180592,
-      "grad_norm": 0.5714851021766663,
+      "grad_norm": 0.5376216769218445,
       "learning_rate": 0.0004792617377226119,
-      "loss": 3.5577,
+      "loss": 3.5611,
       "step": 18750
     },
     {
       "epoch": 2.026954177897574,
-      "grad_norm": 0.5695775151252747,
+      "grad_norm": 0.5663812160491943,
       "learning_rate": 0.0004789379384781435,
-      "loss": 3.5428,
+      "loss": 3.5463,
       "step": 18800
     },
     {
       "epoch": 2.032345013477089,
-      "grad_norm": 0.530637264251709,
+      "grad_norm": 0.5509110689163208,
       "learning_rate": 0.0004786141392336751,
-      "loss": 3.5231,
+      "loss": 3.5305,
       "step": 18850
     },
     {
       "epoch": 2.0377358490566038,
-      "grad_norm": 0.5261335968971252,
+      "grad_norm": 0.5725340247154236,
       "learning_rate": 0.00047829033998920663,
-      "loss": 3.5522,
+      "loss": 3.5575,
       "step": 18900
     },
     {
       "epoch": 2.0431266846361185,
-      "grad_norm": 0.5697975158691406,
+      "grad_norm": 0.567261815071106,
       "learning_rate": 0.00047796654074473824,
-      "loss": 3.5444,
+      "loss": 3.5483,
       "step": 18950
     },
     {
       "epoch": 2.0485175202156336,
-      "grad_norm": 0.6406385898590088,
+      "grad_norm": 0.538179874420166,
       "learning_rate": 0.0004776427415002698,
-      "loss": 3.5593,
+      "loss": 3.5633,
       "step": 19000
     },
     {
       "epoch": 2.0485175202156336,
-      "eval_accuracy": 0.36186920343634316,
-      "eval_loss": 3.57981014251709,
-      "eval_runtime": 183.1277,
-      "eval_samples_per_second": 98.352,
-      "eval_steps_per_second": 6.149,
+      "eval_accuracy": 0.36126226840501147,
+      "eval_loss": 3.5819406509399414,
+      "eval_runtime": 184.5182,
+      "eval_samples_per_second": 97.611,
+      "eval_steps_per_second": 6.102,
       "step": 19000
     },
     {
       "epoch": 2.0539083557951483,
-      "grad_norm": 0.5842219591140747,
+      "grad_norm": 0.6141398549079895,
       "learning_rate": 0.0004773189422558014,
-      "loss": 3.5202,
+      "loss": 3.5224,
       "step": 19050
     },
     {
       "epoch": 2.059299191374663,
-      "grad_norm": 0.5753844380378723,
+      "grad_norm": 0.5971595048904419,
       "learning_rate": 0.00047699514301133294,
-      "loss": 3.5584,
+      "loss": 3.5632,
       "step": 19100
     },
     {
       "epoch": 2.0646900269541777,
-      "grad_norm": 0.5149918794631958,
+      "grad_norm": 0.537144660949707,
       "learning_rate": 0.00047667134376686455,
-      "loss": 3.5434,
+      "loss": 3.5482,
       "step": 19150
     },
     {
       "epoch": 2.070080862533693,
-      "grad_norm": 0.5646060705184937,
-      "learning_rate": 0.0004763540205072854,
-      "loss": 3.5652,
+      "grad_norm": 0.5301309823989868,
+      "learning_rate": 0.0004763475445223961,
+      "loss": 3.5681,
       "step": 19200
     },
     {
       "epoch": 2.0754716981132075,
-      "grad_norm": 0.6100060343742371,
-      "learning_rate": 0.00047603022126281705,
-      "loss": 3.563,
+      "grad_norm": 0.5356302857398987,
+      "learning_rate": 0.0004760237452779276,
+      "loss": 3.5651,
       "step": 19250
     },
     {
       "epoch": 2.0808625336927222,
-      "grad_norm": 0.5924269556999207,
-      "learning_rate": 0.00047570642201834855,
-      "loss": 3.5591,
+      "grad_norm": 0.5818206071853638,
+      "learning_rate": 0.0004756999460334592,
+      "loss": 3.5635,
       "step": 19300
     },
     {
       "epoch": 2.0862533692722374,
-      "grad_norm": 0.5271193385124207,
-      "learning_rate": 0.00047538262277388015,
-      "loss": 3.5502,
+      "grad_norm": 0.5593371987342834,
+      "learning_rate": 0.00047537614678899075,
+      "loss": 3.5531,
       "step": 19350
     },
     {
       "epoch": 2.091644204851752,
-      "grad_norm": 0.559655487537384,
-      "learning_rate": 0.0004750588235294117,
-      "loss": 3.5551,
+      "grad_norm": 0.5617176294326782,
+      "learning_rate": 0.00047505234754452235,
+      "loss": 3.5586,
       "step": 19400
     },
     {
       "epoch": 2.0970350404312668,
-      "grad_norm": 0.5192553997039795,
-      "learning_rate": 0.0004747350242849433,
-      "loss": 3.5419,
+      "grad_norm": 0.547956109046936,
+      "learning_rate": 0.0004747285483000539,
+      "loss": 3.5474,
       "step": 19450
     },
     {
       "epoch": 2.1024258760107815,
-      "grad_norm": 0.5999012589454651,
-      "learning_rate": 0.00047441122504047486,
-      "loss": 3.5376,
+      "grad_norm": 0.5973509550094604,
+      "learning_rate": 0.0004744047490555855,
+      "loss": 3.5412,
       "step": 19500
     },
     {
       "epoch": 2.1078167115902966,
-      "grad_norm": 0.601438581943512,
-      "learning_rate": 0.0004740874257960064,
-      "loss": 3.5554,
+      "grad_norm": 0.5937268137931824,
+      "learning_rate": 0.00047408094981111706,
+      "loss": 3.5597,
       "step": 19550
     },
     {
       "epoch": 2.1132075471698113,
-      "grad_norm": 0.5252161622047424,
-      "learning_rate": 0.000473763626551538,
-      "loss": 3.5518,
+      "grad_norm": 0.5365374684333801,
+      "learning_rate": 0.00047375715056664866,
+      "loss": 3.5539,
       "step": 19600
     },
     {
       "epoch": 2.118598382749326,
-      "grad_norm": 0.568138062953949,
-      "learning_rate": 0.00047343982730706956,
-      "loss": 3.5517,
+      "grad_norm": 0.5774807929992676,
+      "learning_rate": 0.0004734333513221802,
+      "loss": 3.5552,
       "step": 19650
     },
     {
       "epoch": 2.123989218328841,
-      "grad_norm": 0.5670929551124573,
-      "learning_rate": 0.00047311602806260117,
-      "loss": 3.5404,
+      "grad_norm": 0.5723707675933838,
+      "learning_rate": 0.00047310955207771177,
+      "loss": 3.5463,
       "step": 19700
     },
     {
       "epoch": 2.129380053908356,
-      "grad_norm": 0.6419331431388855,
-      "learning_rate": 0.0004727922288181327,
-      "loss": 3.5383,
+      "grad_norm": 0.6286929249763489,
+      "learning_rate": 0.00047278575283324337,
+      "loss": 3.5417,
       "step": 19750
     },
     {
       "epoch": 2.1347708894878705,
-      "grad_norm": 0.5455891489982605,
-      "learning_rate": 0.0004724684295736643,
-      "loss": 3.5673,
+      "grad_norm": 0.5768539309501648,
+      "learning_rate": 0.0004724619535887749,
+      "loss": 3.5725,
       "step": 19800
     },
     {
       "epoch": 2.1401617250673857,
-      "grad_norm": 0.5633212327957153,
-      "learning_rate": 0.0004721446303291959,
-      "loss": 3.5679,
+      "grad_norm": 0.5574427843093872,
+      "learning_rate": 0.0004721381543443065,
+      "loss": 3.5727,
       "step": 19850
     },
     {
       "epoch": 2.1455525606469004,
-      "grad_norm": Infinity,
-      "learning_rate": 0.0004718273070696168,
-      "loss": 3.5567,
+      "grad_norm": 0.7127569913864136,
+      "learning_rate": 0.000471814355099838,
+      "loss": 3.5606,
       "step": 19900
     },
     {
       "epoch": 2.150943396226415,
-      "grad_norm": 0.5624653697013855,
-      "learning_rate": 0.0004715035078251483,
-      "loss": 3.5539,
+      "grad_norm": 0.5295952558517456,
+      "learning_rate": 0.0004714905558553697,
+      "loss": 3.5585,
       "step": 19950
     },
     {
       "epoch": 2.1563342318059298,
-      "grad_norm": 0.5584042072296143,
-      "learning_rate": 0.0004711797085806799,
-      "loss": 3.5552,
+      "grad_norm": 0.5639391541481018,
+      "learning_rate": 0.0004711667566109012,
+      "loss": 3.5575,
       "step": 20000
     },
     {
       "epoch": 2.1563342318059298,
-      "eval_accuracy": 0.3635111658769956,
-      "eval_loss": 3.5678915977478027,
-      "eval_runtime": 183.3667,
-      "eval_samples_per_second": 98.224,
-      "eval_steps_per_second": 6.141,
+      "eval_accuracy": 0.36280861630438715,
+      "eval_loss": 3.5698821544647217,
+      "eval_runtime": 184.6295,
+      "eval_samples_per_second": 97.552,
+      "eval_steps_per_second": 6.099,
       "step": 20000
     },
     {
       "epoch": 2.161725067385445,
-      "grad_norm": 0.5540956258773804,
-      "learning_rate": 0.0004708559093362115,
-      "loss": 3.5543,
+      "grad_norm": 0.5672075152397156,
+      "learning_rate": 0.00047084295736643273,
+      "loss": 3.5578,
       "step": 20050
     },
     {
       "epoch": 2.1671159029649596,
-      "grad_norm": 0.5896965265274048,
-      "learning_rate": 0.0004705321100917431,
-      "loss": 3.5654,
+      "grad_norm": 0.6316539645195007,
+      "learning_rate": 0.00047051915812196433,
+      "loss": 3.5688,
       "step": 20100
     },
     {
       "epoch": 2.1725067385444743,
-      "grad_norm": 0.5863565802574158,
-      "learning_rate": 0.00047020831084727463,
-      "loss": 3.5427,
+      "grad_norm": 0.5680938959121704,
+      "learning_rate": 0.0004701953588774959,
+      "loss": 3.5456,
       "step": 20150
     },
     {
       "epoch": 2.177897574123989,
-      "grad_norm": 0.6331449747085571,
-      "learning_rate": 0.00046988451160280624,
-      "loss": 3.5658,
+      "grad_norm": 0.5884872674942017,
+      "learning_rate": 0.0004698715596330275,
+      "loss": 3.5699,
       "step": 20200
     },
     {
       "epoch": 2.183288409703504,
-      "grad_norm": 0.5432201027870178,
-      "learning_rate": 0.0004695607123583378,
-      "loss": 3.5648,
+      "grad_norm": 0.5326011180877686,
+      "learning_rate": 0.00046954776038855904,
+      "loss": 3.5695,
       "step": 20250
     },
     {
       "epoch": 2.188679245283019,
-      "grad_norm": 0.5611268877983093,
-      "learning_rate": 0.00046923691311386934,
-      "loss": 3.5401,
+      "grad_norm": 0.6190282106399536,
+      "learning_rate": 0.00046922396114409064,
+      "loss": 3.5441,
       "step": 20300
     },
     {
       "epoch": 2.1940700808625335,
-      "grad_norm": 0.5305395722389221,
-      "learning_rate": 0.00046891311386940094,
-      "loss": 3.5583,
+      "grad_norm": 0.5389439463615417,
+      "learning_rate": 0.0004689001618996222,
+      "loss": 3.5609,
       "step": 20350
     },
     {
       "epoch": 2.1994609164420487,
-      "grad_norm": 0.5022453665733337,
-      "learning_rate": 0.0004685893146249325,
-      "loss": 3.5571,
+      "grad_norm": 0.5145541429519653,
+      "learning_rate": 0.0004685763626551538,
+      "loss": 3.5609,
       "step": 20400
     },
     {
       "epoch": 2.2048517520215634,
-      "grad_norm": 0.5850080847740173,
-      "learning_rate": 0.0004682655153804641,
-      "loss": 3.5559,
+      "grad_norm": 0.5490039587020874,
+      "learning_rate": 0.00046825256341068535,
+      "loss": 3.5603,
       "step": 20450
     },
     {
       "epoch": 2.210242587601078,
-      "grad_norm": 0.5080869793891907,
-      "learning_rate": 0.00046794171613599565,
-      "loss": 3.5601,
+      "grad_norm": 0.4999011754989624,
+      "learning_rate": 0.0004679287641662169,
+      "loss": 3.5635,
       "step": 20500
     },
     {
       "epoch": 2.215633423180593,
-      "grad_norm": 0.5989983081817627,
-      "learning_rate": 0.00046761791689152725,
-      "loss": 3.563,
+      "grad_norm": 0.6086052060127258,
+      "learning_rate": 0.0004676049649217485,
+      "loss": 3.5662,
       "step": 20550
     },
     {
       "epoch": 2.221024258760108,
-      "grad_norm": 0.6006690859794617,
-      "learning_rate": 0.00046729411764705875,
-      "loss": 3.5674,
+      "grad_norm": 0.5756791234016418,
+      "learning_rate": 0.00046728764166216946,
+      "loss": 3.572,
       "step": 20600
     },
     {
       "epoch": 2.2264150943396226,
-      "grad_norm": 0.5296274423599243,
-      "learning_rate": 0.0004669703184025904,
-      "loss": 3.5525,
+      "grad_norm": 0.5605027675628662,
+      "learning_rate": 0.00046696384241770095,
+      "loss": 3.5554,
       "step": 20650
     },
     {
       "epoch": 2.2318059299191373,
-      "grad_norm": 0.5273208618164062,
-      "learning_rate": 0.0004666465191581219,
-      "loss": 3.5551,
+      "grad_norm": 0.566175639629364,
+      "learning_rate": 0.00046664004317323256,
+      "loss": 3.5597,
       "step": 20700
     },
     {
       "epoch": 2.2371967654986524,
-      "grad_norm": 0.6000422239303589,
-      "learning_rate": 0.00046632271991365346,
-      "loss": 3.556,
+      "grad_norm": 0.5631338953971863,
+      "learning_rate": 0.0004663162439287641,
+      "loss": 3.5616,
       "step": 20750
     },
     {
       "epoch": 2.242587601078167,
-      "grad_norm": 0.542732298374176,
-      "learning_rate": 0.00046599892066918506,
-      "loss": 3.5593,
+      "grad_norm": 0.5509245991706848,
+      "learning_rate": 0.00046599244468429566,
+      "loss": 3.5634,
       "step": 20800
     },
     {
       "epoch": 2.247978436657682,
-      "grad_norm": 0.5846306085586548,
-      "learning_rate": 0.0004656751214247166,
-      "loss": 3.5424,
+      "grad_norm": 0.5366947650909424,
+      "learning_rate": 0.00046566864543982726,
+      "loss": 3.5469,
       "step": 20850
     },
     {
       "epoch": 2.2533692722371965,
-      "grad_norm": 0.6560302972793579,
-      "learning_rate": 0.0004653513221802482,
-      "loss": 3.554,
+      "grad_norm": 0.6192650198936462,
+      "learning_rate": 0.0004653448461953588,
+      "loss": 3.5571,
       "step": 20900
     },
     {
       "epoch": 2.2587601078167117,
-      "grad_norm": 0.5398501753807068,
-      "learning_rate": 0.00046502752293577977,
-      "loss": 3.5597,
+      "grad_norm": 0.5511881113052368,
+      "learning_rate": 0.0004650210469508904,
+      "loss": 3.5644,
       "step": 20950
     },
     {
       "epoch": 2.2641509433962264,
-      "grad_norm": 0.5567417144775391,
-      "learning_rate": 0.00046470372369131137,
-      "loss": 3.5423,
+      "grad_norm": 0.5829699635505676,
+      "learning_rate": 0.00046469724770642197,
+      "loss": 3.5483,
       "step": 21000
     },
     {
       "epoch": 2.2641509433962264,
-      "eval_accuracy": 0.3644758948679004,
-      "eval_loss": 3.5552797317504883,
-      "eval_runtime": 183.5672,
-      "eval_samples_per_second": 98.117,
-      "eval_steps_per_second": 6.134,
+      "eval_accuracy": 0.3637985527652649,
+      "eval_loss": 3.5596225261688232,
+      "eval_runtime": 183.4209,
+      "eval_samples_per_second": 98.195,
+      "eval_steps_per_second": 6.139,
       "step": 21000
     },
     {
       "epoch": 2.269541778975741,
-      "grad_norm": 0.5651289224624634,
-      "learning_rate": 0.0004643799244468429,
-      "loss": 3.5383,
+      "grad_norm": 0.5781875848770142,
+      "learning_rate": 0.0004643734484619536,
+      "loss": 3.5428,
       "step": 21050
     },
     {
       "epoch": 2.274932614555256,
-      "grad_norm": 0.5148941278457642,
-      "learning_rate": 0.0004640561252023745,
-      "loss": 3.5681,
+      "grad_norm": 0.537315845489502,
+      "learning_rate": 0.0004640496492174851,
+      "loss": 3.5703,
       "step": 21100
     },
     {
       "epoch": 2.280323450134771,
-      "grad_norm": 0.5875728726387024,
-      "learning_rate": 0.0004637323259579061,
-      "loss": 3.5605,
+      "grad_norm": 0.5636802911758423,
+      "learning_rate": 0.00046372584997301673,
+      "loss": 3.5654,
       "step": 21150
     },
     {
       "epoch": 2.2857142857142856,
-      "grad_norm": 0.5935449004173279,
-      "learning_rate": 0.0004634085267134376,
-      "loss": 3.5523,
+      "grad_norm": 0.5732676386833191,
+      "learning_rate": 0.0004634020507285483,
+      "loss": 3.5568,
       "step": 21200
     },
     {
       "epoch": 2.2911051212938007,
-      "grad_norm": 0.5729395747184753,
-      "learning_rate": 0.00046308472746896923,
-      "loss": 3.5244,
+      "grad_norm": 0.5696597695350647,
+      "learning_rate": 0.0004630782514840798,
+      "loss": 3.5295,
       "step": 21250
     },
     {
       "epoch": 2.2964959568733154,
-      "grad_norm": 0.6214632391929626,
-      "learning_rate": 0.00046276092822450073,
-      "loss": 3.5476,
+      "grad_norm": 0.5938985347747803,
+      "learning_rate": 0.0004627544522396114,
+      "loss": 3.5529,
       "step": 21300
     },
     {
       "epoch": 2.30188679245283,
-      "grad_norm": 0.5737814903259277,
-      "learning_rate": 0.00046243712898003233,
-      "loss": 3.5626,
+      "grad_norm": 0.5898323059082031,
+      "learning_rate": 0.00046243065299514293,
+      "loss": 3.5668,
       "step": 21350
     },
     {
       "epoch": 2.3072776280323453,
-      "grad_norm": 0.5795217156410217,
-      "learning_rate": 0.0004621133297355639,
-      "loss": 3.54,
+      "grad_norm": 0.565859317779541,
+      "learning_rate": 0.00046210685375067454,
+      "loss": 3.5428,
       "step": 21400
     },
     {
       "epoch": 2.31266846361186,
-      "grad_norm": 0.5368177890777588,
-      "learning_rate": 0.0004617895304910955,
-      "loss": 3.552,
+      "grad_norm": 0.505027711391449,
+      "learning_rate": 0.0004617830545062061,
+      "loss": 3.5567,
       "step": 21450
     },
     {
       "epoch": 2.3180592991913747,
-      "grad_norm": 0.6149646639823914,
-      "learning_rate": 0.00046146573124662704,
-      "loss": 3.557,
+      "grad_norm": 0.5774626135826111,
+      "learning_rate": 0.0004614592552617377,
+      "loss": 3.558,
       "step": 21500
     },
     {
       "epoch": 2.3234501347708894,
-      "grad_norm": 0.5619512796401978,
-      "learning_rate": 0.00046114193200215864,
-      "loss": 3.5438,
+      "grad_norm": 0.5658170580863953,
+      "learning_rate": 0.00046113545601726924,
+      "loss": 3.5466,
       "step": 21550
     },
     {
       "epoch": 2.3288409703504045,
-      "grad_norm": 0.5701526403427124,
-      "learning_rate": 0.0004608181327576902,
-      "loss": 3.5537,
+      "grad_norm": 0.5221099257469177,
+      "learning_rate": 0.00046081165677280085,
+      "loss": 3.5579,
       "step": 21600
     },
     {
       "epoch": 2.334231805929919,
-      "grad_norm": 0.5332589149475098,
-      "learning_rate": 0.00046049433351322175,
-      "loss": 3.5537,
+      "grad_norm": 0.5070032477378845,
+      "learning_rate": 0.0004604878575283324,
+      "loss": 3.5567,
       "step": 21650
     },
     {
       "epoch": 2.339622641509434,
-      "grad_norm": 0.583706259727478,
-      "learning_rate": 0.00046017053426875335,
-      "loss": 3.5544,
+      "grad_norm": 0.5588931441307068,
+      "learning_rate": 0.00046016405828386395,
+      "loss": 3.5579,
       "step": 21700
     },
     {
       "epoch": 2.3450134770889486,
-      "grad_norm": 0.5967761278152466,
-      "learning_rate": 0.0004598467350242849,
-      "loss": 3.5528,
+      "grad_norm": 0.5342980623245239,
+      "learning_rate": 0.00045984025903939555,
+      "loss": 3.5585,
       "step": 21750
     },
     {
       "epoch": 2.3504043126684637,
-      "grad_norm": 0.5579556822776794,
-      "learning_rate": 0.0004595229357798165,
-      "loss": 3.5632,
+      "grad_norm": 0.5457897782325745,
+      "learning_rate": 0.0004595164597949271,
+      "loss": 3.5663,
       "step": 21800
     },
     {
       "epoch": 2.3557951482479784,
-      "grad_norm": 0.5610327124595642,
-      "learning_rate": 0.00045919913653534806,
-      "loss": 3.5557,
+      "grad_norm": 0.5546021461486816,
+      "learning_rate": 0.0004591926605504587,
+      "loss": 3.562,
       "step": 21850
     },
     {
       "epoch": 2.361185983827493,
-      "grad_norm": 0.5787166953086853,
-      "learning_rate": 0.00045887533729087966,
-      "loss": 3.5439,
+      "grad_norm": 0.521409809589386,
+      "learning_rate": 0.0004588688613059902,
+      "loss": 3.5463,
       "step": 21900
     },
     {
       "epoch": 2.3665768194070083,
-      "grad_norm": 0.540556013584137,
-      "learning_rate": 0.00045855153804641116,
-      "loss": 3.558,
+      "grad_norm": 0.5697400569915771,
+      "learning_rate": 0.00045854506206152186,
+      "loss": 3.5603,
       "step": 21950
     },
     {
       "epoch": 2.371967654986523,
-      "grad_norm": 0.5531706809997559,
-      "learning_rate": 0.0004582277388019427,
-      "loss": 3.5562,
+      "grad_norm": 0.5319057703018188,
+      "learning_rate": 0.00045822126281705336,
+      "loss": 3.559,
       "step": 22000
     },
     {
       "epoch": 2.371967654986523,
-      "eval_accuracy": 0.3654050943644898,
-      "eval_loss": 3.5464489459991455,
-      "eval_runtime": 183.2178,
-      "eval_samples_per_second": 98.304,
-      "eval_steps_per_second": 6.146,
+      "eval_accuracy": 0.36464995679417916,
+      "eval_loss": 3.5488476753234863,
+      "eval_runtime": 183.7772,
+      "eval_samples_per_second": 98.005,
+      "eval_steps_per_second": 6.127,
       "step": 22000
     },
     {
       "epoch": 2.3773584905660377,
-      "grad_norm": 0.5978881120681763,
-      "learning_rate": 0.0004579039395574743,
-      "loss": 3.5342,
+      "grad_norm": 0.6048839092254639,
+      "learning_rate": 0.00045789746357258497,
+      "loss": 3.5371,
       "step": 22050
     },
     {
       "epoch": 2.382749326145553,
-      "grad_norm": 0.5623920559883118,
-      "learning_rate": 0.00045758014031300586,
-      "loss": 3.5555,
+      "grad_norm": 0.5780558586120605,
+      "learning_rate": 0.0004575736643281165,
+      "loss": 3.5578,
       "step": 22100
     },
     {
       "epoch": 2.3881401617250675,
-      "grad_norm": 0.5382367372512817,
-      "learning_rate": 0.00045725634106853747,
-      "loss": 3.5515,
+      "grad_norm": 0.5097941756248474,
+      "learning_rate": 0.00045724986508364807,
+      "loss": 3.5551,
       "step": 22150
     },
     {
       "epoch": 2.393530997304582,
-      "grad_norm": 0.6158401370048523,
-      "learning_rate": 0.000456932541824069,
-      "loss": 3.5763,
+      "grad_norm": 0.5635842680931091,
+      "learning_rate": 0.00045692606583917967,
+      "loss": 3.5788,
       "step": 22200
     },
     {
       "epoch": 2.398921832884097,
-      "grad_norm": 0.5851870775222778,
-      "learning_rate": 0.0004566087425796006,
-      "loss": 3.546,
+      "grad_norm": 0.5743860602378845,
+      "learning_rate": 0.0004566022665947112,
+      "loss": 3.5507,
       "step": 22250
     },
     {
       "epoch": 2.404312668463612,
-      "grad_norm": 0.6621217131614685,
-      "learning_rate": 0.0004562849433351322,
-      "loss": 3.527,
+      "grad_norm": 0.6139578819274902,
+      "learning_rate": 0.0004562784673502428,
+      "loss": 3.5305,
       "step": 22300
     },
     {
       "epoch": 2.4097035040431267,
-      "grad_norm": 0.5817117094993591,
-      "learning_rate": 0.0004559611440906638,
-      "loss": 3.55,
+      "grad_norm": 0.5361505150794983,
+      "learning_rate": 0.0004559546681057744,
+      "loss": 3.5553,
       "step": 22350
     },
     {
       "epoch": 2.4150943396226414,
-      "grad_norm": 0.5394627451896667,
-      "learning_rate": 0.00045563734484619533,
-      "loss": 3.5395,
+      "grad_norm": 0.5341053605079651,
+      "learning_rate": 0.000455630868861306,
+      "loss": 3.5441,
       "step": 22400
     },
     {
       "epoch": 2.420485175202156,
-      "grad_norm": 0.6029559969902039,
-      "learning_rate": 0.0004553135456017269,
-      "loss": 3.569,
+      "grad_norm": 0.5928043723106384,
+      "learning_rate": 0.00045530706961683753,
+      "loss": 3.5724,
       "step": 22450
     },
     {
       "epoch": 2.4258760107816713,
-      "grad_norm": 0.6211172342300415,
-      "learning_rate": 0.0004549897463572585,
-      "loss": 3.5407,
+      "grad_norm": 0.5863630175590515,
+      "learning_rate": 0.00045498327037236914,
+      "loss": 3.5449,
       "step": 22500
     },
     {
       "epoch": 2.431266846361186,
-      "grad_norm": 0.5414947867393494,
-      "learning_rate": 0.00045466594711279,
-      "loss": 3.5477,
+      "grad_norm": 0.5474246740341187,
+      "learning_rate": 0.0004546594711279007,
+      "loss": 3.5508,
       "step": 22550
     },
     {
       "epoch": 2.4366576819407006,
-      "grad_norm": 0.5086063742637634,
-      "learning_rate": 0.00045434214786832164,
-      "loss": 3.5477,
+      "grad_norm": 0.5145190954208374,
+      "learning_rate": 0.0004543356718834322,
+      "loss": 3.5503,
       "step": 22600
     },
     {
       "epoch": 2.442048517520216,
-      "grad_norm": 0.5853644609451294,
-      "learning_rate": 0.00045401834862385314,
-      "loss": 3.5445,
+      "grad_norm": 0.5755578875541687,
+      "learning_rate": 0.0004540118726389638,
+      "loss": 3.5489,
       "step": 22650
     },
     {
       "epoch": 2.4474393530997305,
-      "grad_norm": 0.5894696116447449,
-      "learning_rate": 0.00045369454937938474,
-      "loss": 3.5327,
+      "grad_norm": 0.5219981074333191,
+      "learning_rate": 0.00045368807339449534,
+      "loss": 3.5366,
       "step": 22700
     },
     {
       "epoch": 2.452830188679245,
-      "grad_norm": 0.5951809883117676,
-      "learning_rate": 0.0004533707501349163,
-      "loss": 3.5572,
+      "grad_norm": 0.5541500449180603,
+      "learning_rate": 0.00045336427415002694,
+      "loss": 3.5612,
       "step": 22750
     },
     {
       "epoch": 2.4582210242587603,
-      "grad_norm": 0.5352622270584106,
-      "learning_rate": 0.0004530469508904479,
-      "loss": 3.5671,
+      "grad_norm": 0.5818925499916077,
+      "learning_rate": 0.0004530404749055585,
+      "loss": 3.5717,
       "step": 22800
     },
     {
       "epoch": 2.463611859838275,
-      "grad_norm": 0.6040215492248535,
-      "learning_rate": 0.00045272315164597945,
-      "loss": 3.5349,
+      "grad_norm": 0.5783217549324036,
+      "learning_rate": 0.0004527166756610901,
+      "loss": 3.5369,
       "step": 22850
     },
     {
       "epoch": 2.4690026954177897,
-      "grad_norm": 0.573526918888092,
-      "learning_rate": 0.000452399352401511,
-      "loss": 3.5441,
+      "grad_norm": 0.5465967059135437,
+      "learning_rate": 0.00045239287641662165,
+      "loss": 3.5477,
       "step": 22900
     },
     {
       "epoch": 2.4743935309973044,
-      "grad_norm": 0.5855453610420227,
-      "learning_rate": 0.0004520755531570426,
-      "loss": 3.5475,
+      "grad_norm": 0.5679962635040283,
+      "learning_rate": 0.0004520690771721532,
+      "loss": 3.5515,
       "step": 22950
     },
     {
       "epoch": 2.4797843665768196,
-      "grad_norm": 0.5997734665870667,
-      "learning_rate": 0.00045175175391257415,
-      "loss": 3.5392,
+      "grad_norm": 0.5812360048294067,
+      "learning_rate": 0.0004517452779276848,
+      "loss": 3.5424,
       "step": 23000
     },
     {
       "epoch": 2.4797843665768196,
-      "eval_accuracy": 0.36637493004112187,
-      "eval_loss": 3.5344290733337402,
-      "eval_runtime": 183.3961,
-      "eval_samples_per_second": 98.208,
-      "eval_steps_per_second": 6.14,
+      "eval_accuracy": 0.36583373000691793,
+      "eval_loss": 3.536715269088745,
+      "eval_runtime": 183.4711,
+      "eval_samples_per_second": 98.168,
+      "eval_steps_per_second": 6.137,
       "step": 23000
     },
     {
       "epoch": 2.4851752021563343,
-      "grad_norm": 0.5770864486694336,
-      "learning_rate": 0.00045142795466810576,
-      "loss": 3.5379,
+      "grad_norm": 0.5592971444129944,
+      "learning_rate": 0.00045142147868321636,
+      "loss": 3.5427,
       "step": 23050
     },
     {
       "epoch": 2.490566037735849,
-      "grad_norm": 0.5228535532951355,
-      "learning_rate": 0.0004511041554236373,
-      "loss": 3.5355,
+      "grad_norm": 0.5132983326911926,
+      "learning_rate": 0.00045109767943874796,
+      "loss": 3.5384,
       "step": 23100
     },
     {
       "epoch": 2.4959568733153636,
-      "grad_norm": 0.6147720217704773,
-      "learning_rate": 0.0004507803561791689,
-      "loss": 3.5329,
+      "grad_norm": 0.5916054844856262,
+      "learning_rate": 0.0004507738801942795,
+      "loss": 3.5349,
       "step": 23150
     },
     {
       "epoch": 2.501347708894879,
-      "grad_norm": 0.5598917007446289,
-      "learning_rate": 0.00045045655693470046,
-      "loss": 3.5474,
+      "grad_norm": 0.5748770833015442,
+      "learning_rate": 0.0004504500809498111,
+      "loss": 3.55,
       "step": 23200
     },
     {
       "epoch": 2.5067385444743935,
-      "grad_norm": 0.545137345790863,
-      "learning_rate": 0.00045013275769023207,
-      "loss": 3.5582,
+      "grad_norm": 0.5406561493873596,
+      "learning_rate": 0.0004501262817053426,
+      "loss": 3.5629,
       "step": 23250
     },
     {
       "epoch": 2.512129380053908,
-      "grad_norm": 0.5641994476318359,
-      "learning_rate": 0.00044980895844576356,
-      "loss": 3.5553,
+      "grad_norm": 0.5765764713287354,
+      "learning_rate": 0.00044980248246087427,
+      "loss": 3.559,
       "step": 23300
     },
     {
       "epoch": 2.5175202156334233,
-      "grad_norm": 0.5789488554000854,
-      "learning_rate": 0.0004494851592012951,
-      "loss": 3.5306,
+      "grad_norm": 0.5578004717826843,
+      "learning_rate": 0.00044947868321640577,
+      "loss": 3.5343,
       "step": 23350
     },
     {
       "epoch": 2.522911051212938,
-      "grad_norm": 0.5591729283332825,
-      "learning_rate": 0.0004491613599568267,
-      "loss": 3.5349,
+      "grad_norm": 0.5645178556442261,
+      "learning_rate": 0.0004491548839719373,
+      "loss": 3.5383,
       "step": 23400
     },
     {
       "epoch": 2.5283018867924527,
-      "grad_norm": 0.5694953799247742,
-      "learning_rate": 0.00044883756071235827,
-      "loss": 3.539,
+      "grad_norm": 0.564024806022644,
+      "learning_rate": 0.0004488310847274689,
+      "loss": 3.5424,
       "step": 23450
     },
     {
       "epoch": 2.533692722371968,
-      "grad_norm": 0.5788413286209106,
-      "learning_rate": 0.0004485137614678899,
-      "loss": 3.5186,
+      "grad_norm": 0.52243572473526,
+      "learning_rate": 0.0004485072854830005,
+      "loss": 3.5221,
       "step": 23500
     },
     {
       "epoch": 2.5390835579514826,
-      "grad_norm": 0.5994324088096619,
-      "learning_rate": 0.0004481899622234214,
-      "loss": 3.5555,
+      "grad_norm": 0.5709099173545837,
+      "learning_rate": 0.0004481834862385321,
+      "loss": 3.5575,
       "step": 23550
     },
     {
       "epoch": 2.5444743935309972,
-      "grad_norm": 0.6380375623703003,
-      "learning_rate": 0.00044786616297895303,
-      "loss": 3.5488,
+      "grad_norm": 0.6139874458312988,
+      "learning_rate": 0.00044785968699406363,
+      "loss": 3.5509,
       "step": 23600
     },
     {
       "epoch": 2.5498652291105124,
-      "grad_norm": 0.5977026224136353,
-      "learning_rate": 0.0004475423637344846,
-      "loss": 3.5424,
+      "grad_norm": 0.5239370465278625,
+      "learning_rate": 0.00044753588774959523,
+      "loss": 3.5443,
       "step": 23650
     },
     {
       "epoch": 2.555256064690027,
-      "grad_norm": 0.5507897734642029,
-      "learning_rate": 0.00044721856449001613,
-      "loss": 3.5557,
+      "grad_norm": 0.5880962014198303,
+      "learning_rate": 0.0004472120885051268,
+      "loss": 3.5583,
       "step": 23700
     },
     {
       "epoch": 2.560646900269542,
-      "grad_norm": 0.5803244709968567,
-      "learning_rate": 0.00044689476524554774,
-      "loss": 3.5641,
+      "grad_norm": 0.5482958555221558,
+      "learning_rate": 0.0004468882892606584,
+      "loss": 3.5666,
       "step": 23750
     },
     {
       "epoch": 2.5660377358490565,
-      "grad_norm": 0.609472930431366,
-      "learning_rate": 0.0004465709660010793,
-      "loss": 3.5369,
+      "grad_norm": 0.5640934705734253,
+      "learning_rate": 0.00044656449001618994,
+      "loss": 3.5412,
       "step": 23800
     },
     {
       "epoch": 2.571428571428571,
-      "grad_norm": 0.5822898745536804,
-      "learning_rate": 0.0004462471667566109,
-      "loss": 3.5363,
+      "grad_norm": 0.5506890416145325,
+      "learning_rate": 0.0004462406907717215,
+      "loss": 3.5398,
       "step": 23850
     },
     {
       "epoch": 2.5768194070080863,
-      "grad_norm": 0.5829198360443115,
-      "learning_rate": 0.00044592336751214244,
-      "loss": 3.534,
+      "grad_norm": 0.5650559067726135,
+      "learning_rate": 0.0004459168915272531,
+      "loss": 3.5374,
       "step": 23900
     },
     {
       "epoch": 2.582210242587601,
-      "grad_norm": 0.5264182686805725,
-      "learning_rate": 0.00044560604425256334,
-      "loss": 3.5611,
+      "grad_norm": 0.513624370098114,
+      "learning_rate": 0.0004455930922827846,
+      "loss": 3.5652,
       "step": 23950
     },
     {
       "epoch": 2.5876010781671157,
-      "grad_norm": 0.5896156430244446,
-      "learning_rate": 0.000445282245008095,
-      "loss": 3.541,
+      "grad_norm": 0.5808485150337219,
+      "learning_rate": 0.0004452692930383162,
+      "loss": 3.5433,
       "step": 24000
     },
     {
       "epoch": 2.5876010781671157,
-      "eval_accuracy": 0.3671398463713358,
-      "eval_loss": 3.5273725986480713,
-      "eval_runtime": 183.5998,
-      "eval_samples_per_second": 98.099,
+      "eval_accuracy": 0.36707422002709583,
+      "eval_loss": 3.5275959968566895,
+      "eval_runtime": 183.5894,
+      "eval_samples_per_second": 98.105,
       "eval_steps_per_second": 6.133,
       "step": 24000
     },
     {
       "epoch": 2.592991913746631,
-      "grad_norm": 0.6344112157821655,
-      "learning_rate": 0.0004449584457636265,
-      "loss": 3.5453,
+      "grad_norm": 0.6006309390068054,
+      "learning_rate": 0.00044494549379384775,
+      "loss": 3.5497,
       "step": 24050
     },
     {
       "epoch": 2.5983827493261455,
-      "grad_norm": 0.6354188919067383,
-      "learning_rate": 0.00044463464651915805,
-      "loss": 3.5216,
+      "grad_norm": 0.5975236892700195,
+      "learning_rate": 0.00044462169454937935,
+      "loss": 3.5243,
       "step": 24100
     },
     {
       "epoch": 2.6037735849056602,
-      "grad_norm": 0.5827237963676453,
-      "learning_rate": 0.00044431084727468965,
-      "loss": 3.5458,
+      "grad_norm": 0.5262132883071899,
+      "learning_rate": 0.0004442978953049109,
+      "loss": 3.5503,
       "step": 24150
     },
     {
       "epoch": 2.6091644204851754,
-      "grad_norm": 0.5500607490539551,
-      "learning_rate": 0.0004439870480302212,
-      "loss": 3.5379,
+      "grad_norm": 0.5883579850196838,
+      "learning_rate": 0.0004439740960604425,
+      "loss": 3.5416,
       "step": 24200
     },
     {
       "epoch": 2.61455525606469,
-      "grad_norm": 0.5530035495758057,
-      "learning_rate": 0.0004436632487857528,
-      "loss": 3.5422,
+      "grad_norm": 0.5381089448928833,
+      "learning_rate": 0.00044365029681597406,
+      "loss": 3.5442,
       "step": 24250
     },
     {
       "epoch": 2.6199460916442048,
-      "grad_norm": 0.585949182510376,
-      "learning_rate": 0.00044333944954128436,
-      "loss": 3.5475,
+      "grad_norm": 0.5410563945770264,
+      "learning_rate": 0.0004433264975715056,
+      "loss": 3.5508,
       "step": 24300
     },
     {
       "epoch": 2.62533692722372,
-      "grad_norm": 0.6067348718643188,
-      "learning_rate": 0.00044301565029681596,
-      "loss": 3.5488,
+      "grad_norm": 0.5661271810531616,
+      "learning_rate": 0.0004430026983270372,
+      "loss": 3.5549,
       "step": 24350
     },
     {
       "epoch": 2.6307277628032346,
-      "grad_norm": 0.5916098356246948,
-      "learning_rate": 0.0004426918510523475,
-      "loss": 3.5251,
+      "grad_norm": 0.5446505546569824,
+      "learning_rate": 0.00044267889908256876,
+      "loss": 3.5281,
       "step": 24400
     },
     {
       "epoch": 2.6361185983827493,
-      "grad_norm": 0.5653256177902222,
-      "learning_rate": 0.0004423680518078791,
-      "loss": 3.5518,
+      "grad_norm": 0.5491141676902771,
+      "learning_rate": 0.00044235509983810037,
+      "loss": 3.5542,
       "step": 24450
     },
     {
       "epoch": 2.641509433962264,
-      "grad_norm": 0.6101019978523254,
-      "learning_rate": 0.00044204425256341067,
-      "loss": 3.5162,
+      "grad_norm": 0.5471525192260742,
+      "learning_rate": 0.0004420313005936319,
+      "loss": 3.5195,
       "step": 24500
     },
     {
       "epoch": 2.6469002695417787,
-      "grad_norm": 0.5783448815345764,
-      "learning_rate": 0.0004417204533189422,
-      "loss": 3.5196,
+      "grad_norm": 0.570773720741272,
+      "learning_rate": 0.0004417075013491635,
+      "loss": 3.5276,
       "step": 24550
     },
     {
       "epoch": 2.652291105121294,
-      "grad_norm": 0.5608351230621338,
-      "learning_rate": 0.0004413966540744738,
-      "loss": 3.5275,
+      "grad_norm": 0.5534959435462952,
+      "learning_rate": 0.0004413837021046951,
+      "loss": 3.5323,
       "step": 24600
     },
     {
       "epoch": 2.6576819407008085,
-      "grad_norm": 0.533299446105957,
-      "learning_rate": 0.0004410728548300053,
-      "loss": 3.5358,
+      "grad_norm": 0.5329326391220093,
+      "learning_rate": 0.00044105990286022657,
+      "loss": 3.5385,
       "step": 24650
     },
     {
       "epoch": 2.6630727762803232,
-      "grad_norm": 0.5338042378425598,
-      "learning_rate": 0.0004407490555855369,
-      "loss": 3.5365,
+      "grad_norm": 0.5262351036071777,
+      "learning_rate": 0.0004407425796006475,
+      "loss": 3.5407,
       "step": 24700
     },
     {
       "epoch": 2.6684636118598384,
-      "grad_norm": 0.5940648913383484,
-      "learning_rate": 0.0004404252563410685,
-      "loss": 3.5506,
+      "grad_norm": 0.5490606427192688,
+      "learning_rate": 0.00044041878035617913,
+      "loss": 3.5535,
       "step": 24750
     },
     {
       "epoch": 2.673854447439353,
-      "grad_norm": 0.5929331183433533,
-      "learning_rate": 0.0004401014570966001,
-      "loss": 3.5268,
+      "grad_norm": 0.5724402070045471,
+      "learning_rate": 0.0004400949811117107,
+      "loss": 3.5306,
       "step": 24800
     },
     {
       "epoch": 2.6792452830188678,
-      "grad_norm": 0.5341362953186035,
-      "learning_rate": 0.00043977765785213163,
-      "loss": 3.5281,
+      "grad_norm": 0.5811025500297546,
+      "learning_rate": 0.0004397711818672423,
+      "loss": 3.532,
       "step": 24850
     },
     {
       "epoch": 2.684636118598383,
-      "grad_norm": 0.5370336174964905,
-      "learning_rate": 0.0004394538586076632,
-      "loss": 3.5383,
+      "grad_norm": 0.5541442036628723,
+      "learning_rate": 0.00043944738262277383,
+      "loss": 3.5395,
       "step": 24900
     },
     {
       "epoch": 2.6900269541778976,
-      "grad_norm": 0.5800198316574097,
-      "learning_rate": 0.0004391300593631948,
-      "loss": 3.5321,
+      "grad_norm": 0.5251066088676453,
+      "learning_rate": 0.00043912358337830544,
+      "loss": 3.536,
       "step": 24950
     },
     {
       "epoch": 2.6954177897574123,
-      "grad_norm": 0.5767235159873962,
-      "learning_rate": 0.00043880626011872634,
-      "loss": 3.5477,
+      "grad_norm": 0.6008343696594238,
+      "learning_rate": 0.000438799784133837,
+      "loss": 3.5502,
       "step": 25000
     },
     {
       "epoch": 2.6954177897574123,
-      "eval_accuracy": 0.36847790668477115,
-      "eval_loss": 3.51676607131958,
-      "eval_runtime": 183.255,
-      "eval_samples_per_second": 98.284,
-      "eval_steps_per_second": 6.144,
+      "eval_accuracy": 0.36807730348739914,
+      "eval_loss": 3.5190443992614746,
+      "eval_runtime": 183.1039,
+      "eval_samples_per_second": 98.365,
+      "eval_steps_per_second": 6.15,
       "step": 25000
     },
     {
       "epoch": 2.7008086253369274,
-      "grad_norm": 0.6077183485031128,
-      "learning_rate": 0.00043848246087425794,
-      "loss": 3.5476,
+      "grad_norm": 0.605866551399231,
+      "learning_rate": 0.00043847598488936854,
+      "loss": 3.5493,
       "step": 25050
     },
     {
       "epoch": 2.706199460916442,
-      "grad_norm": 0.5808961391448975,
-      "learning_rate": 0.0004381586616297895,
-      "loss": 3.5578,
+      "grad_norm": 0.5946877598762512,
+      "learning_rate": 0.00043815218564490014,
+      "loss": 3.5611,
       "step": 25100
     },
     {
       "epoch": 2.711590296495957,
-      "grad_norm": 0.6259551048278809,
-      "learning_rate": 0.0004378348623853211,
-      "loss": 3.5405,
+      "grad_norm": 0.5921048521995544,
+      "learning_rate": 0.0004378283864004317,
+      "loss": 3.5425,
       "step": 25150
     },
     {
       "epoch": 2.7169811320754715,
-      "grad_norm": 0.6004087328910828,
-      "learning_rate": 0.00043751106314085265,
-      "loss": 3.5239,
+      "grad_norm": 0.5856859683990479,
+      "learning_rate": 0.0004375045871559633,
+      "loss": 3.5258,
       "step": 25200
     },
     {
       "epoch": 2.7223719676549867,
-      "grad_norm": 0.5496945381164551,
-      "learning_rate": 0.00043718726389638425,
-      "loss": 3.5258,
+      "grad_norm": 0.5377466678619385,
+      "learning_rate": 0.00043718078791149485,
+      "loss": 3.5286,
       "step": 25250
     },
     {
       "epoch": 2.7277628032345014,
-      "grad_norm": 0.6094425916671753,
-      "learning_rate": 0.00043686346465191575,
-      "loss": 3.536,
+      "grad_norm": 0.5315113663673401,
+      "learning_rate": 0.00043685698866702645,
+      "loss": 3.5376,
       "step": 25300
     },
     {
       "epoch": 2.733153638814016,
-      "grad_norm": 0.5401579737663269,
-      "learning_rate": 0.0004365396654074473,
-      "loss": 3.5275,
+      "grad_norm": 0.5452070832252502,
+      "learning_rate": 0.00043653318942255795,
+      "loss": 3.5301,
       "step": 25350
     },
     {
       "epoch": 2.7385444743935308,
-      "grad_norm": 0.5293307304382324,
-      "learning_rate": 0.00043622234214786825,
-      "loss": 3.5194,
+      "grad_norm": 0.5450847744941711,
+      "learning_rate": 0.00043620939017808956,
+      "loss": 3.5206,
       "step": 25400
     },
     {
       "epoch": 2.743935309973046,
-      "grad_norm": 0.5775792598724365,
-      "learning_rate": 0.00043589854290339985,
-      "loss": 3.535,
+      "grad_norm": 0.543219804763794,
+      "learning_rate": 0.0004358855909336211,
+      "loss": 3.5375,
       "step": 25450
     },
     {
       "epoch": 2.7493261455525606,
-      "grad_norm": 0.5909881591796875,
-      "learning_rate": 0.0004355747436589314,
-      "loss": 3.5507,
+      "grad_norm": 0.5815766453742981,
+      "learning_rate": 0.00043556179168915266,
+      "loss": 3.5553,
       "step": 25500
     },
     {
       "epoch": 2.7547169811320753,
-      "grad_norm": 0.5832464098930359,
-      "learning_rate": 0.000435250944414463,
-      "loss": 3.5444,
+      "grad_norm": 0.6215568780899048,
+      "learning_rate": 0.00043523799244468426,
+      "loss": 3.5472,
       "step": 25550
     },
     {
       "epoch": 2.7601078167115904,
-      "grad_norm": 0.5698075294494629,
-      "learning_rate": 0.00043492714516999456,
-      "loss": 3.535,
+      "grad_norm": 0.581833004951477,
+      "learning_rate": 0.0004349141932002158,
+      "loss": 3.5394,
       "step": 25600
     },
     {
       "epoch": 2.765498652291105,
-      "grad_norm": 0.5500668883323669,
-      "learning_rate": 0.0004346033459255261,
-      "loss": 3.5232,
+      "grad_norm": 0.5784794688224792,
+      "learning_rate": 0.0004345903939557474,
+      "loss": 3.5277,
       "step": 25650
     },
     {
       "epoch": 2.77088948787062,
-      "grad_norm": 0.5673616528511047,
-      "learning_rate": 0.0004342795466810577,
-      "loss": 3.5455,
+      "grad_norm": 0.562318742275238,
+      "learning_rate": 0.00043426659471127897,
+      "loss": 3.5494,
       "step": 25700
     },
     {
       "epoch": 2.776280323450135,
-      "grad_norm": 0.5733495354652405,
-      "learning_rate": 0.00043395574743658927,
-      "loss": 3.5295,
+      "grad_norm": 0.5342723727226257,
+      "learning_rate": 0.00043394279546681057,
+      "loss": 3.5321,
       "step": 25750
     },
     {
       "epoch": 2.7816711590296497,
-      "grad_norm": 0.5701019763946533,
-      "learning_rate": 0.00043363194819212087,
-      "loss": 3.5307,
+      "grad_norm": 0.5558409690856934,
+      "learning_rate": 0.0004336189962223421,
+      "loss": 3.5344,
       "step": 25800
     },
     {
       "epoch": 2.7870619946091644,
-      "grad_norm": 0.5702261924743652,
-      "learning_rate": 0.0004333081489476524,
-      "loss": 3.516,
+      "grad_norm": 0.5549172759056091,
+      "learning_rate": 0.0004332951969778737,
+      "loss": 3.5193,
       "step": 25850
     },
     {
       "epoch": 2.7924528301886795,
-      "grad_norm": 0.5822378396987915,
-      "learning_rate": 0.000432984349703184,
-      "loss": 3.5372,
+      "grad_norm": 0.558799147605896,
+      "learning_rate": 0.0004329713977334053,
+      "loss": 3.5399,
       "step": 25900
     },
     {
       "epoch": 2.797843665768194,
-      "grad_norm": 0.550873875617981,
-      "learning_rate": 0.0004326605504587155,
-      "loss": 3.5085,
+      "grad_norm": 0.5150911808013916,
+      "learning_rate": 0.0004326475984889368,
+      "loss": 3.5109,
       "step": 25950
     },
     {
       "epoch": 2.803234501347709,
-      "grad_norm": 0.5707508325576782,
-      "learning_rate": 0.0004323367512142472,
-      "loss": 3.5275,
+      "grad_norm": 0.5915558934211731,
+      "learning_rate": 0.0004323237992444684,
+      "loss": 3.5318,
       "step": 26000
     },
     {
       "epoch": 2.803234501347709,
-      "eval_accuracy": 0.36893414016070414,
-      "eval_loss": 3.510256290435791,
-      "eval_runtime": 183.5418,
-      "eval_samples_per_second": 98.13,
-      "eval_steps_per_second": 6.135,
+      "eval_accuracy": 0.36879256544788186,
+      "eval_loss": 3.5108420848846436,
+      "eval_runtime": 183.4637,
+      "eval_samples_per_second": 98.172,
+      "eval_steps_per_second": 6.137,
       "step": 26000
     },
     {
       "epoch": 2.8086253369272236,
-      "grad_norm": 0.5530628561973572,
-      "learning_rate": 0.0004320129519697787,
-      "loss": 3.5324,
+      "grad_norm": 0.5449238419532776,
+      "learning_rate": 0.00043199999999999993,
+      "loss": 3.535,
       "step": 26050
     },
     {
       "epoch": 2.8140161725067383,
-      "grad_norm": 0.536308228969574,
-      "learning_rate": 0.00043168915272531023,
-      "loss": 3.5192,
+      "grad_norm": 0.4974726140499115,
+      "learning_rate": 0.00043167620075553153,
+      "loss": 3.524,
       "step": 26100
     },
     {
       "epoch": 2.8194070080862534,
-      "grad_norm": 0.5508840680122375,
-      "learning_rate": 0.00043136535348084183,
-      "loss": 3.5308,
+      "grad_norm": 0.577414870262146,
+      "learning_rate": 0.0004313524015110631,
+      "loss": 3.5314,
       "step": 26150
     },
     {
       "epoch": 2.824797843665768,
-      "grad_norm": 0.661864697933197,
-      "learning_rate": 0.0004310415542363734,
-      "loss": 3.5415,
+      "grad_norm": 0.64058917760849,
+      "learning_rate": 0.0004310286022665947,
+      "loss": 3.5444,
       "step": 26200
     },
     {
       "epoch": 2.830188679245283,
-      "grad_norm": 0.5606989860534668,
-      "learning_rate": 0.000430717754991905,
-      "loss": 3.5281,
+      "grad_norm": 0.5493928790092468,
+      "learning_rate": 0.00043070480302212624,
+      "loss": 3.532,
       "step": 26250
     },
     {
       "epoch": 2.835579514824798,
-      "grad_norm": 0.5549330711364746,
-      "learning_rate": 0.00043039395574743654,
-      "loss": 3.5263,
+      "grad_norm": 0.5664352774620056,
+      "learning_rate": 0.0004303810037776578,
+      "loss": 3.527,
       "step": 26300
     },
     {
       "epoch": 2.8409703504043127,
-      "grad_norm": 0.5238580107688904,
-      "learning_rate": 0.00043007015650296814,
-      "loss": 3.5394,
+      "grad_norm": 0.5029638409614563,
+      "learning_rate": 0.0004300572045331894,
+      "loss": 3.5431,
       "step": 26350
     },
     {
       "epoch": 2.8463611859838274,
-      "grad_norm": 0.5641552805900574,
-      "learning_rate": 0.0004297463572584997,
-      "loss": 3.5218,
+      "grad_norm": 0.5343481302261353,
+      "learning_rate": 0.00042973340528872095,
+      "loss": 3.5254,
       "step": 26400
     },
     {
       "epoch": 2.8517520215633425,
-      "grad_norm": 0.5834696888923645,
-      "learning_rate": 0.0004294225580140313,
-      "loss": 3.5309,
+      "grad_norm": 0.5601471662521362,
+      "learning_rate": 0.00042940960604425255,
+      "loss": 3.5343,
       "step": 26450
     },
     {
       "epoch": 2.857142857142857,
-      "grad_norm": 0.6112784147262573,
-      "learning_rate": 0.00042909875876956285,
-      "loss": 3.52,
+      "grad_norm": 0.5483253598213196,
+      "learning_rate": 0.0004290858067997841,
+      "loss": 3.5214,
       "step": 26500
     },
     {
       "epoch": 2.862533692722372,
-      "grad_norm": 0.5841647386550903,
-      "learning_rate": 0.0004287749595250944,
-      "loss": 3.5032,
+      "grad_norm": 0.5550464987754822,
+      "learning_rate": 0.0004287620075553157,
+      "loss": 3.506,
       "step": 26550
     },
     {
       "epoch": 2.867924528301887,
-      "grad_norm": 0.5762807726860046,
-      "learning_rate": 0.000428451160280626,
-      "loss": 3.5271,
+      "grad_norm": 0.5305153131484985,
+      "learning_rate": 0.00042843820831084726,
+      "loss": 3.5299,
       "step": 26600
     },
     {
       "epoch": 2.8733153638814017,
-      "grad_norm": 0.5195743441581726,
-      "learning_rate": 0.0004281273610361575,
-      "loss": 3.5221,
+      "grad_norm": 0.5094022750854492,
+      "learning_rate": 0.00042811440906637886,
+      "loss": 3.5231,
       "step": 26650
     },
     {
       "epoch": 2.8787061994609164,
-      "grad_norm": 0.5398359298706055,
-      "learning_rate": 0.0004278035617916891,
-      "loss": 3.5163,
+      "grad_norm": 0.5317772626876831,
+      "learning_rate": 0.00042779060982191036,
+      "loss": 3.5191,
       "step": 26700
     },
     {
       "epoch": 2.884097035040431,
-      "grad_norm": 0.5633315443992615,
-      "learning_rate": 0.00042747976254722066,
-      "loss": 3.5195,
+      "grad_norm": 0.5471376180648804,
+      "learning_rate": 0.0004274668105774419,
+      "loss": 3.5251,
       "step": 26750
     },
     {
       "epoch": 2.889487870619946,
-      "grad_norm": 0.5828991532325745,
-      "learning_rate": 0.00042715596330275226,
-      "loss": 3.5012,
+      "grad_norm": 0.5802553296089172,
+      "learning_rate": 0.0004271430113329735,
+      "loss": 3.5045,
       "step": 26800
     },
     {
       "epoch": 2.894878706199461,
-      "grad_norm": 0.6273923516273499,
-      "learning_rate": 0.0004268321640582838,
-      "loss": 3.5202,
+      "grad_norm": 0.6292346119880676,
+      "learning_rate": 0.00042681921208850506,
+      "loss": 3.5226,
       "step": 26850
     },
     {
       "epoch": 2.9002695417789757,
-      "grad_norm": 0.5381441712379456,
-      "learning_rate": 0.0004265083648138154,
-      "loss": 3.5087,
+      "grad_norm": 0.513176441192627,
+      "learning_rate": 0.00042649541284403667,
+      "loss": 3.5104,
       "step": 26900
     },
     {
       "epoch": 2.9056603773584904,
-      "grad_norm": 0.596377432346344,
-      "learning_rate": 0.00042618456556934697,
-      "loss": 3.5073,
+      "grad_norm": 0.5755507349967957,
+      "learning_rate": 0.0004261780895844576,
+      "loss": 3.5097,
       "step": 26950
     },
     {
       "epoch": 2.9110512129380055,
-      "grad_norm": 0.5934976935386658,
-      "learning_rate": 0.0004258607663248785,
-      "loss": 3.5341,
+      "grad_norm": 0.5716110467910767,
+      "learning_rate": 0.00042585429033998917,
+      "loss": 3.5367,
       "step": 27000
     },
     {
       "epoch": 2.9110512129380055,
-      "eval_accuracy": 0.3700002422959398,
-      "eval_loss": 3.500957727432251,
-      "eval_runtime": 183.1714,
-      "eval_samples_per_second": 98.329,
-      "eval_steps_per_second": 6.147,
+      "eval_accuracy": 0.3698658386737133,
+      "eval_loss": 3.5031847953796387,
+      "eval_runtime": 181.4003,
+      "eval_samples_per_second": 99.289,
+      "eval_steps_per_second": 6.207,
       "step": 27000
     },
     {
       "epoch": 2.91644204851752,
-      "grad_norm": 0.6017852425575256,
-      "learning_rate": 0.0004255369670804101,
-      "loss": 3.5455,
+      "grad_norm": 0.6250123977661133,
+      "learning_rate": 0.0004255304910955207,
+      "loss": 3.5478,
       "step": 27050
     },
     {
       "epoch": 2.921832884097035,
-      "grad_norm": 0.5310950875282288,
-      "learning_rate": 0.0004252131678359417,
-      "loss": 3.5272,
+      "grad_norm": 0.509468674659729,
+      "learning_rate": 0.0004252066918510523,
+      "loss": 3.5291,
       "step": 27100
     },
     {
       "epoch": 2.92722371967655,
-      "grad_norm": 0.5664175152778625,
-      "learning_rate": 0.0004248893685914733,
-      "loss": 3.5418,
+      "grad_norm": 0.559503436088562,
+      "learning_rate": 0.0004248828926065839,
+      "loss": 3.5441,
       "step": 27150
     },
     {
       "epoch": 2.9326145552560647,
-      "grad_norm": 0.6421070694923401,
-      "learning_rate": 0.00042456556934700483,
-      "loss": 3.5271,
+      "grad_norm": 0.5554913282394409,
+      "learning_rate": 0.0004245590933621155,
+      "loss": 3.5302,
       "step": 27200
     },
     {
       "epoch": 2.9380053908355794,
-      "grad_norm": 0.5426092743873596,
-      "learning_rate": 0.00042424177010253643,
-      "loss": 3.542,
+      "grad_norm": 0.540884256362915,
+      "learning_rate": 0.00042423529411764703,
+      "loss": 3.5451,
       "step": 27250
     },
     {
       "epoch": 2.9433962264150946,
-      "grad_norm": 0.597486138343811,
-      "learning_rate": 0.00042391797085806793,
-      "loss": 3.5224,
+      "grad_norm": 0.5777525305747986,
+      "learning_rate": 0.00042391149487317864,
+      "loss": 3.5266,
       "step": 27300
     },
     {
       "epoch": 2.9487870619946093,
-      "grad_norm": 0.6039035320281982,
-      "learning_rate": 0.0004235941716135995,
-      "loss": 3.5114,
+      "grad_norm": 0.5945996046066284,
+      "learning_rate": 0.00042358769562871013,
+      "loss": 3.5155,
       "step": 27350
     },
     {
       "epoch": 2.954177897574124,
-      "grad_norm": 0.5943632125854492,
-      "learning_rate": 0.0004232703723691311,
-      "loss": 3.5101,
+      "grad_norm": 0.6042043566703796,
+      "learning_rate": 0.00042326389638424174,
+      "loss": 3.5129,
       "step": 27400
     },
     {
       "epoch": 2.9595687331536387,
-      "grad_norm": 0.5611184239387512,
-      "learning_rate": 0.00042294657312466264,
-      "loss": 3.527,
+      "grad_norm": 0.5184352397918701,
+      "learning_rate": 0.0004229400971397733,
+      "loss": 3.5285,
       "step": 27450
     },
     {
       "epoch": 2.964959568733154,
-      "grad_norm": 0.5667484402656555,
-      "learning_rate": 0.00042262277388019424,
-      "loss": 3.5303,
+      "grad_norm": 0.5442382097244263,
+      "learning_rate": 0.00042261629789530484,
+      "loss": 3.5308,
       "step": 27500
     },
     {
       "epoch": 2.9703504043126685,
-      "grad_norm": 0.5914982557296753,
-      "learning_rate": 0.0004222989746357258,
-      "loss": 3.5148,
+      "grad_norm": 0.5574375987052917,
+      "learning_rate": 0.00042229249865083644,
+      "loss": 3.5166,
       "step": 27550
     },
     {
       "epoch": 2.975741239892183,
-      "grad_norm": 0.5547810792922974,
-      "learning_rate": 0.0004219751753912574,
-      "loss": 3.5162,
+      "grad_norm": 0.5667176246643066,
+      "learning_rate": 0.000421968699406368,
+      "loss": 3.519,
       "step": 27600
     },
     {
       "epoch": 2.981132075471698,
-      "grad_norm": 0.5683316588401794,
-      "learning_rate": 0.00042165137614678895,
-      "loss": 3.5384,
+      "grad_norm": 0.5388574004173279,
+      "learning_rate": 0.0004216449001618996,
+      "loss": 3.5412,
       "step": 27650
     },
     {
       "epoch": 2.986522911051213,
-      "grad_norm": 0.5850192904472351,
-      "learning_rate": 0.00042132757690232055,
-      "loss": 3.5048,
+      "grad_norm": 0.5490278601646423,
+      "learning_rate": 0.00042132110091743115,
+      "loss": 3.5056,
       "step": 27700
     },
     {
       "epoch": 2.9919137466307277,
-      "grad_norm": 0.5472140908241272,
-      "learning_rate": 0.0004210037776578521,
-      "loss": 3.4964,
+      "grad_norm": 0.5674181580543518,
+      "learning_rate": 0.00042099730167296275,
+      "loss": 3.4982,
       "step": 27750
     },
     {
       "epoch": 2.9973045822102424,
-      "grad_norm": 0.6173052787780762,
-      "learning_rate": 0.00042067997841338365,
-      "loss": 3.5167,
+      "grad_norm": 0.621146023273468,
+      "learning_rate": 0.0004206735024284943,
+      "loss": 3.5196,
       "step": 27800
     },
     {
       "epoch": 3.0026954177897576,
-      "grad_norm": 0.5687317848205566,
-      "learning_rate": 0.00042035617916891526,
-      "loss": 3.4765,
+      "grad_norm": 0.5646306872367859,
+      "learning_rate": 0.0004203497031840259,
+      "loss": 3.4793,
       "step": 27850
     },
     {
       "epoch": 3.0080862533692723,
-      "grad_norm": 0.5316717624664307,
-      "learning_rate": 0.0004200323799244468,
-      "loss": 3.4207,
+      "grad_norm": 0.5279334783554077,
+      "learning_rate": 0.00042002590393955746,
+      "loss": 3.4244,
       "step": 27900
     },
     {
       "epoch": 3.013477088948787,
-      "grad_norm": 0.6342532634735107,
-      "learning_rate": 0.0004197085806799784,
-      "loss": 3.3998,
+      "grad_norm": 0.6317477226257324,
+      "learning_rate": 0.00041970210469508896,
+      "loss": 3.4038,
       "step": 27950
     },
     {
       "epoch": 3.018867924528302,
-      "grad_norm": 0.5721489787101746,
-      "learning_rate": 0.0004193847814355099,
-      "loss": 3.4218,
+      "grad_norm": 0.5460816621780396,
+      "learning_rate": 0.00041937830545062056,
+      "loss": 3.4248,
       "step": 28000
     },
     {
       "epoch": 3.018867924528302,
-      "eval_accuracy": 0.37088456814986187,
-      "eval_loss": 3.4964616298675537,
-      "eval_runtime": 183.4604,
-      "eval_samples_per_second": 98.174,
-      "eval_steps_per_second": 6.138,
+      "eval_accuracy": 0.3704118194349811,
+      "eval_loss": 3.498478412628174,
+      "eval_runtime": 181.3201,
+      "eval_samples_per_second": 99.333,
+      "eval_steps_per_second": 6.21,
       "step": 28000
     },
     {
       "epoch": 3.024258760107817,
-      "grad_norm": 0.5991702079772949,
-      "learning_rate": 0.0004190609821910415,
-      "loss": 3.4134,
+      "grad_norm": 0.5552032589912415,
+      "learning_rate": 0.0004190545062061521,
+      "loss": 3.4178,
       "step": 28050
     },
     {
       "epoch": 3.0296495956873315,
-      "grad_norm": 0.5457400679588318,
-      "learning_rate": 0.00041873718294657306,
-      "loss": 3.4297,
+      "grad_norm": 0.6345612406730652,
+      "learning_rate": 0.0004187307069616837,
+      "loss": 3.433,
       "step": 28100
     },
     {
       "epoch": 3.035040431266846,
-      "grad_norm": 0.5663076043128967,
-      "learning_rate": 0.00041841338370210467,
-      "loss": 3.431,
+      "grad_norm": 0.5480504631996155,
+      "learning_rate": 0.00041840690771721527,
+      "loss": 3.4346,
       "step": 28150
     },
     {
       "epoch": 3.0404312668463613,
-      "grad_norm": 0.6150392889976501,
-      "learning_rate": 0.0004180895844576362,
-      "loss": 3.4296,
+      "grad_norm": 0.6065757274627686,
+      "learning_rate": 0.00041808310847274687,
+      "loss": 3.4327,
       "step": 28200
     },
     {
       "epoch": 3.045822102425876,
-      "grad_norm": 0.5990293622016907,
-      "learning_rate": 0.00041776578521316777,
-      "loss": 3.435,
+      "grad_norm": 0.5909293293952942,
+      "learning_rate": 0.0004177593092282784,
+      "loss": 3.4369,
       "step": 28250
     },
     {
       "epoch": 3.0512129380053907,
-      "grad_norm": 0.5907416939735413,
-      "learning_rate": 0.0004174419859686994,
-      "loss": 3.4342,
+      "grad_norm": 0.6524730324745178,
+      "learning_rate": 0.00041743550998381,
+      "loss": 3.4363,
       "step": 28300
     },
     {
       "epoch": 3.056603773584906,
-      "grad_norm": 0.5918059945106506,
-      "learning_rate": 0.0004171181867242309,
-      "loss": 3.4428,
+      "grad_norm": 0.5694872736930847,
+      "learning_rate": 0.0004171117107393416,
+      "loss": 3.4449,
       "step": 28350
     },
     {
       "epoch": 3.0619946091644206,
-      "grad_norm": 0.6297253966331482,
-      "learning_rate": 0.00041679438747976253,
-      "loss": 3.4174,
+      "grad_norm": 0.576894998550415,
+      "learning_rate": 0.00041678791149487313,
+      "loss": 3.4225,
       "step": 28400
     },
     {
       "epoch": 3.0673854447439353,
-      "grad_norm": 0.6004771590232849,
-      "learning_rate": 0.0004164705882352941,
-      "loss": 3.4398,
+      "grad_norm": 0.6022789478302002,
+      "learning_rate": 0.00041646411225040473,
+      "loss": 3.4429,
       "step": 28450
     },
     {
       "epoch": 3.07277628032345,
-      "grad_norm": 0.6342185139656067,
-      "learning_rate": 0.0004161467889908257,
-      "loss": 3.4255,
+      "grad_norm": 0.5837662220001221,
+      "learning_rate": 0.0004161403130059363,
+      "loss": 3.4301,
       "step": 28500
     },
     {
       "epoch": 3.078167115902965,
-      "grad_norm": 0.5759187340736389,
-      "learning_rate": 0.00041582298974635724,
-      "loss": 3.4223,
+      "grad_norm": 0.6072590351104736,
+      "learning_rate": 0.0004158165137614679,
+      "loss": 3.425,
       "step": 28550
     },
     {
       "epoch": 3.08355795148248,
-      "grad_norm": 0.5629438757896423,
-      "learning_rate": 0.00041549919050188884,
-      "loss": 3.4434,
+      "grad_norm": 0.5637064576148987,
+      "learning_rate": 0.00041549271451699944,
+      "loss": 3.448,
       "step": 28600
     },
     {
       "epoch": 3.0889487870619945,
-      "grad_norm": 0.6492778658866882,
-      "learning_rate": 0.00041517539125742034,
-      "loss": 3.4449,
+      "grad_norm": 0.605414092540741,
+      "learning_rate": 0.00041516891527253104,
+      "loss": 3.4492,
       "step": 28650
     },
     {
       "epoch": 3.0943396226415096,
-      "grad_norm": 0.5692011713981628,
-      "learning_rate": 0.0004148515920129519,
-      "loss": 3.447,
+      "grad_norm": 0.5644150972366333,
+      "learning_rate": 0.00041484511602806254,
+      "loss": 3.4516,
       "step": 28700
     },
     {
       "epoch": 3.0997304582210243,
-      "grad_norm": 0.9911279678344727,
-      "learning_rate": 0.0004145277927684835,
-      "loss": 3.443,
+      "grad_norm": 0.5911921858787537,
+      "learning_rate": 0.0004145213167835941,
+      "loss": 3.444,
       "step": 28750
     },
     {
       "epoch": 3.105121293800539,
-      "grad_norm": 0.6052446365356445,
-      "learning_rate": 0.00041420399352401504,
-      "loss": 3.4396,
+      "grad_norm": 0.5797194838523865,
+      "learning_rate": 0.0004141975175391257,
+      "loss": 3.4434,
       "step": 28800
     },
     {
       "epoch": 3.1105121293800537,
-      "grad_norm": 0.5958963632583618,
-      "learning_rate": 0.00041388019427954665,
-      "loss": 3.4439,
+      "grad_norm": 0.5931008458137512,
+      "learning_rate": 0.00041387371829465725,
+      "loss": 3.4485,
       "step": 28850
     },
     {
       "epoch": 3.115902964959569,
-      "grad_norm": 0.5762249231338501,
-      "learning_rate": 0.0004135563950350782,
-      "loss": 3.4425,
+      "grad_norm": 0.5824803113937378,
+      "learning_rate": 0.00041354991905018885,
+      "loss": 3.4454,
       "step": 28900
     },
     {
       "epoch": 3.1212938005390836,
-      "grad_norm": 0.5552219152450562,
-      "learning_rate": 0.0004132325957906098,
-      "loss": 3.4359,
+      "grad_norm": 0.5844389200210571,
+      "learning_rate": 0.0004132261198057204,
+      "loss": 3.4383,
       "step": 28950
     },
     {
       "epoch": 3.1266846361185983,
-      "grad_norm": 0.59885573387146,
-      "learning_rate": 0.00041290879654614135,
-      "loss": 3.4461,
+      "grad_norm": 0.5910643935203552,
+      "learning_rate": 0.000412902320561252,
+      "loss": 3.4478,
       "step": 29000
     },
     {
       "epoch": 3.1266846361185983,
-      "eval_accuracy": 0.37161156462222855,
-      "eval_loss": 3.489034414291382,
-      "eval_runtime": 183.4689,
-      "eval_samples_per_second": 98.169,
-      "eval_steps_per_second": 6.137,
+      "eval_accuracy": 0.3713936069292727,
+      "eval_loss": 3.4914348125457764,
+      "eval_runtime": 181.3977,
+      "eval_samples_per_second": 99.29,
+      "eval_steps_per_second": 6.207,
       "step": 29000
     },
     {
       "epoch": 3.1320754716981134,
-      "grad_norm": 0.5445829033851624,
-      "learning_rate": 0.00041258499730167296,
-      "loss": 3.4258,
+      "grad_norm": 0.5493391752243042,
+      "learning_rate": 0.00041257852131678356,
+      "loss": 3.4271,
       "step": 29050
     },
     {
       "epoch": 3.137466307277628,
-      "grad_norm": 0.5851534008979797,
-      "learning_rate": 0.0004122611980572045,
-      "loss": 3.4514,
+      "grad_norm": 0.5826569199562073,
+      "learning_rate": 0.00041225472207231516,
+      "loss": 3.4537,
       "step": 29100
     },
     {
       "epoch": 3.142857142857143,
-      "grad_norm": 0.5788666605949402,
-      "learning_rate": 0.00041193739881273606,
-      "loss": 3.4466,
+      "grad_norm": 0.5545570254325867,
+      "learning_rate": 0.0004119309228278467,
+      "loss": 3.4485,
       "step": 29150
     },
     {
       "epoch": 3.1482479784366575,
-      "grad_norm": 0.6072415113449097,
-      "learning_rate": 0.00041161359956826766,
-      "loss": 3.4602,
+      "grad_norm": 0.5585206151008606,
+      "learning_rate": 0.00041160712358337826,
+      "loss": 3.464,
       "step": 29200
     },
     {
       "epoch": 3.1536388140161726,
-      "grad_norm": 0.6163262724876404,
+      "grad_norm": 0.5983799695968628,
       "learning_rate": 0.0004112898003237992,
-      "loss": 3.4614,
+      "loss": 3.4643,
       "step": 29250
     },
     {
       "epoch": 3.1590296495956873,
-      "grad_norm": 0.612508237361908,
+      "grad_norm": 0.6112026572227478,
       "learning_rate": 0.0004109660010793308,
-      "loss": 3.4541,
+      "loss": 3.4583,
       "step": 29300
     },
     {
       "epoch": 3.164420485175202,
-      "grad_norm": 0.5615676045417786,
+      "grad_norm": 0.5398700833320618,
       "learning_rate": 0.0004106422018348623,
-      "loss": 3.4416,
+      "loss": 3.4439,
       "step": 29350
     },
     {
       "epoch": 3.169811320754717,
-      "grad_norm": 0.6066969633102417,
+      "grad_norm": 0.6425313949584961,
       "learning_rate": 0.0004103184025903939,
-      "loss": 3.4516,
+      "loss": 3.4548,
       "step": 29400
     },
     {
       "epoch": 3.175202156334232,
-      "grad_norm": 0.603179395198822,
-      "learning_rate": 0.00040999460334592547,
-      "loss": 3.4643,
+      "grad_norm": 0.6146963238716125,
+      "learning_rate": 0.0004100010793308148,
+      "loss": 3.4678,
       "step": 29450
     },
     {
       "epoch": 3.1805929919137466,
-      "grad_norm": 0.5823822021484375,
-      "learning_rate": 0.000409670804101457,
-      "loss": 3.4381,
+      "grad_norm": 0.5854633450508118,
+      "learning_rate": 0.0004096772800863464,
+      "loss": 3.4382,
       "step": 29500
     },
     {
       "epoch": 3.1859838274932613,
-      "grad_norm": 0.5800785422325134,
+      "grad_norm": 0.5792513489723206,
       "learning_rate": 0.000409353480841878,
-      "loss": 3.4505,
+      "loss": 3.4524,
       "step": 29550
     },
     {
       "epoch": 3.1913746630727764,
-      "grad_norm": 0.5989530086517334,
+      "grad_norm": 0.5601498484611511,
       "learning_rate": 0.0004090296815974096,
-      "loss": 3.4613,
+      "loss": 3.4647,
       "step": 29600
     },
     {
       "epoch": 3.196765498652291,
-      "grad_norm": 0.5782071352005005,
+      "grad_norm": 0.5658991932868958,
       "learning_rate": 0.00040870588235294113,
-      "loss": 3.4469,
+      "loss": 3.448,
       "step": 29650
     },
     {
       "epoch": 3.202156334231806,
-      "grad_norm": 0.6211130023002625,
+      "grad_norm": 0.5900035500526428,
       "learning_rate": 0.00040838208310847273,
-      "loss": 3.4652,
+      "loss": 3.4684,
       "step": 29700
     },
     {
       "epoch": 3.207547169811321,
-      "grad_norm": 0.6505089998245239,
+      "grad_norm": 0.6197194457054138,
       "learning_rate": 0.0004080582838640043,
-      "loss": 3.4618,
+      "loss": 3.4643,
       "step": 29750
     },
     {
       "epoch": 3.2129380053908356,
-      "grad_norm": 0.676662027835846,
+      "grad_norm": 0.651862621307373,
       "learning_rate": 0.0004077344846195359,
-      "loss": 3.437,
+      "loss": 3.4407,
       "step": 29800
     },
     {
       "epoch": 3.2183288409703503,
-      "grad_norm": 0.6046519875526428,
+      "grad_norm": 0.6026725769042969,
       "learning_rate": 0.00040741068537506744,
-      "loss": 3.4592,
+      "loss": 3.4622,
       "step": 29850
     },
     {
       "epoch": 3.223719676549865,
-      "grad_norm": 0.6096304655075073,
+      "grad_norm": 0.5765792727470398,
       "learning_rate": 0.000407086886130599,
-      "loss": 3.4508,
+      "loss": 3.4538,
       "step": 29900
     },
     {
       "epoch": 3.22911051212938,
-      "grad_norm": 0.5660011172294617,
+      "grad_norm": 0.6115339398384094,
       "learning_rate": 0.0004067630868861306,
-      "loss": 3.4471,
+      "loss": 3.4504,
       "step": 29950
     },
     {
       "epoch": 3.234501347708895,
-      "grad_norm": 0.6080735921859741,
+      "grad_norm": 0.6550230979919434,
       "learning_rate": 0.0004064392876416621,
-      "loss": 3.4554,
+      "loss": 3.4584,
       "step": 30000
     },
     {
       "epoch": 3.234501347708895,
-      "eval_accuracy": 0.3721109332946239,
-      "eval_loss": 3.484154224395752,
-      "eval_runtime": 183.5489,
-      "eval_samples_per_second": 98.126,
-      "eval_steps_per_second": 6.135,
+      "eval_accuracy": 0.37159211575530976,
+      "eval_loss": 3.487363576889038,
+      "eval_runtime": 181.8349,
+      "eval_samples_per_second": 99.051,
+      "eval_steps_per_second": 6.192,
       "step": 30000
     },
     {
       "epoch": 3.2398921832884096,
-      "grad_norm": 0.5689899325370789,
+      "grad_norm": 0.5956474542617798,
       "learning_rate": 0.0004061154883971937,
-      "loss": 3.4345,
+      "loss": 3.437,
       "step": 30050
     },
     {
       "epoch": 3.2452830188679247,
-      "grad_norm": 0.6104323267936707,
+      "grad_norm": 0.779757022857666,
       "learning_rate": 0.00040579168915272525,
-      "loss": 3.451,
+      "loss": 3.4532,
       "step": 30100
     },
     {
       "epoch": 3.2506738544474394,
-      "grad_norm": 0.5824128985404968,
+      "grad_norm": 0.5633816123008728,
       "learning_rate": 0.00040546788990825685,
-      "loss": 3.4518,
+      "loss": 3.4564,
       "step": 30150
     },
     {
       "epoch": 3.256064690026954,
-      "grad_norm": 0.5787423849105835,
+      "grad_norm": 0.5868045091629028,
       "learning_rate": 0.0004051440906637884,
-      "loss": 3.4673,
+      "loss": 3.4693,
       "step": 30200
     },
     {
       "epoch": 3.2614555256064692,
-      "grad_norm": 0.5913103222846985,
+      "grad_norm": 0.6109861731529236,
       "learning_rate": 0.00040482029141931995,
-      "loss": 3.4537,
+      "loss": 3.4583,
       "step": 30250
     },
     {
       "epoch": 3.266846361185984,
-      "grad_norm": 0.6089011430740356,
+      "grad_norm": 0.6622506380081177,
       "learning_rate": 0.00040449649217485156,
-      "loss": 3.4319,
+      "loss": 3.4353,
       "step": 30300
     },
     {
       "epoch": 3.2722371967654986,
-      "grad_norm": 0.5752812623977661,
+      "grad_norm": 0.5871800184249878,
       "learning_rate": 0.0004041726929303831,
-      "loss": 3.4487,
+      "loss": 3.4535,
       "step": 30350
     },
     {
       "epoch": 3.2776280323450133,
-      "grad_norm": 0.6006338596343994,
+      "grad_norm": 0.5843096971511841,
       "learning_rate": 0.0004038488936859147,
-      "loss": 3.4607,
+      "loss": 3.4644,
       "step": 30400
     },
     {
       "epoch": 3.2830188679245285,
-      "grad_norm": 0.7376881241798401,
+      "grad_norm": 0.6638559103012085,
       "learning_rate": 0.00040352509444144626,
-      "loss": 3.45,
+      "loss": 3.452,
       "step": 30450
     },
     {
       "epoch": 3.288409703504043,
-      "grad_norm": 0.6478651165962219,
+      "grad_norm": 0.6473684906959534,
       "learning_rate": 0.00040320129519697787,
-      "loss": 3.4595,
+      "loss": 3.4645,
       "step": 30500
     },
     {
       "epoch": 3.293800539083558,
-      "grad_norm": 0.6603450775146484,
+      "grad_norm": 0.6173699498176575,
       "learning_rate": 0.0004028774959525094,
-      "loss": 3.4558,
+      "loss": 3.4601,
       "step": 30550
     },
     {
       "epoch": 3.2991913746630726,
-      "grad_norm": 0.5964590311050415,
+      "grad_norm": 0.603931188583374,
       "learning_rate": 0.000402553696708041,
-      "loss": 3.4487,
+      "loss": 3.4506,
       "step": 30600
     },
     {
       "epoch": 3.3045822102425877,
-      "grad_norm": 0.5935302376747131,
+      "grad_norm": 0.5841691493988037,
       "learning_rate": 0.0004022298974635726,
-      "loss": 3.4468,
+      "loss": 3.4499,
       "step": 30650
     },
     {
       "epoch": 3.3099730458221024,
-      "grad_norm": 0.6866962313652039,
+      "grad_norm": 0.6147717237472534,
       "learning_rate": 0.00040190609821910407,
-      "loss": 3.4606,
+      "loss": 3.4629,
       "step": 30700
     },
     {
       "epoch": 3.315363881401617,
-      "grad_norm": 0.5697274804115295,
+      "grad_norm": 0.5787293910980225,
       "learning_rate": 0.0004015822989746357,
-      "loss": 3.4381,
+      "loss": 3.4411,
       "step": 30750
     },
     {
       "epoch": 3.3207547169811322,
-      "grad_norm": 0.5480004549026489,
+      "grad_norm": 0.565758466720581,
       "learning_rate": 0.0004012584997301672,
-      "loss": 3.4342,
+      "loss": 3.4378,
       "step": 30800
     },
     {
       "epoch": 3.326145552560647,
-      "grad_norm": 0.591159999370575,
+      "grad_norm": 0.5544254183769226,
       "learning_rate": 0.00040093470048569883,
-      "loss": 3.4336,
+      "loss": 3.4379,
       "step": 30850
     },
     {
       "epoch": 3.3315363881401616,
-      "grad_norm": 0.564811646938324,
+      "grad_norm": 0.5848394632339478,
       "learning_rate": 0.0004006109012412304,
-      "loss": 3.4658,
+      "loss": 3.4697,
       "step": 30900
     },
     {
       "epoch": 3.3369272237196768,
-      "grad_norm": 0.6144384145736694,
+      "grad_norm": 0.5993809700012207,
       "learning_rate": 0.000400287101996762,
-      "loss": 3.4619,
+      "loss": 3.4655,
       "step": 30950
     },
     {
       "epoch": 3.3423180592991915,
-      "grad_norm": 0.5722789168357849,
+      "grad_norm": 0.5726505517959595,
       "learning_rate": 0.00039996330275229354,
-      "loss": 3.4404,
+      "loss": 3.4439,
       "step": 31000
     },
     {
       "epoch": 3.3423180592991915,
-      "eval_accuracy": 0.37290725030941624,
-      "eval_loss": 3.476534605026245,
-      "eval_runtime": 183.314,
-      "eval_samples_per_second": 98.252,
-      "eval_steps_per_second": 6.142,
+      "eval_accuracy": 0.3722968383856745,
+      "eval_loss": 3.4804699420928955,
+      "eval_runtime": 181.2749,
+      "eval_samples_per_second": 99.357,
+      "eval_steps_per_second": 6.212,
       "step": 31000
     },
     {
       "epoch": 3.347708894878706,
-      "grad_norm": 0.5752816200256348,
+      "grad_norm": 0.6000190377235413,
       "learning_rate": 0.00039963950350782514,
-      "loss": 3.4498,
+      "loss": 3.4535,
       "step": 31050
     },
     {
       "epoch": 3.353099730458221,
-      "grad_norm": 0.6340233087539673,
+      "grad_norm": 0.6834233403205872,
       "learning_rate": 0.0003993157042633567,
-      "loss": 3.4681,
+      "loss": 3.4712,
       "step": 31100
     },
     {
       "epoch": 3.358490566037736,
-      "grad_norm": 0.6339342594146729,
+      "grad_norm": 0.6108174920082092,
       "learning_rate": 0.00039899190501888824,
-      "loss": 3.4499,
+      "loss": 3.4528,
       "step": 31150
     },
     {
       "epoch": 3.3638814016172507,
-      "grad_norm": 0.5568013191223145,
+      "grad_norm": 0.5592135190963745,
       "learning_rate": 0.00039866810577441985,
-      "loss": 3.4723,
+      "loss": 3.475,
       "step": 31200
     },
     {
       "epoch": 3.3692722371967654,
-      "grad_norm": 0.5924887657165527,
+      "grad_norm": 0.6018789410591125,
       "learning_rate": 0.0003983443065299514,
-      "loss": 3.4685,
+      "loss": 3.4719,
       "step": 31250
     },
     {
       "epoch": 3.37466307277628,
-      "grad_norm": 0.6027880907058716,
+      "grad_norm": 0.5522273778915405,
       "learning_rate": 0.000398020507285483,
-      "loss": 3.4603,
+      "loss": 3.4626,
       "step": 31300
     },
     {
       "epoch": 3.3800539083557952,
-      "grad_norm": 0.5847180485725403,
+      "grad_norm": 0.5645899772644043,
       "learning_rate": 0.0003976967080410145,
-      "loss": 3.4554,
+      "loss": 3.4592,
       "step": 31350
     },
     {
       "epoch": 3.38544474393531,
-      "grad_norm": 0.6057071089744568,
+      "grad_norm": 0.5799838304519653,
       "learning_rate": 0.0003973729087965461,
-      "loss": 3.4646,
+      "loss": 3.4655,
       "step": 31400
     },
     {
       "epoch": 3.3908355795148246,
-      "grad_norm": 0.6064717769622803,
+      "grad_norm": 0.5917410850524902,
       "learning_rate": 0.00039704910955207765,
-      "loss": 3.4559,
+      "loss": 3.4569,
       "step": 31450
     },
     {
       "epoch": 3.3962264150943398,
-      "grad_norm": 0.7252898216247559,
+      "grad_norm": 0.7213910222053528,
       "learning_rate": 0.00039672531030760926,
-      "loss": 3.4556,
+      "loss": 3.4593,
       "step": 31500
     },
     {
       "epoch": 3.4016172506738545,
-      "grad_norm": 0.5483014583587646,
-      "learning_rate": 0.00039640798704803016,
-      "loss": 3.4635,
+      "grad_norm": 0.5657336115837097,
+      "learning_rate": 0.0003964015110631408,
+      "loss": 3.4647,
       "step": 31550
     },
     {
       "epoch": 3.407008086253369,
-      "grad_norm": 0.5969088673591614,
-      "learning_rate": 0.00039608418780356176,
-      "loss": 3.4526,
+      "grad_norm": 0.5951988697052002,
+      "learning_rate": 0.00039607771181867236,
+      "loss": 3.4544,
       "step": 31600
     },
     {
       "epoch": 3.4123989218328843,
-      "grad_norm": 0.6174740791320801,
-      "learning_rate": 0.0003957603885590933,
-      "loss": 3.4417,
+      "grad_norm": 0.5717089176177979,
+      "learning_rate": 0.00039575391257420397,
+      "loss": 3.4435,
       "step": 31650
     },
     {
       "epoch": 3.417789757412399,
-      "grad_norm": 0.5688740611076355,
-      "learning_rate": 0.0003954365893146249,
-      "loss": 3.4588,
+      "grad_norm": 0.5992870330810547,
+      "learning_rate": 0.0003954301133297355,
+      "loss": 3.4616,
       "step": 31700
     },
     {
       "epoch": 3.4231805929919137,
-      "grad_norm": 0.6028859615325928,
-      "learning_rate": 0.00039511279007015647,
-      "loss": 3.4622,
+      "grad_norm": 0.5768448710441589,
+      "learning_rate": 0.0003951063140852671,
+      "loss": 3.4643,
       "step": 31750
     },
     {
       "epoch": 3.4285714285714284,
-      "grad_norm": 0.5561661124229431,
-      "learning_rate": 0.00039478899082568807,
-      "loss": 3.4419,
+      "grad_norm": 0.5669674873352051,
+      "learning_rate": 0.00039478251484079867,
+      "loss": 3.4443,
       "step": 31800
     },
     {
       "epoch": 3.4339622641509435,
-      "grad_norm": 0.6191286444664001,
-      "learning_rate": 0.0003944651915812196,
-      "loss": 3.4556,
+      "grad_norm": 0.5871769785881042,
+      "learning_rate": 0.0003944587155963303,
+      "loss": 3.4606,
       "step": 31850
     },
     {
       "epoch": 3.439353099730458,
-      "grad_norm": 0.6074486374855042,
-      "learning_rate": 0.0003941413923367512,
-      "loss": 3.468,
+      "grad_norm": 0.5850691199302673,
+      "learning_rate": 0.0003941349163518618,
+      "loss": 3.4699,
       "step": 31900
     },
     {
       "epoch": 3.444743935309973,
-      "grad_norm": 0.5516985058784485,
-      "learning_rate": 0.0003938175930922828,
-      "loss": 3.4587,
+      "grad_norm": 0.5405361652374268,
+      "learning_rate": 0.00039381111710739343,
+      "loss": 3.4611,
       "step": 31950
     },
     {
       "epoch": 3.450134770889488,
-      "grad_norm": 0.5780602097511292,
-      "learning_rate": 0.0003934937938478143,
-      "loss": 3.4539,
+      "grad_norm": 0.5631510615348816,
+      "learning_rate": 0.000393487317862925,
+      "loss": 3.4572,
       "step": 32000
     },
     {
       "epoch": 3.450134770889488,
-      "eval_accuracy": 0.3738237374181314,
-      "eval_loss": 3.4704768657684326,
-      "eval_runtime": 183.3826,
-      "eval_samples_per_second": 98.215,
-      "eval_steps_per_second": 6.14,
+      "eval_accuracy": 0.3733245860515117,
+      "eval_loss": 3.473914384841919,
+      "eval_runtime": 181.48,
+      "eval_samples_per_second": 99.245,
+      "eval_steps_per_second": 6.205,
       "step": 32000
     },
     {
       "epoch": 3.4555256064690028,
-      "grad_norm": 0.5890681147575378,
-      "learning_rate": 0.0003931699946033459,
-      "loss": 3.4567,
+      "grad_norm": 0.628905713558197,
+      "learning_rate": 0.0003931635186184565,
+      "loss": 3.46,
       "step": 32050
     },
     {
       "epoch": 3.4609164420485174,
-      "grad_norm": 0.6089186668395996,
-      "learning_rate": 0.00039284619535887743,
-      "loss": 3.4616,
+      "grad_norm": 0.5995890498161316,
+      "learning_rate": 0.0003928397193739881,
+      "loss": 3.465,
       "step": 32100
     },
     {
       "epoch": 3.466307277628032,
-      "grad_norm": 0.6278788447380066,
-      "learning_rate": 0.00039252239611440904,
-      "loss": 3.4596,
+      "grad_norm": 0.5921915769577026,
+      "learning_rate": 0.00039251592012951963,
+      "loss": 3.4608,
       "step": 32150
     },
     {
       "epoch": 3.4716981132075473,
-      "grad_norm": 0.6116218566894531,
-      "learning_rate": 0.0003921985968699406,
-      "loss": 3.439,
+      "grad_norm": 0.5565003752708435,
+      "learning_rate": 0.00039219212088505124,
+      "loss": 3.4431,
       "step": 32200
     },
     {
       "epoch": 3.477088948787062,
-      "grad_norm": 0.5771948099136353,
-      "learning_rate": 0.0003918747976254722,
-      "loss": 3.4533,
+      "grad_norm": 0.5746980905532837,
+      "learning_rate": 0.0003918683216405828,
+      "loss": 3.4567,
       "step": 32250
     },
     {
       "epoch": 3.4824797843665767,
-      "grad_norm": 0.5892335176467896,
-      "learning_rate": 0.00039155099838100374,
-      "loss": 3.4757,
+      "grad_norm": 0.591099202632904,
+      "learning_rate": 0.0003915445223961144,
+      "loss": 3.4792,
       "step": 32300
     },
     {
       "epoch": 3.487870619946092,
-      "grad_norm": 0.6450402140617371,
-      "learning_rate": 0.0003912271991365353,
-      "loss": 3.4477,
+      "grad_norm": 0.5964375734329224,
+      "learning_rate": 0.00039122072315164594,
+      "loss": 3.4498,
       "step": 32350
     },
     {
       "epoch": 3.4932614555256065,
-      "grad_norm": 0.585907518863678,
-      "learning_rate": 0.0003909033998920669,
-      "loss": 3.4573,
+      "grad_norm": 0.5887673497200012,
+      "learning_rate": 0.0003908969239071775,
+      "loss": 3.461,
       "step": 32400
     },
     {
       "epoch": 3.498652291105121,
-      "grad_norm": 0.6161054372787476,
-      "learning_rate": 0.00039057960064759845,
-      "loss": 3.4465,
+      "grad_norm": 0.7257962822914124,
+      "learning_rate": 0.0003905731246627091,
+      "loss": 3.4486,
       "step": 32450
     },
     {
       "epoch": 3.5040431266846364,
-      "grad_norm": 0.5852574110031128,
-      "learning_rate": 0.00039025580140313005,
-      "loss": 3.4551,
+      "grad_norm": 0.6065286993980408,
+      "learning_rate": 0.00039024932541824065,
+      "loss": 3.4582,
       "step": 32500
     },
     {
       "epoch": 3.509433962264151,
-      "grad_norm": 0.6066722273826599,
-      "learning_rate": 0.0003899320021586616,
-      "loss": 3.4501,
+      "grad_norm": 0.6183472871780396,
+      "learning_rate": 0.00038992552617377225,
+      "loss": 3.4535,
       "step": 32550
     },
     {
       "epoch": 3.5148247978436657,
-      "grad_norm": 0.5780394077301025,
-      "learning_rate": 0.0003896082029141932,
-      "loss": 3.4549,
+      "grad_norm": 0.5770750045776367,
+      "learning_rate": 0.0003896017269293038,
+      "loss": 3.4578,
       "step": 32600
     },
     {
       "epoch": 3.5202156334231804,
-      "grad_norm": 0.6035788059234619,
-      "learning_rate": 0.00038928440366972476,
-      "loss": 3.448,
+      "grad_norm": 0.5831050872802734,
+      "learning_rate": 0.0003892779276848354,
+      "loss": 3.4497,
       "step": 32650
     },
     {
       "epoch": 3.525606469002695,
-      "grad_norm": 0.6049900650978088,
-      "learning_rate": 0.00038896060442525636,
-      "loss": 3.4605,
+      "grad_norm": 0.6157300472259521,
+      "learning_rate": 0.0003889541284403669,
+      "loss": 3.463,
       "step": 32700
     },
     {
       "epoch": 3.5309973045822103,
-      "grad_norm": 0.6193360686302185,
-      "learning_rate": 0.00038863680518078786,
-      "loss": 3.4481,
+      "grad_norm": 0.6285575032234192,
+      "learning_rate": 0.0003886303291958985,
+      "loss": 3.4518,
       "step": 32750
     },
     {
       "epoch": 3.536388140161725,
-      "grad_norm": 0.6224262714385986,
-      "learning_rate": 0.0003883130059363194,
-      "loss": 3.4517,
+      "grad_norm": 0.6076500415802002,
+      "learning_rate": 0.00038830652995143006,
+      "loss": 3.4541,
       "step": 32800
     },
     {
       "epoch": 3.5417789757412397,
-      "grad_norm": 0.6505341529846191,
-      "learning_rate": 0.000387989206691851,
-      "loss": 3.4337,
+      "grad_norm": 0.5972782969474792,
+      "learning_rate": 0.0003879827307069616,
+      "loss": 3.435,
       "step": 32850
     },
     {
       "epoch": 3.547169811320755,
-      "grad_norm": 0.5855464935302734,
-      "learning_rate": 0.00038766540744738256,
-      "loss": 3.4484,
+      "grad_norm": 0.5806552171707153,
+      "learning_rate": 0.0003876589314624932,
+      "loss": 3.4519,
       "step": 32900
     },
     {
       "epoch": 3.5525606469002695,
-      "grad_norm": 0.5817187428474426,
-      "learning_rate": 0.00038734160820291417,
-      "loss": 3.4388,
+      "grad_norm": 0.5712008476257324,
+      "learning_rate": 0.00038733513221802477,
+      "loss": 3.4401,
       "step": 32950
     },
     {
       "epoch": 3.557951482479784,
-      "grad_norm": 0.6249986290931702,
-      "learning_rate": 0.0003870178089584457,
-      "loss": 3.4407,
+      "grad_norm": 0.6120150685310364,
+      "learning_rate": 0.00038701133297355637,
+      "loss": 3.4453,
       "step": 33000
     },
     {
       "epoch": 3.557951482479784,
-      "eval_accuracy": 0.3741863121047683,
-      "eval_loss": 3.4653964042663574,
-      "eval_runtime": 183.4614,
-      "eval_samples_per_second": 98.173,
-      "eval_steps_per_second": 6.138,
+      "eval_accuracy": 0.37386209188752995,
+      "eval_loss": 3.4680962562561035,
+      "eval_runtime": 181.3487,
+      "eval_samples_per_second": 99.317,
+      "eval_steps_per_second": 6.209,
       "step": 33000
     },
     {
       "epoch": 3.5633423180592994,
-      "grad_norm": 0.6825558543205261,
-      "learning_rate": 0.0003866940097139773,
-      "loss": 3.4488,
+      "grad_norm": 0.6286499500274658,
+      "learning_rate": 0.0003866875337290879,
+      "loss": 3.4498,
       "step": 33050
     },
     {
       "epoch": 3.568733153638814,
-      "grad_norm": 0.6727454662322998,
-      "learning_rate": 0.0003863702104695089,
-      "loss": 3.4594,
+      "grad_norm": 0.6690121293067932,
+      "learning_rate": 0.00038636373448461953,
+      "loss": 3.4617,
       "step": 33100
     },
     {
       "epoch": 3.5741239892183287,
-      "grad_norm": 0.5715111494064331,
-      "learning_rate": 0.0003860464112250404,
-      "loss": 3.4315,
+      "grad_norm": 0.6036093235015869,
+      "learning_rate": 0.0003860399352401511,
+      "loss": 3.4354,
       "step": 33150
     },
     {
       "epoch": 3.579514824797844,
-      "grad_norm": 0.5947990417480469,
-      "learning_rate": 0.00038572261198057203,
-      "loss": 3.4395,
+      "grad_norm": 0.5803263187408447,
+      "learning_rate": 0.0003857161359956827,
+      "loss": 3.4433,
       "step": 33200
     },
     {
       "epoch": 3.5849056603773586,
-      "grad_norm": 0.6582793593406677,
-      "learning_rate": 0.0003853988127361036,
-      "loss": 3.4704,
+      "grad_norm": 0.6099799275398254,
+      "learning_rate": 0.00038539233675121423,
+      "loss": 3.4729,
       "step": 33250
     },
     {
       "epoch": 3.5902964959568733,
-      "grad_norm": 0.6481404900550842,
+      "grad_norm": 0.6689486503601074,
       "learning_rate": 0.0003850750134916352,
-      "loss": 3.4585,
+      "loss": 3.4637,
       "step": 33300
     },
     {
       "epoch": 3.595687331536388,
-      "grad_norm": 0.5785871148109436,
+      "grad_norm": 0.555364191532135,
       "learning_rate": 0.0003847512142471667,
-      "loss": 3.446,
+      "loss": 3.4498,
       "step": 33350
     },
     {
       "epoch": 3.601078167115903,
-      "grad_norm": 0.6385973691940308,
+      "grad_norm": 0.6476346254348755,
       "learning_rate": 0.0003844274150026983,
-      "loss": 3.4475,
+      "loss": 3.4508,
       "step": 33400
     },
     {
       "epoch": 3.606469002695418,
-      "grad_norm": 0.6048685908317566,
+      "grad_norm": 0.5849195122718811,
       "learning_rate": 0.00038410361575822984,
-      "loss": 3.4571,
+      "loss": 3.4601,
       "step": 33450
     },
     {
       "epoch": 3.6118598382749325,
-      "grad_norm": 0.5991024971008301,
+      "grad_norm": 0.6262844204902649,
       "learning_rate": 0.00038377981651376144,
-      "loss": 3.4378,
+      "loss": 3.4415,
       "step": 33500
     },
     {
       "epoch": 3.617250673854447,
-      "grad_norm": 0.6056102514266968,
-      "learning_rate": 0.00038346249325418234,
-      "loss": 3.4473,
+      "grad_norm": 0.6381199955940247,
+      "learning_rate": 0.000383456017269293,
+      "loss": 3.4501,
       "step": 33550
     },
     {
       "epoch": 3.6226415094339623,
-      "grad_norm": 0.6247057914733887,
-      "learning_rate": 0.00038313869400971395,
-      "loss": 3.4526,
+      "grad_norm": 0.6402269601821899,
+      "learning_rate": 0.00038313221802482454,
+      "loss": 3.4547,
       "step": 33600
     },
     {
       "epoch": 3.628032345013477,
-      "grad_norm": 0.6171483397483826,
-      "learning_rate": 0.0003828148947652455,
-      "loss": 3.4612,
+      "grad_norm": 0.6093237996101379,
+      "learning_rate": 0.00038280841878035615,
+      "loss": 3.4673,
       "step": 33650
     },
     {
       "epoch": 3.6334231805929917,
-      "grad_norm": 0.5629798769950867,
-      "learning_rate": 0.0003824910955207771,
-      "loss": 3.4562,
+      "grad_norm": 0.5662837028503418,
+      "learning_rate": 0.0003824846195358877,
+      "loss": 3.4594,
       "step": 33700
     },
     {
       "epoch": 3.638814016172507,
-      "grad_norm": 0.5999249815940857,
-      "learning_rate": 0.00038216729627630865,
-      "loss": 3.4415,
+      "grad_norm": 0.6111090183258057,
+      "learning_rate": 0.0003821608202914193,
+      "loss": 3.4442,
       "step": 33750
     },
     {
       "epoch": 3.6442048517520216,
-      "grad_norm": 0.5683233737945557,
-      "learning_rate": 0.00038184349703184026,
-      "loss": 3.4568,
+      "grad_norm": 0.584557294845581,
+      "learning_rate": 0.00038183702104695085,
+      "loss": 3.4601,
       "step": 33800
     },
     {
       "epoch": 3.6495956873315363,
-      "grad_norm": 0.6162096858024597,
-      "learning_rate": 0.0003815196977873718,
-      "loss": 3.4557,
+      "grad_norm": 0.5788835287094116,
+      "learning_rate": 0.00038151322180248246,
+      "loss": 3.4573,
       "step": 33850
     },
     {
       "epoch": 3.6549865229110514,
-      "grad_norm": 0.6410558223724365,
-      "learning_rate": 0.0003811958985429034,
-      "loss": 3.4458,
+      "grad_norm": 0.6080313324928284,
+      "learning_rate": 0.000381189422558014,
+      "loss": 3.4496,
       "step": 33900
     },
     {
       "epoch": 3.660377358490566,
-      "grad_norm": 0.6777019500732422,
-      "learning_rate": 0.00038087209929843496,
-      "loss": 3.4405,
+      "grad_norm": 0.6473999619483948,
+      "learning_rate": 0.0003808656233135456,
+      "loss": 3.4463,
       "step": 33950
     },
     {
       "epoch": 3.665768194070081,
-      "grad_norm": 0.6722299456596375,
-      "learning_rate": 0.00038054830005396646,
-      "loss": 3.4573,
+      "grad_norm": 0.6274574398994446,
+      "learning_rate": 0.00038054182406907716,
+      "loss": 3.4593,
       "step": 34000
     },
     {
       "epoch": 3.665768194070081,
-      "eval_accuracy": 0.37471262669334165,
-      "eval_loss": 3.458420991897583,
-      "eval_runtime": 183.3402,
-      "eval_samples_per_second": 98.238,
-      "eval_steps_per_second": 6.142,
+      "eval_accuracy": 0.3744651154149003,
+      "eval_loss": 3.459097146987915,
+      "eval_runtime": 181.9057,
+      "eval_samples_per_second": 99.013,
+      "eval_steps_per_second": 6.19,
       "step": 34000
     },
     {
       "epoch": 3.671159029649596,
-      "grad_norm": 0.6154153943061829,
-      "learning_rate": 0.00038022450080949806,
-      "loss": 3.4406,
+      "grad_norm": 0.6133659482002258,
+      "learning_rate": 0.00038021802482460866,
+      "loss": 3.4433,
       "step": 34050
     },
     {
       "epoch": 3.6765498652291106,
-      "grad_norm": 0.5900728702545166,
-      "learning_rate": 0.0003799007015650296,
-      "loss": 3.4567,
+      "grad_norm": 0.615942120552063,
+      "learning_rate": 0.00037989422558014027,
+      "loss": 3.4605,
       "step": 34100
     },
     {
       "epoch": 3.6819407008086253,
-      "grad_norm": 0.6302698850631714,
-      "learning_rate": 0.0003795769023205612,
-      "loss": 3.4545,
+      "grad_norm": 0.6161052584648132,
+      "learning_rate": 0.0003795704263356718,
+      "loss": 3.4584,
       "step": 34150
     },
     {
       "epoch": 3.68733153638814,
-      "grad_norm": 0.5984343886375427,
-      "learning_rate": 0.00037925310307609277,
-      "loss": 3.4368,
+      "grad_norm": 0.6069619655609131,
+      "learning_rate": 0.0003792466270912034,
+      "loss": 3.4393,
       "step": 34200
     },
     {
       "epoch": 3.6927223719676547,
-      "grad_norm": 0.6989060044288635,
-      "learning_rate": 0.0003789293038316244,
-      "loss": 3.4501,
+      "grad_norm": 0.6923947930335999,
+      "learning_rate": 0.00037892282784673497,
+      "loss": 3.4533,
       "step": 34250
     },
     {
       "epoch": 3.69811320754717,
-      "grad_norm": 0.6396805644035339,
-      "learning_rate": 0.0003786055045871559,
-      "loss": 3.4404,
+      "grad_norm": 0.6208213567733765,
+      "learning_rate": 0.0003785990286022666,
+      "loss": 3.4417,
       "step": 34300
     },
     {
       "epoch": 3.7035040431266846,
-      "grad_norm": 0.560369074344635,
-      "learning_rate": 0.0003782817053426875,
-      "loss": 3.4632,
+      "grad_norm": 0.6071147322654724,
+      "learning_rate": 0.00037827522935779813,
+      "loss": 3.4664,
       "step": 34350
     },
     {
       "epoch": 3.7088948787061993,
-      "grad_norm": 0.6261559128761292,
-      "learning_rate": 0.0003779579060982191,
-      "loss": 3.4469,
+      "grad_norm": 0.6345399022102356,
+      "learning_rate": 0.00037795143011332973,
+      "loss": 3.4509,
       "step": 34400
     },
     {
       "epoch": 3.7142857142857144,
-      "grad_norm": 0.6551036238670349,
-      "learning_rate": 0.00037763410685375063,
-      "loss": 3.433,
+      "grad_norm": 0.667558491230011,
+      "learning_rate": 0.0003776276308688613,
+      "loss": 3.437,
       "step": 34450
     },
     {
       "epoch": 3.719676549865229,
-      "grad_norm": 0.7240360975265503,
-      "learning_rate": 0.00037731030760928223,
-      "loss": 3.4508,
+      "grad_norm": 0.7233009338378906,
+      "learning_rate": 0.00037730383162439283,
+      "loss": 3.4531,
       "step": 34500
     },
     {
       "epoch": 3.725067385444744,
-      "grad_norm": 0.6055688858032227,
-      "learning_rate": 0.0003769865083648138,
-      "loss": 3.4537,
+      "grad_norm": 0.6252279877662659,
+      "learning_rate": 0.00037698003237992444,
+      "loss": 3.4579,
       "step": 34550
     },
     {
       "epoch": 3.730458221024259,
-      "grad_norm": 0.6346741318702698,
-      "learning_rate": 0.0003766627091203454,
-      "loss": 3.4566,
+      "grad_norm": 0.6751020550727844,
+      "learning_rate": 0.000376656233135456,
+      "loss": 3.4587,
       "step": 34600
     },
     {
       "epoch": 3.7358490566037736,
-      "grad_norm": 0.5869565010070801,
-      "learning_rate": 0.00037633890987587694,
-      "loss": 3.4455,
+      "grad_norm": 0.574543297290802,
+      "learning_rate": 0.0003763324338909876,
+      "loss": 3.4478,
       "step": 34650
     },
     {
       "epoch": 3.7412398921832883,
-      "grad_norm": 0.606854259967804,
-      "learning_rate": 0.00037601511063140855,
-      "loss": 3.4511,
+      "grad_norm": 0.5911687016487122,
+      "learning_rate": 0.0003760086346465191,
+      "loss": 3.4548,
       "step": 34700
     },
     {
       "epoch": 3.7466307277628035,
-      "grad_norm": 0.6614265441894531,
-      "learning_rate": 0.00037569131138694004,
-      "loss": 3.4331,
+      "grad_norm": 0.6213070750236511,
+      "learning_rate": 0.0003756848354020507,
+      "loss": 3.436,
       "step": 34750
     },
     {
       "epoch": 3.752021563342318,
-      "grad_norm": 0.6091951727867126,
-      "learning_rate": 0.0003753675121424716,
-      "loss": 3.4687,
+      "grad_norm": 0.5843448638916016,
+      "learning_rate": 0.00037536103615758224,
+      "loss": 3.4707,
       "step": 34800
     },
     {
       "epoch": 3.757412398921833,
-      "grad_norm": 0.6363885402679443,
-      "learning_rate": 0.0003750437128980032,
-      "loss": 3.454,
+      "grad_norm": 0.6959250569343567,
+      "learning_rate": 0.0003750372369131138,
+      "loss": 3.4563,
       "step": 34850
     },
     {
       "epoch": 3.7628032345013476,
-      "grad_norm": 0.6237848997116089,
-      "learning_rate": 0.00037471991365353475,
-      "loss": 3.4517,
+      "grad_norm": 0.6216655969619751,
+      "learning_rate": 0.0003747134376686454,
+      "loss": 3.4554,
       "step": 34900
     },
     {
       "epoch": 3.7681940700808623,
-      "grad_norm": 0.6054222583770752,
-      "learning_rate": 0.00037439611440906635,
-      "loss": 3.4515,
+      "grad_norm": 0.6173912882804871,
+      "learning_rate": 0.00037438963842417695,
+      "loss": 3.4562,
       "step": 34950
     },
     {
       "epoch": 3.7735849056603774,
-      "grad_norm": 0.6497372984886169,
-      "learning_rate": 0.0003740723151645979,
-      "loss": 3.4392,
+      "grad_norm": 0.6174741983413696,
+      "learning_rate": 0.00037406583917970856,
+      "loss": 3.4427,
       "step": 35000
     },
     {
       "epoch": 3.7735849056603774,
-      "eval_accuracy": 0.3750332613652822,
-      "eval_loss": 3.4538044929504395,
-      "eval_runtime": 183.7049,
-      "eval_samples_per_second": 98.043,
-      "eval_steps_per_second": 6.129,
+      "eval_accuracy": 0.3749664398392763,
+      "eval_loss": 3.4547739028930664,
+      "eval_runtime": 181.792,
+      "eval_samples_per_second": 99.075,
+      "eval_steps_per_second": 6.194,
       "step": 35000
     },
     {
       "epoch": 3.778975741239892,
-      "grad_norm": 0.7446255683898926,
-      "learning_rate": 0.0003737485159201295,
-      "loss": 3.4523,
+      "grad_norm": 0.7106022834777832,
+      "learning_rate": 0.0003737420399352401,
+      "loss": 3.4553,
       "step": 35050
     },
     {
       "epoch": 3.784366576819407,
-      "grad_norm": 0.5870745778083801,
-      "learning_rate": 0.00037342471667566106,
-      "loss": 3.4448,
+      "grad_norm": 0.6293195486068726,
+      "learning_rate": 0.0003734182406907717,
+      "loss": 3.4497,
       "step": 35100
     },
     {
       "epoch": 3.789757412398922,
-      "grad_norm": 0.6146149635314941,
-      "learning_rate": 0.00037310091743119266,
-      "loss": 3.4419,
+      "grad_norm": 0.6353211402893066,
+      "learning_rate": 0.00037309444144630326,
+      "loss": 3.4435,
       "step": 35150
     },
     {
       "epoch": 3.7951482479784366,
-      "grad_norm": 0.7114392518997192,
-      "learning_rate": 0.0003727771181867242,
-      "loss": 3.4663,
+      "grad_norm": 0.6555863618850708,
+      "learning_rate": 0.00037277064220183487,
+      "loss": 3.469,
       "step": 35200
     },
     {
       "epoch": 3.8005390835579513,
-      "grad_norm": 0.6365261077880859,
-      "learning_rate": 0.00037245331894225576,
-      "loss": 3.4394,
+      "grad_norm": 0.5852487087249756,
+      "learning_rate": 0.0003724468429573664,
+      "loss": 3.4423,
       "step": 35250
     },
     {
       "epoch": 3.8059299191374665,
-      "grad_norm": 0.6266224980354309,
-      "learning_rate": 0.00037212951969778737,
-      "loss": 3.4338,
+      "grad_norm": 0.6362602710723877,
+      "learning_rate": 0.0003721230437128979,
+      "loss": 3.4369,
       "step": 35300
     },
     {
       "epoch": 3.811320754716981,
-      "grad_norm": 0.5985236167907715,
-      "learning_rate": 0.0003718121964382083,
-      "loss": 3.4362,
+      "grad_norm": 0.6013461351394653,
+      "learning_rate": 0.00037179924446842957,
+      "loss": 3.4376,
       "step": 35350
     },
     {
       "epoch": 3.816711590296496,
-      "grad_norm": 0.6070831418037415,
-      "learning_rate": 0.0003714883971937398,
-      "loss": 3.4484,
+      "grad_norm": 0.6238158345222473,
+      "learning_rate": 0.00037147544522396107,
+      "loss": 3.4532,
       "step": 35400
     },
     {
       "epoch": 3.822102425876011,
-      "grad_norm": 0.556121826171875,
-      "learning_rate": 0.0003711645979492714,
-      "loss": 3.4579,
+      "grad_norm": 0.5739532709121704,
+      "learning_rate": 0.0003711516459794927,
+      "loss": 3.4626,
       "step": 35450
     },
     {
       "epoch": 3.8274932614555257,
-      "grad_norm": 0.5937209129333496,
-      "learning_rate": 0.00037084079870480297,
-      "loss": 3.4385,
+      "grad_norm": 0.6169940829277039,
+      "learning_rate": 0.0003708278467350242,
+      "loss": 3.4422,
       "step": 35500
     },
     {
       "epoch": 3.8328840970350404,
-      "grad_norm": 0.5834438800811768,
-      "learning_rate": 0.0003705169994603345,
-      "loss": 3.4426,
+      "grad_norm": 0.5878958106040955,
+      "learning_rate": 0.00037050404749055583,
+      "loss": 3.4466,
       "step": 35550
     },
     {
       "epoch": 3.838274932614555,
-      "grad_norm": 0.6240514516830444,
-      "learning_rate": 0.00037019320021586613,
-      "loss": 3.4437,
+      "grad_norm": 0.6152637600898743,
+      "learning_rate": 0.0003701802482460874,
+      "loss": 3.445,
       "step": 35600
     },
     {
       "epoch": 3.8436657681940702,
-      "grad_norm": 0.6120893955230713,
-      "learning_rate": 0.0003698694009713977,
-      "loss": 3.4429,
+      "grad_norm": 0.6067671775817871,
+      "learning_rate": 0.000369856449001619,
+      "loss": 3.4465,
       "step": 35650
     },
     {
       "epoch": 3.849056603773585,
-      "grad_norm": 0.6248950958251953,
-      "learning_rate": 0.0003695456017269293,
-      "loss": 3.4407,
+      "grad_norm": 0.5947751402854919,
+      "learning_rate": 0.00036953264975715053,
+      "loss": 3.4454,
       "step": 35700
     },
     {
       "epoch": 3.8544474393530996,
-      "grad_norm": 0.5935930013656616,
-      "learning_rate": 0.00036922180248246083,
-      "loss": 3.45,
+      "grad_norm": 0.5876570343971252,
+      "learning_rate": 0.0003692088505126821,
+      "loss": 3.4532,
       "step": 35750
     },
     {
       "epoch": 3.8598382749326143,
-      "grad_norm": 0.643417239189148,
-      "learning_rate": 0.00036889800323799244,
-      "loss": 3.4408,
+      "grad_norm": 0.6134797930717468,
+      "learning_rate": 0.0003688850512682137,
+      "loss": 3.4419,
       "step": 35800
     },
     {
       "epoch": 3.8652291105121295,
-      "grad_norm": 0.6277262568473816,
-      "learning_rate": 0.000368574203993524,
-      "loss": 3.4397,
+      "grad_norm": 0.5943005681037903,
+      "learning_rate": 0.00036856125202374524,
+      "loss": 3.4443,
       "step": 35850
     },
     {
       "epoch": 3.870619946091644,
-      "grad_norm": 0.6452029347419739,
-      "learning_rate": 0.0003682504047490556,
-      "loss": 3.4507,
+      "grad_norm": 0.6293517351150513,
+      "learning_rate": 0.00036823745277927685,
+      "loss": 3.4538,
       "step": 35900
     },
     {
       "epoch": 3.876010781671159,
-      "grad_norm": 0.645775556564331,
-      "learning_rate": 0.00036792660550458714,
-      "loss": 3.4488,
+      "grad_norm": 0.5938091278076172,
+      "learning_rate": 0.0003679136535348084,
+      "loss": 3.4515,
       "step": 35950
     },
     {
       "epoch": 3.881401617250674,
-      "grad_norm": 0.6617048382759094,
-      "learning_rate": 0.00036760280626011864,
-      "loss": 3.4464,
+      "grad_norm": 0.6512495875358582,
+      "learning_rate": 0.00036758985429034,
+      "loss": 3.4495,
       "step": 36000
     },
     {
       "epoch": 3.881401617250674,
-      "eval_accuracy": 0.37615086496934197,
-      "eval_loss": 3.445986032485962,
-      "eval_runtime": 183.1416,
-      "eval_samples_per_second": 98.345,
-      "eval_steps_per_second": 6.148,
+      "eval_accuracy": 0.37573092155793897,
+      "eval_loss": 3.447737455368042,
+      "eval_runtime": 181.7497,
+      "eval_samples_per_second": 99.098,
+      "eval_steps_per_second": 6.195,
       "step": 36000
     },
     {
       "epoch": 3.8867924528301887,
-      "grad_norm": 0.6022095084190369,
-      "learning_rate": 0.0003672790070156503,
-      "loss": 3.4428,
+      "grad_norm": 0.6184289455413818,
+      "learning_rate": 0.0003672660550458715,
+      "loss": 3.4449,
       "step": 36050
     },
     {
       "epoch": 3.8921832884097034,
-      "grad_norm": 0.5766339302062988,
-      "learning_rate": 0.0003669552077711818,
-      "loss": 3.4499,
+      "grad_norm": 0.5750420093536377,
+      "learning_rate": 0.00036694225580140316,
+      "loss": 3.4553,
       "step": 36100
     },
     {
       "epoch": 3.8975741239892185,
-      "grad_norm": 0.6311817765235901,
-      "learning_rate": 0.0003666314085267134,
-      "loss": 3.4279,
+      "grad_norm": 0.6374450922012329,
+      "learning_rate": 0.00036661845655693465,
+      "loss": 3.4293,
       "step": 36150
     },
     {
       "epoch": 3.9029649595687332,
-      "grad_norm": 0.6049012541770935,
-      "learning_rate": 0.00036630760928224495,
-      "loss": 3.4384,
+      "grad_norm": 0.6292021870613098,
+      "learning_rate": 0.0003662946573124662,
+      "loss": 3.4422,
       "step": 36200
     },
     {
       "epoch": 3.908355795148248,
-      "grad_norm": 0.6929218769073486,
-      "learning_rate": 0.00036598381003777656,
-      "loss": 3.4577,
+      "grad_norm": 0.6224103569984436,
+      "learning_rate": 0.0003659708580679978,
+      "loss": 3.461,
       "step": 36250
     },
     {
       "epoch": 3.913746630727763,
-      "grad_norm": 0.6345332264900208,
-      "learning_rate": 0.0003656600107933081,
-      "loss": 3.4422,
+      "grad_norm": 0.6218449473381042,
+      "learning_rate": 0.00036564705882352936,
+      "loss": 3.4465,
       "step": 36300
     },
     {
       "epoch": 3.9191374663072778,
-      "grad_norm": 0.6367397308349609,
-      "learning_rate": 0.0003653362115488397,
-      "loss": 3.4492,
+      "grad_norm": 0.6312861442565918,
+      "learning_rate": 0.00036532325957906096,
+      "loss": 3.453,
       "step": 36350
     },
     {
       "epoch": 3.9245283018867925,
-      "grad_norm": 0.6182951331138611,
-      "learning_rate": 0.00036501241230437126,
-      "loss": 3.4355,
+      "grad_norm": 0.6503349542617798,
+      "learning_rate": 0.0003649994603345925,
+      "loss": 3.4393,
       "step": 36400
     },
     {
       "epoch": 3.929919137466307,
-      "grad_norm": 0.5866326093673706,
-      "learning_rate": 0.0003646886130599028,
-      "loss": 3.458,
+      "grad_norm": 0.6077538132667542,
+      "learning_rate": 0.0003646756610901241,
+      "loss": 3.4618,
       "step": 36450
     },
     {
       "epoch": 3.935309973045822,
-      "grad_norm": 0.6007795333862305,
-      "learning_rate": 0.0003643648138154344,
-      "loss": 3.4518,
+      "grad_norm": 0.6427134275436401,
+      "learning_rate": 0.00036435186184565567,
+      "loss": 3.456,
       "step": 36500
     },
     {
       "epoch": 3.940700808625337,
-      "grad_norm": 0.575502336025238,
-      "learning_rate": 0.00036404101457096597,
-      "loss": 3.4513,
+      "grad_norm": 0.582518994808197,
+      "learning_rate": 0.0003640280626011873,
+      "loss": 3.4558,
       "step": 36550
     },
     {
       "epoch": 3.9460916442048517,
-      "grad_norm": 0.5908992290496826,
-      "learning_rate": 0.00036371721532649757,
-      "loss": 3.452,
+      "grad_norm": 0.6387560367584229,
+      "learning_rate": 0.0003637042633567188,
+      "loss": 3.4558,
       "step": 36600
     },
     {
       "epoch": 3.9514824797843664,
-      "grad_norm": 0.6322728395462036,
-      "learning_rate": 0.0003633934160820291,
-      "loss": 3.4509,
+      "grad_norm": 0.6564193964004517,
+      "learning_rate": 0.0003633804641122503,
+      "loss": 3.4522,
       "step": 36650
     },
     {
       "epoch": 3.9568733153638815,
-      "grad_norm": 0.7075260877609253,
-      "learning_rate": 0.00036306961683756073,
-      "loss": 3.4427,
+      "grad_norm": 0.674395740032196,
+      "learning_rate": 0.000363056664867782,
+      "loss": 3.4463,
       "step": 36700
     },
     {
       "epoch": 3.9622641509433962,
-      "grad_norm": 0.6351099610328674,
-      "learning_rate": 0.0003627458175930922,
-      "loss": 3.474,
+      "grad_norm": 0.6570578813552856,
+      "learning_rate": 0.0003627328656233135,
+      "loss": 3.477,
       "step": 36750
     },
     {
       "epoch": 3.967654986522911,
-      "grad_norm": 0.6212638020515442,
-      "learning_rate": 0.0003624220183486238,
-      "loss": 3.4598,
+      "grad_norm": 0.6231279969215393,
+      "learning_rate": 0.0003624090663788451,
+      "loss": 3.4593,
       "step": 36800
     },
     {
       "epoch": 3.973045822102426,
-      "grad_norm": 0.6004298329353333,
-      "learning_rate": 0.0003620982191041554,
-      "loss": 3.4321,
+      "grad_norm": 0.6319435834884644,
+      "learning_rate": 0.00036208526713437663,
+      "loss": 3.436,
       "step": 36850
     },
     {
       "epoch": 3.9784366576819408,
-      "grad_norm": 0.6405156254768372,
-      "learning_rate": 0.00036177441985968693,
-      "loss": 3.4454,
+      "grad_norm": 0.6328497529029846,
+      "learning_rate": 0.00036176146788990824,
+      "loss": 3.4489,
       "step": 36900
     },
     {
       "epoch": 3.9838274932614555,
-      "grad_norm": 0.6019781231880188,
-      "learning_rate": 0.00036145062061521854,
-      "loss": 3.424,
+      "grad_norm": 0.6450760960578918,
+      "learning_rate": 0.0003614376686454398,
+      "loss": 3.4272,
       "step": 36950
     },
     {
       "epoch": 3.9892183288409706,
-      "grad_norm": 0.6448103785514832,
-      "learning_rate": 0.0003611268213707501,
-      "loss": 3.4344,
+      "grad_norm": 0.5970215797424316,
+      "learning_rate": 0.00036111386940097134,
+      "loss": 3.4392,
       "step": 37000
     },
     {
       "epoch": 3.9892183288409706,
-      "eval_accuracy": 0.37672487817566586,
-      "eval_loss": 3.4396376609802246,
-      "eval_runtime": 183.5533,
-      "eval_samples_per_second": 98.124,
-      "eval_steps_per_second": 6.134,
+      "eval_accuracy": 0.3762634293611178,
+      "eval_loss": 3.442335844039917,
+      "eval_runtime": 181.834,
+      "eval_samples_per_second": 99.052,
+      "eval_steps_per_second": 6.192,
       "step": 37000
     },
     {
       "epoch": 3.9946091644204853,
-      "grad_norm": 0.6120318174362183,
-      "learning_rate": 0.0003608030221262817,
-      "loss": 3.4419,
+      "grad_norm": 0.6506046056747437,
+      "learning_rate": 0.00036079007015650294,
+      "loss": 3.4447,
       "step": 37050
     },
     {
       "epoch": 4.0,
-      "grad_norm": 1.1697500944137573,
-      "learning_rate": 0.00036047922288181324,
-      "loss": 3.4236,
+      "grad_norm": 1.155403971672058,
+      "learning_rate": 0.0003604662709120345,
+      "loss": 3.4277,
       "step": 37100
     },
     {
       "epoch": 4.005390835579515,
-      "grad_norm": 0.6020103693008423,
-      "learning_rate": 0.00036015542363734485,
-      "loss": 3.3492,
+      "grad_norm": 0.580362856388092,
+      "learning_rate": 0.0003601424716675661,
+      "loss": 3.3523,
       "step": 37150
     },
     {
       "epoch": 4.010781671159029,
-      "grad_norm": 0.5951403975486755,
-      "learning_rate": 0.0003598316243928764,
-      "loss": 3.3509,
+      "grad_norm": 0.6231250762939453,
+      "learning_rate": 0.00035981867242309765,
+      "loss": 3.3545,
       "step": 37200
     },
     {
       "epoch": 4.0161725067385445,
-      "grad_norm": 0.57025545835495,
-      "learning_rate": 0.00035950782514840795,
-      "loss": 3.3444,
+      "grad_norm": 0.6124888062477112,
+      "learning_rate": 0.00035949487317862925,
+      "loss": 3.3514,
       "step": 37250
     },
     {
       "epoch": 4.02156334231806,
-      "grad_norm": 0.6555331349372864,
-      "learning_rate": 0.00035918402590393955,
-      "loss": 3.3651,
+      "grad_norm": 0.6419339179992676,
+      "learning_rate": 0.0003591775499190502,
+      "loss": 3.3697,
       "step": 37300
     },
     {
       "epoch": 4.026954177897574,
-      "grad_norm": 0.5830883979797363,
-      "learning_rate": 0.00035886022665947105,
-      "loss": 3.3525,
+      "grad_norm": 0.6085278987884521,
+      "learning_rate": 0.00035885375067458176,
+      "loss": 3.3555,
       "step": 37350
     },
     {
       "epoch": 4.032345013477089,
-      "grad_norm": 0.6118924021720886,
-      "learning_rate": 0.0003585364274150027,
-      "loss": 3.3481,
+      "grad_norm": 0.601249635219574,
+      "learning_rate": 0.00035852995143011325,
+      "loss": 3.3513,
       "step": 37400
     },
     {
       "epoch": 4.037735849056604,
-      "grad_norm": 0.6575861573219299,
-      "learning_rate": 0.0003582126281705342,
-      "loss": 3.3609,
+      "grad_norm": 0.6509155035018921,
+      "learning_rate": 0.00035820615218564486,
+      "loss": 3.3671,
       "step": 37450
     },
     {
       "epoch": 4.0431266846361185,
-      "grad_norm": 0.664242684841156,
-      "learning_rate": 0.0003578888289260658,
-      "loss": 3.3625,
+      "grad_norm": 0.6398265957832336,
+      "learning_rate": 0.0003578823529411764,
+      "loss": 3.3661,
       "step": 37500
     },
     {
       "epoch": 4.048517520215634,
-      "grad_norm": 0.5893779993057251,
+      "grad_norm": 0.5987088084220886,
       "learning_rate": 0.00035756502968159736,
-      "loss": 3.3602,
+      "loss": 3.3621,
       "step": 37550
     },
     {
       "epoch": 4.053908355795148,
-      "grad_norm": 0.662519633769989,
+      "grad_norm": 0.651912271976471,
       "learning_rate": 0.00035724123043712896,
-      "loss": 3.3475,
+      "loss": 3.3507,
       "step": 37600
     },
     {
       "epoch": 4.059299191374663,
-      "grad_norm": 0.609507143497467,
+      "grad_norm": 0.595187246799469,
       "learning_rate": 0.0003569174311926605,
-      "loss": 3.3682,
+      "loss": 3.3707,
       "step": 37650
     },
     {
       "epoch": 4.064690026954178,
-      "grad_norm": 0.6340688467025757,
+      "grad_norm": 0.593060314655304,
       "learning_rate": 0.00035659363194819206,
-      "loss": 3.3747,
+      "loss": 3.3786,
       "step": 37700
     },
     {
       "epoch": 4.070080862533692,
-      "grad_norm": 0.6548017263412476,
+      "grad_norm": 0.6573312282562256,
       "learning_rate": 0.00035626983270372367,
-      "loss": 3.3722,
+      "loss": 3.3751,
       "step": 37750
     },
     {
       "epoch": 4.0754716981132075,
-      "grad_norm": 0.6166157126426697,
+      "grad_norm": 0.6254971027374268,
       "learning_rate": 0.0003559460334592552,
-      "loss": 3.3431,
+      "loss": 3.3473,
       "step": 37800
     },
     {
       "epoch": 4.080862533692723,
-      "grad_norm": 0.6283932328224182,
+      "grad_norm": 0.6234529614448547,
       "learning_rate": 0.0003556222342147868,
-      "loss": 3.3594,
+      "loss": 3.3629,
       "step": 37850
     },
     {
       "epoch": 4.086253369272237,
-      "grad_norm": 0.6251431703567505,
+      "grad_norm": 0.6303468346595764,
       "learning_rate": 0.0003552984349703184,
-      "loss": 3.3586,
+      "loss": 3.3615,
       "step": 37900
     },
     {
       "epoch": 4.091644204851752,
-      "grad_norm": 0.6172202825546265,
+      "grad_norm": 0.5979152917861938,
       "learning_rate": 0.00035497463572585,
-      "loss": 3.3612,
+      "loss": 3.3631,
       "step": 37950
     },
     {
       "epoch": 4.097035040431267,
-      "grad_norm": 0.655680775642395,
+      "grad_norm": 0.6547856330871582,
       "learning_rate": 0.00035465083648138153,
-      "loss": 3.3725,
+      "loss": 3.3764,
       "step": 38000
     },
     {
       "epoch": 4.097035040431267,
-      "eval_accuracy": 0.37726716473874794,
-      "eval_loss": 3.4422571659088135,
-      "eval_runtime": 183.6362,
-      "eval_samples_per_second": 98.08,
-      "eval_steps_per_second": 6.132,
+      "eval_accuracy": 0.3769157812995559,
+      "eval_loss": 3.4456403255462646,
+      "eval_runtime": 181.7587,
+      "eval_samples_per_second": 99.093,
+      "eval_steps_per_second": 6.195,
       "step": 38000
     },
     {
       "epoch": 4.1024258760107815,
-      "grad_norm": 0.6230634450912476,
+      "grad_norm": 0.6505630612373352,
       "learning_rate": 0.00035432703723691314,
-      "loss": 3.3661,
+      "loss": 3.3706,
       "step": 38050
     },
     {
       "epoch": 4.107816711590297,
-      "grad_norm": 0.6373627781867981,
+      "grad_norm": 0.7523139119148254,
       "learning_rate": 0.00035400323799244463,
-      "loss": 3.3758,
+      "loss": 3.3784,
       "step": 38100
     },
     {
       "epoch": 4.113207547169812,
-      "grad_norm": 0.6033377051353455,
+      "grad_norm": 0.6740533709526062,
       "learning_rate": 0.0003536794387479762,
-      "loss": 3.3743,
+      "loss": 3.3765,
       "step": 38150
     },
     {
       "epoch": 4.118598382749326,
-      "grad_norm": 0.6158263683319092,
+      "grad_norm": 0.6086987257003784,
       "learning_rate": 0.0003533556395035078,
-      "loss": 3.3557,
+      "loss": 3.3589,
       "step": 38200
     },
     {
       "epoch": 4.123989218328841,
-      "grad_norm": 0.688728928565979,
+      "grad_norm": 0.6515759825706482,
       "learning_rate": 0.00035303184025903934,
-      "loss": 3.3649,
+      "loss": 3.3668,
       "step": 38250
     },
     {
       "epoch": 4.129380053908355,
-      "grad_norm": 0.6338489055633545,
+      "grad_norm": 0.6497345566749573,
       "learning_rate": 0.00035270804101457094,
-      "loss": 3.3688,
+      "loss": 3.3741,
       "step": 38300
     },
     {
       "epoch": 4.1347708894878705,
-      "grad_norm": 0.6235848069190979,
+      "grad_norm": 0.6250614523887634,
       "learning_rate": 0.0003523842417701025,
-      "loss": 3.3523,
+      "loss": 3.3547,
       "step": 38350
     },
     {
       "epoch": 4.140161725067386,
-      "grad_norm": 0.6039944887161255,
+      "grad_norm": 0.5884137749671936,
       "learning_rate": 0.0003520604425256341,
-      "loss": 3.3666,
+      "loss": 3.3729,
       "step": 38400
     },
     {
       "epoch": 4.1455525606469,
-      "grad_norm": 0.6759158372879028,
+      "grad_norm": 0.6577835083007812,
       "learning_rate": 0.00035173664328116565,
-      "loss": 3.3516,
+      "loss": 3.3561,
       "step": 38450
     },
     {
       "epoch": 4.150943396226415,
-      "grad_norm": 0.6314166188240051,
+      "grad_norm": 0.5884889364242554,
       "learning_rate": 0.00035141284403669725,
-      "loss": 3.3835,
+      "loss": 3.3863,
       "step": 38500
     },
     {
       "epoch": 4.15633423180593,
-      "grad_norm": 0.6417436003684998,
+      "grad_norm": 0.6204766631126404,
       "learning_rate": 0.0003510890447922288,
-      "loss": 3.3594,
+      "loss": 3.3646,
       "step": 38550
     },
     {
       "epoch": 4.1617250673854445,
-      "grad_norm": 0.6634682416915894,
+      "grad_norm": 0.6417945623397827,
       "learning_rate": 0.00035076524554776035,
-      "loss": 3.368,
+      "loss": 3.3709,
       "step": 38600
     },
     {
       "epoch": 4.16711590296496,
-      "grad_norm": 0.6313776969909668,
+      "grad_norm": 0.6710885763168335,
       "learning_rate": 0.00035044144630329196,
-      "loss": 3.3669,
+      "loss": 3.3709,
       "step": 38650
     },
     {
       "epoch": 4.172506738544475,
-      "grad_norm": 0.616958737373352,
+      "grad_norm": 0.6059690713882446,
       "learning_rate": 0.00035011764705882346,
-      "loss": 3.3885,
+      "loss": 3.3935,
       "step": 38700
     },
     {
       "epoch": 4.177897574123989,
-      "grad_norm": 0.7168732285499573,
+      "grad_norm": 0.6476007699966431,
       "learning_rate": 0.0003497938478143551,
-      "loss": 3.3589,
+      "loss": 3.3636,
       "step": 38750
     },
     {
       "epoch": 4.183288409703504,
-      "grad_norm": 0.7030311226844788,
+      "grad_norm": 0.6382569670677185,
       "learning_rate": 0.0003494700485698866,
-      "loss": 3.3839,
+      "loss": 3.3884,
       "step": 38800
     },
     {
       "epoch": 4.188679245283019,
-      "grad_norm": 0.6451276540756226,
+      "grad_norm": 0.6265844106674194,
       "learning_rate": 0.0003491462493254182,
-      "loss": 3.374,
+      "loss": 3.3775,
       "step": 38850
     },
     {
       "epoch": 4.1940700808625335,
-      "grad_norm": 0.6581581234931946,
+      "grad_norm": 0.655836820602417,
       "learning_rate": 0.00034882245008094977,
-      "loss": 3.3815,
+      "loss": 3.3857,
       "step": 38900
     },
     {
       "epoch": 4.199460916442049,
-      "grad_norm": 0.6221701502799988,
+      "grad_norm": 0.6442790031433105,
       "learning_rate": 0.0003484986508364813,
-      "loss": 3.3756,
+      "loss": 3.3809,
       "step": 38950
     },
     {
       "epoch": 4.204851752021563,
-      "grad_norm": 0.7239381074905396,
+      "grad_norm": 0.6702572107315063,
       "learning_rate": 0.0003481748515920129,
-      "loss": 3.371,
+      "loss": 3.3744,
       "step": 39000
     },
     {
       "epoch": 4.204851752021563,
-      "eval_accuracy": 0.37740939136889706,
-      "eval_loss": 3.438514232635498,
-      "eval_runtime": 183.1338,
-      "eval_samples_per_second": 98.349,
-      "eval_steps_per_second": 6.149,
+      "eval_accuracy": 0.3769865143295231,
+      "eval_loss": 3.4400267601013184,
+      "eval_runtime": 181.6857,
+      "eval_samples_per_second": 99.133,
+      "eval_steps_per_second": 6.198,
       "step": 39000
     },
     {
       "epoch": 4.210242587601078,
-      "grad_norm": 0.6811919212341309,
+      "grad_norm": 0.6564778685569763,
       "learning_rate": 0.00034785105234754447,
-      "loss": 3.3755,
+      "loss": 3.3791,
       "step": 39050
     },
     {
       "epoch": 4.215633423180593,
-      "grad_norm": 0.6211300492286682,
+      "grad_norm": 0.6518337726593018,
       "learning_rate": 0.0003475272531030761,
-      "loss": 3.3774,
+      "loss": 3.3795,
       "step": 39100
     },
     {
       "epoch": 4.2210242587601075,
-      "grad_norm": 0.6302964091300964,
+      "grad_norm": 0.6220804452896118,
       "learning_rate": 0.00034720345385860763,
-      "loss": 3.3873,
+      "loss": 3.3917,
       "step": 39150
     },
     {
       "epoch": 4.226415094339623,
-      "grad_norm": 0.6861122250556946,
+      "grad_norm": 0.6643537878990173,
       "learning_rate": 0.00034687965461413923,
-      "loss": 3.3823,
+      "loss": 3.3843,
       "step": 39200
     },
     {
       "epoch": 4.231805929919138,
-      "grad_norm": 0.6081944108009338,
+      "grad_norm": 0.6522789597511292,
       "learning_rate": 0.0003465558553696708,
-      "loss": 3.38,
+      "loss": 3.3836,
       "step": 39250
     },
     {
       "epoch": 4.237196765498652,
-      "grad_norm": 0.6504591703414917,
+      "grad_norm": 0.6553415060043335,
       "learning_rate": 0.0003462320561252024,
-      "loss": 3.375,
+      "loss": 3.3799,
       "step": 39300
     },
     {
       "epoch": 4.242587601078167,
-      "grad_norm": 0.6743220090866089,
+      "grad_norm": 0.6793872714042664,
       "learning_rate": 0.00034590825688073394,
-      "loss": 3.3905,
+      "loss": 3.3937,
       "step": 39350
     },
     {
       "epoch": 4.247978436657682,
-      "grad_norm": 0.6235594749450684,
-      "learning_rate": 0.0003455909336211549,
-      "loss": 3.3754,
+      "grad_norm": 0.6650474071502686,
+      "learning_rate": 0.00034558445763626543,
+      "loss": 3.3788,
       "step": 39400
     },
     {
       "epoch": 4.2533692722371965,
-      "grad_norm": 0.6357383728027344,
-      "learning_rate": 0.0003452671343766864,
-      "loss": 3.3785,
+      "grad_norm": 0.6935107111930847,
+      "learning_rate": 0.00034526065839179704,
+      "loss": 3.3817,
       "step": 39450
     },
     {
       "epoch": 4.258760107816712,
-      "grad_norm": 0.6747174859046936,
-      "learning_rate": 0.000344943335132218,
-      "loss": 3.377,
+      "grad_norm": 0.6611354351043701,
+      "learning_rate": 0.0003449368591473286,
+      "loss": 3.3828,
       "step": 39500
     },
     {
       "epoch": 4.264150943396227,
-      "grad_norm": 0.6351191997528076,
-      "learning_rate": 0.00034461953588774954,
-      "loss": 3.3805,
+      "grad_norm": 0.6441178321838379,
+      "learning_rate": 0.0003446130599028602,
+      "loss": 3.3833,
       "step": 39550
     },
     {
       "epoch": 4.269541778975741,
-      "grad_norm": 0.5940690636634827,
-      "learning_rate": 0.00034429573664328115,
-      "loss": 3.3745,
+      "grad_norm": 0.6164142489433289,
+      "learning_rate": 0.00034428926065839174,
+      "loss": 3.3798,
       "step": 39600
     },
     {
       "epoch": 4.274932614555256,
-      "grad_norm": 0.5747135281562805,
-      "learning_rate": 0.0003439719373988127,
-      "loss": 3.3735,
+      "grad_norm": 0.6198307871818542,
+      "learning_rate": 0.00034396546141392335,
+      "loss": 3.3748,
       "step": 39650
     },
     {
       "epoch": 4.280323450134771,
-      "grad_norm": 0.6357155442237854,
-      "learning_rate": 0.00034364813815434425,
-      "loss": 3.3891,
+      "grad_norm": 0.6396060585975647,
+      "learning_rate": 0.0003436416621694549,
+      "loss": 3.3917,
       "step": 39700
     },
     {
       "epoch": 4.285714285714286,
-      "grad_norm": 0.6541160345077515,
-      "learning_rate": 0.00034332433890987585,
-      "loss": 3.3779,
+      "grad_norm": 0.621781587600708,
+      "learning_rate": 0.0003433178629249865,
+      "loss": 3.3808,
       "step": 39750
     },
     {
       "epoch": 4.291105121293801,
-      "grad_norm": 0.6745419502258301,
-      "learning_rate": 0.0003430005396654074,
-      "loss": 3.384,
+      "grad_norm": 0.6394072771072388,
+      "learning_rate": 0.00034299406368051806,
+      "loss": 3.3854,
       "step": 39800
     },
     {
       "epoch": 4.296495956873315,
-      "grad_norm": 0.641961395740509,
-      "learning_rate": 0.000342676740420939,
-      "loss": 3.3535,
+      "grad_norm": 0.6566154956817627,
+      "learning_rate": 0.0003426702644360496,
+      "loss": 3.3563,
       "step": 39850
     },
     {
       "epoch": 4.30188679245283,
-      "grad_norm": 0.6052359342575073,
-      "learning_rate": 0.00034235294117647056,
-      "loss": 3.3919,
+      "grad_norm": 0.6247180104255676,
+      "learning_rate": 0.0003423464651915812,
+      "loss": 3.3958,
       "step": 39900
     },
     {
       "epoch": 4.307277628032345,
-      "grad_norm": 0.6415327787399292,
-      "learning_rate": 0.00034202914193200216,
-      "loss": 3.3692,
+      "grad_norm": 0.6729653477668762,
+      "learning_rate": 0.00034202266594711276,
+      "loss": 3.3724,
       "step": 39950
     },
     {
       "epoch": 4.3126684636118595,
-      "grad_norm": 0.6692045331001282,
-      "learning_rate": 0.0003417053426875337,
-      "loss": 3.3901,
+      "grad_norm": 0.6515873670578003,
+      "learning_rate": 0.00034169886670264437,
+      "loss": 3.393,
       "step": 40000
     },
     {
       "epoch": 4.3126684636118595,
-      "eval_accuracy": 0.3779461366347006,
-      "eval_loss": 3.4338619709014893,
-      "eval_runtime": 183.59,
-      "eval_samples_per_second": 98.104,
-      "eval_steps_per_second": 6.133,
+      "eval_accuracy": 0.3776187654837156,
+      "eval_loss": 3.435962200164795,
+      "eval_runtime": 181.74,
+      "eval_samples_per_second": 99.103,
+      "eval_steps_per_second": 6.196,
       "step": 40000
     },
     {
       "epoch": 4.318059299191375,
-      "grad_norm": 0.659393846988678,
-      "learning_rate": 0.0003413815434430653,
-      "loss": 3.4026,
+      "grad_norm": 0.6128860116004944,
+      "learning_rate": 0.00034137506745817586,
+      "loss": 3.4036,
       "step": 40050
     },
     {
       "epoch": 4.32345013477089,
-      "grad_norm": 0.6064386963844299,
-      "learning_rate": 0.0003410577441985968,
-      "loss": 3.3698,
+      "grad_norm": 0.6145449280738831,
+      "learning_rate": 0.0003410512682137075,
+      "loss": 3.372,
       "step": 40100
     },
     {
       "epoch": 4.328840970350404,
-      "grad_norm": 0.6234272718429565,
-      "learning_rate": 0.00034073394495412837,
-      "loss": 3.3865,
+      "grad_norm": 0.6391635537147522,
+      "learning_rate": 0.000340727468969239,
+      "loss": 3.3898,
       "step": 40150
     },
     {
       "epoch": 4.334231805929919,
-      "grad_norm": 0.6467458009719849,
-      "learning_rate": 0.00034041014570965997,
-      "loss": 3.4039,
+      "grad_norm": 0.625508725643158,
+      "learning_rate": 0.0003404036697247706,
+      "loss": 3.4076,
       "step": 40200
     },
     {
       "epoch": 4.339622641509434,
-      "grad_norm": 0.5988481640815735,
-      "learning_rate": 0.0003400863464651915,
-      "loss": 3.3805,
+      "grad_norm": 0.608378529548645,
+      "learning_rate": 0.0003400798704803022,
+      "loss": 3.386,
       "step": 40250
     },
     {
       "epoch": 4.345013477088949,
-      "grad_norm": 0.660705029964447,
-      "learning_rate": 0.0003397625472207231,
-      "loss": 3.3948,
+      "grad_norm": 0.6230303645133972,
+      "learning_rate": 0.0003397560712358337,
+      "loss": 3.3976,
       "step": 40300
     },
     {
       "epoch": 4.350404312668464,
-      "grad_norm": 0.6406025886535645,
-      "learning_rate": 0.0003394387479762547,
-      "loss": 3.401,
+      "grad_norm": 0.664487898349762,
+      "learning_rate": 0.00033943227199136533,
+      "loss": 3.4035,
       "step": 40350
     },
     {
       "epoch": 4.355795148247978,
-      "grad_norm": 0.6672471761703491,
-      "learning_rate": 0.0003391149487317863,
-      "loss": 3.3882,
+      "grad_norm": 0.6623818278312683,
+      "learning_rate": 0.0003391084727468969,
+      "loss": 3.3918,
       "step": 40400
     },
     {
       "epoch": 4.361185983827493,
-      "grad_norm": 0.6221401691436768,
-      "learning_rate": 0.00033879114948731783,
-      "loss": 3.3945,
+      "grad_norm": 0.6926326751708984,
+      "learning_rate": 0.0003387846735024285,
+      "loss": 3.3979,
       "step": 40450
     },
     {
       "epoch": 4.366576819407008,
-      "grad_norm": 0.6297544240951538,
-      "learning_rate": 0.00033846735024284944,
-      "loss": 3.3631,
+      "grad_norm": 0.6583217978477478,
+      "learning_rate": 0.00033846087425796003,
+      "loss": 3.3663,
       "step": 40500
     },
     {
       "epoch": 4.3719676549865225,
-      "grad_norm": 0.6644108295440674,
-      "learning_rate": 0.000338143550998381,
-      "loss": 3.3732,
+      "grad_norm": 0.6462557911872864,
+      "learning_rate": 0.00033813707501349164,
+      "loss": 3.3772,
       "step": 40550
     },
     {
       "epoch": 4.377358490566038,
-      "grad_norm": 0.633611798286438,
-      "learning_rate": 0.00033781975175391254,
-      "loss": 3.3804,
+      "grad_norm": 0.6221892833709717,
+      "learning_rate": 0.0003378132757690232,
+      "loss": 3.3811,
       "step": 40600
     },
     {
       "epoch": 4.382749326145553,
-      "grad_norm": 0.641505241394043,
-      "learning_rate": 0.00033749595250944414,
-      "loss": 3.3789,
+      "grad_norm": 0.6604007482528687,
+      "learning_rate": 0.00033748947652455474,
+      "loss": 3.3833,
       "step": 40650
     },
     {
       "epoch": 4.388140161725067,
-      "grad_norm": 0.6398807764053345,
-      "learning_rate": 0.00033717215326497564,
-      "loss": 3.3847,
+      "grad_norm": 0.648074746131897,
+      "learning_rate": 0.00033716567728008635,
+      "loss": 3.3892,
       "step": 40700
     },
     {
       "epoch": 4.393530997304582,
-      "grad_norm": 0.6694504022598267,
-      "learning_rate": 0.0003368483540205073,
-      "loss": 3.3853,
+      "grad_norm": 0.6675060987472534,
+      "learning_rate": 0.00033684187803561784,
+      "loss": 3.3868,
       "step": 40750
     },
     {
       "epoch": 4.398921832884097,
-      "grad_norm": 0.6023000478744507,
-      "learning_rate": 0.0003365245547760388,
-      "loss": 3.3843,
+      "grad_norm": 0.5989810824394226,
+      "learning_rate": 0.00033651807879114945,
+      "loss": 3.3891,
       "step": 40800
     },
     {
       "epoch": 4.404312668463612,
-      "grad_norm": 0.6796262860298157,
-      "learning_rate": 0.0003362007555315704,
-      "loss": 3.3754,
+      "grad_norm": 0.6883215308189392,
+      "learning_rate": 0.000336194279546681,
+      "loss": 3.3793,
       "step": 40850
     },
     {
       "epoch": 4.409703504043127,
-      "grad_norm": 0.6454067230224609,
-      "learning_rate": 0.00033587695628710195,
-      "loss": 3.3825,
+      "grad_norm": 0.6383180618286133,
+      "learning_rate": 0.0003358704803022126,
+      "loss": 3.3858,
       "step": 40900
     },
     {
       "epoch": 4.415094339622642,
-      "grad_norm": 0.6159910559654236,
-      "learning_rate": 0.00033555315704263355,
-      "loss": 3.3901,
+      "grad_norm": 0.6031454801559448,
+      "learning_rate": 0.00033554668105774415,
+      "loss": 3.3936,
       "step": 40950
     },
     {
       "epoch": 4.420485175202156,
-      "grad_norm": 0.6237287521362305,
-      "learning_rate": 0.0003352293577981651,
-      "loss": 3.3831,
+      "grad_norm": 0.6764903664588928,
+      "learning_rate": 0.00033522288181327576,
+      "loss": 3.3865,
       "step": 41000
     },
     {
       "epoch": 4.420485175202156,
-      "eval_accuracy": 0.378417364209152,
-      "eval_loss": 3.4264814853668213,
-      "eval_runtime": 183.3491,
-      "eval_samples_per_second": 98.233,
-      "eval_steps_per_second": 6.141,
+      "eval_accuracy": 0.3779634124438631,
+      "eval_loss": 3.4306559562683105,
+      "eval_runtime": 184.1481,
+      "eval_samples_per_second": 97.807,
+      "eval_steps_per_second": 6.115,
       "step": 41000
     },
     {
       "epoch": 4.425876010781671,
-      "grad_norm": 0.6706305742263794,
-      "learning_rate": 0.00033490555855369665,
-      "loss": 3.3701,
+      "grad_norm": 0.635355532169342,
+      "learning_rate": 0.0003348990825688073,
+      "loss": 3.3737,
       "step": 41050
     },
     {
       "epoch": 4.431266846361186,
-      "grad_norm": 0.6600586175918579,
-      "learning_rate": 0.00033458175930922826,
-      "loss": 3.3875,
+      "grad_norm": 0.6517626047134399,
+      "learning_rate": 0.00033457528332433886,
+      "loss": 3.3894,
       "step": 41100
     },
     {
       "epoch": 4.436657681940701,
-      "grad_norm": 0.5768634080886841,
-      "learning_rate": 0.0003342579600647598,
-      "loss": 3.383,
+      "grad_norm": 0.6000182628631592,
+      "learning_rate": 0.00033425148407987046,
+      "loss": 3.386,
       "step": 41150
     },
     {
       "epoch": 4.442048517520216,
-      "grad_norm": 0.6009304523468018,
-      "learning_rate": 0.0003339341608202914,
-      "loss": 3.3686,
+      "grad_norm": 0.6046862602233887,
+      "learning_rate": 0.000333927684835402,
+      "loss": 3.3711,
       "step": 41200
     },
     {
       "epoch": 4.44743935309973,
-      "grad_norm": 0.6819839477539062,
-      "learning_rate": 0.00033361036157582297,
-      "loss": 3.3808,
+      "grad_norm": 0.6689609289169312,
+      "learning_rate": 0.0003336038855909336,
+      "loss": 3.3847,
       "step": 41250
     },
     {
       "epoch": 4.452830188679245,
-      "grad_norm": 0.63370281457901,
-      "learning_rate": 0.00033328656233135457,
-      "loss": 3.3742,
+      "grad_norm": 0.639434278011322,
+      "learning_rate": 0.00033328008634646517,
+      "loss": 3.3765,
       "step": 41300
     },
     {
       "epoch": 4.45822102425876,
-      "grad_norm": 0.6805122494697571,
-      "learning_rate": 0.0003329627630868861,
-      "loss": 3.3999,
+      "grad_norm": 0.6322727203369141,
+      "learning_rate": 0.0003329562871019968,
+      "loss": 3.404,
       "step": 41350
     },
     {
       "epoch": 4.463611859838275,
-      "grad_norm": 0.6374489068984985,
-      "learning_rate": 0.0003326454398273071,
-      "loss": 3.3923,
+      "grad_norm": 0.6922193169593811,
+      "learning_rate": 0.00033263248785752827,
+      "loss": 3.3953,
       "step": 41400
     },
     {
       "epoch": 4.46900269541779,
-      "grad_norm": 0.632504403591156,
-      "learning_rate": 0.00033232164058283857,
-      "loss": 3.3802,
+      "grad_norm": 0.6255041360855103,
+      "learning_rate": 0.00033230868861305993,
+      "loss": 3.3834,
       "step": 41450
     },
     {
       "epoch": 4.474393530997305,
-      "grad_norm": 0.6276558637619019,
-      "learning_rate": 0.0003319978413383702,
-      "loss": 3.371,
+      "grad_norm": 0.6385671496391296,
+      "learning_rate": 0.0003319848893685914,
+      "loss": 3.3717,
       "step": 41500
     },
     {
       "epoch": 4.479784366576819,
-      "grad_norm": 0.6790854334831238,
-      "learning_rate": 0.0003316740420939017,
-      "loss": 3.3857,
+      "grad_norm": 0.6567233800888062,
+      "learning_rate": 0.000331661090124123,
+      "loss": 3.3882,
       "step": 41550
     },
     {
       "epoch": 4.485175202156334,
-      "grad_norm": 0.6570220589637756,
-      "learning_rate": 0.00033135024284943333,
-      "loss": 3.3908,
+      "grad_norm": 0.6391835808753967,
+      "learning_rate": 0.0003313372908796546,
+      "loss": 3.3943,
       "step": 41600
     },
     {
       "epoch": 4.490566037735849,
-      "grad_norm": 0.6306306719779968,
-      "learning_rate": 0.0003310264436049649,
-      "loss": 3.3758,
+      "grad_norm": 0.6709011197090149,
+      "learning_rate": 0.00033101349163518613,
+      "loss": 3.3781,
       "step": 41650
     },
     {
       "epoch": 4.495956873315364,
-      "grad_norm": 0.6574743986129761,
-      "learning_rate": 0.0003307026443604965,
-      "loss": 3.3988,
+      "grad_norm": 0.6142731308937073,
+      "learning_rate": 0.0003306961683756071,
+      "loss": 3.4012,
       "step": 41700
     },
     {
       "epoch": 4.501347708894879,
-      "grad_norm": 0.6541933417320251,
-      "learning_rate": 0.00033037884511602804,
-      "loss": 3.379,
+      "grad_norm": 0.6658247113227844,
+      "learning_rate": 0.0003303723691311387,
+      "loss": 3.383,
       "step": 41750
     },
     {
       "epoch": 4.506738544474393,
-      "grad_norm": 0.6046768426895142,
-      "learning_rate": 0.0003300550458715596,
-      "loss": 3.3775,
+      "grad_norm": 0.635217547416687,
+      "learning_rate": 0.00033004856988667024,
+      "loss": 3.3811,
       "step": 41800
     },
     {
       "epoch": 4.512129380053908,
-      "grad_norm": 0.6767274737358093,
-      "learning_rate": 0.0003297312466270912,
-      "loss": 3.3878,
+      "grad_norm": 0.6755160689353943,
+      "learning_rate": 0.0003297247706422018,
+      "loss": 3.3906,
       "step": 41850
     },
     {
       "epoch": 4.517520215633423,
-      "grad_norm": 0.6689441204071045,
-      "learning_rate": 0.00032940744738262274,
-      "loss": 3.3872,
+      "grad_norm": 0.6707085371017456,
+      "learning_rate": 0.0003294009713977334,
+      "loss": 3.3915,
       "step": 41900
     },
     {
       "epoch": 4.5229110512129385,
-      "grad_norm": 0.6398838758468628,
-      "learning_rate": 0.00032908364813815435,
-      "loss": 3.3898,
+      "grad_norm": 0.6344310641288757,
+      "learning_rate": 0.00032907717215326494,
+      "loss": 3.391,
       "step": 41950
     },
     {
       "epoch": 4.528301886792453,
-      "grad_norm": 0.6058347225189209,
-      "learning_rate": 0.0003287598488936859,
-      "loss": 3.3797,
+      "grad_norm": 0.6044669151306152,
+      "learning_rate": 0.00032875337290879655,
+      "loss": 3.3815,
       "step": 42000
     },
     {
       "epoch": 4.528301886792453,
-      "eval_accuracy": 0.37927583067577425,
-      "eval_loss": 3.4238638877868652,
-      "eval_runtime": 183.4895,
-      "eval_samples_per_second": 98.158,
-      "eval_steps_per_second": 6.137,
+      "eval_accuracy": 0.3791368636322595,
+      "eval_loss": 3.4250853061676025,
+      "eval_runtime": 184.5525,
+      "eval_samples_per_second": 97.593,
+      "eval_steps_per_second": 6.101,
       "step": 42000
     },
     {
       "epoch": 4.533692722371968,
-      "grad_norm": 0.5938351154327393,
-      "learning_rate": 0.0003284360496492175,
-      "loss": 3.3765,
+      "grad_norm": 0.6335576176643372,
+      "learning_rate": 0.00032842957366432805,
+      "loss": 3.3803,
       "step": 42050
     },
     {
       "epoch": 4.539083557951482,
-      "grad_norm": 0.6558327674865723,
-      "learning_rate": 0.000328112250404749,
-      "loss": 3.4059,
+      "grad_norm": 0.6365965008735657,
+      "learning_rate": 0.0003281057744198597,
+      "loss": 3.4088,
       "step": 42100
     },
     {
       "epoch": 4.544474393530997,
-      "grad_norm": 0.64605313539505,
-      "learning_rate": 0.00032778845116028066,
-      "loss": 3.3917,
+      "grad_norm": 0.6515207886695862,
+      "learning_rate": 0.0003277819751753912,
+      "loss": 3.3944,
       "step": 42150
     },
     {
       "epoch": 4.549865229110512,
-      "grad_norm": 0.5833234190940857,
-      "learning_rate": 0.00032746465191581215,
-      "loss": 3.3712,
+      "grad_norm": 0.6068034172058105,
+      "learning_rate": 0.0003274581759309228,
+      "loss": 3.3743,
       "step": 42200
     },
     {
       "epoch": 4.555256064690027,
-      "grad_norm": 0.822323203086853,
+      "grad_norm": 0.8428053855895996,
       "learning_rate": 0.0003271408526713437,
-      "loss": 3.3953,
+      "loss": 3.3997,
       "step": 42250
     },
     {
       "epoch": 4.560646900269542,
-      "grad_norm": 0.6433358192443848,
+      "grad_norm": 0.6403253078460693,
       "learning_rate": 0.0003268170534268753,
-      "loss": 3.3939,
+      "loss": 3.3963,
       "step": 42300
     },
     {
       "epoch": 4.566037735849057,
-      "grad_norm": 0.7177082300186157,
+      "grad_norm": 0.6786686778068542,
       "learning_rate": 0.00032649325418240686,
-      "loss": 3.3757,
+      "loss": 3.378,
       "step": 42350
     },
     {
       "epoch": 4.571428571428571,
-      "grad_norm": 0.6373773217201233,
+      "grad_norm": 0.6787482500076294,
       "learning_rate": 0.00032616945493793846,
-      "loss": 3.4017,
+      "loss": 3.4033,
       "step": 42400
     },
     {
       "epoch": 4.576819407008086,
-      "grad_norm": 0.6122283935546875,
+      "grad_norm": 0.6195935606956482,
       "learning_rate": 0.00032584565569347,
-      "loss": 3.3703,
+      "loss": 3.372,
       "step": 42450
     },
     {
       "epoch": 4.5822102425876015,
-      "grad_norm": 0.6813517212867737,
+      "grad_norm": 0.6464322805404663,
       "learning_rate": 0.0003255218564490016,
-      "loss": 3.3709,
+      "loss": 3.3725,
       "step": 42500
     },
     {
       "epoch": 4.587601078167116,
-      "grad_norm": 0.6224530935287476,
+      "grad_norm": 0.6381585001945496,
       "learning_rate": 0.00032519805720453317,
-      "loss": 3.3845,
+      "loss": 3.3863,
       "step": 42550
     },
     {
       "epoch": 4.592991913746631,
-      "grad_norm": 0.6742523312568665,
+      "grad_norm": 0.6735817790031433,
       "learning_rate": 0.0003248742579600647,
-      "loss": 3.3823,
+      "loss": 3.3856,
       "step": 42600
     },
     {
       "epoch": 4.598382749326145,
-      "grad_norm": 0.7158631682395935,
+      "grad_norm": 0.7214176654815674,
       "learning_rate": 0.0003245504587155963,
-      "loss": 3.3956,
+      "loss": 3.3971,
       "step": 42650
     },
     {
       "epoch": 4.60377358490566,
-      "grad_norm": 0.6356601715087891,
+      "grad_norm": 0.6393851637840271,
       "learning_rate": 0.0003242266594711278,
-      "loss": 3.3887,
+      "loss": 3.3914,
       "step": 42700
     },
     {
       "epoch": 4.609164420485175,
-      "grad_norm": 0.6447049975395203,
+      "grad_norm": 0.6101807355880737,
       "learning_rate": 0.0003239028602266595,
-      "loss": 3.3758,
+      "loss": 3.3803,
       "step": 42750
     },
     {
       "epoch": 4.6145552560646905,
-      "grad_norm": 0.6369867324829102,
+      "grad_norm": 0.6338185667991638,
       "learning_rate": 0.000323579060982191,
-      "loss": 3.3985,
+      "loss": 3.4001,
       "step": 42800
     },
     {
       "epoch": 4.619946091644205,
-      "grad_norm": 0.6150829195976257,
+      "grad_norm": 0.625145673751831,
       "learning_rate": 0.0003232552617377226,
-      "loss": 3.3909,
+      "loss": 3.3948,
       "step": 42850
     },
     {
       "epoch": 4.62533692722372,
-      "grad_norm": 0.5987628102302551,
+      "grad_norm": 0.6058762073516846,
       "learning_rate": 0.00032293146249325413,
-      "loss": 3.378,
+      "loss": 3.382,
       "step": 42900
     },
     {
       "epoch": 4.630727762803234,
-      "grad_norm": 0.6027454733848572,
+      "grad_norm": 0.6360901594161987,
       "learning_rate": 0.00032260766324878574,
-      "loss": 3.382,
+      "loss": 3.3865,
       "step": 42950
     },
     {
       "epoch": 4.636118598382749,
-      "grad_norm": 0.6479946374893188,
+      "grad_norm": 0.6292373538017273,
       "learning_rate": 0.0003222838640043173,
-      "loss": 3.3968,
+      "loss": 3.4003,
       "step": 43000
     },
     {
       "epoch": 4.636118598382749,
-      "eval_accuracy": 0.3795472455895348,
-      "eval_loss": 3.419617176055908,
-      "eval_runtime": 183.1797,
-      "eval_samples_per_second": 98.324,
-      "eval_steps_per_second": 6.147,
+      "eval_accuracy": 0.37946651649388863,
+      "eval_loss": 3.4206013679504395,
+      "eval_runtime": 181.2518,
+      "eval_samples_per_second": 99.37,
+      "eval_steps_per_second": 6.212,
       "step": 43000
     },
     {
       "epoch": 4.6415094339622645,
-      "grad_norm": 0.6356445550918579,
+      "grad_norm": 0.6332093477249146,
       "learning_rate": 0.00032196006475984884,
-      "loss": 3.3766,
+      "loss": 3.3808,
       "step": 43050
     },
     {
       "epoch": 4.646900269541779,
-      "grad_norm": 0.6246060132980347,
+      "grad_norm": 0.7153672575950623,
       "learning_rate": 0.00032163626551538044,
-      "loss": 3.3818,
+      "loss": 3.3831,
       "step": 43100
     },
     {
       "epoch": 4.652291105121294,
-      "grad_norm": 0.6289001107215881,
+      "grad_norm": 0.6388234496116638,
       "learning_rate": 0.000321312466270912,
-      "loss": 3.3773,
+      "loss": 3.3789,
       "step": 43150
     },
     {
       "epoch": 4.657681940700809,
-      "grad_norm": 0.6760928630828857,
+      "grad_norm": 0.6893576383590698,
       "learning_rate": 0.0003209886670264436,
-      "loss": 3.3847,
+      "loss": 3.3877,
       "step": 43200
     },
     {
       "epoch": 4.663072776280323,
-      "grad_norm": 0.6528955698013306,
+      "grad_norm": 0.6786460280418396,
       "learning_rate": 0.00032066486778197515,
-      "loss": 3.3837,
+      "loss": 3.3859,
       "step": 43250
     },
     {
       "epoch": 4.668463611859838,
-      "grad_norm": 0.6557238101959229,
+      "grad_norm": 0.6445801258087158,
       "learning_rate": 0.00032034106853750675,
-      "loss": 3.3919,
+      "loss": 3.3949,
       "step": 43300
     },
     {
       "epoch": 4.6738544474393535,
-      "grad_norm": 0.6291058659553528,
+      "grad_norm": 0.6880573630332947,
       "learning_rate": 0.0003200172692930383,
-      "loss": 3.3674,
+      "loss": 3.3714,
       "step": 43350
     },
     {
       "epoch": 4.679245283018868,
-      "grad_norm": 0.7018590569496155,
+      "grad_norm": 0.6739255785942078,
       "learning_rate": 0.0003196934700485699,
-      "loss": 3.377,
+      "loss": 3.3789,
       "step": 43400
     },
     {
       "epoch": 4.684636118598383,
-      "grad_norm": 0.6758108139038086,
-      "learning_rate": 0.00031937614678899075,
-      "loss": 3.3833,
+      "grad_norm": 0.677082359790802,
+      "learning_rate": 0.0003193696708041014,
+      "loss": 3.3876,
       "step": 43450
     },
     {
       "epoch": 4.690026954177897,
-      "grad_norm": 0.6351593136787415,
-      "learning_rate": 0.00031905234754452236,
-      "loss": 3.383,
+      "grad_norm": 0.629194438457489,
+      "learning_rate": 0.00031904587155963296,
+      "loss": 3.3863,
       "step": 43500
     },
     {
       "epoch": 4.695417789757412,
-      "grad_norm": 0.639167308807373,
-      "learning_rate": 0.0003187285483000539,
-      "loss": 3.3697,
+      "grad_norm": 0.6427448987960815,
+      "learning_rate": 0.00031872207231516456,
+      "loss": 3.3735,
       "step": 43550
     },
     {
       "epoch": 4.7008086253369274,
-      "grad_norm": 0.6248657703399658,
-      "learning_rate": 0.0003184047490555855,
-      "loss": 3.3745,
+      "grad_norm": 0.6634519100189209,
+      "learning_rate": 0.0003183982730706961,
+      "loss": 3.3778,
       "step": 43600
     },
     {
       "epoch": 4.706199460916442,
-      "grad_norm": 0.633437991142273,
-      "learning_rate": 0.00031808094981111706,
-      "loss": 3.3931,
+      "grad_norm": 0.6442071199417114,
+      "learning_rate": 0.0003180744738262277,
+      "loss": 3.3963,
       "step": 43650
     },
     {
       "epoch": 4.711590296495957,
-      "grad_norm": 0.627947211265564,
-      "learning_rate": 0.00031775715056664867,
-      "loss": 3.384,
+      "grad_norm": 0.632140576839447,
+      "learning_rate": 0.00031775067458175927,
+      "loss": 3.3843,
       "step": 43700
     },
     {
       "epoch": 4.716981132075472,
-      "grad_norm": 0.6561819911003113,
-      "learning_rate": 0.0003174333513221802,
-      "loss": 3.3817,
+      "grad_norm": 0.6474012136459351,
+      "learning_rate": 0.00031742687533729087,
+      "loss": 3.3846,
       "step": 43750
     },
     {
       "epoch": 4.722371967654986,
-      "grad_norm": 0.6273093819618225,
-      "learning_rate": 0.00031710955207771177,
-      "loss": 3.3727,
+      "grad_norm": 0.6319801807403564,
+      "learning_rate": 0.0003171030760928224,
+      "loss": 3.3775,
       "step": 43800
     },
     {
       "epoch": 4.727762803234501,
-      "grad_norm": 0.6125487089157104,
-      "learning_rate": 0.0003167857528332434,
-      "loss": 3.3952,
+      "grad_norm": 0.6332365870475769,
+      "learning_rate": 0.000316779276848354,
+      "loss": 3.3972,
       "step": 43850
     },
     {
       "epoch": 4.7331536388140165,
-      "grad_norm": 0.6168906092643738,
-      "learning_rate": 0.0003164619535887749,
-      "loss": 3.4017,
+      "grad_norm": 0.6384433507919312,
+      "learning_rate": 0.0003164554776038856,
+      "loss": 3.4054,
       "step": 43900
     },
     {
       "epoch": 4.738544474393531,
-      "grad_norm": 0.631703794002533,
-      "learning_rate": 0.00031613815434430653,
-      "loss": 3.4001,
+      "grad_norm": 0.6466211080551147,
+      "learning_rate": 0.00031613167835941713,
+      "loss": 3.4035,
       "step": 43950
     },
     {
       "epoch": 4.743935309973046,
-      "grad_norm": 0.6847707629203796,
-      "learning_rate": 0.0003158143550998381,
-      "loss": 3.3652,
+      "grad_norm": 0.6958954334259033,
+      "learning_rate": 0.00031580787911494873,
+      "loss": 3.3676,
       "step": 44000
     },
     {
       "epoch": 4.743935309973046,
-      "eval_accuracy": 0.38033150213377975,
-      "eval_loss": 3.4119884967803955,
-      "eval_runtime": 183.7579,
-      "eval_samples_per_second": 98.015,
-      "eval_steps_per_second": 6.128,
+      "eval_accuracy": 0.38001727798222024,
+      "eval_loss": 3.415374517440796,
+      "eval_runtime": 181.445,
+      "eval_samples_per_second": 99.264,
+      "eval_steps_per_second": 6.206,
       "step": 44000
     },
     {
       "epoch": 4.74932614555256,
-      "grad_norm": 0.6523143649101257,
-      "learning_rate": 0.0003154905558553697,
-      "loss": 3.3999,
+      "grad_norm": 0.6475406289100647,
+      "learning_rate": 0.00031548407987048023,
+      "loss": 3.4047,
       "step": 44050
     },
     {
       "epoch": 4.754716981132075,
-      "grad_norm": 0.6526023149490356,
-      "learning_rate": 0.0003151667566109012,
-      "loss": 3.3786,
+      "grad_norm": 0.648854672908783,
+      "learning_rate": 0.0003151602806260119,
+      "loss": 3.3804,
       "step": 44100
     },
     {
       "epoch": 4.7601078167115904,
-      "grad_norm": 0.6396304368972778,
-      "learning_rate": 0.00031484295736643284,
-      "loss": 3.3723,
+      "grad_norm": 0.6378458738327026,
+      "learning_rate": 0.0003148364813815434,
+      "loss": 3.3753,
       "step": 44150
     },
     {
       "epoch": 4.765498652291106,
-      "grad_norm": 0.6429187059402466,
-      "learning_rate": 0.00031451915812196434,
-      "loss": 3.3932,
+      "grad_norm": 0.6204406023025513,
+      "learning_rate": 0.000314512682137075,
+      "loss": 3.3965,
       "step": 44200
     },
     {
       "epoch": 4.77088948787062,
-      "grad_norm": 0.6210240721702576,
-      "learning_rate": 0.0003141953588774959,
-      "loss": 3.3862,
+      "grad_norm": 0.6212044358253479,
+      "learning_rate": 0.00031418888289260654,
+      "loss": 3.3889,
       "step": 44250
     },
     {
       "epoch": 4.776280323450135,
-      "grad_norm": 0.6462406516075134,
-      "learning_rate": 0.0003138715596330275,
-      "loss": 3.3807,
+      "grad_norm": 0.6488358974456787,
+      "learning_rate": 0.0003138650836481381,
+      "loss": 3.3836,
       "step": 44300
     },
     {
       "epoch": 4.781671159029649,
-      "grad_norm": 0.6584225296974182,
-      "learning_rate": 0.00031354776038855904,
-      "loss": 3.3916,
+      "grad_norm": 0.6549113988876343,
+      "learning_rate": 0.0003135412844036697,
+      "loss": 3.3948,
       "step": 44350
     },
     {
       "epoch": 4.787061994609164,
-      "grad_norm": 0.6552765369415283,
-      "learning_rate": 0.00031322396114409065,
-      "loss": 3.3723,
+      "grad_norm": 0.619624137878418,
+      "learning_rate": 0.00031321748515920124,
+      "loss": 3.3753,
       "step": 44400
     },
     {
       "epoch": 4.7924528301886795,
-      "grad_norm": 0.6333467364311218,
-      "learning_rate": 0.0003129001618996222,
-      "loss": 3.39,
+      "grad_norm": 0.6465790867805481,
+      "learning_rate": 0.00031289368591473285,
+      "loss": 3.3919,
       "step": 44450
     },
     {
       "epoch": 4.797843665768194,
-      "grad_norm": 0.6195804476737976,
-      "learning_rate": 0.0003125763626551538,
-      "loss": 3.3721,
+      "grad_norm": 0.6547673940658569,
+      "learning_rate": 0.0003125698866702644,
+      "loss": 3.3745,
       "step": 44500
     },
     {
       "epoch": 4.803234501347709,
-      "grad_norm": 0.6451464891433716,
-      "learning_rate": 0.00031225256341068535,
-      "loss": 3.3719,
+      "grad_norm": 0.6841105222702026,
+      "learning_rate": 0.000312246087425796,
+      "loss": 3.3729,
       "step": 44550
     },
     {
       "epoch": 4.808625336927224,
-      "grad_norm": 0.6498815417289734,
-      "learning_rate": 0.00031192876416621696,
-      "loss": 3.3853,
+      "grad_norm": 0.6424994468688965,
+      "learning_rate": 0.00031192228818132756,
+      "loss": 3.3885,
       "step": 44600
     },
     {
       "epoch": 4.814016172506738,
-      "grad_norm": 0.6606132984161377,
-      "learning_rate": 0.0003116049649217485,
-      "loss": 3.375,
+      "grad_norm": 0.6680660247802734,
+      "learning_rate": 0.00031159848893685916,
+      "loss": 3.377,
       "step": 44650
     },
     {
       "epoch": 4.819407008086253,
-      "grad_norm": 0.6065442562103271,
-      "learning_rate": 0.00031128116567728,
-      "loss": 3.3695,
+      "grad_norm": 0.6277469396591187,
+      "learning_rate": 0.0003112746896923907,
+      "loss": 3.3709,
       "step": 44700
     },
     {
       "epoch": 4.824797843665769,
-      "grad_norm": 0.6994802355766296,
-      "learning_rate": 0.00031095736643281166,
-      "loss": 3.3718,
+      "grad_norm": 0.7602716088294983,
+      "learning_rate": 0.0003109508904479222,
+      "loss": 3.373,
       "step": 44750
     },
     {
       "epoch": 4.830188679245283,
-      "grad_norm": 0.6376444697380066,
-      "learning_rate": 0.00031063356718834316,
-      "loss": 3.3855,
+      "grad_norm": 0.6440214514732361,
+      "learning_rate": 0.0003106270912034538,
+      "loss": 3.3882,
       "step": 44800
     },
     {
       "epoch": 4.835579514824798,
-      "grad_norm": 0.6514849662780762,
-      "learning_rate": 0.00031030976794387476,
-      "loss": 3.3803,
+      "grad_norm": 0.613705039024353,
+      "learning_rate": 0.00031030329195898536,
+      "loss": 3.3834,
       "step": 44850
     },
     {
       "epoch": 4.840970350404312,
-      "grad_norm": 0.6474654078483582,
-      "learning_rate": 0.0003099859686994063,
-      "loss": 3.3833,
+      "grad_norm": 0.6359333992004395,
+      "learning_rate": 0.00030997949271451697,
+      "loss": 3.3843,
       "step": 44900
     },
     {
       "epoch": 4.846361185983827,
-      "grad_norm": 0.6419128179550171,
-      "learning_rate": 0.0003096621694549379,
-      "loss": 3.3786,
+      "grad_norm": 0.6831448078155518,
+      "learning_rate": 0.0003096556934700485,
+      "loss": 3.3816,
       "step": 44950
     },
     {
       "epoch": 4.8517520215633425,
-      "grad_norm": 0.6200140714645386,
-      "learning_rate": 0.00030933837021046947,
-      "loss": 3.3886,
+      "grad_norm": 0.6147156953811646,
+      "learning_rate": 0.0003093318942255801,
+      "loss": 3.394,
       "step": 45000
     },
     {
       "epoch": 4.8517520215633425,
-      "eval_accuracy": 0.38061834575760994,
-      "eval_loss": 3.4101431369781494,
-      "eval_runtime": 183.1471,
-      "eval_samples_per_second": 98.342,
-      "eval_steps_per_second": 6.148,
+      "eval_accuracy": 0.3804664490204453,
+      "eval_loss": 3.409999132156372,
+      "eval_runtime": 181.6296,
+      "eval_samples_per_second": 99.163,
+      "eval_steps_per_second": 6.199,
       "step": 45000
     },
     {
       "epoch": 4.857142857142857,
-      "grad_norm": 0.6373332738876343,
-      "learning_rate": 0.0003090145709660011,
-      "loss": 3.3643,
+      "grad_norm": 0.6765714287757874,
+      "learning_rate": 0.0003090080949811117,
+      "loss": 3.3678,
       "step": 45050
     },
     {
       "epoch": 4.862533692722372,
-      "grad_norm": 0.6662492752075195,
-      "learning_rate": 0.0003086907717215326,
-      "loss": 3.3895,
+      "grad_norm": 0.6566330194473267,
+      "learning_rate": 0.0003086842957366433,
+      "loss": 3.3938,
       "step": 45100
     },
     {
       "epoch": 4.867924528301887,
-      "grad_norm": 0.7503253817558289,
-      "learning_rate": 0.0003083669724770642,
-      "loss": 3.3651,
+      "grad_norm": 0.6916565299034119,
+      "learning_rate": 0.00030836049649217483,
+      "loss": 3.3667,
       "step": 45150
     },
     {
       "epoch": 4.873315363881401,
-      "grad_norm": 0.6341877579689026,
-      "learning_rate": 0.0003080431732325958,
-      "loss": 3.3864,
+      "grad_norm": 0.6514751315116882,
+      "learning_rate": 0.0003080366972477064,
+      "loss": 3.3901,
       "step": 45200
     },
     {
       "epoch": 4.878706199460916,
-      "grad_norm": 0.6354518532752991,
-      "learning_rate": 0.00030771937398812733,
-      "loss": 3.3839,
+      "grad_norm": 0.6567890048027039,
+      "learning_rate": 0.000307712898003238,
+      "loss": 3.3857,
       "step": 45250
     },
     {
       "epoch": 4.884097035040432,
-      "grad_norm": 0.6411907076835632,
-      "learning_rate": 0.00030739557474365894,
-      "loss": 3.389,
+      "grad_norm": 0.6464932560920715,
+      "learning_rate": 0.00030738909875876953,
+      "loss": 3.3918,
       "step": 45300
     },
     {
       "epoch": 4.889487870619946,
-      "grad_norm": 0.6452317833900452,
-      "learning_rate": 0.0003070717754991905,
-      "loss": 3.3803,
+      "grad_norm": 0.663248598575592,
+      "learning_rate": 0.00030706529951430114,
+      "loss": 3.3821,
       "step": 45350
     },
     {
       "epoch": 4.894878706199461,
-      "grad_norm": 0.61771160364151,
-      "learning_rate": 0.0003067479762547221,
-      "loss": 3.3768,
+      "grad_norm": 0.6422716975212097,
+      "learning_rate": 0.0003067415002698327,
+      "loss": 3.3793,
       "step": 45400
     },
     {
       "epoch": 4.900269541778976,
-      "grad_norm": 0.6439537405967712,
-      "learning_rate": 0.00030643065299514294,
-      "loss": 3.3857,
+      "grad_norm": 0.631613552570343,
+      "learning_rate": 0.0003064177010253643,
+      "loss": 3.3886,
       "step": 45450
     },
     {
       "epoch": 4.90566037735849,
-      "grad_norm": 0.6568528413772583,
-      "learning_rate": 0.00030610685375067454,
-      "loss": 3.3843,
+      "grad_norm": 0.6880245804786682,
+      "learning_rate": 0.0003060939017808958,
+      "loss": 3.3877,
       "step": 45500
     },
     {
       "epoch": 4.9110512129380055,
-      "grad_norm": 0.6745019555091858,
-      "learning_rate": 0.0003057830545062061,
-      "loss": 3.3737,
+      "grad_norm": 0.6527613997459412,
+      "learning_rate": 0.0003057701025364274,
+      "loss": 3.3761,
       "step": 45550
     },
     {
       "epoch": 4.916442048517521,
-      "grad_norm": 0.6615657806396484,
-      "learning_rate": 0.0003054592552617377,
-      "loss": 3.3996,
+      "grad_norm": 0.6681221723556519,
+      "learning_rate": 0.00030544630329195895,
+      "loss": 3.4008,
       "step": 45600
     },
     {
       "epoch": 4.921832884097035,
-      "grad_norm": 0.6774879097938538,
-      "learning_rate": 0.00030513545601726925,
-      "loss": 3.3726,
+      "grad_norm": 0.6331242322921753,
+      "learning_rate": 0.0003051225040474905,
+      "loss": 3.3738,
       "step": 45650
     },
     {
       "epoch": 4.92722371967655,
-      "grad_norm": 0.6349773406982422,
-      "learning_rate": 0.00030481165677280085,
-      "loss": 3.3848,
+      "grad_norm": 0.6249701380729675,
+      "learning_rate": 0.0003047987048030221,
+      "loss": 3.3868,
       "step": 45700
     },
     {
       "epoch": 4.932614555256064,
-      "grad_norm": 0.6601877808570862,
-      "learning_rate": 0.0003044878575283324,
-      "loss": 3.3864,
+      "grad_norm": 0.6285362839698792,
+      "learning_rate": 0.00030447490555855365,
+      "loss": 3.39,
       "step": 45750
     },
     {
       "epoch": 4.938005390835579,
-      "grad_norm": 0.7116173505783081,
-      "learning_rate": 0.000304164058283864,
-      "loss": 3.3814,
+      "grad_norm": 0.7011100053787231,
+      "learning_rate": 0.00030415110631408526,
+      "loss": 3.3845,
       "step": 45800
     },
     {
       "epoch": 4.943396226415095,
-      "grad_norm": 0.7644145488739014,
-      "learning_rate": 0.00030384025903939556,
-      "loss": 3.382,
+      "grad_norm": 0.7112923860549927,
+      "learning_rate": 0.0003038273070696168,
+      "loss": 3.3846,
       "step": 45850
     },
     {
       "epoch": 4.948787061994609,
-      "grad_norm": 0.6855260729789734,
-      "learning_rate": 0.0003035164597949271,
-      "loss": 3.3844,
+      "grad_norm": 0.6628667712211609,
+      "learning_rate": 0.0003035035078251484,
+      "loss": 3.3886,
       "step": 45900
     },
     {
       "epoch": 4.954177897574124,
-      "grad_norm": 0.6151649951934814,
-      "learning_rate": 0.0003031926605504587,
-      "loss": 3.3674,
+      "grad_norm": 0.6680319905281067,
+      "learning_rate": 0.00030317970858067996,
+      "loss": 3.3696,
       "step": 45950
     },
     {
       "epoch": 4.959568733153639,
-      "grad_norm": 0.6668580174446106,
-      "learning_rate": 0.00030286886130599026,
-      "loss": 3.3804,
+      "grad_norm": 0.6501636505126953,
+      "learning_rate": 0.00030285590933621157,
+      "loss": 3.3828,
       "step": 46000
     },
     {
       "epoch": 4.959568733153639,
-      "eval_accuracy": 0.3810672994900594,
-      "eval_loss": 3.4056036472320557,
-      "eval_runtime": 183.5372,
-      "eval_samples_per_second": 98.133,
-      "eval_steps_per_second": 6.135,
+      "eval_accuracy": 0.38101308169903997,
+      "eval_loss": 3.4067649841308594,
+      "eval_runtime": 181.2801,
+      "eval_samples_per_second": 99.355,
+      "eval_steps_per_second": 6.211,
       "step": 46000
     },
     {
       "epoch": 4.964959568733153,
-      "grad_norm": 0.6548580527305603,
-      "learning_rate": 0.00030254506206152187,
-      "loss": 3.387,
+      "grad_norm": 0.6265591382980347,
+      "learning_rate": 0.0003025321100917431,
+      "loss": 3.389,
       "step": 46050
     },
     {
       "epoch": 4.9703504043126685,
-      "grad_norm": 0.6375557780265808,
-      "learning_rate": 0.00030222126281705336,
-      "loss": 3.3795,
+      "grad_norm": 2.630911111831665,
+      "learning_rate": 0.0003022083108472746,
+      "loss": 3.3831,
       "step": 46100
     },
     {
       "epoch": 4.975741239892184,
-      "grad_norm": 0.6740067601203918,
-      "learning_rate": 0.000301897463572585,
-      "loss": 3.3812,
+      "grad_norm": 0.6849364042282104,
+      "learning_rate": 0.0003018845116028062,
+      "loss": 3.383,
       "step": 46150
     },
     {
       "epoch": 4.981132075471698,
-      "grad_norm": 0.6744655966758728,
-      "learning_rate": 0.0003015736643281165,
-      "loss": 3.3782,
+      "grad_norm": 0.664171040058136,
+      "learning_rate": 0.00030156071235833777,
+      "loss": 3.3813,
       "step": 46200
     },
     {
       "epoch": 4.986522911051213,
-      "grad_norm": 0.6689214110374451,
-      "learning_rate": 0.00030124986508364807,
-      "loss": 3.4003,
+      "grad_norm": 0.6433027386665344,
+      "learning_rate": 0.0003012369131138694,
+      "loss": 3.4033,
       "step": 46250
     },
     {
       "epoch": 4.991913746630727,
-      "grad_norm": 0.6952337026596069,
-      "learning_rate": 0.0003009260658391797,
-      "loss": 3.3866,
+      "grad_norm": 0.6849321126937866,
+      "learning_rate": 0.0003009195898542903,
+      "loss": 3.3904,
       "step": 46300
     },
     {
       "epoch": 4.997304582210242,
-      "grad_norm": 0.6591714024543762,
-      "learning_rate": 0.0003006022665947112,
-      "loss": 3.3759,
+      "grad_norm": 0.6644103527069092,
+      "learning_rate": 0.0003005957906098219,
+      "loss": 3.3814,
       "step": 46350
     },
     {
       "epoch": 5.002695417789758,
-      "grad_norm": 0.6524887084960938,
-      "learning_rate": 0.00030027846735024283,
-      "loss": 3.3332,
+      "grad_norm": 0.7095720171928406,
+      "learning_rate": 0.00030027199136535343,
+      "loss": 3.3362,
       "step": 46400
     },
     {
       "epoch": 5.008086253369272,
-      "grad_norm": 0.6842234134674072,
-      "learning_rate": 0.00029995466810577443,
-      "loss": 3.2811,
+      "grad_norm": 0.654776930809021,
+      "learning_rate": 0.00029994819212088503,
+      "loss": 3.2852,
       "step": 46450
     },
     {
       "epoch": 5.013477088948787,
-      "grad_norm": 0.6671633720397949,
-      "learning_rate": 0.00029963086886130593,
-      "loss": 3.2851,
+      "grad_norm": 0.658161997795105,
+      "learning_rate": 0.0002996243928764166,
+      "loss": 3.2895,
       "step": 46500
     },
     {
       "epoch": 5.018867924528302,
-      "grad_norm": 0.6657042503356934,
-      "learning_rate": 0.00029930706961683754,
-      "loss": 3.2927,
+      "grad_norm": 0.6847811341285706,
+      "learning_rate": 0.0002993005936319482,
+      "loss": 3.2957,
       "step": 46550
     },
     {
       "epoch": 5.024258760107816,
-      "grad_norm": 0.6508234739303589,
-      "learning_rate": 0.0002989832703723691,
-      "loss": 3.2926,
+      "grad_norm": 0.6701328158378601,
+      "learning_rate": 0.00029897679438747974,
+      "loss": 3.2963,
       "step": 46600
     },
     {
       "epoch": 5.0296495956873315,
-      "grad_norm": 0.7433764934539795,
-      "learning_rate": 0.0002986594711279007,
-      "loss": 3.2969,
+      "grad_norm": 0.6993874907493591,
+      "learning_rate": 0.0002986529951430113,
+      "loss": 3.2989,
       "step": 46650
     },
     {
       "epoch": 5.035040431266847,
-      "grad_norm": 0.6221592426300049,
-      "learning_rate": 0.00029833567188343224,
-      "loss": 3.2891,
+      "grad_norm": 0.6440631151199341,
+      "learning_rate": 0.0002983291958985429,
+      "loss": 3.2932,
       "step": 46700
     },
     {
       "epoch": 5.040431266846361,
-      "grad_norm": 0.6607252955436707,
-      "learning_rate": 0.00029801187263896385,
-      "loss": 3.3061,
+      "grad_norm": 0.65876305103302,
+      "learning_rate": 0.00029800539665407444,
+      "loss": 3.3075,
       "step": 46750
     },
     {
       "epoch": 5.045822102425876,
-      "grad_norm": 0.7137832045555115,
-      "learning_rate": 0.0002976880733944954,
-      "loss": 3.2899,
+      "grad_norm": 0.7072811126708984,
+      "learning_rate": 0.000297681597409606,
+      "loss": 3.2925,
       "step": 46800
     },
     {
       "epoch": 5.051212938005391,
-      "grad_norm": 0.6454117894172668,
-      "learning_rate": 0.00029736427415002695,
-      "loss": 3.308,
+      "grad_norm": 0.7308153510093689,
+      "learning_rate": 0.0002973577981651376,
+      "loss": 3.3111,
       "step": 46850
     },
     {
       "epoch": 5.056603773584905,
-      "grad_norm": 0.6954532265663147,
-      "learning_rate": 0.0002970404749055585,
-      "loss": 3.296,
+      "grad_norm": 0.6857506632804871,
+      "learning_rate": 0.00029703399892066915,
+      "loss": 3.2955,
       "step": 46900
     },
     {
       "epoch": 5.061994609164421,
-      "grad_norm": 0.6978551149368286,
-      "learning_rate": 0.0002967166756610901,
-      "loss": 3.2855,
+      "grad_norm": 0.6599889993667603,
+      "learning_rate": 0.00029671019967620076,
+      "loss": 3.2879,
       "step": 46950
     },
     {
       "epoch": 5.067385444743936,
-      "grad_norm": 0.7152319550514221,
-      "learning_rate": 0.00029639287641662165,
-      "loss": 3.3041,
+      "grad_norm": 0.7296129465103149,
+      "learning_rate": 0.0002963864004317323,
+      "loss": 3.3074,
       "step": 47000
     },
     {
       "epoch": 5.067385444743936,
-      "eval_accuracy": 0.3810717542584598,
-      "eval_loss": 3.4087512493133545,
-      "eval_runtime": 183.5405,
-      "eval_samples_per_second": 98.131,
-      "eval_steps_per_second": 6.135,
+      "eval_accuracy": 0.38087237620932024,
+      "eval_loss": 3.4135961532592773,
+      "eval_runtime": 181.4602,
+      "eval_samples_per_second": 99.256,
+      "eval_steps_per_second": 6.205,
       "step": 47000
     },
     {
       "epoch": 5.07277628032345,
-      "grad_norm": 0.6855849623680115,
-      "learning_rate": 0.00029606907717215326,
-      "loss": 3.3057,
+      "grad_norm": 0.6860654950141907,
+      "learning_rate": 0.00029606260118726386,
+      "loss": 3.3086,
       "step": 47050
     },
     {
       "epoch": 5.078167115902965,
-      "grad_norm": 0.6449735164642334,
-      "learning_rate": 0.0002957452779276848,
-      "loss": 3.3095,
+      "grad_norm": 0.6684079766273499,
+      "learning_rate": 0.0002957388019427954,
+      "loss": 3.311,
       "step": 47100
     },
     {
       "epoch": 5.083557951482479,
-      "grad_norm": 0.668123722076416,
-      "learning_rate": 0.00029542147868321636,
-      "loss": 3.304,
+      "grad_norm": 0.6741335391998291,
+      "learning_rate": 0.000295415002698327,
+      "loss": 3.3059,
       "step": 47150
     },
     {
       "epoch": 5.0889487870619945,
-      "grad_norm": 0.6669653058052063,
-      "learning_rate": 0.00029509767943874796,
-      "loss": 3.2993,
+      "grad_norm": 0.6535823941230774,
+      "learning_rate": 0.00029509120345385856,
+      "loss": 3.3028,
       "step": 47200
     },
     {
       "epoch": 5.09433962264151,
-      "grad_norm": 0.6730850338935852,
-      "learning_rate": 0.0002947738801942795,
-      "loss": 3.3039,
+      "grad_norm": 0.747099757194519,
+      "learning_rate": 0.00029476740420939017,
+      "loss": 3.306,
       "step": 47250
     },
     {
       "epoch": 5.099730458221024,
-      "grad_norm": 0.6958271265029907,
-      "learning_rate": 0.0002944500809498111,
-      "loss": 3.2981,
+      "grad_norm": 0.6633142828941345,
+      "learning_rate": 0.0002944436049649217,
+      "loss": 3.2998,
       "step": 47300
     },
     {
       "epoch": 5.105121293800539,
-      "grad_norm": 0.6824030876159668,
-      "learning_rate": 0.00029412628170534267,
-      "loss": 3.2891,
+      "grad_norm": 0.6420027613639832,
+      "learning_rate": 0.0002941198057204533,
+      "loss": 3.2917,
       "step": 47350
     },
     {
       "epoch": 5.110512129380054,
-      "grad_norm": 0.6391562223434448,
-      "learning_rate": 0.0002938024824608742,
-      "loss": 3.3064,
+      "grad_norm": 0.6478297710418701,
+      "learning_rate": 0.00029379600647598487,
+      "loss": 3.3098,
       "step": 47400
     },
     {
       "epoch": 5.115902964959568,
-      "grad_norm": 0.6703659296035767,
-      "learning_rate": 0.00029347868321640577,
-      "loss": 3.3213,
+      "grad_norm": 0.6421354413032532,
+      "learning_rate": 0.0002934722072315164,
+      "loss": 3.3242,
       "step": 47450
     },
     {
       "epoch": 5.121293800539084,
-      "grad_norm": 0.6760079860687256,
-      "learning_rate": 0.0002931548839719374,
-      "loss": 3.3149,
+      "grad_norm": 0.6514407992362976,
+      "learning_rate": 0.000293148407987048,
+      "loss": 3.3191,
       "step": 47500
     },
     {
       "epoch": 5.126684636118599,
-      "grad_norm": 0.6816515326499939,
-      "learning_rate": 0.00029283756071235833,
-      "loss": 3.3225,
+      "grad_norm": 0.6773439645767212,
+      "learning_rate": 0.0002928246087425796,
+      "loss": 3.3251,
       "step": 47550
     },
     {
       "epoch": 5.132075471698113,
-      "grad_norm": 0.6747937202453613,
-      "learning_rate": 0.0002925137614678899,
-      "loss": 3.314,
+      "grad_norm": 0.681089460849762,
+      "learning_rate": 0.00029250080949811113,
+      "loss": 3.3165,
       "step": 47600
     },
     {
       "epoch": 5.137466307277628,
-      "grad_norm": 0.6805335879325867,
-      "learning_rate": 0.00029218996222342143,
-      "loss": 3.3041,
+      "grad_norm": 0.6679985523223877,
+      "learning_rate": 0.00029217701025364273,
+      "loss": 3.306,
       "step": 47650
     },
     {
       "epoch": 5.142857142857143,
-      "grad_norm": 0.6687912940979004,
-      "learning_rate": 0.00029186616297895303,
-      "loss": 3.3185,
+      "grad_norm": 0.6983124017715454,
+      "learning_rate": 0.0002918532110091743,
+      "loss": 3.3206,
       "step": 47700
     },
     {
       "epoch": 5.1482479784366575,
-      "grad_norm": 0.6776725649833679,
-      "learning_rate": 0.0002915423637344846,
-      "loss": 3.311,
+      "grad_norm": 0.63779616355896,
+      "learning_rate": 0.0002915294117647059,
+      "loss": 3.3135,
       "step": 47750
     },
     {
       "epoch": 5.153638814016173,
-      "grad_norm": 0.6747115850448608,
-      "learning_rate": 0.00029121856449001613,
-      "loss": 3.3006,
+      "grad_norm": 0.6387830972671509,
+      "learning_rate": 0.00029120561252023744,
+      "loss": 3.3049,
       "step": 47800
     },
     {
       "epoch": 5.159029649595688,
-      "grad_norm": 0.6835746169090271,
-      "learning_rate": 0.00029089476524554774,
-      "loss": 3.3243,
+      "grad_norm": 0.6785860061645508,
+      "learning_rate": 0.000290881813275769,
+      "loss": 3.3287,
       "step": 47850
     },
     {
       "epoch": 5.164420485175202,
-      "grad_norm": 0.6365933418273926,
-      "learning_rate": 0.0002905709660010793,
-      "loss": 3.3159,
+      "grad_norm": 0.6840365529060364,
+      "learning_rate": 0.00029055801403130054,
+      "loss": 3.319,
       "step": 47900
     },
     {
       "epoch": 5.169811320754717,
-      "grad_norm": 0.8081470727920532,
-      "learning_rate": 0.0002902471667566109,
-      "loss": 3.3033,
+      "grad_norm": 0.7495551109313965,
+      "learning_rate": 0.00029023421478683215,
+      "loss": 3.307,
       "step": 47950
     },
     {
       "epoch": 5.175202156334231,
-      "grad_norm": 0.6865705251693726,
-      "learning_rate": 0.00028992336751214245,
-      "loss": 3.3075,
+      "grad_norm": 0.658063530921936,
+      "learning_rate": 0.0002899104155423637,
+      "loss": 3.3102,
       "step": 48000
     },
     {
       "epoch": 5.175202156334231,
-      "eval_accuracy": 0.3815551509563464,
-      "eval_loss": 3.405240297317505,
-      "eval_runtime": 183.5473,
-      "eval_samples_per_second": 98.127,
-      "eval_steps_per_second": 6.135,
+      "eval_accuracy": 0.3813590324938413,
+      "eval_loss": 3.405421733856201,
+      "eval_runtime": 181.2524,
+      "eval_samples_per_second": 99.37,
+      "eval_steps_per_second": 6.212,
       "step": 48000
     },
     {
       "epoch": 5.180592991913747,
-      "grad_norm": 0.6026649475097656,
-      "learning_rate": 0.00028959956826767405,
-      "loss": 3.3132,
+      "grad_norm": 0.6241575479507446,
+      "learning_rate": 0.0002895866162978953,
+      "loss": 3.3163,
       "step": 48050
     },
     {
       "epoch": 5.185983827493262,
-      "grad_norm": 0.7102674245834351,
-      "learning_rate": 0.00028927576902320555,
-      "loss": 3.3234,
+      "grad_norm": 0.6943466067314148,
+      "learning_rate": 0.00028926281705342685,
+      "loss": 3.3266,
       "step": 48100
     },
     {
       "epoch": 5.191374663072776,
-      "grad_norm": 0.6844705939292908,
-      "learning_rate": 0.00028895196977873715,
-      "loss": 3.319,
+      "grad_norm": 0.6717433929443359,
+      "learning_rate": 0.0002889390178089584,
+      "loss": 3.3226,
       "step": 48150
     },
     {
       "epoch": 5.196765498652291,
-      "grad_norm": 0.733917236328125,
-      "learning_rate": 0.0002886281705342687,
-      "loss": 3.329,
+      "grad_norm": 0.6995725631713867,
+      "learning_rate": 0.00028861521856449,
+      "loss": 3.3313,
       "step": 48200
     },
     {
       "epoch": 5.202156334231806,
-      "grad_norm": 0.6499011516571045,
-      "learning_rate": 0.0002883043712898003,
-      "loss": 3.3221,
+      "grad_norm": 0.6772176623344421,
+      "learning_rate": 0.00028829141932002156,
+      "loss": 3.3241,
       "step": 48250
     },
     {
       "epoch": 5.2075471698113205,
-      "grad_norm": 0.6283177733421326,
-      "learning_rate": 0.00028798057204533186,
-      "loss": 3.2958,
+      "grad_norm": 0.6531539559364319,
+      "learning_rate": 0.0002879676200755531,
+      "loss": 3.2994,
       "step": 48300
     },
     {
       "epoch": 5.212938005390836,
-      "grad_norm": 0.6744676232337952,
-      "learning_rate": 0.00028765677280086346,
-      "loss": 3.3312,
+      "grad_norm": 0.6652123928070068,
+      "learning_rate": 0.00028765029681597406,
+      "loss": 3.3346,
       "step": 48350
     },
     {
       "epoch": 5.218328840970351,
-      "grad_norm": 0.6724770665168762,
-      "learning_rate": 0.000287332973556395,
-      "loss": 3.3213,
+      "grad_norm": 0.6726841330528259,
+      "learning_rate": 0.00028732649757150566,
+      "loss": 3.3254,
       "step": 48400
     },
     {
       "epoch": 5.223719676549865,
-      "grad_norm": 0.7276390790939331,
-      "learning_rate": 0.0002870091743119266,
-      "loss": 3.3264,
+      "grad_norm": 0.7053751945495605,
+      "learning_rate": 0.0002870026983270372,
+      "loss": 3.329,
       "step": 48450
     },
     {
       "epoch": 5.22911051212938,
-      "grad_norm": 0.6782853007316589,
-      "learning_rate": 0.00028668537506745817,
-      "loss": 3.3324,
+      "grad_norm": 0.6710510849952698,
+      "learning_rate": 0.00028667889908256877,
+      "loss": 3.3345,
       "step": 48500
     },
     {
       "epoch": 5.234501347708895,
-      "grad_norm": 0.7115477323532104,
-      "learning_rate": 0.0002863615758229897,
-      "loss": 3.3095,
+      "grad_norm": 0.8056950569152832,
+      "learning_rate": 0.00028635509983810037,
+      "loss": 3.3121,
       "step": 48550
     },
     {
       "epoch": 5.2398921832884096,
-      "grad_norm": 0.6378598213195801,
-      "learning_rate": 0.00028603777657852127,
-      "loss": 3.3119,
+      "grad_norm": 0.6724224090576172,
+      "learning_rate": 0.0002860313005936319,
+      "loss": 3.3151,
       "step": 48600
     },
     {
       "epoch": 5.245283018867925,
-      "grad_norm": 0.678990364074707,
-      "learning_rate": 0.0002857139773340529,
-      "loss": 3.3342,
+      "grad_norm": 0.7094789743423462,
+      "learning_rate": 0.00028570750134916347,
+      "loss": 3.3369,
       "step": 48650
     },
     {
       "epoch": 5.250673854447439,
-      "grad_norm": 0.7339763045310974,
-      "learning_rate": 0.0002853901780895844,
-      "loss": 3.3229,
+      "grad_norm": 0.7018639445304871,
+      "learning_rate": 0.0002853837021046951,
+      "loss": 3.3247,
       "step": 48700
     },
     {
       "epoch": 5.256064690026954,
-      "grad_norm": 0.7002367973327637,
-      "learning_rate": 0.00028506637884511603,
-      "loss": 3.3221,
+      "grad_norm": 0.7365012764930725,
+      "learning_rate": 0.00028505990286022663,
+      "loss": 3.3246,
       "step": 48750
     },
     {
       "epoch": 5.261455525606469,
-      "grad_norm": 0.6659910082817078,
-      "learning_rate": 0.0002847425796006476,
-      "loss": 3.3332,
+      "grad_norm": 0.6922176480293274,
+      "learning_rate": 0.0002847361036157582,
+      "loss": 3.337,
       "step": 48800
     },
     {
       "epoch": 5.2668463611859835,
-      "grad_norm": 0.6622328758239746,
-      "learning_rate": 0.00028441878035617913,
-      "loss": 3.3246,
+      "grad_norm": 0.6411733627319336,
+      "learning_rate": 0.0002844123043712898,
+      "loss": 3.3282,
       "step": 48850
     },
     {
       "epoch": 5.272237196765499,
-      "grad_norm": 0.6887483596801758,
-      "learning_rate": 0.00028409498111171073,
-      "loss": 3.3156,
+      "grad_norm": 0.6878524422645569,
+      "learning_rate": 0.00028408850512682133,
+      "loss": 3.3165,
       "step": 48900
     },
     {
       "epoch": 5.277628032345014,
-      "grad_norm": 0.6723185181617737,
-      "learning_rate": 0.0002837711818672423,
-      "loss": 3.326,
+      "grad_norm": 0.6813527345657349,
+      "learning_rate": 0.00028376470588235294,
+      "loss": 3.3307,
       "step": 48950
     },
     {
       "epoch": 5.283018867924528,
-      "grad_norm": 0.6477317214012146,
-      "learning_rate": 0.00028344738262277384,
-      "loss": 3.3062,
+      "grad_norm": 0.6502499580383301,
+      "learning_rate": 0.0002834409066378845,
+      "loss": 3.3085,
       "step": 49000
     },
     {
       "epoch": 5.283018867924528,
-      "eval_accuracy": 0.3819458667409273,
-      "eval_loss": 3.401156425476074,
-      "eval_runtime": 183.2869,
-      "eval_samples_per_second": 98.267,
-      "eval_steps_per_second": 6.143,
+      "eval_accuracy": 0.381347515287733,
+      "eval_loss": 3.4066929817199707,
+      "eval_runtime": 181.4637,
+      "eval_samples_per_second": 99.254,
+      "eval_steps_per_second": 6.205,
       "step": 49000
     },
     {
       "epoch": 5.288409703504043,
-      "grad_norm": 0.7285588979721069,
-      "learning_rate": 0.00028312358337830544,
-      "loss": 3.3276,
+      "grad_norm": 0.7151468396186829,
+      "learning_rate": 0.00028311710739341604,
+      "loss": 3.3298,
       "step": 49050
     },
     {
       "epoch": 5.293800539083558,
-      "grad_norm": 0.6508597731590271,
-      "learning_rate": 0.000282799784133837,
-      "loss": 3.3288,
+      "grad_norm": 0.629274845123291,
+      "learning_rate": 0.00028279330814894764,
+      "loss": 3.3339,
       "step": 49100
     },
     {
       "epoch": 5.2991913746630726,
-      "grad_norm": 0.6885204911231995,
-      "learning_rate": 0.00028247598488936854,
-      "loss": 3.3244,
+      "grad_norm": 0.7136619687080383,
+      "learning_rate": 0.0002824695089044792,
+      "loss": 3.3274,
       "step": 49150
     },
     {
       "epoch": 5.304582210242588,
-      "grad_norm": 0.6657274961471558,
-      "learning_rate": 0.00028215218564490015,
-      "loss": 3.3197,
+      "grad_norm": 0.6803702712059021,
+      "learning_rate": 0.00028214570966001075,
+      "loss": 3.3232,
       "step": 49200
     },
     {
       "epoch": 5.309973045822103,
-      "grad_norm": 0.6937828063964844,
-      "learning_rate": 0.0002818283864004317,
-      "loss": 3.2995,
+      "grad_norm": 0.6569246649742126,
+      "learning_rate": 0.00028182191041554235,
+      "loss": 3.3035,
       "step": 49250
     },
     {
       "epoch": 5.315363881401617,
-      "grad_norm": 0.6877214312553406,
-      "learning_rate": 0.0002815045871559633,
-      "loss": 3.3058,
+      "grad_norm": 0.6785882115364075,
+      "learning_rate": 0.0002814981111710739,
+      "loss": 3.3089,
       "step": 49300
     },
     {
       "epoch": 5.320754716981132,
-      "grad_norm": 0.7068207859992981,
-      "learning_rate": 0.00028118078791149485,
-      "loss": 3.3351,
+      "grad_norm": 0.6798893213272095,
+      "learning_rate": 0.0002811743119266055,
+      "loss": 3.339,
       "step": 49350
     },
     {
       "epoch": 5.3261455525606465,
-      "grad_norm": 0.6211855411529541,
-      "learning_rate": 0.0002808569886670264,
-      "loss": 3.3069,
+      "grad_norm": 0.6131676435470581,
+      "learning_rate": 0.00028085051268213706,
+      "loss": 3.3087,
       "step": 49400
     },
     {
       "epoch": 5.331536388140162,
-      "grad_norm": 0.7042295932769775,
-      "learning_rate": 0.00028053318942255795,
-      "loss": 3.322,
+      "grad_norm": 0.7191921472549438,
+      "learning_rate": 0.0002805267134376686,
+      "loss": 3.3258,
       "step": 49450
     },
     {
       "epoch": 5.336927223719677,
-      "grad_norm": 0.6894974708557129,
-      "learning_rate": 0.00028020939017808956,
-      "loss": 3.3321,
+      "grad_norm": 0.6881919503211975,
+      "learning_rate": 0.00028020291419320016,
+      "loss": 3.3359,
       "step": 49500
     },
     {
       "epoch": 5.342318059299191,
-      "grad_norm": 0.645094633102417,
-      "learning_rate": 0.0002798920669185105,
-      "loss": 3.3203,
+      "grad_norm": 0.6387270092964172,
+      "learning_rate": 0.00027987911494873176,
+      "loss": 3.3224,
       "step": 49550
     },
     {
       "epoch": 5.347708894878706,
-      "grad_norm": 0.6860967874526978,
-      "learning_rate": 0.00027956826767404206,
-      "loss": 3.3143,
+      "grad_norm": 0.6720103621482849,
+      "learning_rate": 0.0002795553157042633,
+      "loss": 3.3184,
       "step": 49600
     },
     {
       "epoch": 5.353099730458221,
-      "grad_norm": 0.7084428668022156,
-      "learning_rate": 0.00027924446842957367,
-      "loss": 3.3198,
+      "grad_norm": 0.673838198184967,
+      "learning_rate": 0.0002792315164597949,
+      "loss": 3.3216,
       "step": 49650
     },
     {
       "epoch": 5.3584905660377355,
-      "grad_norm": 0.6508020162582397,
-      "learning_rate": 0.0002789206691851052,
-      "loss": 3.2959,
+      "grad_norm": 0.6432577967643738,
+      "learning_rate": 0.00027890771721532647,
+      "loss": 3.2985,
       "step": 49700
     },
     {
       "epoch": 5.363881401617251,
-      "grad_norm": 0.651019811630249,
-      "learning_rate": 0.00027859686994063677,
-      "loss": 3.3288,
+      "grad_norm": 0.6672797203063965,
+      "learning_rate": 0.00027858391797085807,
+      "loss": 3.3332,
       "step": 49750
     },
     {
       "epoch": 5.369272237196766,
-      "grad_norm": 0.6705125570297241,
-      "learning_rate": 0.0002782730706961683,
-      "loss": 3.3182,
+      "grad_norm": 0.7149954438209534,
+      "learning_rate": 0.0002782601187263896,
+      "loss": 3.3234,
       "step": 49800
     },
     {
       "epoch": 5.37466307277628,
-      "grad_norm": 0.6464545130729675,
-      "learning_rate": 0.0002779492714516999,
-      "loss": 3.3219,
+      "grad_norm": 0.660434901714325,
+      "learning_rate": 0.0002779363194819212,
+      "loss": 3.3229,
       "step": 49850
     },
     {
       "epoch": 5.380053908355795,
-      "grad_norm": 0.6712095737457275,
-      "learning_rate": 0.00027762547220723147,
-      "loss": 3.3218,
+      "grad_norm": 0.6786943078041077,
+      "learning_rate": 0.0002776125202374527,
+      "loss": 3.3256,
       "step": 49900
     },
     {
       "epoch": 5.38544474393531,
-      "grad_norm": 0.7088808417320251,
-      "learning_rate": 0.0002773016729627631,
-      "loss": 3.32,
+      "grad_norm": 0.6622052192687988,
+      "learning_rate": 0.00027728872099298433,
+      "loss": 3.3235,
       "step": 49950
     },
     {
       "epoch": 5.390835579514825,
-      "grad_norm": 0.6843079924583435,
-      "learning_rate": 0.00027697787371829463,
-      "loss": 3.331,
+      "grad_norm": 0.6635069251060486,
+      "learning_rate": 0.0002769649217485159,
+      "loss": 3.3344,
       "step": 50000
     },
     {
       "epoch": 5.390835579514825,
-      "eval_accuracy": 0.3825648622428064,
-      "eval_loss": 3.397858142852783,
-      "eval_runtime": 183.2184,
-      "eval_samples_per_second": 98.303,
-      "eval_steps_per_second": 6.146,
+      "eval_accuracy": 0.3822336882407478,
+      "eval_loss": 3.4002845287323,
+      "eval_runtime": 181.2768,
+      "eval_samples_per_second": 99.356,
+      "eval_steps_per_second": 6.211,
       "step": 50000
     },
     {
       "epoch": 5.39622641509434,
-      "grad_norm": 0.6787253618240356,
-      "learning_rate": 0.00027665407447382623,
-      "loss": 3.324,
+      "grad_norm": 0.6638054251670837,
+      "learning_rate": 0.0002766411225040475,
+      "loss": 3.3265,
       "step": 50050
     },
     {
       "epoch": 5.401617250673855,
-      "grad_norm": 0.7045350074768066,
-      "learning_rate": 0.0002763302752293578,
-      "loss": 3.3343,
+      "grad_norm": 0.6489728093147278,
+      "learning_rate": 0.00027631732325957903,
+      "loss": 3.3372,
       "step": 50100
     },
     {
       "epoch": 5.407008086253369,
-      "grad_norm": 0.6545983552932739,
-      "learning_rate": 0.00027600647598488933,
-      "loss": 3.3073,
+      "grad_norm": 0.6705710887908936,
+      "learning_rate": 0.0002759935240151106,
+      "loss": 3.3109,
       "step": 50150
     },
     {
       "epoch": 5.412398921832884,
-      "grad_norm": 0.7040044665336609,
-      "learning_rate": 0.0002756826767404209,
-      "loss": 3.3056,
+      "grad_norm": 0.7629367709159851,
+      "learning_rate": 0.0002756697247706422,
+      "loss": 3.3092,
       "step": 50200
     },
     {
       "epoch": 5.4177897574123985,
-      "grad_norm": 0.682117760181427,
-      "learning_rate": 0.0002753588774959525,
-      "loss": 3.3306,
+      "grad_norm": 0.7149102687835693,
+      "learning_rate": 0.00027534592552617374,
+      "loss": 3.3331,
       "step": 50250
     },
     {
       "epoch": 5.423180592991914,
-      "grad_norm": 0.6643507480621338,
-      "learning_rate": 0.00027503507825148404,
-      "loss": 3.3126,
+      "grad_norm": 0.6840925216674805,
+      "learning_rate": 0.00027502212628170535,
+      "loss": 3.3145,
       "step": 50300
     },
     {
       "epoch": 5.428571428571429,
-      "grad_norm": 0.6264309287071228,
-      "learning_rate": 0.00027471127900701564,
-      "loss": 3.3265,
+      "grad_norm": 0.6455720663070679,
+      "learning_rate": 0.0002746983270372369,
+      "loss": 3.3294,
       "step": 50350
     },
     {
       "epoch": 5.433962264150943,
-      "grad_norm": 0.6541180610656738,
-      "learning_rate": 0.0002743874797625472,
-      "loss": 3.3153,
+      "grad_norm": 0.6796095371246338,
+      "learning_rate": 0.00027437452779276845,
+      "loss": 3.3186,
       "step": 50400
     },
     {
       "epoch": 5.439353099730458,
-      "grad_norm": 0.7172690629959106,
-      "learning_rate": 0.0002740636805180788,
-      "loss": 3.327,
+      "grad_norm": 0.6487014889717102,
+      "learning_rate": 0.0002740572045331894,
+      "loss": 3.3308,
       "step": 50450
     },
     {
       "epoch": 5.444743935309973,
-      "grad_norm": 0.6622113585472107,
-      "learning_rate": 0.00027373988127361035,
-      "loss": 3.3246,
+      "grad_norm": 0.6597819924354553,
+      "learning_rate": 0.00027373340528872095,
+      "loss": 3.3308,
       "step": 50500
     },
     {
       "epoch": 5.450134770889488,
-      "grad_norm": 0.6717416644096375,
-      "learning_rate": 0.0002734160820291419,
-      "loss": 3.3573,
+      "grad_norm": 0.6617118120193481,
+      "learning_rate": 0.00027340960604425255,
+      "loss": 3.3584,
       "step": 50550
     },
     {
       "epoch": 5.455525606469003,
-      "grad_norm": 0.6707860231399536,
-      "learning_rate": 0.00027309228278467345,
-      "loss": 3.3132,
+      "grad_norm": 0.6936394572257996,
+      "learning_rate": 0.0002730858067997841,
+      "loss": 3.3182,
       "step": 50600
     },
     {
       "epoch": 5.460916442048518,
-      "grad_norm": 0.6682204008102417,
-      "learning_rate": 0.00027276848354020506,
-      "loss": 3.3331,
+      "grad_norm": 0.6671001315116882,
+      "learning_rate": 0.00027276200755531565,
+      "loss": 3.3354,
       "step": 50650
     },
     {
       "epoch": 5.466307277628032,
-      "grad_norm": 0.6391171813011169,
-      "learning_rate": 0.0002724446842957366,
-      "loss": 3.3307,
+      "grad_norm": 0.6388653516769409,
+      "learning_rate": 0.00027243820831084726,
+      "loss": 3.3328,
       "step": 50700
     },
     {
       "epoch": 5.471698113207547,
-      "grad_norm": 0.681674063205719,
-      "learning_rate": 0.0002721208850512682,
-      "loss": 3.3208,
+      "grad_norm": 0.6863663196563721,
+      "learning_rate": 0.0002721144090663788,
+      "loss": 3.3233,
       "step": 50750
     },
     {
       "epoch": 5.4770889487870615,
-      "grad_norm": 0.6989498138427734,
-      "learning_rate": 0.00027179708580679976,
-      "loss": 3.3355,
+      "grad_norm": 0.677651047706604,
+      "learning_rate": 0.00027179060982191036,
+      "loss": 3.3362,
       "step": 50800
     },
     {
       "epoch": 5.482479784366577,
-      "grad_norm": 0.6760035753250122,
-      "learning_rate": 0.0002714732865623313,
-      "loss": 3.336,
+      "grad_norm": 0.7276838421821594,
+      "learning_rate": 0.00027146681057744197,
+      "loss": 3.3382,
       "step": 50850
     },
     {
       "epoch": 5.487870619946092,
-      "grad_norm": 0.6494047045707703,
-      "learning_rate": 0.0002711494873178629,
-      "loss": 3.328,
+      "grad_norm": 0.6539101004600525,
+      "learning_rate": 0.0002711430113329735,
+      "loss": 3.3311,
       "step": 50900
     },
     {
       "epoch": 5.493261455525606,
-      "grad_norm": 0.7357152104377747,
-      "learning_rate": 0.00027082568807339447,
-      "loss": 3.3348,
+      "grad_norm": 0.6710454821586609,
+      "learning_rate": 0.0002708192120885051,
+      "loss": 3.3363,
       "step": 50950
     },
     {
       "epoch": 5.498652291105121,
-      "grad_norm": 0.6467098593711853,
-      "learning_rate": 0.000270501888828926,
-      "loss": 3.3158,
+      "grad_norm": 0.6733425855636597,
+      "learning_rate": 0.00027049541284403667,
+      "loss": 3.3177,
       "step": 51000
     },
     {
       "epoch": 5.498652291105121,
-      "eval_accuracy": 0.3829662260103931,
-      "eval_loss": 3.393737316131592,
-      "eval_runtime": 183.6994,
-      "eval_samples_per_second": 98.046,
-      "eval_steps_per_second": 6.13,
+      "eval_accuracy": 0.38276782583724384,
+      "eval_loss": 3.394947052001953,
+      "eval_runtime": 181.7426,
+      "eval_samples_per_second": 99.102,
+      "eval_steps_per_second": 6.196,
       "step": 51000
     },
     {
       "epoch": 5.504043126684636,
-      "grad_norm": 0.6844212412834167,
-      "learning_rate": 0.0002701780895844576,
-      "loss": 3.3327,
+      "grad_norm": 0.6891338229179382,
+      "learning_rate": 0.0002701716135995683,
+      "loss": 3.338,
       "step": 51050
     },
     {
       "epoch": 5.509433962264151,
-      "grad_norm": 0.6608554124832153,
-      "learning_rate": 0.0002698542903399892,
-      "loss": 3.3068,
+      "grad_norm": 0.6948062181472778,
+      "learning_rate": 0.0002698478143550998,
+      "loss": 3.3096,
       "step": 51100
     },
     {
       "epoch": 5.514824797843666,
-      "grad_norm": 0.7180224061012268,
-      "learning_rate": 0.0002695304910955207,
-      "loss": 3.3203,
+      "grad_norm": 0.7040722370147705,
+      "learning_rate": 0.0002695240151106314,
+      "loss": 3.3223,
       "step": 51150
     },
     {
       "epoch": 5.520215633423181,
-      "grad_norm": 0.7351436614990234,
-      "learning_rate": 0.00026920669185105233,
-      "loss": 3.3291,
+      "grad_norm": 0.6579620838165283,
+      "learning_rate": 0.00026920021586616293,
+      "loss": 3.3315,
       "step": 51200
     },
     {
       "epoch": 5.525606469002695,
-      "grad_norm": 0.6572839021682739,
-      "learning_rate": 0.0002688828926065839,
-      "loss": 3.3288,
+      "grad_norm": 0.6677024364471436,
+      "learning_rate": 0.00026887641662169453,
+      "loss": 3.332,
       "step": 51250
     },
     {
       "epoch": 5.53099730458221,
-      "grad_norm": 0.7058694958686829,
-      "learning_rate": 0.0002685590933621155,
-      "loss": 3.3398,
+      "grad_norm": 0.7141261100769043,
+      "learning_rate": 0.0002685526173772261,
+      "loss": 3.3436,
       "step": 51300
     },
     {
       "epoch": 5.536388140161725,
-      "grad_norm": 0.6541451811790466,
-      "learning_rate": 0.00026823529411764704,
-      "loss": 3.3093,
+      "grad_norm": 0.6406843066215515,
+      "learning_rate": 0.0002682288181327577,
+      "loss": 3.3116,
       "step": 51350
     },
     {
       "epoch": 5.54177897574124,
-      "grad_norm": 0.7691888809204102,
-      "learning_rate": 0.0002679114948731786,
-      "loss": 3.3288,
+      "grad_norm": 0.7111983895301819,
+      "learning_rate": 0.00026790501888828924,
+      "loss": 3.3329,
       "step": 51400
     },
     {
       "epoch": 5.547169811320755,
-      "grad_norm": 0.6912489533424377,
-      "learning_rate": 0.0002675876956287102,
-      "loss": 3.3276,
+      "grad_norm": 0.7381031513214111,
+      "learning_rate": 0.00026758121964382084,
+      "loss": 3.3309,
       "step": 51450
     },
     {
       "epoch": 5.55256064690027,
-      "grad_norm": 0.6621717810630798,
-      "learning_rate": 0.00026726389638424174,
-      "loss": 3.3131,
+      "grad_norm": 0.6562272906303406,
+      "learning_rate": 0.00026725742039935234,
+      "loss": 3.3171,
       "step": 51500
     },
     {
       "epoch": 5.557951482479784,
-      "grad_norm": 0.674435555934906,
-      "learning_rate": 0.0002669400971397733,
-      "loss": 3.3392,
+      "grad_norm": 0.6938887238502502,
+      "learning_rate": 0.00026693362115488394,
+      "loss": 3.3438,
       "step": 51550
     },
     {
       "epoch": 5.563342318059299,
-      "grad_norm": 0.7033376097679138,
-      "learning_rate": 0.00026662277388019424,
-      "loss": 3.3262,
+      "grad_norm": 0.6981541514396667,
+      "learning_rate": 0.0002666098219104155,
+      "loss": 3.3301,
       "step": 51600
     },
     {
       "epoch": 5.568733153638814,
-      "grad_norm": 0.6880776882171631,
-      "learning_rate": 0.00026629897463572585,
-      "loss": 3.3215,
+      "grad_norm": 0.6889672875404358,
+      "learning_rate": 0.0002662860226659471,
+      "loss": 3.322,
       "step": 51650
     },
     {
       "epoch": 5.574123989218329,
-      "grad_norm": 0.6662737727165222,
-      "learning_rate": 0.0002659751753912574,
-      "loss": 3.3392,
+      "grad_norm": 0.6821058392524719,
+      "learning_rate": 0.00026596222342147865,
+      "loss": 3.3428,
       "step": 51700
     },
     {
       "epoch": 5.579514824797844,
-      "grad_norm": 0.6773297190666199,
-      "learning_rate": 0.00026565137614678895,
-      "loss": 3.328,
+      "grad_norm": 0.6806004643440247,
+      "learning_rate": 0.00026563842417701026,
+      "loss": 3.332,
       "step": 51750
     },
     {
       "epoch": 5.584905660377358,
-      "grad_norm": 0.7141923308372498,
-      "learning_rate": 0.0002653275769023205,
-      "loss": 3.3329,
+      "grad_norm": 0.7079453468322754,
+      "learning_rate": 0.0002653146249325418,
+      "loss": 3.3344,
       "step": 51800
     },
     {
       "epoch": 5.590296495956873,
-      "grad_norm": 0.6370801329612732,
-      "learning_rate": 0.0002650037776578521,
-      "loss": 3.3387,
+      "grad_norm": 0.6341474056243896,
+      "learning_rate": 0.00026499082568807336,
+      "loss": 3.3416,
       "step": 51850
     },
     {
       "epoch": 5.595687331536388,
-      "grad_norm": 0.7268838286399841,
-      "learning_rate": 0.00026468645439827306,
-      "loss": 3.3367,
+      "grad_norm": 0.6999854445457458,
+      "learning_rate": 0.00026466702644360496,
+      "loss": 3.3396,
       "step": 51900
     },
     {
       "epoch": 5.601078167115903,
-      "grad_norm": 0.6624352335929871,
-      "learning_rate": 0.0002643626551538046,
-      "loss": 3.3439,
+      "grad_norm": 0.6550791263580322,
+      "learning_rate": 0.0002643432271991365,
+      "loss": 3.3488,
       "step": 51950
     },
     {
       "epoch": 5.606469002695418,
-      "grad_norm": 0.6981469988822937,
-      "learning_rate": 0.0002640388559093362,
-      "loss": 3.3354,
+      "grad_norm": 0.7080801129341125,
+      "learning_rate": 0.00026401942795466806,
+      "loss": 3.3391,
       "step": 52000
     },
     {
       "epoch": 5.606469002695418,
-      "eval_accuracy": 0.38319341919881317,
-      "eval_loss": 3.390268087387085,
-      "eval_runtime": 183.3396,
-      "eval_samples_per_second": 98.238,
-      "eval_steps_per_second": 6.142,
+      "eval_accuracy": 0.38279259869566556,
+      "eval_loss": 3.392232894897461,
+      "eval_runtime": 181.1112,
+      "eval_samples_per_second": 99.447,
+      "eval_steps_per_second": 6.217,
       "step": 52000
     },
     {
       "epoch": 5.611859838274933,
-      "grad_norm": 0.7103446125984192,
-      "learning_rate": 0.00026371505666486776,
-      "loss": 3.3376,
+      "grad_norm": 0.7377049326896667,
+      "learning_rate": 0.00026369562871019967,
+      "loss": 3.3398,
       "step": 52050
     },
     {
       "epoch": 5.617250673854447,
-      "grad_norm": 0.6647046804428101,
-      "learning_rate": 0.0002633912574203993,
-      "loss": 3.3324,
+      "grad_norm": 0.6713391542434692,
+      "learning_rate": 0.0002633718294657312,
+      "loss": 3.3347,
       "step": 52100
     },
     {
       "epoch": 5.622641509433962,
-      "grad_norm": 0.6633039712905884,
-      "learning_rate": 0.00026306745817593086,
-      "loss": 3.3342,
+      "grad_norm": 0.7107684016227722,
+      "learning_rate": 0.0002630480302212628,
+      "loss": 3.3378,
       "step": 52150
     },
     {
       "epoch": 5.628032345013477,
-      "grad_norm": 0.7168214917182922,
-      "learning_rate": 0.00026274365893146247,
-      "loss": 3.3202,
+      "grad_norm": 0.6784622669219971,
+      "learning_rate": 0.00026272423097679437,
+      "loss": 3.3234,
       "step": 52200
     },
     {
       "epoch": 5.633423180592992,
-      "grad_norm": 0.6797349452972412,
-      "learning_rate": 0.000262419859686994,
-      "loss": 3.3205,
+      "grad_norm": 0.6633272767066956,
+      "learning_rate": 0.0002624004317323259,
+      "loss": 3.3234,
       "step": 52250
     },
     {
       "epoch": 5.638814016172507,
-      "grad_norm": 0.7016252279281616,
-      "learning_rate": 0.0002620960604425256,
-      "loss": 3.341,
+      "grad_norm": 0.6873083710670471,
+      "learning_rate": 0.00026207663248785753,
+      "loss": 3.3442,
       "step": 52300
     },
     {
       "epoch": 5.644204851752022,
-      "grad_norm": 0.6899751424789429,
-      "learning_rate": 0.0002617722611980572,
-      "loss": 3.3142,
+      "grad_norm": 0.6982221603393555,
+      "learning_rate": 0.0002617528332433891,
+      "loss": 3.316,
       "step": 52350
     },
     {
       "epoch": 5.649595687331536,
-      "grad_norm": 0.6652300953865051,
-      "learning_rate": 0.0002614484619535888,
-      "loss": 3.3168,
+      "grad_norm": 0.6670834422111511,
+      "learning_rate": 0.00026142903399892063,
+      "loss": 3.319,
       "step": 52400
     },
     {
       "epoch": 5.654986522911051,
-      "grad_norm": 0.6907793283462524,
-      "learning_rate": 0.00026112466270912033,
-      "loss": 3.3208,
+      "grad_norm": 0.7044607400894165,
+      "learning_rate": 0.00026110523475445223,
+      "loss": 3.3234,
       "step": 52450
     },
     {
       "epoch": 5.660377358490566,
-      "grad_norm": 0.700583279132843,
-      "learning_rate": 0.0002608008634646519,
-      "loss": 3.3349,
+      "grad_norm": 0.7109695672988892,
+      "learning_rate": 0.00026078791149487313,
+      "loss": 3.3369,
       "step": 52500
     },
     {
       "epoch": 5.665768194070081,
-      "grad_norm": 0.7041873335838318,
-      "learning_rate": 0.00026047706422018343,
-      "loss": 3.3405,
+      "grad_norm": 0.6892924904823303,
+      "learning_rate": 0.00026046411225040474,
+      "loss": 3.343,
       "step": 52550
     },
     {
       "epoch": 5.671159029649596,
-      "grad_norm": 0.7494111657142639,
-      "learning_rate": 0.00026015326497571504,
-      "loss": 3.3385,
+      "grad_norm": 0.7476372718811035,
+      "learning_rate": 0.0002601403130059363,
+      "loss": 3.3422,
       "step": 52600
     },
     {
       "epoch": 5.67654986522911,
-      "grad_norm": 0.645006000995636,
-      "learning_rate": 0.0002598294657312466,
-      "loss": 3.3303,
+      "grad_norm": 0.6502141952514648,
+      "learning_rate": 0.0002598165137614679,
+      "loss": 3.3336,
       "step": 52650
     },
     {
       "epoch": 5.681940700808625,
-      "grad_norm": 0.6822366118431091,
-      "learning_rate": 0.0002595056664867782,
-      "loss": 3.3312,
+      "grad_norm": 0.6780169010162354,
+      "learning_rate": 0.00025949271451699944,
+      "loss": 3.3334,
       "step": 52700
     },
     {
       "epoch": 5.6873315363881405,
-      "grad_norm": 0.6748678088188171,
-      "learning_rate": 0.00025918186724230974,
+      "grad_norm": 0.6777785420417786,
+      "learning_rate": 0.000259168915272531,
       "loss": 3.3285,
       "step": 52750
     },
     {
       "epoch": 5.692722371967655,
-      "grad_norm": 0.6696112155914307,
-      "learning_rate": 0.00025885806799784135,
-      "loss": 3.3249,
+      "grad_norm": 0.6609834432601929,
+      "learning_rate": 0.0002588451160280626,
+      "loss": 3.3268,
       "step": 52800
     },
     {
       "epoch": 5.69811320754717,
-      "grad_norm": 0.6802454590797424,
-      "learning_rate": 0.0002585342687533729,
-      "loss": 3.3251,
+      "grad_norm": 0.6979050636291504,
+      "learning_rate": 0.00025852131678359415,
+      "loss": 3.329,
       "step": 52850
     },
     {
       "epoch": 5.703504043126685,
-      "grad_norm": 0.7200360894203186,
-      "learning_rate": 0.00025821046950890445,
-      "loss": 3.3194,
+      "grad_norm": 0.6647638082504272,
+      "learning_rate": 0.0002581975175391257,
+      "loss": 3.3241,
       "step": 52900
     },
     {
       "epoch": 5.708894878706199,
-      "grad_norm": 0.6851761937141418,
-      "learning_rate": 0.000257886670264436,
-      "loss": 3.3145,
+      "grad_norm": 0.6817635297775269,
+      "learning_rate": 0.0002578737182946573,
+      "loss": 3.3175,
       "step": 52950
     },
     {
       "epoch": 5.714285714285714,
-      "grad_norm": 0.6522359251976013,
-      "learning_rate": 0.0002575628710199676,
-      "loss": 3.3422,
+      "grad_norm": 0.6869781017303467,
+      "learning_rate": 0.00025754991905018885,
+      "loss": 3.3462,
       "step": 53000
     },
     {
       "epoch": 5.714285714285714,
-      "eval_accuracy": 0.38380991568427253,
-      "eval_loss": 3.3851399421691895,
-      "eval_runtime": 183.4753,
-      "eval_samples_per_second": 98.166,
-      "eval_steps_per_second": 6.137,
+      "eval_accuracy": 0.3835276354817305,
+      "eval_loss": 3.386016607284546,
+      "eval_runtime": 181.4604,
+      "eval_samples_per_second": 99.256,
+      "eval_steps_per_second": 6.205,
       "step": 53000
     },
     {
       "epoch": 5.719676549865229,
-      "grad_norm": 0.6580410003662109,
-      "learning_rate": 0.00025723907177549915,
-      "loss": 3.3208,
+      "grad_norm": 0.6572978496551514,
+      "learning_rate": 0.00025722611980572046,
+      "loss": 3.3234,
       "step": 53050
     },
     {
       "epoch": 5.725067385444744,
-      "grad_norm": 0.6901311278343201,
-      "learning_rate": 0.00025691527253103076,
-      "loss": 3.314,
+      "grad_norm": 0.7044622302055359,
+      "learning_rate": 0.000256902320561252,
+      "loss": 3.3175,
       "step": 53100
     },
     {
       "epoch": 5.730458221024259,
-      "grad_norm": 0.7142326235771179,
-      "learning_rate": 0.0002565914732865623,
-      "loss": 3.3408,
+      "grad_norm": 0.7395718097686768,
+      "learning_rate": 0.00025657852131678356,
+      "loss": 3.3445,
       "step": 53150
     },
     {
       "epoch": 5.735849056603773,
-      "grad_norm": 0.6859725117683411,
-      "learning_rate": 0.00025626767404209386,
-      "loss": 3.3055,
+      "grad_norm": 0.682083249092102,
+      "learning_rate": 0.0002562547220723151,
+      "loss": 3.3083,
       "step": 53200
     },
     {
       "epoch": 5.741239892183288,
-      "grad_norm": 0.6842724680900574,
-      "learning_rate": 0.00025594387479762546,
-      "loss": 3.3248,
+      "grad_norm": 0.6786825656890869,
+      "learning_rate": 0.0002559309228278467,
+      "loss": 3.3281,
       "step": 53250
     },
     {
       "epoch": 5.7466307277628035,
-      "grad_norm": 0.6995661854743958,
-      "learning_rate": 0.000255620075553157,
-      "loss": 3.3282,
+      "grad_norm": 0.6872519850730896,
+      "learning_rate": 0.00025560712358337827,
+      "loss": 3.3315,
       "step": 53300
     },
     {
       "epoch": 5.752021563342318,
-      "grad_norm": 0.6869672536849976,
-      "learning_rate": 0.00025529627630868857,
-      "loss": 3.3208,
+      "grad_norm": 0.6882497072219849,
+      "learning_rate": 0.00025528332433890987,
+      "loss": 3.3246,
       "step": 53350
     },
     {
       "epoch": 5.757412398921833,
-      "grad_norm": 0.6568567156791687,
-      "learning_rate": 0.00025497247706422017,
-      "loss": 3.3467,
+      "grad_norm": 0.6687334775924683,
+      "learning_rate": 0.0002549595250944414,
+      "loss": 3.3491,
       "step": 53400
     },
     {
       "epoch": 5.762803234501348,
-      "grad_norm": 0.6627490520477295,
-      "learning_rate": 0.0002546486778197517,
-      "loss": 3.3176,
+      "grad_norm": 0.6696929335594177,
+      "learning_rate": 0.000254635725849973,
+      "loss": 3.3219,
       "step": 53450
     },
     {
       "epoch": 5.768194070080862,
-      "grad_norm": 0.6492809653282166,
-      "learning_rate": 0.00025432487857528327,
-      "loss": 3.3171,
+      "grad_norm": 0.692789614200592,
+      "learning_rate": 0.0002543119266055046,
+      "loss": 3.3202,
       "step": 53500
     },
     {
       "epoch": 5.773584905660377,
-      "grad_norm": 0.7379694581031799,
-      "learning_rate": 0.0002540010793308149,
-      "loss": 3.3445,
+      "grad_norm": 0.7241188883781433,
+      "learning_rate": 0.00025398812736103613,
+      "loss": 3.3465,
       "step": 53550
     },
     {
       "epoch": 5.7789757412398925,
-      "grad_norm": 0.7134669423103333,
-      "learning_rate": 0.0002536772800863464,
-      "loss": 3.3258,
+      "grad_norm": 0.721319854259491,
+      "learning_rate": 0.0002536643281165677,
+      "loss": 3.3281,
       "step": 53600
     },
     {
       "epoch": 5.784366576819407,
-      "grad_norm": 0.7085931897163391,
-      "learning_rate": 0.00025335348084187803,
-      "loss": 3.3441,
+      "grad_norm": 0.6907305717468262,
+      "learning_rate": 0.0002533405288720993,
+      "loss": 3.3459,
       "step": 53650
     },
     {
       "epoch": 5.789757412398922,
-      "grad_norm": 0.7263320088386536,
-      "learning_rate": 0.0002530296815974096,
-      "loss": 3.3446,
+      "grad_norm": 0.7575922608375549,
+      "learning_rate": 0.00025301672962763083,
+      "loss": 3.3465,
       "step": 53700
     },
     {
       "epoch": 5.795148247978437,
-      "grad_norm": 0.7074159383773804,
-      "learning_rate": 0.0002527058823529412,
-      "loss": 3.3256,
+      "grad_norm": 0.6928850412368774,
+      "learning_rate": 0.00025269293038316244,
+      "loss": 3.3272,
       "step": 53750
     },
     {
       "epoch": 5.800539083557951,
-      "grad_norm": 0.6964407563209534,
-      "learning_rate": 0.0002523820831084727,
-      "loss": 3.3323,
+      "grad_norm": 0.7083280086517334,
+      "learning_rate": 0.000252369131138694,
+      "loss": 3.3351,
       "step": 53800
     },
     {
       "epoch": 5.8059299191374665,
-      "grad_norm": 0.6850724220275879,
-      "learning_rate": 0.0002520582838640043,
-      "loss": 3.3159,
+      "grad_norm": 0.7181123495101929,
+      "learning_rate": 0.00025204533189422554,
+      "loss": 3.3181,
       "step": 53850
     },
     {
       "epoch": 5.811320754716981,
-      "grad_norm": 0.6719462871551514,
-      "learning_rate": 0.00025173448461953584,
-      "loss": 3.3493,
+      "grad_norm": 0.6715301871299744,
+      "learning_rate": 0.00025172153264975714,
+      "loss": 3.352,
       "step": 53900
     },
     {
       "epoch": 5.816711590296496,
-      "grad_norm": 0.6726512908935547,
-      "learning_rate": 0.00025141068537506744,
-      "loss": 3.3207,
+      "grad_norm": 0.6676560044288635,
+      "learning_rate": 0.0002513977334052887,
+      "loss": 3.3242,
       "step": 53950
     },
     {
       "epoch": 5.822102425876011,
-      "grad_norm": 0.6462234258651733,
-      "learning_rate": 0.000251086886130599,
-      "loss": 3.3243,
+      "grad_norm": 0.6840689182281494,
+      "learning_rate": 0.00025107393416082025,
+      "loss": 3.3272,
       "step": 54000
     },
     {
       "epoch": 5.822102425876011,
-      "eval_accuracy": 0.38394768754602127,
-      "eval_loss": 3.3802895545959473,
-      "eval_runtime": 183.3426,
-      "eval_samples_per_second": 98.237,
-      "eval_steps_per_second": 6.142,
+      "eval_accuracy": 0.3836229240643438,
+      "eval_loss": 3.3817358016967773,
+      "eval_runtime": 193.169,
+      "eval_samples_per_second": 93.24,
+      "eval_steps_per_second": 5.829,
       "step": 54000
     },
     {
       "epoch": 5.827493261455525,
-      "grad_norm": 0.7041391134262085,
-      "learning_rate": 0.0002507630868861306,
-      "loss": 3.319,
+      "grad_norm": 0.7343254685401917,
+      "learning_rate": 0.00025075013491635185,
+      "loss": 3.3219,
       "step": 54050
     },
     {
       "epoch": 5.83288409703504,
-      "grad_norm": 0.671320915222168,
-      "learning_rate": 0.00025043928764166215,
-      "loss": 3.3248,
+      "grad_norm": 0.6892924308776855,
+      "learning_rate": 0.0002504263356718834,
+      "loss": 3.3291,
       "step": 54100
     },
     {
       "epoch": 5.8382749326145555,
-      "grad_norm": 0.6599231958389282,
-      "learning_rate": 0.00025011548839719375,
-      "loss": 3.342,
+      "grad_norm": 0.6996431350708008,
+      "learning_rate": 0.000250102536427415,
+      "loss": 3.3445,
       "step": 54150
     },
     {
       "epoch": 5.84366576819407,
-      "grad_norm": 0.6870060563087463,
-      "learning_rate": 0.0002497916891527253,
-      "loss": 3.3316,
+      "grad_norm": 0.6944709420204163,
+      "learning_rate": 0.00024977873718294656,
+      "loss": 3.3337,
       "step": 54200
     },
     {
       "epoch": 5.849056603773585,
-      "grad_norm": 0.6556626558303833,
-      "learning_rate": 0.00024946788990825686,
-      "loss": 3.3394,
+      "grad_norm": 0.6422256827354431,
+      "learning_rate": 0.0002494549379384781,
+      "loss": 3.3412,
       "step": 54250
     },
     {
       "epoch": 5.8544474393531,
-      "grad_norm": 0.7121956944465637,
-      "learning_rate": 0.0002491440906637884,
-      "loss": 3.3378,
+      "grad_norm": 0.7230224013328552,
+      "learning_rate": 0.0002491311386940097,
+      "loss": 3.3401,
       "step": 54300
     },
     {
       "epoch": 5.859838274932614,
-      "grad_norm": 0.6935588717460632,
-      "learning_rate": 0.00024882029141932,
-      "loss": 3.3287,
+      "grad_norm": 0.679324746131897,
+      "learning_rate": 0.00024880733944954126,
+      "loss": 3.333,
       "step": 54350
     },
     {
       "epoch": 5.8652291105121295,
-      "grad_norm": 0.6818873882293701,
-      "learning_rate": 0.00024849649217485156,
-      "loss": 3.3147,
+      "grad_norm": 0.7162814736366272,
+      "learning_rate": 0.0002484835402050728,
+      "loss": 3.317,
       "step": 54400
     },
     {
       "epoch": 5.870619946091644,
-      "grad_norm": 0.6630216836929321,
-      "learning_rate": 0.00024817269293038317,
-      "loss": 3.3099,
+      "grad_norm": 0.658562421798706,
+      "learning_rate": 0.0002481597409606044,
+      "loss": 3.3128,
       "step": 54450
     },
     {
       "epoch": 5.876010781671159,
-      "grad_norm": 0.7189406752586365,
-      "learning_rate": 0.0002478488936859147,
-      "loss": 3.3263,
+      "grad_norm": 0.7305333614349365,
+      "learning_rate": 0.00024784241770102537,
+      "loss": 3.3282,
       "step": 54500
     },
     {
       "epoch": 5.881401617250674,
-      "grad_norm": 0.708378791809082,
-      "learning_rate": 0.00024752509444144627,
-      "loss": 3.3216,
+      "grad_norm": 0.7414852976799011,
+      "learning_rate": 0.0002475186184565569,
+      "loss": 3.3243,
       "step": 54550
     },
     {
       "epoch": 5.886792452830189,
-      "grad_norm": 0.696700394153595,
-      "learning_rate": 0.00024720129519697787,
-      "loss": 3.3531,
+      "grad_norm": 0.6888782978057861,
+      "learning_rate": 0.00024719481921208847,
+      "loss": 3.3544,
       "step": 54600
     },
     {
       "epoch": 5.892183288409703,
-      "grad_norm": 0.6870036125183105,
-      "learning_rate": 0.0002468774959525094,
-      "loss": 3.3366,
+      "grad_norm": 0.6861412525177002,
+      "learning_rate": 0.0002468710199676201,
+      "loss": 3.3381,
       "step": 54650
     },
     {
       "epoch": 5.8975741239892185,
-      "grad_norm": 0.7101106643676758,
-      "learning_rate": 0.00024655369670804097,
-      "loss": 3.3365,
+      "grad_norm": 0.6980893611907959,
+      "learning_rate": 0.0002465472207231516,
+      "loss": 3.3387,
       "step": 54700
     },
     {
       "epoch": 5.902964959568733,
-      "grad_norm": 0.7199245691299438,
-      "learning_rate": 0.0002462298974635726,
-      "loss": 3.3377,
+      "grad_norm": 0.7005370855331421,
+      "learning_rate": 0.0002462234214786832,
+      "loss": 3.3397,
       "step": 54750
     },
     {
       "epoch": 5.908355795148248,
-      "grad_norm": 0.714850127696991,
-      "learning_rate": 0.00024590609821910413,
-      "loss": 3.3418,
+      "grad_norm": 0.6975410580635071,
+      "learning_rate": 0.0002458996222342148,
+      "loss": 3.3434,
       "step": 54800
     },
     {
       "epoch": 5.913746630727763,
-      "grad_norm": 0.7096270322799683,
-      "learning_rate": 0.0002455822989746357,
-      "loss": 3.3353,
+      "grad_norm": 0.7340664267539978,
+      "learning_rate": 0.00024557582298974633,
+      "loss": 3.3387,
       "step": 54850
     },
     {
       "epoch": 5.919137466307277,
-      "grad_norm": 0.691370964050293,
-      "learning_rate": 0.0002452584997301673,
-      "loss": 3.3158,
+      "grad_norm": 0.6792736649513245,
+      "learning_rate": 0.0002452520237452779,
+      "loss": 3.3196,
       "step": 54900
     },
     {
       "epoch": 5.9245283018867925,
-      "grad_norm": 0.6646804213523865,
-      "learning_rate": 0.00024493470048569883,
-      "loss": 3.3219,
+      "grad_norm": 0.6899210810661316,
+      "learning_rate": 0.0002449282245008095,
+      "loss": 3.3234,
       "step": 54950
     },
     {
       "epoch": 5.929919137466308,
-      "grad_norm": 0.6864519119262695,
-      "learning_rate": 0.00024461090124123044,
-      "loss": 3.3245,
+      "grad_norm": 0.6393539309501648,
+      "learning_rate": 0.00024460442525634104,
+      "loss": 3.3284,
       "step": 55000
     },
     {
       "epoch": 5.929919137466308,
-      "eval_accuracy": 0.3845000788276701,
-      "eval_loss": 3.375938892364502,
-      "eval_runtime": 183.427,
-      "eval_samples_per_second": 98.192,
-      "eval_steps_per_second": 6.139,
+      "eval_accuracy": 0.3842817951760508,
+      "eval_loss": 3.377272367477417,
+      "eval_runtime": 192.7259,
+      "eval_samples_per_second": 93.454,
+      "eval_steps_per_second": 5.842,
       "step": 55000
     },
     {
       "epoch": 5.935309973045822,
-      "grad_norm": 0.6417977809906006,
-      "learning_rate": 0.000244287101996762,
-      "loss": 3.3377,
+      "grad_norm": 0.6344804763793945,
+      "learning_rate": 0.00024428062601187264,
+      "loss": 3.3404,
       "step": 55050
     },
     {
       "epoch": 5.940700808625337,
-      "grad_norm": 0.7369255423545837,
-      "learning_rate": 0.00024396330275229354,
-      "loss": 3.3254,
+      "grad_norm": 0.7260132431983948,
+      "learning_rate": 0.0002439568267674042,
+      "loss": 3.3284,
       "step": 55100
     },
     {
       "epoch": 5.946091644204852,
-      "grad_norm": 0.6987436413764954,
-      "learning_rate": 0.00024363950350782512,
-      "loss": 3.3251,
+      "grad_norm": 0.7179984450340271,
+      "learning_rate": 0.00024363302752293574,
+      "loss": 3.3253,
       "step": 55150
     },
     {
       "epoch": 5.951482479784366,
-      "grad_norm": 0.689603865146637,
-      "learning_rate": 0.0002433157042633567,
-      "loss": 3.3442,
+      "grad_norm": 0.7092376947402954,
+      "learning_rate": 0.00024330922827846732,
+      "loss": 3.3465,
       "step": 55200
     },
     {
       "epoch": 5.9568733153638815,
-      "grad_norm": 0.7123790979385376,
-      "learning_rate": 0.00024299190501888827,
-      "loss": 3.3154,
+      "grad_norm": 0.667742133140564,
+      "learning_rate": 0.0002429854290339989,
+      "loss": 3.3194,
       "step": 55250
     },
     {
       "epoch": 5.962264150943396,
-      "grad_norm": 0.666307270526886,
-      "learning_rate": 0.00024266810577441985,
-      "loss": 3.3173,
+      "grad_norm": 0.6692838072776794,
+      "learning_rate": 0.00024266162978953048,
+      "loss": 3.3196,
       "step": 55300
     },
     {
       "epoch": 5.967654986522911,
-      "grad_norm": 0.6958657503128052,
-      "learning_rate": 0.00024234430652995143,
-      "loss": 3.3256,
+      "grad_norm": 0.7141178250312805,
+      "learning_rate": 0.00024233783054506203,
+      "loss": 3.3277,
       "step": 55350
     },
     {
       "epoch": 5.973045822102426,
-      "grad_norm": 0.7178813219070435,
-      "learning_rate": 0.00024202050728548298,
-      "loss": 3.3286,
+      "grad_norm": 0.7486997842788696,
+      "learning_rate": 0.0002420140313005936,
+      "loss": 3.3322,
       "step": 55400
     },
     {
       "epoch": 5.97843665768194,
-      "grad_norm": 0.690979540348053,
-      "learning_rate": 0.00024169670804101456,
-      "loss": 3.3219,
+      "grad_norm": 0.6665425896644592,
+      "learning_rate": 0.00024169023205612518,
+      "loss": 3.3243,
       "step": 55450
     },
     {
       "epoch": 5.9838274932614555,
-      "grad_norm": 0.6696831583976746,
-      "learning_rate": 0.0002413729087965461,
-      "loss": 3.3019,
+      "grad_norm": 0.7089661359786987,
+      "learning_rate": 0.00024136643281165676,
+      "loss": 3.3036,
       "step": 55500
     },
     {
       "epoch": 5.989218328840971,
-      "grad_norm": 0.6762212514877319,
-      "learning_rate": 0.00024104910955207768,
-      "loss": 3.3312,
+      "grad_norm": 0.6869831681251526,
+      "learning_rate": 0.00024104263356718834,
+      "loss": 3.3346,
       "step": 55550
     },
     {
       "epoch": 5.994609164420485,
-      "grad_norm": 0.6979408264160156,
-      "learning_rate": 0.00024072531030760926,
-      "loss": 3.3208,
+      "grad_norm": 0.7549130320549011,
+      "learning_rate": 0.0002407188343227199,
+      "loss": 3.3224,
       "step": 55600
     },
     {
       "epoch": 6.0,
-      "grad_norm": 1.4232317209243774,
-      "learning_rate": 0.00024040151106314084,
-      "loss": 3.3492,
+      "grad_norm": 1.4368031024932861,
+      "learning_rate": 0.00024039503507825147,
+      "loss": 3.3519,
       "step": 55650
     },
     {
       "epoch": 6.005390835579515,
-      "grad_norm": 0.6714563965797424,
-      "learning_rate": 0.0002400777118186724,
-      "loss": 3.2286,
+      "grad_norm": 0.6394692659378052,
+      "learning_rate": 0.00024007123583378302,
+      "loss": 3.2314,
       "step": 55700
     },
     {
       "epoch": 6.010781671159029,
-      "grad_norm": 0.6843574047088623,
-      "learning_rate": 0.00023975391257420397,
-      "loss": 3.2305,
+      "grad_norm": 0.68421870470047,
+      "learning_rate": 0.0002397474365893146,
+      "loss": 3.2345,
       "step": 55750
     },
     {
       "epoch": 6.0161725067385445,
-      "grad_norm": 0.7195513248443604,
-      "learning_rate": 0.00023943011332973555,
-      "loss": 3.2422,
+      "grad_norm": 0.6972149610519409,
+      "learning_rate": 0.00023942363734484617,
+      "loss": 3.2439,
       "step": 55800
     },
     {
       "epoch": 6.02156334231806,
-      "grad_norm": 0.6866133809089661,
-      "learning_rate": 0.00023910631408526712,
-      "loss": 3.2445,
+      "grad_norm": 0.6584705710411072,
+      "learning_rate": 0.00023909983810037775,
+      "loss": 3.2473,
       "step": 55850
     },
     {
       "epoch": 6.026954177897574,
-      "grad_norm": 0.7409161925315857,
-      "learning_rate": 0.0002387825148407987,
-      "loss": 3.2428,
+      "grad_norm": 0.7260363101959229,
+      "learning_rate": 0.00023877603885590933,
+      "loss": 3.2467,
       "step": 55900
     },
     {
       "epoch": 6.032345013477089,
-      "grad_norm": 0.7081006169319153,
-      "learning_rate": 0.00023845871559633025,
-      "loss": 3.2432,
+      "grad_norm": 0.6996325254440308,
+      "learning_rate": 0.0002384522396114409,
+      "loss": 3.2454,
       "step": 55950
     },
     {
       "epoch": 6.037735849056604,
-      "grad_norm": 0.736912727355957,
-      "learning_rate": 0.0002381413923367512,
-      "loss": 3.2569,
+      "grad_norm": 0.7425010800361633,
+      "learning_rate": 0.00023812844036697248,
+      "loss": 3.2601,
       "step": 56000
     },
     {
       "epoch": 6.037735849056604,
-      "eval_accuracy": 0.38479583198830114,
-      "eval_loss": 3.3801846504211426,
-      "eval_runtime": 183.4392,
-      "eval_samples_per_second": 98.185,
-      "eval_steps_per_second": 6.138,
+      "eval_accuracy": 0.38408187386247217,
+      "eval_loss": 3.3832037448883057,
+      "eval_runtime": 192.4051,
+      "eval_samples_per_second": 93.61,
+      "eval_steps_per_second": 5.852,
       "step": 56000
     },
     {
       "epoch": 6.0431266846361185,
-      "grad_norm": 0.7223060131072998,
-      "learning_rate": 0.00023781759309228275,
-      "loss": 3.2541,
+      "grad_norm": 0.7906090617179871,
+      "learning_rate": 0.000237804641122504,
+      "loss": 3.2573,
       "step": 56050
     },
     {
       "epoch": 6.048517520215634,
-      "grad_norm": 0.853127658367157,
-      "learning_rate": 0.00023749379384781433,
-      "loss": 3.2419,
+      "grad_norm": 0.7113637924194336,
+      "learning_rate": 0.00023748084187803558,
+      "loss": 3.2461,
       "step": 56100
     },
     {
       "epoch": 6.053908355795148,
-      "grad_norm": 0.7139176726341248,
-      "learning_rate": 0.0002371699946033459,
-      "loss": 3.2564,
+      "grad_norm": 0.7238534688949585,
+      "learning_rate": 0.00023715704263356716,
+      "loss": 3.2588,
       "step": 56150
     },
     {
       "epoch": 6.059299191374663,
-      "grad_norm": 0.7075513005256653,
-      "learning_rate": 0.0002368461953588775,
-      "loss": 3.2494,
+      "grad_norm": 0.7175899147987366,
+      "learning_rate": 0.00023683324338909874,
+      "loss": 3.2526,
       "step": 56200
     },
     {
       "epoch": 6.064690026954178,
-      "grad_norm": 0.7035233974456787,
-      "learning_rate": 0.00023652239611440904,
-      "loss": 3.2336,
+      "grad_norm": 0.6635222434997559,
+      "learning_rate": 0.00023650944414463032,
+      "loss": 3.2355,
       "step": 56250
     },
     {
       "epoch": 6.070080862533692,
-      "grad_norm": 0.7118614912033081,
-      "learning_rate": 0.00023619859686994062,
-      "loss": 3.2595,
+      "grad_norm": 0.7341075539588928,
+      "learning_rate": 0.0002361856449001619,
+      "loss": 3.2623,
       "step": 56300
     },
     {
       "epoch": 6.0754716981132075,
-      "grad_norm": 0.7053843140602112,
-      "learning_rate": 0.00023587479762547217,
-      "loss": 3.2453,
+      "grad_norm": 0.7347299456596375,
+      "learning_rate": 0.00023586184565569347,
+      "loss": 3.2475,
       "step": 56350
     },
     {
       "epoch": 6.080862533692723,
-      "grad_norm": 0.7055346369743347,
-      "learning_rate": 0.00023555099838100374,
-      "loss": 3.2494,
+      "grad_norm": 0.6943380832672119,
+      "learning_rate": 0.00023553804641122502,
+      "loss": 3.2508,
       "step": 56400
     },
     {
       "epoch": 6.086253369272237,
-      "grad_norm": 0.739841878414154,
-      "learning_rate": 0.00023522719913653532,
-      "loss": 3.2442,
+      "grad_norm": 0.7580903172492981,
+      "learning_rate": 0.00023521424716675657,
+      "loss": 3.2458,
       "step": 56450
     },
     {
       "epoch": 6.091644204851752,
-      "grad_norm": 0.7180924415588379,
-      "learning_rate": 0.0002349033998920669,
-      "loss": 3.2631,
+      "grad_norm": 0.7593175172805786,
+      "learning_rate": 0.00023489692390717752,
+      "loss": 3.2674,
       "step": 56500
     },
     {
       "epoch": 6.097035040431267,
-      "grad_norm": 0.7231001853942871,
-      "learning_rate": 0.00023457960064759848,
-      "loss": 3.2418,
+      "grad_norm": 0.7214465141296387,
+      "learning_rate": 0.0002345731246627091,
+      "loss": 3.2436,
       "step": 56550
     },
     {
       "epoch": 6.1024258760107815,
-      "grad_norm": 0.706662118434906,
-      "learning_rate": 0.00023425580140313005,
-      "loss": 3.2493,
+      "grad_norm": 0.7108901143074036,
+      "learning_rate": 0.00023424932541824068,
+      "loss": 3.2511,
       "step": 56600
     },
     {
       "epoch": 6.107816711590297,
-      "grad_norm": 0.6997068524360657,
-      "learning_rate": 0.00023393200215866163,
-      "loss": 3.2453,
+      "grad_norm": 0.7234324216842651,
+      "learning_rate": 0.00023392552617377226,
+      "loss": 3.2504,
       "step": 56650
     },
     {
       "epoch": 6.113207547169812,
-      "grad_norm": 0.6997630000114441,
-      "learning_rate": 0.00023360820291419316,
-      "loss": 3.2687,
+      "grad_norm": 0.7587084174156189,
+      "learning_rate": 0.00023360172692930384,
+      "loss": 3.2728,
       "step": 56700
     },
     {
       "epoch": 6.118598382749326,
-      "grad_norm": 0.7305110692977905,
-      "learning_rate": 0.00023328440366972473,
-      "loss": 3.2583,
+      "grad_norm": 0.6924833059310913,
+      "learning_rate": 0.00023327792768483539,
+      "loss": 3.2619,
       "step": 56750
     },
     {
       "epoch": 6.123989218328841,
-      "grad_norm": 0.7409155368804932,
-      "learning_rate": 0.0002329606044252563,
-      "loss": 3.2608,
+      "grad_norm": 0.6919812560081482,
+      "learning_rate": 0.00023295412844036694,
+      "loss": 3.2643,
       "step": 56800
     },
     {
       "epoch": 6.129380053908355,
-      "grad_norm": 0.6931422352790833,
-      "learning_rate": 0.0002326368051807879,
-      "loss": 3.2628,
+      "grad_norm": 0.7377097606658936,
+      "learning_rate": 0.00023263032919589851,
+      "loss": 3.2655,
       "step": 56850
     },
     {
       "epoch": 6.1347708894878705,
-      "grad_norm": 0.7668306827545166,
-      "learning_rate": 0.00023231300593631947,
-      "loss": 3.2696,
+      "grad_norm": 0.7574697136878967,
+      "learning_rate": 0.0002323065299514301,
+      "loss": 3.273,
       "step": 56900
     },
     {
       "epoch": 6.140161725067386,
-      "grad_norm": 0.7603601217269897,
-      "learning_rate": 0.00023198920669185104,
-      "loss": 3.2466,
+      "grad_norm": 0.7605524063110352,
+      "learning_rate": 0.00023198273070696167,
+      "loss": 3.2501,
       "step": 56950
     },
     {
       "epoch": 6.1455525606469,
-      "grad_norm": 0.7167591452598572,
-      "learning_rate": 0.00023166540744738262,
-      "loss": 3.2452,
+      "grad_norm": 0.7264750599861145,
+      "learning_rate": 0.00023165893146249325,
+      "loss": 3.2484,
       "step": 57000
     },
     {
       "epoch": 6.1455525606469,
-      "eval_accuracy": 0.38466523121714813,
-      "eval_loss": 3.3798298835754395,
-      "eval_runtime": 183.6772,
-      "eval_samples_per_second": 98.058,
-      "eval_steps_per_second": 6.13,
+      "eval_accuracy": 0.3842636501437858,
+      "eval_loss": 3.380316972732544,
+      "eval_runtime": 192.1964,
+      "eval_samples_per_second": 93.711,
+      "eval_steps_per_second": 5.859,
       "step": 57000
     },
     {
       "epoch": 6.150943396226415,
-      "grad_norm": 0.6705722808837891,
-      "learning_rate": 0.0002313416082029142,
-      "loss": 3.2731,
+      "grad_norm": 0.7119517922401428,
+      "learning_rate": 0.0002313351322180248,
+      "loss": 3.2774,
       "step": 57050
     },
     {
       "epoch": 6.15633423180593,
-      "grad_norm": 0.691725492477417,
-      "learning_rate": 0.00023101780895844572,
-      "loss": 3.2644,
+      "grad_norm": 0.7369308471679688,
+      "learning_rate": 0.00023101133297355638,
+      "loss": 3.2666,
       "step": 57100
     },
     {
       "epoch": 6.1617250673854445,
-      "grad_norm": 0.6815258264541626,
-      "learning_rate": 0.0002306940097139773,
-      "loss": 3.247,
+      "grad_norm": 0.6724468469619751,
+      "learning_rate": 0.00023068753372908795,
+      "loss": 3.248,
       "step": 57150
     },
     {
       "epoch": 6.16711590296496,
-      "grad_norm": 0.6972390413284302,
-      "learning_rate": 0.00023037021046950888,
-      "loss": 3.2524,
+      "grad_norm": 0.694419801235199,
+      "learning_rate": 0.0002303637344846195,
+      "loss": 3.2568,
       "step": 57200
     },
     {
       "epoch": 6.172506738544475,
-      "grad_norm": 0.695257842540741,
-      "learning_rate": 0.00023004641122504046,
-      "loss": 3.256,
+      "grad_norm": 0.7059666514396667,
+      "learning_rate": 0.00023003993524015108,
+      "loss": 3.2599,
       "step": 57250
     },
     {
       "epoch": 6.177897574123989,
-      "grad_norm": 0.8920659422874451,
-      "learning_rate": 0.00022972261198057203,
-      "loss": 3.255,
+      "grad_norm": 0.6996408104896545,
+      "learning_rate": 0.00022971613599568266,
+      "loss": 3.2575,
       "step": 57300
     },
     {
       "epoch": 6.183288409703504,
-      "grad_norm": 0.7532164454460144,
-      "learning_rate": 0.0002293988127361036,
-      "loss": 3.271,
+      "grad_norm": 0.7600948810577393,
+      "learning_rate": 0.0002293923367512142,
+      "loss": 3.2761,
       "step": 57350
     },
     {
       "epoch": 6.188679245283019,
-      "grad_norm": 0.6974930167198181,
-      "learning_rate": 0.00022907501349163516,
-      "loss": 3.2621,
+      "grad_norm": 0.6894692182540894,
+      "learning_rate": 0.0002290685375067458,
+      "loss": 3.2654,
       "step": 57400
     },
     {
       "epoch": 6.1940700808625335,
-      "grad_norm": 0.7437677979469299,
-      "learning_rate": 0.00022875121424716674,
-      "loss": 3.2682,
+      "grad_norm": 0.7871859073638916,
+      "learning_rate": 0.00022874473826227736,
+      "loss": 3.2691,
       "step": 57450
     },
     {
       "epoch": 6.199460916442049,
-      "grad_norm": 0.7247287631034851,
-      "learning_rate": 0.00022842741500269832,
-      "loss": 3.2651,
+      "grad_norm": 0.7082929611206055,
+      "learning_rate": 0.00022842093901780894,
+      "loss": 3.2687,
       "step": 57500
     },
     {
       "epoch": 6.204851752021563,
-      "grad_norm": 0.7260898947715759,
-      "learning_rate": 0.00022810361575822987,
-      "loss": 3.2551,
+      "grad_norm": 0.734351396560669,
+      "learning_rate": 0.00022809713977334052,
+      "loss": 3.2565,
       "step": 57550
     },
     {
       "epoch": 6.210242587601078,
-      "grad_norm": 0.783756673336029,
-      "learning_rate": 0.00022777981651376145,
-      "loss": 3.2437,
+      "grad_norm": 0.7525688409805298,
+      "learning_rate": 0.0002277733405288721,
+      "loss": 3.2453,
       "step": 57600
     },
     {
       "epoch": 6.215633423180593,
-      "grad_norm": 0.7135037779808044,
-      "learning_rate": 0.00022745601726929302,
-      "loss": 3.2631,
+      "grad_norm": 0.7365319132804871,
+      "learning_rate": 0.00022744954128440365,
+      "loss": 3.2682,
       "step": 57650
     },
     {
       "epoch": 6.2210242587601075,
-      "grad_norm": 0.7173543572425842,
-      "learning_rate": 0.00022713221802482457,
-      "loss": 3.2685,
+      "grad_norm": 0.7436929941177368,
+      "learning_rate": 0.0002271257420399352,
+      "loss": 3.2712,
       "step": 57700
     },
     {
       "epoch": 6.226415094339623,
-      "grad_norm": 0.7847826480865479,
-      "learning_rate": 0.00022680841878035615,
-      "loss": 3.283,
+      "grad_norm": 0.7687820196151733,
+      "learning_rate": 0.00022680194279546678,
+      "loss": 3.2876,
       "step": 57750
     },
     {
       "epoch": 6.231805929919138,
-      "grad_norm": 0.7096986770629883,
-      "learning_rate": 0.00022648461953588773,
-      "loss": 3.28,
+      "grad_norm": 0.7185782194137573,
+      "learning_rate": 0.00022647814355099835,
+      "loss": 3.2836,
       "step": 57800
     },
     {
       "epoch": 6.237196765498652,
-      "grad_norm": 0.7000459432601929,
-      "learning_rate": 0.0002261608202914193,
-      "loss": 3.2509,
+      "grad_norm": 0.7000666260719299,
+      "learning_rate": 0.00022615434430652993,
+      "loss": 3.2538,
       "step": 57850
     },
     {
       "epoch": 6.242587601078167,
-      "grad_norm": 0.6716017127037048,
-      "learning_rate": 0.00022583702104695088,
-      "loss": 3.2718,
+      "grad_norm": 0.6874049305915833,
+      "learning_rate": 0.0002258305450620615,
+      "loss": 3.2769,
       "step": 57900
     },
     {
       "epoch": 6.247978436657682,
-      "grad_norm": 0.712790310382843,
-      "learning_rate": 0.00022551322180248246,
-      "loss": 3.2622,
+      "grad_norm": 0.7027003169059753,
+      "learning_rate": 0.0002255067458175931,
+      "loss": 3.2634,
       "step": 57950
     },
     {
       "epoch": 6.2533692722371965,
-      "grad_norm": 0.7777376770973206,
-      "learning_rate": 0.00022518942255801399,
-      "loss": 3.2473,
+      "grad_norm": 0.7743560075759888,
+      "learning_rate": 0.00022518294657312467,
+      "loss": 3.2507,
       "step": 58000
     },
     {
       "epoch": 6.2533692722371965,
-      "eval_accuracy": 0.3850733314637793,
-      "eval_loss": 3.378077983856201,
-      "eval_runtime": 183.3483,
-      "eval_samples_per_second": 98.234,
-      "eval_steps_per_second": 6.141,
+      "eval_accuracy": 0.3847341171480225,
+      "eval_loss": 3.3779726028442383,
+      "eval_runtime": 193.435,
+      "eval_samples_per_second": 93.111,
+      "eval_steps_per_second": 5.821,
       "step": 58000
     },
     {
       "epoch": 6.258760107816712,
-      "grad_norm": 0.7233048677444458,
-      "learning_rate": 0.00022486562331354556,
-      "loss": 3.2652,
+      "grad_norm": 0.7026686072349548,
+      "learning_rate": 0.0002248591473286562,
+      "loss": 3.2677,
       "step": 58050
     },
     {
       "epoch": 6.264150943396227,
-      "grad_norm": 0.7433217167854309,
-      "learning_rate": 0.00022454182406907714,
-      "loss": 3.2765,
+      "grad_norm": 0.7107704877853394,
+      "learning_rate": 0.00022453534808418777,
+      "loss": 3.2788,
       "step": 58100
     },
     {
       "epoch": 6.269541778975741,
-      "grad_norm": 0.7410855889320374,
-      "learning_rate": 0.00022421802482460872,
-      "loss": 3.2542,
+      "grad_norm": 0.7971775531768799,
+      "learning_rate": 0.00022421154883971934,
+      "loss": 3.2586,
       "step": 58150
     },
     {
       "epoch": 6.274932614555256,
-      "grad_norm": 0.7226539850234985,
-      "learning_rate": 0.00022390070156502967,
-      "loss": 3.2707,
+      "grad_norm": 0.7312342524528503,
+      "learning_rate": 0.00022388774959525092,
+      "loss": 3.2748,
       "step": 58200
     },
     {
       "epoch": 6.280323450134771,
-      "grad_norm": 0.7170883417129517,
-      "learning_rate": 0.00022357690232056125,
-      "loss": 3.2763,
+      "grad_norm": 0.7458978295326233,
+      "learning_rate": 0.0002235639503507825,
+      "loss": 3.2783,
       "step": 58250
     },
     {
       "epoch": 6.285714285714286,
-      "grad_norm": 0.7725226879119873,
-      "learning_rate": 0.0002232531030760928,
-      "loss": 3.2611,
+      "grad_norm": 0.7790622115135193,
+      "learning_rate": 0.00022324015110631408,
+      "loss": 3.2638,
       "step": 58300
     },
     {
       "epoch": 6.291105121293801,
-      "grad_norm": 0.766638457775116,
-      "learning_rate": 0.00022292930383162435,
-      "loss": 3.2514,
+      "grad_norm": 0.7516189217567444,
+      "learning_rate": 0.00022291635186184565,
+      "loss": 3.2535,
       "step": 58350
     },
     {
       "epoch": 6.296495956873315,
-      "grad_norm": 0.7184855341911316,
-      "learning_rate": 0.00022260550458715593,
-      "loss": 3.2592,
+      "grad_norm": 0.7125251293182373,
+      "learning_rate": 0.0002225925526173772,
+      "loss": 3.2629,
       "step": 58400
     },
     {
       "epoch": 6.30188679245283,
-      "grad_norm": 0.7369925379753113,
-      "learning_rate": 0.0002222817053426875,
-      "loss": 3.2782,
+      "grad_norm": 0.7433750629425049,
+      "learning_rate": 0.00022226875337290878,
+      "loss": 3.281,
       "step": 58450
     },
     {
       "epoch": 6.307277628032345,
-      "grad_norm": 0.6875489354133606,
-      "learning_rate": 0.00022195790609821908,
-      "loss": 3.2501,
+      "grad_norm": 0.6968837380409241,
+      "learning_rate": 0.00022194495412844033,
+      "loss": 3.2532,
       "step": 58500
     },
     {
       "epoch": 6.3126684636118595,
-      "grad_norm": 0.7300560474395752,
-      "learning_rate": 0.00022163410685375066,
-      "loss": 3.2723,
+      "grad_norm": 0.7056468725204468,
+      "learning_rate": 0.0002216211548839719,
+      "loss": 3.2739,
       "step": 58550
     },
     {
       "epoch": 6.318059299191375,
-      "grad_norm": 0.7411653995513916,
-      "learning_rate": 0.00022131030760928224,
-      "loss": 3.2799,
+      "grad_norm": 0.7373625040054321,
+      "learning_rate": 0.0002212973556395035,
+      "loss": 3.2826,
       "step": 58600
     },
     {
       "epoch": 6.32345013477089,
-      "grad_norm": 0.7001060843467712,
-      "learning_rate": 0.00022098650836481382,
-      "loss": 3.2727,
+      "grad_norm": 0.6784160733222961,
+      "learning_rate": 0.00022097355639503507,
+      "loss": 3.2768,
       "step": 58650
     },
     {
       "epoch": 6.328840970350404,
-      "grad_norm": 0.7271462678909302,
-      "learning_rate": 0.0002206627091203454,
-      "loss": 3.2883,
+      "grad_norm": 0.7374270558357239,
+      "learning_rate": 0.00022064975715056664,
+      "loss": 3.2913,
       "step": 58700
     },
     {
       "epoch": 6.334231805929919,
-      "grad_norm": 0.7320724129676819,
-      "learning_rate": 0.00022033890987587692,
-      "loss": 3.2852,
+      "grad_norm": 0.7141253352165222,
+      "learning_rate": 0.0002203259579060982,
+      "loss": 3.2874,
       "step": 58750
     },
     {
       "epoch": 6.339622641509434,
-      "grad_norm": 0.7684124112129211,
-      "learning_rate": 0.0002200151106314085,
-      "loss": 3.2706,
+      "grad_norm": 0.7287693023681641,
+      "learning_rate": 0.00022000863464651915,
+      "loss": 3.2745,
       "step": 58800
     },
     {
       "epoch": 6.345013477088949,
-      "grad_norm": 0.7218632102012634,
-      "learning_rate": 0.00021969131138694007,
-      "loss": 3.2767,
+      "grad_norm": 0.7196800708770752,
+      "learning_rate": 0.0002196848354020507,
+      "loss": 3.2781,
       "step": 58850
     },
     {
       "epoch": 6.350404312668464,
-      "grad_norm": 0.7296848297119141,
-      "learning_rate": 0.00021936751214247165,
-      "loss": 3.2619,
+      "grad_norm": 0.7499868869781494,
+      "learning_rate": 0.00021936103615758227,
+      "loss": 3.266,
       "step": 58900
     },
     {
       "epoch": 6.355795148247978,
-      "grad_norm": 0.7030173540115356,
-      "learning_rate": 0.00021904371289800323,
-      "loss": 3.2828,
+      "grad_norm": 0.7269318699836731,
+      "learning_rate": 0.00021903723691311385,
+      "loss": 3.2854,
       "step": 58950
     },
     {
       "epoch": 6.361185983827493,
-      "grad_norm": 0.6917963624000549,
-      "learning_rate": 0.0002187199136535348,
-      "loss": 3.2807,
+      "grad_norm": 0.7103339433670044,
+      "learning_rate": 0.00021871343766864543,
+      "loss": 3.2835,
       "step": 59000
     },
     {
       "epoch": 6.361185983827493,
-      "eval_accuracy": 0.38561768243172984,
-      "eval_loss": 3.3727059364318848,
-      "eval_runtime": 183.1739,
-      "eval_samples_per_second": 98.327,
-      "eval_steps_per_second": 6.147,
+      "eval_accuracy": 0.3855278264935073,
+      "eval_loss": 3.3729255199432373,
+      "eval_runtime": 192.3148,
+      "eval_samples_per_second": 93.654,
+      "eval_steps_per_second": 5.855,
       "step": 59000
     },
     {
       "epoch": 6.366576819407008,
-      "grad_norm": 0.7581689357757568,
-      "learning_rate": 0.00021839611440906638,
-      "loss": 3.2838,
+      "grad_norm": 0.7235169410705566,
+      "learning_rate": 0.00021838963842417698,
+      "loss": 3.2866,
       "step": 59050
     },
     {
       "epoch": 6.3719676549865225,
-      "grad_norm": 0.7609435319900513,
-      "learning_rate": 0.00021807231516459793,
-      "loss": 3.2745,
+      "grad_norm": 0.7528637647628784,
+      "learning_rate": 0.00021806583917970856,
+      "loss": 3.2782,
       "step": 59100
     },
     {
       "epoch": 6.377358490566038,
-      "grad_norm": 0.747968316078186,
-      "learning_rate": 0.00021774851592012948,
-      "loss": 3.2765,
+      "grad_norm": 0.8133290410041809,
+      "learning_rate": 0.00021774203993524014,
+      "loss": 3.2784,
       "step": 59150
     },
     {
       "epoch": 6.382749326145553,
-      "grad_norm": 0.733305037021637,
-      "learning_rate": 0.00021742471667566106,
-      "loss": 3.2944,
+      "grad_norm": 0.7441257834434509,
+      "learning_rate": 0.00021741824069077171,
+      "loss": 3.2988,
       "step": 59200
     },
     {
       "epoch": 6.388140161725067,
-      "grad_norm": 0.7445916533470154,
-      "learning_rate": 0.00021710091743119264,
-      "loss": 3.2692,
+      "grad_norm": 0.7935904860496521,
+      "learning_rate": 0.00021709444144630326,
+      "loss": 3.2734,
       "step": 59250
     },
     {
       "epoch": 6.393530997304582,
-      "grad_norm": 0.7506402730941772,
-      "learning_rate": 0.00021677711818672422,
-      "loss": 3.2604,
+      "grad_norm": 0.7205762267112732,
+      "learning_rate": 0.00021677064220183484,
+      "loss": 3.2624,
       "step": 59300
     },
     {
       "epoch": 6.398921832884097,
-      "grad_norm": 0.7963356971740723,
-      "learning_rate": 0.0002164533189422558,
-      "loss": 3.2751,
+      "grad_norm": 0.8088115453720093,
+      "learning_rate": 0.00021644684295736642,
+      "loss": 3.2771,
       "step": 59350
     },
     {
       "epoch": 6.404312668463612,
-      "grad_norm": 0.7167596817016602,
-      "learning_rate": 0.00021612951969778734,
-      "loss": 3.2961,
+      "grad_norm": 0.7442873120307922,
+      "learning_rate": 0.00021612304371289797,
+      "loss": 3.2981,
       "step": 59400
     },
     {
       "epoch": 6.409703504043127,
-      "grad_norm": 0.7183796763420105,
-      "learning_rate": 0.00021580572045331892,
-      "loss": 3.2764,
+      "grad_norm": 0.724807858467102,
+      "learning_rate": 0.00021579924446842955,
+      "loss": 3.2771,
       "step": 59450
     },
     {
       "epoch": 6.415094339622642,
-      "grad_norm": 0.6938195824623108,
-      "learning_rate": 0.0002154819212088505,
-      "loss": 3.2624,
+      "grad_norm": 0.7131649255752563,
+      "learning_rate": 0.00021547544522396113,
+      "loss": 3.2643,
       "step": 59500
     },
     {
       "epoch": 6.420485175202156,
-      "grad_norm": 0.702420175075531,
-      "learning_rate": 0.00021515812196438208,
-      "loss": 3.2971,
+      "grad_norm": 0.7176441550254822,
+      "learning_rate": 0.0002151516459794927,
+      "loss": 3.299,
       "step": 59550
     },
     {
       "epoch": 6.425876010781671,
-      "grad_norm": 0.7254041433334351,
-      "learning_rate": 0.00021483432271991363,
-      "loss": 3.2668,
+      "grad_norm": 0.732309877872467,
+      "learning_rate": 0.00021482784673502428,
+      "loss": 3.2688,
       "step": 59600
     },
     {
       "epoch": 6.431266846361186,
-      "grad_norm": 0.7375361323356628,
-      "learning_rate": 0.0002145105234754452,
-      "loss": 3.2671,
+      "grad_norm": 0.7032806277275085,
+      "learning_rate": 0.00021450404749055586,
+      "loss": 3.2689,
       "step": 59650
     },
     {
       "epoch": 6.436657681940701,
-      "grad_norm": 0.7460655570030212,
-      "learning_rate": 0.00021418672423097676,
-      "loss": 3.305,
+      "grad_norm": 0.758945882320404,
+      "learning_rate": 0.00021418024824608738,
+      "loss": 3.3085,
       "step": 59700
     },
     {
       "epoch": 6.442048517520216,
-      "grad_norm": 0.7111104130744934,
-      "learning_rate": 0.00021386292498650833,
-      "loss": 3.286,
+      "grad_norm": 0.6998205184936523,
+      "learning_rate": 0.00021385644900161896,
+      "loss": 3.2888,
       "step": 59750
     },
     {
       "epoch": 6.44743935309973,
-      "grad_norm": 0.7258356809616089,
-      "learning_rate": 0.0002135391257420399,
-      "loss": 3.2734,
+      "grad_norm": 0.7199980020523071,
+      "learning_rate": 0.00021353264975715054,
+      "loss": 3.2778,
       "step": 59800
     },
     {
       "epoch": 6.452830188679245,
-      "grad_norm": 0.7276678085327148,
-      "learning_rate": 0.0002132153264975715,
-      "loss": 3.2684,
+      "grad_norm": 0.7492018938064575,
+      "learning_rate": 0.00021320885051268211,
+      "loss": 3.2706,
       "step": 59850
     },
     {
       "epoch": 6.45822102425876,
-      "grad_norm": 0.7468143701553345,
-      "learning_rate": 0.00021289152725310307,
-      "loss": 3.2648,
+      "grad_norm": 0.752880334854126,
+      "learning_rate": 0.0002128850512682137,
+      "loss": 3.2673,
       "step": 59900
     },
     {
       "epoch": 6.463611859838275,
-      "grad_norm": 0.7350443005561829,
-      "learning_rate": 0.00021256772800863464,
-      "loss": 3.2631,
+      "grad_norm": 0.6855457425117493,
+      "learning_rate": 0.00021256125202374527,
+      "loss": 3.2668,
       "step": 59950
     },
     {
       "epoch": 6.46900269541779,
-      "grad_norm": 0.6993089318275452,
-      "learning_rate": 0.0002122439287641662,
-      "loss": 3.2817,
+      "grad_norm": 0.7635577321052551,
+      "learning_rate": 0.00021223745277927685,
+      "loss": 3.2854,
       "step": 60000
     },
     {
       "epoch": 6.46900269541779,
-      "eval_accuracy": 0.38607826202317547,
-      "eval_loss": 3.366489887237549,
-      "eval_runtime": 183.7834,
-      "eval_samples_per_second": 98.001,
-      "eval_steps_per_second": 6.127,
+      "eval_accuracy": 0.3857989154486044,
+      "eval_loss": 3.368502616882324,
+      "eval_runtime": 193.0288,
+      "eval_samples_per_second": 93.307,
+      "eval_steps_per_second": 5.833,
       "step": 60000
     },
     {
       "epoch": 6.474393530997305,
-      "grad_norm": 0.7127420902252197,
-      "learning_rate": 0.00021192012951969775,
-      "loss": 3.2576,
+      "grad_norm": 0.7124029397964478,
+      "learning_rate": 0.00021191365353480843,
+      "loss": 3.2627,
       "step": 60050
     },
     {
       "epoch": 6.479784366576819,
-      "grad_norm": 0.7203495502471924,
-      "learning_rate": 0.00021159633027522932,
-      "loss": 3.2822,
+      "grad_norm": 0.7748430371284485,
+      "learning_rate": 0.00021158985429033995,
+      "loss": 3.2842,
       "step": 60100
     },
     {
       "epoch": 6.485175202156334,
-      "grad_norm": 0.7426835298538208,
-      "learning_rate": 0.0002112725310307609,
-      "loss": 3.2841,
+      "grad_norm": 0.7648136615753174,
+      "learning_rate": 0.00021126605504587153,
+      "loss": 3.2864,
       "step": 60150
     },
     {
       "epoch": 6.490566037735849,
-      "grad_norm": 0.7064744234085083,
-      "learning_rate": 0.00021094873178629248,
-      "loss": 3.2526,
+      "grad_norm": 0.7680351138114929,
+      "learning_rate": 0.0002109422558014031,
+      "loss": 3.2544,
       "step": 60200
     },
     {
       "epoch": 6.495956873315364,
-      "grad_norm": 0.7300220131874084,
-      "learning_rate": 0.00021063140852671343,
-      "loss": 3.2619,
+      "grad_norm": 0.7273554801940918,
+      "learning_rate": 0.00021061845655693468,
+      "loss": 3.2658,
       "step": 60250
     },
     {
       "epoch": 6.501347708894879,
-      "grad_norm": 0.7384023666381836,
-      "learning_rate": 0.000210307609282245,
-      "loss": 3.297,
+      "grad_norm": 0.7001141905784607,
+      "learning_rate": 0.00021029465731246626,
+      "loss": 3.2999,
       "step": 60300
     },
     {
       "epoch": 6.506738544474393,
-      "grad_norm": 0.757164478302002,
-      "learning_rate": 0.00020998381003777653,
-      "loss": 3.2567,
+      "grad_norm": 0.7441627979278564,
+      "learning_rate": 0.00020997085806799784,
+      "loss": 3.2593,
       "step": 60350
     },
     {
       "epoch": 6.512129380053908,
-      "grad_norm": 0.7513059377670288,
-      "learning_rate": 0.0002096600107933081,
-      "loss": 3.2715,
+      "grad_norm": 0.7646014094352722,
+      "learning_rate": 0.0002096470588235294,
+      "loss": 3.2732,
       "step": 60400
     },
     {
       "epoch": 6.517520215633423,
-      "grad_norm": 0.7424511313438416,
-      "learning_rate": 0.0002093362115488397,
-      "loss": 3.275,
+      "grad_norm": 0.7458738088607788,
+      "learning_rate": 0.00020932325957906097,
+      "loss": 3.2775,
       "step": 60450
     },
     {
       "epoch": 6.5229110512129385,
-      "grad_norm": 0.704623281955719,
-      "learning_rate": 0.00020901888828926064,
-      "loss": 3.28,
+      "grad_norm": 0.7185546159744263,
+      "learning_rate": 0.00020899946033459254,
+      "loss": 3.2821,
       "step": 60500
     },
     {
       "epoch": 6.528301886792453,
-      "grad_norm": 0.7265028953552246,
-      "learning_rate": 0.00020869508904479222,
-      "loss": 3.2796,
+      "grad_norm": 0.7477337121963501,
+      "learning_rate": 0.0002086756610901241,
+      "loss": 3.2813,
       "step": 60550
     },
     {
       "epoch": 6.533692722371968,
-      "grad_norm": 0.7257518172264099,
-      "learning_rate": 0.0002083712898003238,
-      "loss": 3.2759,
+      "grad_norm": 0.744189977645874,
+      "learning_rate": 0.00020835186184565567,
+      "loss": 3.2783,
       "step": 60600
     },
     {
       "epoch": 6.539083557951482,
-      "grad_norm": 0.7516768574714661,
-      "learning_rate": 0.00020804749055585537,
-      "loss": 3.2809,
+      "grad_norm": 0.7393600344657898,
+      "learning_rate": 0.00020802806260118725,
+      "loss": 3.2835,
       "step": 60650
     },
     {
       "epoch": 6.544474393530997,
-      "grad_norm": 0.7546630501747131,
-      "learning_rate": 0.0002077236913113869,
-      "loss": 3.2721,
+      "grad_norm": 0.7577639818191528,
+      "learning_rate": 0.00020770426335671883,
+      "loss": 3.2763,
       "step": 60700
     },
     {
       "epoch": 6.549865229110512,
-      "grad_norm": 0.8262805342674255,
-      "learning_rate": 0.00020739989206691847,
-      "loss": 3.2756,
+      "grad_norm": 0.7860512733459473,
+      "learning_rate": 0.00020738046411225038,
+      "loss": 3.2792,
       "step": 60750
     },
     {
       "epoch": 6.555256064690027,
-      "grad_norm": 0.7627807259559631,
-      "learning_rate": 0.00020707609282245005,
-      "loss": 3.2692,
+      "grad_norm": 0.7300195097923279,
+      "learning_rate": 0.00020705666486778196,
+      "loss": 3.2717,
       "step": 60800
     },
     {
       "epoch": 6.560646900269542,
-      "grad_norm": 0.7664798498153687,
-      "learning_rate": 0.00020675229357798163,
-      "loss": 3.2687,
+      "grad_norm": 0.7282647490501404,
+      "learning_rate": 0.00020673286562331353,
+      "loss": 3.2709,
       "step": 60850
     },
     {
       "epoch": 6.566037735849057,
-      "grad_norm": 0.75236576795578,
-      "learning_rate": 0.0002064284943335132,
-      "loss": 3.266,
+      "grad_norm": 0.7394574284553528,
+      "learning_rate": 0.00020641554236373446,
+      "loss": 3.2698,
       "step": 60900
     },
     {
       "epoch": 6.571428571428571,
-      "grad_norm": 0.7668713331222534,
-      "learning_rate": 0.00020610469508904478,
-      "loss": 3.2746,
+      "grad_norm": 0.7292279601097107,
+      "learning_rate": 0.00020609174311926604,
+      "loss": 3.2793,
       "step": 60950
     },
     {
       "epoch": 6.576819407008086,
-      "grad_norm": 0.6870357990264893,
-      "learning_rate": 0.00020578089584457636,
-      "loss": 3.2753,
+      "grad_norm": 0.7076910138130188,
+      "learning_rate": 0.0002057679438747976,
+      "loss": 3.2781,
       "step": 61000
     },
     {
       "epoch": 6.576819407008086,
-      "eval_accuracy": 0.38637314596070393,
-      "eval_loss": 3.363420248031616,
-      "eval_runtime": 183.2196,
-      "eval_samples_per_second": 98.303,
-      "eval_steps_per_second": 6.146,
+      "eval_accuracy": 0.3862648190315529,
+      "eval_loss": 3.365630626678467,
+      "eval_runtime": 181.0252,
+      "eval_samples_per_second": 99.494,
+      "eval_steps_per_second": 6.22,
       "step": 61000
     },
     {
       "epoch": 6.5822102425876015,
-      "grad_norm": 0.7615857720375061,
-      "learning_rate": 0.00020545709660010794,
-      "loss": 3.2695,
+      "grad_norm": 0.7544005513191223,
+      "learning_rate": 0.00020544414463032916,
+      "loss": 3.2701,
       "step": 61050
     },
     {
       "epoch": 6.587601078167116,
-      "grad_norm": 0.7549526691436768,
-      "learning_rate": 0.00020513329735563946,
-      "loss": 3.288,
+      "grad_norm": 0.7452325820922852,
+      "learning_rate": 0.00020512034538586074,
+      "loss": 3.2901,
       "step": 61100
     },
     {
       "epoch": 6.592991913746631,
-      "grad_norm": 0.76313716173172,
-      "learning_rate": 0.00020480949811117104,
-      "loss": 3.2877,
+      "grad_norm": 0.7330053448677063,
+      "learning_rate": 0.00020479654614139232,
+      "loss": 3.2909,
       "step": 61150
     },
     {
       "epoch": 6.598382749326145,
-      "grad_norm": 0.7764831185340881,
-      "learning_rate": 0.00020448569886670262,
-      "loss": 3.2824,
+      "grad_norm": 0.7455123066902161,
+      "learning_rate": 0.0002044727468969239,
+      "loss": 3.2867,
       "step": 61200
     },
     {
       "epoch": 6.60377358490566,
-      "grad_norm": 0.7454062104225159,
-      "learning_rate": 0.0002041618996222342,
-      "loss": 3.2825,
+      "grad_norm": 0.770001232624054,
+      "learning_rate": 0.00020414894765245547,
+      "loss": 3.2835,
       "step": 61250
     },
     {
       "epoch": 6.609164420485175,
-      "grad_norm": 0.7208166718482971,
-      "learning_rate": 0.00020383810037776577,
-      "loss": 3.2659,
+      "grad_norm": 0.7624261379241943,
+      "learning_rate": 0.00020382514840798702,
+      "loss": 3.2706,
       "step": 61300
     },
     {
       "epoch": 6.6145552560646905,
-      "grad_norm": 0.7168710827827454,
-      "learning_rate": 0.00020351430113329735,
-      "loss": 3.2976,
+      "grad_norm": 0.7015934586524963,
+      "learning_rate": 0.0002035013491635186,
+      "loss": 3.3005,
       "step": 61350
     },
     {
       "epoch": 6.619946091644205,
-      "grad_norm": 0.7853072285652161,
-      "learning_rate": 0.00020319050188882893,
-      "loss": 3.2847,
+      "grad_norm": 0.7578527331352234,
+      "learning_rate": 0.00020317754991905015,
+      "loss": 3.2863,
       "step": 61400
     },
     {
       "epoch": 6.62533692722372,
-      "grad_norm": 0.7360592484474182,
-      "learning_rate": 0.00020286670264436048,
-      "loss": 3.2946,
+      "grad_norm": 0.7222304940223694,
+      "learning_rate": 0.00020285375067458173,
+      "loss": 3.2974,
       "step": 61450
     },
     {
       "epoch": 6.630727762803234,
-      "grad_norm": 0.7565436363220215,
-      "learning_rate": 0.00020254290339989206,
-      "loss": 3.281,
+      "grad_norm": 0.7630794644355774,
+      "learning_rate": 0.0002025299514301133,
+      "loss": 3.2838,
       "step": 61500
     },
     {
       "epoch": 6.636118598382749,
-      "grad_norm": 0.7261871099472046,
-      "learning_rate": 0.0002022191041554236,
-      "loss": 3.2848,
+      "grad_norm": 0.7180784940719604,
+      "learning_rate": 0.00020220615218564489,
+      "loss": 3.2909,
       "step": 61550
     },
     {
       "epoch": 6.6415094339622645,
-      "grad_norm": 0.7736920714378357,
-      "learning_rate": 0.00020189530491095519,
-      "loss": 3.2826,
+      "grad_norm": 0.7515360116958618,
+      "learning_rate": 0.00020188235294117646,
+      "loss": 3.2846,
       "step": 61600
     },
     {
       "epoch": 6.646900269541779,
-      "grad_norm": 0.74223393201828,
-      "learning_rate": 0.00020157150566648676,
-      "loss": 3.28,
+      "grad_norm": 0.7437397241592407,
+      "learning_rate": 0.00020155855369670804,
+      "loss": 3.2835,
       "step": 61650
     },
     {
       "epoch": 6.652291105121294,
-      "grad_norm": 0.777579128742218,
-      "learning_rate": 0.00020124770642201834,
-      "loss": 3.2633,
+      "grad_norm": 0.7911229133605957,
+      "learning_rate": 0.00020123475445223956,
+      "loss": 3.265,
       "step": 61700
     },
     {
       "epoch": 6.657681940700809,
-      "grad_norm": 0.764687180519104,
-      "learning_rate": 0.0002009239071775499,
-      "loss": 3.2872,
+      "grad_norm": 0.7381808161735535,
+      "learning_rate": 0.00020091095520777114,
+      "loss": 3.2885,
       "step": 61750
     },
     {
       "epoch": 6.663072776280323,
-      "grad_norm": 0.7332888245582581,
-      "learning_rate": 0.00020060010793308147,
-      "loss": 3.2934,
+      "grad_norm": 0.7466468214988708,
+      "learning_rate": 0.00020058715596330272,
+      "loss": 3.2939,
       "step": 61800
     },
     {
       "epoch": 6.668463611859838,
-      "grad_norm": 0.8394750356674194,
-      "learning_rate": 0.00020027630868861305,
-      "loss": 3.2917,
+      "grad_norm": 0.789730429649353,
+      "learning_rate": 0.0002002633567188343,
+      "loss": 3.2921,
       "step": 61850
     },
     {
       "epoch": 6.6738544474393535,
-      "grad_norm": 0.7316187620162964,
-      "learning_rate": 0.00019995250944414462,
-      "loss": 3.2639,
+      "grad_norm": 0.7421731352806091,
+      "learning_rate": 0.00019993955747436588,
+      "loss": 3.2654,
       "step": 61900
     },
     {
       "epoch": 6.679245283018868,
-      "grad_norm": 0.7434121370315552,
-      "learning_rate": 0.00019962871019967617,
-      "loss": 3.2656,
+      "grad_norm": 0.7725925445556641,
+      "learning_rate": 0.00019961575822989745,
+      "loss": 3.2688,
       "step": 61950
     },
     {
       "epoch": 6.684636118598383,
-      "grad_norm": 0.8291236162185669,
-      "learning_rate": 0.00019930491095520775,
-      "loss": 3.2679,
+      "grad_norm": 0.802435576915741,
+      "learning_rate": 0.00019929195898542903,
+      "loss": 3.2689,
       "step": 62000
     },
     {
       "epoch": 6.684636118598383,
-      "eval_accuracy": 0.3871309998532099,
-      "eval_loss": 3.358170986175537,
-      "eval_runtime": 183.5019,
-      "eval_samples_per_second": 98.152,
-      "eval_steps_per_second": 6.136,
+      "eval_accuracy": 0.3867482157294395,
+      "eval_loss": 3.3587045669555664,
+      "eval_runtime": 181.5196,
+      "eval_samples_per_second": 99.223,
+      "eval_steps_per_second": 6.203,
       "step": 62000
     },
     {
       "epoch": 6.690026954177897,
-      "grad_norm": 0.7411676645278931,
-      "learning_rate": 0.0001989811117107393,
-      "loss": 3.2909,
+      "grad_norm": 0.7587479948997498,
+      "learning_rate": 0.0001989681597409606,
+      "loss": 3.2921,
       "step": 62050
     },
     {
       "epoch": 6.695417789757412,
-      "grad_norm": 0.7574923038482666,
-      "learning_rate": 0.00019865731246627088,
-      "loss": 3.2834,
+      "grad_norm": 0.7678079605102539,
+      "learning_rate": 0.00019864436049649216,
+      "loss": 3.2861,
       "step": 62100
     },
     {
       "epoch": 6.7008086253369274,
-      "grad_norm": 0.7495836615562439,
-      "learning_rate": 0.00019833351322180246,
-      "loss": 3.2719,
+      "grad_norm": 0.727103054523468,
+      "learning_rate": 0.0001983205612520237,
+      "loss": 3.2741,
       "step": 62150
     },
     {
       "epoch": 6.706199460916442,
-      "grad_norm": 0.7210440635681152,
-      "learning_rate": 0.00019800971397733404,
-      "loss": 3.2732,
+      "grad_norm": 0.7572033405303955,
+      "learning_rate": 0.0001979967620075553,
+      "loss": 3.2772,
       "step": 62200
     },
     {
       "epoch": 6.711590296495957,
-      "grad_norm": 0.7707087397575378,
-      "learning_rate": 0.00019768591473286561,
-      "loss": 3.2867,
+      "grad_norm": 0.7538331747055054,
+      "learning_rate": 0.00019767296276308686,
+      "loss": 3.2882,
       "step": 62250
     },
     {
       "epoch": 6.716981132075472,
-      "grad_norm": 0.7962491512298584,
-      "learning_rate": 0.0001973621154883972,
-      "loss": 3.2703,
+      "grad_norm": 0.7675634026527405,
+      "learning_rate": 0.00019734916351861844,
+      "loss": 3.2708,
       "step": 62300
     },
     {
       "epoch": 6.722371967654986,
-      "grad_norm": 0.7390112280845642,
-      "learning_rate": 0.00019703831624392877,
-      "loss": 3.2883,
+      "grad_norm": 0.733335554599762,
+      "learning_rate": 0.00019702536427415002,
+      "loss": 3.2913,
       "step": 62350
     },
     {
       "epoch": 6.727762803234501,
-      "grad_norm": 0.8107832670211792,
-      "learning_rate": 0.0001967145169994603,
-      "loss": 3.271,
+      "grad_norm": 0.8393965363502502,
+      "learning_rate": 0.0001967015650296816,
+      "loss": 3.2727,
       "step": 62400
     },
     {
       "epoch": 6.7331536388140165,
-      "grad_norm": 0.741752028465271,
-      "learning_rate": 0.00019639071775499187,
-      "loss": 3.2754,
+      "grad_norm": 0.7474007606506348,
+      "learning_rate": 0.00019637776578521315,
+      "loss": 3.2796,
       "step": 62450
     },
     {
       "epoch": 6.738544474393531,
-      "grad_norm": 0.7602443695068359,
-      "learning_rate": 0.00019606691851052345,
-      "loss": 3.2785,
+      "grad_norm": 0.7601961493492126,
+      "learning_rate": 0.00019605396654074473,
+      "loss": 3.2807,
       "step": 62500
     },
     {
       "epoch": 6.743935309973046,
-      "grad_norm": 0.716080904006958,
-      "learning_rate": 0.00019574311926605503,
-      "loss": 3.2864,
+      "grad_norm": 0.7366898059844971,
+      "learning_rate": 0.0001957301672962763,
+      "loss": 3.2889,
       "step": 62550
     },
     {
       "epoch": 6.74932614555256,
-      "grad_norm": 0.7163423299789429,
-      "learning_rate": 0.0001954193200215866,
-      "loss": 3.269,
+      "grad_norm": 0.730017364025116,
+      "learning_rate": 0.00019540636805180785,
+      "loss": 3.2715,
       "step": 62600
     },
     {
       "epoch": 6.754716981132075,
-      "grad_norm": 0.7376225590705872,
-      "learning_rate": 0.00019509552077711818,
-      "loss": 3.274,
+      "grad_norm": 0.7351442575454712,
+      "learning_rate": 0.00019508256880733943,
+      "loss": 3.2754,
       "step": 62650
     },
     {
       "epoch": 6.7601078167115904,
-      "grad_norm": 0.7619684338569641,
-      "learning_rate": 0.00019477172153264976,
-      "loss": 3.2638,
+      "grad_norm": 0.7444887757301331,
+      "learning_rate": 0.000194758769562871,
+      "loss": 3.266,
       "step": 62700
     },
     {
       "epoch": 6.765498652291106,
-      "grad_norm": 0.7855158448219299,
-      "learning_rate": 0.00019444792228818134,
-      "loss": 3.2921,
+      "grad_norm": 0.7755415439605713,
+      "learning_rate": 0.00019443497031840256,
+      "loss": 3.2965,
       "step": 62750
     },
     {
       "epoch": 6.77088948787062,
-      "grad_norm": 0.7887711524963379,
-      "learning_rate": 0.00019412412304371286,
-      "loss": 3.2783,
+      "grad_norm": 0.7490450143814087,
+      "learning_rate": 0.00019411117107393414,
+      "loss": 3.2801,
       "step": 62800
     },
     {
       "epoch": 6.776280323450135,
-      "grad_norm": 0.7664616107940674,
-      "learning_rate": 0.00019380032379924444,
-      "loss": 3.2934,
+      "grad_norm": 0.7514929175376892,
+      "learning_rate": 0.00019378737182946572,
+      "loss": 3.2944,
       "step": 62850
     },
     {
       "epoch": 6.781671159029649,
-      "grad_norm": 0.7111510634422302,
-      "learning_rate": 0.00019347652455477602,
-      "loss": 3.2909,
+      "grad_norm": 0.7469158172607422,
+      "learning_rate": 0.0001934635725849973,
+      "loss": 3.2928,
       "step": 62900
     },
     {
       "epoch": 6.787061994609164,
-      "grad_norm": 0.7316784262657166,
-      "learning_rate": 0.0001931527253103076,
-      "loss": 3.2634,
+      "grad_norm": 0.7205702662467957,
+      "learning_rate": 0.00019313977334052887,
+      "loss": 3.2643,
       "step": 62950
     },
     {
       "epoch": 6.7924528301886795,
-      "grad_norm": 0.7745811939239502,
-      "learning_rate": 0.00019282892606583917,
-      "loss": 3.2727,
+      "grad_norm": 0.7833849191665649,
+      "learning_rate": 0.00019281597409606042,
+      "loss": 3.275,
       "step": 63000
     },
     {
       "epoch": 6.7924528301886795,
-      "eval_accuracy": 0.3873692756361871,
-      "eval_loss": 3.3546576499938965,
-      "eval_runtime": 183.7912,
-      "eval_samples_per_second": 97.997,
-      "eval_steps_per_second": 6.127,
+      "eval_accuracy": 0.38716087939735894,
+      "eval_loss": 3.3561055660247803,
+      "eval_runtime": 181.462,
+      "eval_samples_per_second": 99.255,
+      "eval_steps_per_second": 6.205,
       "step": 63000
     },
     {
       "epoch": 6.797843665768194,
-      "grad_norm": 0.7474355101585388,
-      "learning_rate": 0.00019250512682137075,
-      "loss": 3.2738,
+      "grad_norm": 0.7507427334785461,
+      "learning_rate": 0.00019249217485159197,
+      "loss": 3.2757,
       "step": 63050
     },
     {
       "epoch": 6.803234501347709,
-      "grad_norm": 0.7750080823898315,
-      "learning_rate": 0.0001921813275769023,
-      "loss": 3.2733,
+      "grad_norm": 0.7834129929542542,
+      "learning_rate": 0.00019216837560712355,
+      "loss": 3.2778,
       "step": 63100
     },
     {
       "epoch": 6.808625336927224,
-      "grad_norm": 0.7421109676361084,
-      "learning_rate": 0.00019185752833243388,
-      "loss": 3.2766,
+      "grad_norm": 0.7286310791969299,
+      "learning_rate": 0.00019184457636265513,
+      "loss": 3.2781,
       "step": 63150
     },
     {
       "epoch": 6.814016172506738,
-      "grad_norm": 0.7296343445777893,
-      "learning_rate": 0.00019153372908796545,
-      "loss": 3.2777,
+      "grad_norm": 0.752474844455719,
+      "learning_rate": 0.0001915207771181867,
+      "loss": 3.28,
       "step": 63200
     },
     {
       "epoch": 6.819407008086253,
-      "grad_norm": 0.7462671995162964,
-      "learning_rate": 0.000191209929843497,
-      "loss": 3.27,
+      "grad_norm": 0.7446063160896301,
+      "learning_rate": 0.00019119697787371828,
+      "loss": 3.2735,
       "step": 63250
     },
     {
       "epoch": 6.824797843665769,
-      "grad_norm": 0.7307702898979187,
-      "learning_rate": 0.00019088613059902858,
-      "loss": 3.2594,
+      "grad_norm": 0.7489317059516907,
+      "learning_rate": 0.00019087317862924986,
+      "loss": 3.2619,
       "step": 63300
     },
     {
       "epoch": 6.830188679245283,
-      "grad_norm": 0.751537561416626,
-      "learning_rate": 0.00019056233135456016,
-      "loss": 3.2779,
+      "grad_norm": 0.7200451493263245,
+      "learning_rate": 0.00019054937938478144,
+      "loss": 3.2838,
       "step": 63350
     },
     {
       "epoch": 6.835579514824798,
-      "grad_norm": 0.7643833756446838,
-      "learning_rate": 0.0001902385321100917,
-      "loss": 3.2702,
+      "grad_norm": 0.7453355193138123,
+      "learning_rate": 0.00019022558014031302,
+      "loss": 3.2733,
       "step": 63400
     },
     {
       "epoch": 6.840970350404312,
-      "grad_norm": 0.7389995455741882,
-      "learning_rate": 0.0001899147328656233,
-      "loss": 3.2886,
+      "grad_norm": 0.7682626247406006,
+      "learning_rate": 0.00018990178089584454,
+      "loss": 3.2911,
       "step": 63450
     },
     {
       "epoch": 6.846361185983827,
-      "grad_norm": 0.7349757552146912,
-      "learning_rate": 0.00018959093362115487,
-      "loss": 3.2749,
+      "grad_norm": 0.7468233108520508,
+      "learning_rate": 0.00018957798165137612,
+      "loss": 3.2786,
       "step": 63500
     },
     {
       "epoch": 6.8517520215633425,
-      "grad_norm": 0.7448405623435974,
-      "learning_rate": 0.00018926713437668644,
-      "loss": 3.2778,
+      "grad_norm": 0.7568632364273071,
+      "learning_rate": 0.0001892541824069077,
+      "loss": 3.2813,
       "step": 63550
     },
     {
       "epoch": 6.857142857142857,
-      "grad_norm": 0.7943175435066223,
-      "learning_rate": 0.00018894333513221802,
-      "loss": 3.271,
+      "grad_norm": 0.7588648796081543,
+      "learning_rate": 0.00018893038316243927,
+      "loss": 3.2738,
       "step": 63600
     },
     {
       "epoch": 6.862533692722372,
-      "grad_norm": 0.7844904661178589,
-      "learning_rate": 0.00018861953588774957,
-      "loss": 3.2804,
+      "grad_norm": 0.7854700684547424,
+      "learning_rate": 0.00018860658391797085,
+      "loss": 3.2832,
       "step": 63650
     },
     {
       "epoch": 6.867924528301887,
-      "grad_norm": 0.7307280898094177,
-      "learning_rate": 0.00018829573664328115,
-      "loss": 3.2785,
+      "grad_norm": 0.728640615940094,
+      "learning_rate": 0.00018828278467350243,
+      "loss": 3.2818,
       "step": 63700
     },
     {
       "epoch": 6.873315363881401,
-      "grad_norm": 0.7532712817192078,
-      "learning_rate": 0.0001879719373988127,
-      "loss": 3.2801,
+      "grad_norm": 0.7183127403259277,
+      "learning_rate": 0.000187958985429034,
+      "loss": 3.282,
       "step": 63750
     },
     {
       "epoch": 6.878706199460916,
-      "grad_norm": 0.7567185759544373,
-      "learning_rate": 0.00018764813815434428,
-      "loss": 3.2661,
+      "grad_norm": 0.7702310085296631,
+      "learning_rate": 0.00018763518618456556,
+      "loss": 3.2684,
       "step": 63800
     },
     {
       "epoch": 6.884097035040432,
-      "grad_norm": 0.7488755583763123,
-      "learning_rate": 0.00018732433890987586,
-      "loss": 3.2661,
+      "grad_norm": 0.7492479085922241,
+      "learning_rate": 0.0001873113869400971,
+      "loss": 3.2701,
       "step": 63850
     },
     {
       "epoch": 6.889487870619946,
-      "grad_norm": 0.7576258778572083,
-      "learning_rate": 0.00018700053966540743,
-      "loss": 3.2932,
+      "grad_norm": 0.7798910140991211,
+      "learning_rate": 0.00018698758769562868,
+      "loss": 3.2978,
       "step": 63900
     },
     {
       "epoch": 6.894878706199461,
-      "grad_norm": 0.7660182118415833,
-      "learning_rate": 0.000186676740420939,
-      "loss": 3.2827,
+      "grad_norm": 0.7670497298240662,
+      "learning_rate": 0.00018666378845116026,
+      "loss": 3.2835,
       "step": 63950
     },
     {
       "epoch": 6.900269541778976,
-      "grad_norm": 0.7318092584609985,
-      "learning_rate": 0.0001863529411764706,
-      "loss": 3.2896,
+      "grad_norm": 0.746772050857544,
+      "learning_rate": 0.00018633998920669184,
+      "loss": 3.2919,
       "step": 64000
     },
     {
       "epoch": 6.900269541778976,
-      "eval_accuracy": 0.3878326802027159,
-      "eval_loss": 3.352012872695923,
-      "eval_runtime": 183.2112,
-      "eval_samples_per_second": 98.307,
-      "eval_steps_per_second": 6.146,
+      "eval_accuracy": 0.38746956225163987,
+      "eval_loss": 3.3524253368377686,
+      "eval_runtime": 181.4771,
+      "eval_samples_per_second": 99.247,
+      "eval_steps_per_second": 6.205,
       "step": 64000
     },
     {
       "epoch": 6.90566037735849,
-      "grad_norm": 0.7451673150062561,
-      "learning_rate": 0.00018603561791689149,
-      "loss": 3.273,
+      "grad_norm": 0.7522950172424316,
+      "learning_rate": 0.00018601618996222342,
+      "loss": 3.2767,
       "step": 64050
     },
     {
       "epoch": 6.9110512129380055,
-      "grad_norm": 0.7861278653144836,
-      "learning_rate": 0.00018571181867242306,
-      "loss": 3.2591,
+      "grad_norm": 0.8034794330596924,
+      "learning_rate": 0.00018569239071775497,
+      "loss": 3.2624,
       "step": 64100
     },
     {
       "epoch": 6.916442048517521,
-      "grad_norm": 0.7036826610565186,
-      "learning_rate": 0.00018538801942795464,
-      "loss": 3.2688,
+      "grad_norm": 0.7302206754684448,
+      "learning_rate": 0.00018536859147328655,
+      "loss": 3.2708,
       "step": 64150
     },
     {
       "epoch": 6.921832884097035,
-      "grad_norm": 0.7634555697441101,
-      "learning_rate": 0.00018506422018348622,
-      "loss": 3.2769,
+      "grad_norm": 0.7536947131156921,
+      "learning_rate": 0.00018504479222881812,
+      "loss": 3.279,
       "step": 64200
     },
     {
       "epoch": 6.92722371967655,
-      "grad_norm": 0.7458739280700684,
-      "learning_rate": 0.0001847404209390178,
-      "loss": 3.2709,
+      "grad_norm": 0.7437476515769958,
+      "learning_rate": 0.0001847209929843497,
+      "loss": 3.2746,
       "step": 64250
     },
     {
       "epoch": 6.932614555256064,
-      "grad_norm": 0.8086719512939453,
-      "learning_rate": 0.00018441662169454937,
-      "loss": 3.2676,
+      "grad_norm": 0.8075777888298035,
+      "learning_rate": 0.00018439719373988125,
+      "loss": 3.2706,
       "step": 64300
     },
     {
       "epoch": 6.938005390835579,
-      "grad_norm": 0.716335654258728,
-      "learning_rate": 0.00018409282245008095,
-      "loss": 3.2789,
+      "grad_norm": 0.7226145267486572,
+      "learning_rate": 0.00018407339449541283,
+      "loss": 3.2831,
       "step": 64350
     },
     {
       "epoch": 6.943396226415095,
-      "grad_norm": 0.7624198794364929,
-      "learning_rate": 0.00018376902320561253,
-      "loss": 3.2732,
+      "grad_norm": 0.7236364483833313,
+      "learning_rate": 0.00018374959525094438,
+      "loss": 3.2741,
       "step": 64400
     },
     {
       "epoch": 6.948787061994609,
-      "grad_norm": 0.77959144115448,
-      "learning_rate": 0.00018344522396114405,
-      "loss": 3.2667,
+      "grad_norm": 0.7727348804473877,
+      "learning_rate": 0.00018342579600647596,
+      "loss": 3.2719,
       "step": 64450
     },
     {
       "epoch": 6.954177897574124,
-      "grad_norm": 0.766917884349823,
-      "learning_rate": 0.00018312142471667563,
-      "loss": 3.2833,
+      "grad_norm": 0.7388468980789185,
+      "learning_rate": 0.00018310199676200753,
+      "loss": 3.2865,
       "step": 64500
     },
     {
       "epoch": 6.959568733153639,
-      "grad_norm": 0.7285516858100891,
-      "learning_rate": 0.0001827976254722072,
-      "loss": 3.275,
+      "grad_norm": 0.7288505434989929,
+      "learning_rate": 0.0001827781975175391,
+      "loss": 3.2787,
       "step": 64550
     },
     {
       "epoch": 6.964959568733153,
-      "grad_norm": 0.7674565315246582,
-      "learning_rate": 0.00018247382622773879,
-      "loss": 3.2768,
+      "grad_norm": 0.7576027512550354,
+      "learning_rate": 0.0001824543982730707,
+      "loss": 3.2813,
       "step": 64600
     },
     {
       "epoch": 6.9703504043126685,
-      "grad_norm": 0.7331922054290771,
-      "learning_rate": 0.00018215002698327036,
-      "loss": 3.2847,
+      "grad_norm": 0.7288236021995544,
+      "learning_rate": 0.00018213059902860227,
+      "loss": 3.2867,
       "step": 64650
     },
     {
       "epoch": 6.975741239892184,
-      "grad_norm": 0.7738147377967834,
-      "learning_rate": 0.00018182622773880194,
-      "loss": 3.2892,
+      "grad_norm": 0.7492437362670898,
+      "learning_rate": 0.00018180679978413382,
+      "loss": 3.2933,
       "step": 64700
     },
     {
       "epoch": 6.981132075471698,
-      "grad_norm": 0.7373318672180176,
-      "learning_rate": 0.00018150242849433352,
-      "loss": 3.2808,
+      "grad_norm": 0.7648993730545044,
+      "learning_rate": 0.00018148300053966537,
+      "loss": 3.2844,
       "step": 64750
     },
     {
       "epoch": 6.986522911051213,
-      "grad_norm": 0.7882824540138245,
-      "learning_rate": 0.00018117862924986507,
-      "loss": 3.2797,
+      "grad_norm": 0.8067795038223267,
+      "learning_rate": 0.00018115920129519695,
+      "loss": 3.2836,
       "step": 64800
     },
     {
       "epoch": 6.991913746630727,
-      "grad_norm": 0.718296468257904,
-      "learning_rate": 0.00018085483000539662,
-      "loss": 3.2775,
+      "grad_norm": 0.7241319417953491,
+      "learning_rate": 0.00018083540205072852,
+      "loss": 3.2801,
       "step": 64850
     },
     {
       "epoch": 6.997304582210242,
-      "grad_norm": 0.7915265560150146,
-      "learning_rate": 0.0001805310307609282,
-      "loss": 3.2826,
+      "grad_norm": 0.8056755661964417,
+      "learning_rate": 0.00018051807879114948,
+      "loss": 3.2849,
       "step": 64900
     },
     {
       "epoch": 7.002695417789758,
-      "grad_norm": 0.7813780903816223,
-      "learning_rate": 0.00018020723151645978,
-      "loss": 3.228,
+      "grad_norm": 0.8495665192604065,
+      "learning_rate": 0.00018019427954668105,
+      "loss": 3.2314,
       "step": 64950
     },
     {
       "epoch": 7.008086253369272,
-      "grad_norm": 0.8058916926383972,
-      "learning_rate": 0.00017988343227199135,
-      "loss": 3.1934,
+      "grad_norm": 0.7899782657623291,
+      "learning_rate": 0.00017987048030221263,
+      "loss": 3.1959,
       "step": 65000
     },
     {
       "epoch": 7.008086253369272,
-      "eval_accuracy": 0.387832028285389,
-      "eval_loss": 3.354559898376465,
-      "eval_runtime": 183.3424,
-      "eval_samples_per_second": 98.237,
-      "eval_steps_per_second": 6.142,
+      "eval_accuracy": 0.3877698788335591,
+      "eval_loss": 3.354261636734009,
+      "eval_runtime": 181.1797,
+      "eval_samples_per_second": 99.41,
+      "eval_steps_per_second": 6.215,
       "step": 65000
     },
     {
       "epoch": 7.013477088948787,
-      "grad_norm": 0.7725360989570618,
-      "learning_rate": 0.00017955963302752293,
-      "loss": 3.1876,
+      "grad_norm": 0.7633707523345947,
+      "learning_rate": 0.00017954668105774416,
+      "loss": 3.1922,
       "step": 65050
     },
     {
       "epoch": 7.018867924528302,
-      "grad_norm": 0.7770508527755737,
-      "learning_rate": 0.00017923583378305448,
-      "loss": 3.1881,
+      "grad_norm": 0.798157274723053,
+      "learning_rate": 0.00017922288181327573,
+      "loss": 3.1908,
       "step": 65100
     },
     {
       "epoch": 7.024258760107816,
-      "grad_norm": 0.8097355961799622,
-      "learning_rate": 0.00017891203453858606,
-      "loss": 3.1979,
+      "grad_norm": 0.7584009170532227,
+      "learning_rate": 0.0001788990825688073,
+      "loss": 3.1994,
       "step": 65150
     },
     {
       "epoch": 7.0296495956873315,
-      "grad_norm": 0.7231821417808533,
-      "learning_rate": 0.00017858823529411764,
-      "loss": 3.175,
+      "grad_norm": 0.7545470595359802,
+      "learning_rate": 0.0001785752833243389,
+      "loss": 3.1772,
       "step": 65200
     },
     {
       "epoch": 7.035040431266847,
-      "grad_norm": 0.8374466300010681,
-      "learning_rate": 0.00017826443604964921,
-      "loss": 3.1885,
+      "grad_norm": 0.8252199292182922,
+      "learning_rate": 0.00017825148407987047,
+      "loss": 3.1907,
       "step": 65250
     },
     {
       "epoch": 7.040431266846361,
-      "grad_norm": 0.8206698298454285,
-      "learning_rate": 0.00017794063680518077,
-      "loss": 3.1949,
+      "grad_norm": 0.7860397696495056,
+      "learning_rate": 0.00017792768483540204,
+      "loss": 3.1965,
       "step": 65300
     },
     {
       "epoch": 7.045822102425876,
-      "grad_norm": 0.735885500907898,
-      "learning_rate": 0.00017761683756071234,
-      "loss": 3.1847,
+      "grad_norm": 0.7742920517921448,
+      "learning_rate": 0.00017760388559093362,
+      "loss": 3.1862,
       "step": 65350
     },
     {
       "epoch": 7.051212938005391,
-      "grad_norm": 0.8094024062156677,
-      "learning_rate": 0.00017729303831624392,
-      "loss": 3.2144,
+      "grad_norm": 0.838625967502594,
+      "learning_rate": 0.0001772800863464652,
+      "loss": 3.2164,
       "step": 65400
     },
     {
       "epoch": 7.056603773584905,
-      "grad_norm": 0.7802730798721313,
-      "learning_rate": 0.00017696923907177547,
-      "loss": 3.2029,
+      "grad_norm": 0.7871566414833069,
+      "learning_rate": 0.00017695628710199672,
+      "loss": 3.2045,
       "step": 65450
     },
     {
       "epoch": 7.061994609164421,
-      "grad_norm": 0.7770246863365173,
-      "learning_rate": 0.00017664543982730705,
-      "loss": 3.1913,
+      "grad_norm": 0.8190498352050781,
+      "learning_rate": 0.0001766324878575283,
+      "loss": 3.1957,
       "step": 65500
     },
     {
       "epoch": 7.067385444743936,
-      "grad_norm": 0.7353746891021729,
-      "learning_rate": 0.00017632164058283863,
-      "loss": 3.2044,
+      "grad_norm": 0.7641475796699524,
+      "learning_rate": 0.00017630868861305988,
+      "loss": 3.2073,
       "step": 65550
     },
     {
       "epoch": 7.07277628032345,
-      "grad_norm": 0.7481196522712708,
-      "learning_rate": 0.0001759978413383702,
-      "loss": 3.1822,
+      "grad_norm": 0.8046497702598572,
+      "learning_rate": 0.00017598488936859146,
+      "loss": 3.1848,
       "step": 65600
     },
     {
       "epoch": 7.078167115902965,
-      "grad_norm": 0.7949692010879517,
-      "learning_rate": 0.00017567404209390178,
-      "loss": 3.2052,
+      "grad_norm": 0.7494258880615234,
+      "learning_rate": 0.00017566109012412303,
+      "loss": 3.2082,
       "step": 65650
     },
     {
       "epoch": 7.083557951482479,
-      "grad_norm": 0.7633875012397766,
-      "learning_rate": 0.00017535024284943333,
-      "loss": 3.2092,
+      "grad_norm": 0.7675015330314636,
+      "learning_rate": 0.0001753372908796546,
+      "loss": 3.2127,
       "step": 65700
     },
     {
       "epoch": 7.0889487870619945,
-      "grad_norm": 0.8302983641624451,
-      "learning_rate": 0.00017502644360496488,
-      "loss": 3.2087,
+      "grad_norm": 0.7822027802467346,
+      "learning_rate": 0.0001750134916351862,
+      "loss": 3.2116,
       "step": 65750
     },
     {
       "epoch": 7.09433962264151,
-      "grad_norm": 0.794335126876831,
-      "learning_rate": 0.00017470264436049646,
-      "loss": 3.2061,
+      "grad_norm": 0.8151496648788452,
+      "learning_rate": 0.00017468969239071774,
+      "loss": 3.2103,
       "step": 65800
     },
     {
       "epoch": 7.099730458221024,
-      "grad_norm": 0.7204697132110596,
-      "learning_rate": 0.00017437884511602804,
-      "loss": 3.1899,
+      "grad_norm": 0.7523123025894165,
+      "learning_rate": 0.00017436589314624932,
+      "loss": 3.1921,
       "step": 65850
     },
     {
       "epoch": 7.105121293800539,
-      "grad_norm": 0.7534080743789673,
-      "learning_rate": 0.00017405504587155962,
-      "loss": 3.2178,
+      "grad_norm": 0.7588797807693481,
+      "learning_rate": 0.00017404209390178087,
+      "loss": 3.2211,
       "step": 65900
     },
     {
       "epoch": 7.110512129380054,
-      "grad_norm": 0.7708962559700012,
-      "learning_rate": 0.0001737312466270912,
-      "loss": 3.1813,
+      "grad_norm": 0.7903715372085571,
+      "learning_rate": 0.00017371829465731244,
+      "loss": 3.1842,
       "step": 65950
     },
     {
       "epoch": 7.115902964959568,
-      "grad_norm": 0.7770982384681702,
-      "learning_rate": 0.00017340744738262277,
-      "loss": 3.2125,
+      "grad_norm": 0.7794401049613953,
+      "learning_rate": 0.00017339449541284402,
+      "loss": 3.2156,
       "step": 66000
     },
     {
       "epoch": 7.115902964959568,
-      "eval_accuracy": 0.3879341619999346,
-      "eval_loss": 3.3549797534942627,
-      "eval_runtime": 183.6028,
-      "eval_samples_per_second": 98.098,
-      "eval_steps_per_second": 6.133,
+      "eval_accuracy": 0.38759538229572904,
+      "eval_loss": 3.3558542728424072,
+      "eval_runtime": 181.6201,
+      "eval_samples_per_second": 99.169,
+      "eval_steps_per_second": 6.2,
       "step": 66000
     },
     {
       "epoch": 7.121293800539084,
-      "grad_norm": 0.8400378823280334,
-      "learning_rate": 0.00017308364813815435,
-      "loss": 3.204,
+      "grad_norm": 0.7734689116477966,
+      "learning_rate": 0.0001730706961683756,
+      "loss": 3.2062,
       "step": 66050
     },
     {
       "epoch": 7.126684636118599,
-      "grad_norm": 0.783427357673645,
-      "learning_rate": 0.00017275984889368593,
-      "loss": 3.1943,
+      "grad_norm": 0.7667518258094788,
+      "learning_rate": 0.00017274689692390715,
+      "loss": 3.1981,
       "step": 66100
     },
     {
       "epoch": 7.132075471698113,
-      "grad_norm": 0.7737600207328796,
-      "learning_rate": 0.00017243604964921745,
-      "loss": 3.2306,
+      "grad_norm": 0.7898749709129333,
+      "learning_rate": 0.00017242309767943873,
+      "loss": 3.2339,
       "step": 66150
     },
     {
       "epoch": 7.137466307277628,
-      "grad_norm": 0.7946293354034424,
-      "learning_rate": 0.00017211225040474903,
-      "loss": 3.2256,
+      "grad_norm": 0.7846260070800781,
+      "learning_rate": 0.0001720992984349703,
+      "loss": 3.2279,
       "step": 66200
     },
     {
       "epoch": 7.142857142857143,
-      "grad_norm": 0.797230064868927,
-      "learning_rate": 0.0001717884511602806,
-      "loss": 3.2133,
+      "grad_norm": 0.7452735304832458,
+      "learning_rate": 0.00017177549919050188,
+      "loss": 3.217,
       "step": 66250
     },
     {
       "epoch": 7.1482479784366575,
-      "grad_norm": 0.817846953868866,
-      "learning_rate": 0.00017146465191581218,
-      "loss": 3.2025,
+      "grad_norm": 0.8026751279830933,
+      "learning_rate": 0.0001714581759309228,
+      "loss": 3.2072,
       "step": 66300
     },
     {
       "epoch": 7.153638814016173,
-      "grad_norm": 0.786282479763031,
-      "learning_rate": 0.00017114085267134376,
-      "loss": 3.21,
+      "grad_norm": 0.8096675276756287,
+      "learning_rate": 0.00017113437668645439,
+      "loss": 3.2124,
       "step": 66350
     },
     {
       "epoch": 7.159029649595688,
-      "grad_norm": 0.7998514771461487,
-      "learning_rate": 0.00017081705342687534,
-      "loss": 3.2079,
+      "grad_norm": 0.7887595295906067,
+      "learning_rate": 0.00017081057744198596,
+      "loss": 3.2097,
       "step": 66400
     },
     {
       "epoch": 7.164420485175202,
-      "grad_norm": 0.7554528117179871,
-      "learning_rate": 0.0001704932541824069,
-      "loss": 3.2061,
+      "grad_norm": 0.7669116854667664,
+      "learning_rate": 0.00017048677819751751,
+      "loss": 3.2099,
       "step": 66450
     },
     {
       "epoch": 7.169811320754717,
-      "grad_norm": 0.7405669093132019,
-      "learning_rate": 0.00017016945493793847,
-      "loss": 3.2116,
+      "grad_norm": 0.7745186686515808,
+      "learning_rate": 0.0001701629789530491,
+      "loss": 3.2135,
       "step": 66500
     },
     {
       "epoch": 7.175202156334231,
-      "grad_norm": 0.7673985958099365,
-      "learning_rate": 0.00016984565569347002,
-      "loss": 3.2082,
+      "grad_norm": 0.8053719997406006,
+      "learning_rate": 0.00016983917970858067,
+      "loss": 3.2106,
       "step": 66550
     },
     {
       "epoch": 7.180592991913747,
-      "grad_norm": 0.7783784866333008,
-      "learning_rate": 0.0001695218564490016,
-      "loss": 3.2171,
+      "grad_norm": 0.7639967799186707,
+      "learning_rate": 0.00016951538046411225,
+      "loss": 3.2181,
       "step": 66600
     },
     {
       "epoch": 7.185983827493262,
-      "grad_norm": 0.7714428901672363,
-      "learning_rate": 0.00016919805720453317,
-      "loss": 3.2155,
+      "grad_norm": 0.7497126460075378,
+      "learning_rate": 0.0001691915812196438,
+      "loss": 3.2186,
       "step": 66650
     },
     {
       "epoch": 7.191374663072776,
-      "grad_norm": 0.7873608469963074,
-      "learning_rate": 0.00016887425796006475,
-      "loss": 3.2122,
+      "grad_norm": 0.764004111289978,
+      "learning_rate": 0.00016886778197517538,
+      "loss": 3.2151,
       "step": 66700
     },
     {
       "epoch": 7.196765498652291,
-      "grad_norm": 0.8219302892684937,
-      "learning_rate": 0.00016855045871559633,
-      "loss": 3.1983,
+      "grad_norm": 0.8434079885482788,
+      "learning_rate": 0.00016854398273070693,
+      "loss": 3.2023,
       "step": 66750
     },
     {
       "epoch": 7.202156334231806,
-      "grad_norm": 0.802437961101532,
-      "learning_rate": 0.00016822665947112788,
-      "loss": 3.2066,
+      "grad_norm": 0.7766229510307312,
+      "learning_rate": 0.0001682201834862385,
+      "loss": 3.2097,
       "step": 66800
     },
     {
       "epoch": 7.2075471698113205,
-      "grad_norm": 0.7578773498535156,
-      "learning_rate": 0.00016790286022665946,
-      "loss": 3.2035,
+      "grad_norm": 0.7469687461853027,
+      "learning_rate": 0.00016789638424177008,
+      "loss": 3.2088,
       "step": 66850
     },
     {
       "epoch": 7.212938005390836,
-      "grad_norm": 0.7701559662818909,
-      "learning_rate": 0.00016757906098219103,
-      "loss": 3.2044,
+      "grad_norm": 0.7752885222434998,
+      "learning_rate": 0.00016757258499730166,
+      "loss": 3.2081,
       "step": 66900
     },
     {
       "epoch": 7.218328840970351,
-      "grad_norm": 0.8019431829452515,
-      "learning_rate": 0.0001672552617377226,
-      "loss": 3.207,
+      "grad_norm": 0.7647175192832947,
+      "learning_rate": 0.00016724878575283324,
+      "loss": 3.2104,
       "step": 66950
     },
     {
       "epoch": 7.223719676549865,
-      "grad_norm": 0.7901089787483215,
-      "learning_rate": 0.00016693146249325416,
-      "loss": 3.215,
+      "grad_norm": 0.8040725588798523,
+      "learning_rate": 0.00016692498650836481,
+      "loss": 3.2173,
       "step": 67000
     },
     {
       "epoch": 7.223719676549865,
-      "eval_accuracy": 0.3883080365869042,
-      "eval_loss": 3.351943016052246,
-      "eval_runtime": 183.3094,
-      "eval_samples_per_second": 98.255,
-      "eval_steps_per_second": 6.143,
+      "eval_accuracy": 0.3880533542178671,
+      "eval_loss": 3.352846145629883,
+      "eval_runtime": 181.2513,
+      "eval_samples_per_second": 99.37,
+      "eval_steps_per_second": 6.212,
       "step": 67000
     },
     {
       "epoch": 7.22911051212938,
-      "grad_norm": 0.7946869134902954,
-      "learning_rate": 0.00016660766324878574,
-      "loss": 3.2258,
+      "grad_norm": 0.7994017601013184,
+      "learning_rate": 0.0001666011872638964,
+      "loss": 3.2293,
       "step": 67050
     },
     {
       "epoch": 7.234501347708895,
-      "grad_norm": 0.77195143699646,
-      "learning_rate": 0.0001662838640043173,
-      "loss": 3.2286,
+      "grad_norm": 0.8026636242866516,
+      "learning_rate": 0.00016627738801942792,
+      "loss": 3.2305,
       "step": 67100
     },
     {
       "epoch": 7.2398921832884096,
-      "grad_norm": 0.782701849937439,
-      "learning_rate": 0.00016596006475984887,
-      "loss": 3.186,
+      "grad_norm": 0.8015336394309998,
+      "learning_rate": 0.0001659535887749595,
+      "loss": 3.1878,
       "step": 67150
     },
     {
       "epoch": 7.245283018867925,
-      "grad_norm": 0.756524384021759,
-      "learning_rate": 0.00016563626551538045,
-      "loss": 3.2187,
+      "grad_norm": 0.7893655300140381,
+      "learning_rate": 0.00016562978953049107,
+      "loss": 3.2201,
       "step": 67200
     },
     {
       "epoch": 7.250673854447439,
-      "grad_norm": 0.7699859142303467,
-      "learning_rate": 0.00016531246627091202,
-      "loss": 3.2101,
+      "grad_norm": 0.7803918123245239,
+      "learning_rate": 0.00016530599028602265,
+      "loss": 3.2136,
       "step": 67250
     },
     {
       "epoch": 7.256064690026954,
-      "grad_norm": 0.7793636322021484,
-      "learning_rate": 0.0001649886670264436,
-      "loss": 3.2176,
+      "grad_norm": 0.7691161036491394,
+      "learning_rate": 0.00016498219104155423,
+      "loss": 3.2191,
       "step": 67300
     },
     {
       "epoch": 7.261455525606469,
-      "grad_norm": 0.7925220727920532,
-      "learning_rate": 0.00016466486778197518,
-      "loss": 3.232,
+      "grad_norm": 0.7607215642929077,
+      "learning_rate": 0.0001646583917970858,
+      "loss": 3.2326,
       "step": 67350
     },
     {
       "epoch": 7.2668463611859835,
-      "grad_norm": 0.7722257375717163,
-      "learning_rate": 0.0001643410685375067,
-      "loss": 3.2253,
+      "grad_norm": 0.7985602617263794,
+      "learning_rate": 0.00016433459255261738,
+      "loss": 3.2288,
       "step": 67400
     },
     {
       "epoch": 7.272237196765499,
-      "grad_norm": 0.8224971294403076,
-      "learning_rate": 0.00016401726929303828,
-      "loss": 3.237,
+      "grad_norm": 0.7763144969940186,
+      "learning_rate": 0.00016401079330814896,
+      "loss": 3.2401,
       "step": 67450
     },
     {
       "epoch": 7.277628032345014,
-      "grad_norm": 0.7844073176383972,
-      "learning_rate": 0.00016369347004856986,
-      "loss": 3.2074,
+      "grad_norm": 0.7595257759094238,
+      "learning_rate": 0.00016368699406368048,
+      "loss": 3.21,
       "step": 67500
     },
     {
       "epoch": 7.283018867924528,
-      "grad_norm": 0.7575783133506775,
+      "grad_norm": 0.761127233505249,
       "learning_rate": 0.00016336967080410143,
-      "loss": 3.1997,
+      "loss": 3.2036,
       "step": 67550
     },
     {
       "epoch": 7.288409703504043,
-      "grad_norm": 0.7861126661300659,
+      "grad_norm": 0.7509574294090271,
       "learning_rate": 0.000163045871559633,
-      "loss": 3.2356,
+      "loss": 3.2382,
       "step": 67600
     },
     {
       "epoch": 7.293800539083558,
-      "grad_norm": 0.8623127341270447,
+      "grad_norm": 0.8485333919525146,
       "learning_rate": 0.0001627220723151646,
-      "loss": 3.2232,
+      "loss": 3.2262,
       "step": 67650
     },
     {
       "epoch": 7.2991913746630726,
-      "grad_norm": 0.7551688551902771,
+      "grad_norm": 0.7620570659637451,
       "learning_rate": 0.00016239827307069617,
-      "loss": 3.2158,
+      "loss": 3.2198,
       "step": 67700
     },
     {
       "epoch": 7.304582210242588,
-      "grad_norm": 0.7738710641860962,
+      "grad_norm": 0.8164886832237244,
       "learning_rate": 0.00016207447382622775,
-      "loss": 3.223,
+      "loss": 3.2244,
       "step": 67750
     },
     {
       "epoch": 7.309973045822103,
-      "grad_norm": 0.7757263779640198,
+      "grad_norm": 0.8140173554420471,
       "learning_rate": 0.00016175067458175932,
-      "loss": 3.2344,
+      "loss": 3.2358,
       "step": 67800
     },
     {
       "epoch": 7.315363881401617,
-      "grad_norm": 0.7811141014099121,
+      "grad_norm": 0.7857149839401245,
       "learning_rate": 0.00016142687533729085,
-      "loss": 3.219,
+      "loss": 3.2214,
       "step": 67850
     },
     {
       "epoch": 7.320754716981132,
-      "grad_norm": 0.8079759478569031,
+      "grad_norm": 0.7988697290420532,
       "learning_rate": 0.00016110307609282242,
-      "loss": 3.2148,
+      "loss": 3.218,
       "step": 67900
     },
     {
       "epoch": 7.3261455525606465,
-      "grad_norm": 0.8070315718650818,
+      "grad_norm": 0.7990771532058716,
       "learning_rate": 0.000160779276848354,
-      "loss": 3.2236,
+      "loss": 3.225,
       "step": 67950
     },
     {
       "epoch": 7.331536388140162,
-      "grad_norm": 0.8166431784629822,
+      "grad_norm": 0.8031650185585022,
       "learning_rate": 0.00016045547760388558,
-      "loss": 3.2193,
+      "loss": 3.2208,
       "step": 68000
     },
     {
       "epoch": 7.331536388140162,
-      "eval_accuracy": 0.3887239598414581,
-      "eval_loss": 3.3503804206848145,
-      "eval_runtime": 183.4566,
-      "eval_samples_per_second": 98.176,
-      "eval_steps_per_second": 6.138,
+      "eval_accuracy": 0.388393655062502,
+      "eval_loss": 3.3500962257385254,
+      "eval_runtime": 181.3683,
+      "eval_samples_per_second": 99.306,
+      "eval_steps_per_second": 6.208,
       "step": 68000
     },
     {
       "epoch": 7.336927223719677,
-      "grad_norm": 0.7583974003791809,
-      "learning_rate": 0.00016013815434430653,
-      "loss": 3.2139,
+      "grad_norm": 0.7975978851318359,
+      "learning_rate": 0.00016013167835941716,
+      "loss": 3.2186,
       "step": 68050
     },
     {
       "epoch": 7.342318059299191,
-      "grad_norm": 0.757961630821228,
-      "learning_rate": 0.0001598143550998381,
-      "loss": 3.2173,
+      "grad_norm": 0.7547180652618408,
+      "learning_rate": 0.00015980787911494873,
+      "loss": 3.222,
       "step": 68100
     },
     {
       "epoch": 7.347708894878706,
-      "grad_norm": 0.7688447833061218,
-      "learning_rate": 0.00015949055585536966,
-      "loss": 3.2238,
+      "grad_norm": 0.8072742223739624,
+      "learning_rate": 0.00015948407987048029,
+      "loss": 3.2246,
       "step": 68150
     },
     {
       "epoch": 7.353099730458221,
-      "grad_norm": 0.7891886830329895,
-      "learning_rate": 0.0001591667566109012,
-      "loss": 3.2162,
+      "grad_norm": 0.7690461277961731,
+      "learning_rate": 0.00015916028062601186,
+      "loss": 3.2192,
       "step": 68200
     },
     {
       "epoch": 7.3584905660377355,
-      "grad_norm": 0.8004134297370911,
-      "learning_rate": 0.0001588429573664328,
-      "loss": 3.2212,
+      "grad_norm": 0.8002843856811523,
+      "learning_rate": 0.00015883648138154344,
+      "loss": 3.2233,
       "step": 68250
     },
     {
       "epoch": 7.363881401617251,
-      "grad_norm": 0.8025893568992615,
-      "learning_rate": 0.00015851915812196437,
-      "loss": 3.2367,
+      "grad_norm": 0.8228212594985962,
+      "learning_rate": 0.000158512682137075,
+      "loss": 3.2392,
       "step": 68300
     },
     {
       "epoch": 7.369272237196766,
-      "grad_norm": 0.8140208125114441,
-      "learning_rate": 0.00015819535887749594,
-      "loss": 3.2145,
+      "grad_norm": 0.7962360382080078,
+      "learning_rate": 0.00015818888289260657,
+      "loss": 3.2182,
       "step": 68350
     },
     {
       "epoch": 7.37466307277628,
-      "grad_norm": 0.7988510727882385,
-      "learning_rate": 0.00015787155963302752,
-      "loss": 3.2117,
+      "grad_norm": 0.8624733686447144,
+      "learning_rate": 0.00015786508364813815,
+      "loss": 3.2153,
       "step": 68400
     },
     {
       "epoch": 7.380053908355795,
-      "grad_norm": 0.8174982666969299,
-      "learning_rate": 0.0001575477603885591,
-      "loss": 3.2268,
+      "grad_norm": 0.8155392408370972,
+      "learning_rate": 0.0001575412844036697,
+      "loss": 3.2283,
       "step": 68450
     },
     {
       "epoch": 7.38544474393531,
-      "grad_norm": 0.8240893483161926,
-      "learning_rate": 0.00015722396114409065,
-      "loss": 3.199,
+      "grad_norm": 0.9194267392158508,
+      "learning_rate": 0.00015721748515920127,
+      "loss": 3.2026,
       "step": 68500
     },
     {
       "epoch": 7.390835579514825,
-      "grad_norm": 0.7737301588058472,
-      "learning_rate": 0.00015690016189962223,
-      "loss": 3.1935,
+      "grad_norm": 0.7518168687820435,
+      "learning_rate": 0.00015689368591473285,
+      "loss": 3.1952,
       "step": 68550
     },
     {
       "epoch": 7.39622641509434,
-      "grad_norm": 0.7589561939239502,
-      "learning_rate": 0.00015657636265515378,
-      "loss": 3.2208,
+      "grad_norm": 0.795861542224884,
+      "learning_rate": 0.00015656988667026443,
+      "loss": 3.2224,
       "step": 68600
     },
     {
       "epoch": 7.401617250673855,
-      "grad_norm": 0.8396160006523132,
-      "learning_rate": 0.00015625256341068536,
-      "loss": 3.2331,
+      "grad_norm": 0.7989816665649414,
+      "learning_rate": 0.000156246087425796,
+      "loss": 3.2367,
       "step": 68650
     },
     {
       "epoch": 7.407008086253369,
-      "grad_norm": 0.7733933925628662,
-      "learning_rate": 0.00015592876416621693,
-      "loss": 3.2099,
+      "grad_norm": 0.7558537721633911,
+      "learning_rate": 0.00015592228818132756,
+      "loss": 3.2146,
       "step": 68700
     },
     {
       "epoch": 7.412398921832884,
-      "grad_norm": 0.8380154967308044,
-      "learning_rate": 0.0001556049649217485,
-      "loss": 3.2235,
+      "grad_norm": 0.861491858959198,
+      "learning_rate": 0.0001555984889368591,
+      "loss": 3.2293,
       "step": 68750
     },
     {
       "epoch": 7.4177897574123985,
-      "grad_norm": 0.799769401550293,
-      "learning_rate": 0.00015528116567728006,
-      "loss": 3.2154,
+      "grad_norm": 0.7963483333587646,
+      "learning_rate": 0.0001552746896923907,
+      "loss": 3.2186,
       "step": 68800
     },
     {
       "epoch": 7.423180592991914,
-      "grad_norm": 0.794215977191925,
-      "learning_rate": 0.00015495736643281164,
-      "loss": 3.2324,
+      "grad_norm": 0.8001631498336792,
+      "learning_rate": 0.00015495089044792226,
+      "loss": 3.2336,
       "step": 68850
     },
     {
       "epoch": 7.428571428571429,
-      "grad_norm": 0.8496611714363098,
-      "learning_rate": 0.00015463356718834322,
-      "loss": 3.2261,
+      "grad_norm": 0.8393190503120422,
+      "learning_rate": 0.00015462709120345384,
+      "loss": 3.2304,
       "step": 68900
     },
     {
       "epoch": 7.433962264150943,
-      "grad_norm": 0.7774752378463745,
-      "learning_rate": 0.0001543097679438748,
-      "loss": 3.2289,
+      "grad_norm": 0.7578170895576477,
+      "learning_rate": 0.00015430329195898542,
+      "loss": 3.2328,
       "step": 68950
     },
     {
       "epoch": 7.439353099730458,
-      "grad_norm": 0.8549684286117554,
-      "learning_rate": 0.00015398596869940637,
-      "loss": 3.2403,
+      "grad_norm": 0.8381150960922241,
+      "learning_rate": 0.000153979492714517,
+      "loss": 3.2452,
       "step": 69000
     },
     {
       "epoch": 7.439353099730458,
-      "eval_accuracy": 0.38884412993538087,
-      "eval_loss": 3.345785617828369,
-      "eval_runtime": 183.4217,
-      "eval_samples_per_second": 98.195,
-      "eval_steps_per_second": 6.139,
+      "eval_accuracy": 0.38859574943383696,
+      "eval_loss": 3.346266984939575,
+      "eval_runtime": 181.8406,
+      "eval_samples_per_second": 99.048,
+      "eval_steps_per_second": 6.192,
       "step": 69000
     },
     {
       "epoch": 7.444743935309973,
-      "grad_norm": 0.766400933265686,
-      "learning_rate": 0.00015366216945493792,
-      "loss": 3.2498,
+      "grad_norm": 0.7726186513900757,
+      "learning_rate": 0.00015365569347004858,
+      "loss": 3.2522,
       "step": 69050
     },
     {
       "epoch": 7.450134770889488,
-      "grad_norm": 0.7826684713363647,
-      "learning_rate": 0.00015333837021046947,
-      "loss": 3.2286,
+      "grad_norm": 0.8026612401008606,
+      "learning_rate": 0.00015333189422558015,
+      "loss": 3.233,
       "step": 69100
     },
     {
       "epoch": 7.455525606469003,
-      "grad_norm": 0.7703137993812561,
-      "learning_rate": 0.00015301457096600105,
-      "loss": 3.2104,
+      "grad_norm": 0.7991743087768555,
+      "learning_rate": 0.00015300809498111168,
+      "loss": 3.212,
       "step": 69150
     },
     {
       "epoch": 7.460916442048518,
-      "grad_norm": 0.7931674718856812,
-      "learning_rate": 0.00015269077172153263,
-      "loss": 3.2337,
+      "grad_norm": 0.8170613646507263,
+      "learning_rate": 0.00015268429573664325,
+      "loss": 3.2373,
       "step": 69200
     },
     {
       "epoch": 7.466307277628032,
-      "grad_norm": 0.8100343346595764,
-      "learning_rate": 0.0001523669724770642,
-      "loss": 3.2269,
+      "grad_norm": 0.8246504664421082,
+      "learning_rate": 0.00015236049649217483,
+      "loss": 3.2296,
       "step": 69250
     },
     {
       "epoch": 7.471698113207547,
-      "grad_norm": 0.7942655682563782,
-      "learning_rate": 0.00015204317323259578,
-      "loss": 3.2206,
+      "grad_norm": 0.7863410115242004,
+      "learning_rate": 0.0001520366972477064,
+      "loss": 3.2231,
       "step": 69300
     },
     {
       "epoch": 7.4770889487870615,
-      "grad_norm": 0.8322016000747681,
-      "learning_rate": 0.00015171937398812736,
-      "loss": 3.221,
+      "grad_norm": 0.8042203783988953,
+      "learning_rate": 0.000151712898003238,
+      "loss": 3.2227,
       "step": 69350
     },
     {
       "epoch": 7.482479784366577,
-      "grad_norm": 0.7898223400115967,
-      "learning_rate": 0.00015139557474365894,
-      "loss": 3.248,
+      "grad_norm": 0.7908916473388672,
+      "learning_rate": 0.00015138909875876956,
+      "loss": 3.2511,
       "step": 69400
     },
     {
       "epoch": 7.487870619946092,
-      "grad_norm": 0.8208786845207214,
-      "learning_rate": 0.00015107177549919046,
-      "loss": 3.2311,
+      "grad_norm": 0.7854745984077454,
+      "learning_rate": 0.00015106529951430114,
+      "loss": 3.2344,
       "step": 69450
     },
     {
       "epoch": 7.493261455525606,
-      "grad_norm": 0.7814972400665283,
-      "learning_rate": 0.00015074797625472204,
-      "loss": 3.2196,
+      "grad_norm": 0.761583685874939,
+      "learning_rate": 0.0001507415002698327,
+      "loss": 3.2243,
       "step": 69500
     },
     {
       "epoch": 7.498652291105121,
-      "grad_norm": 0.8326282501220703,
-      "learning_rate": 0.00015042417701025362,
-      "loss": 3.2347,
+      "grad_norm": 0.7651922702789307,
+      "learning_rate": 0.00015041770102536424,
+      "loss": 3.2375,
       "step": 69550
     },
     {
       "epoch": 7.504043126684636,
-      "grad_norm": 0.8042081594467163,
-      "learning_rate": 0.0001501003777657852,
-      "loss": 3.2242,
+      "grad_norm": 0.7790371775627136,
+      "learning_rate": 0.00015009390178089582,
+      "loss": 3.2284,
       "step": 69600
     },
     {
       "epoch": 7.509433962264151,
-      "grad_norm": 0.7709265947341919,
-      "learning_rate": 0.00014977657852131677,
-      "loss": 3.2092,
+      "grad_norm": 0.7677376866340637,
+      "learning_rate": 0.0001497701025364274,
+      "loss": 3.2113,
       "step": 69650
     },
     {
       "epoch": 7.514824797843666,
-      "grad_norm": 0.7810226678848267,
-      "learning_rate": 0.0001494592552617377,
-      "loss": 3.2464,
+      "grad_norm": 0.8034192323684692,
+      "learning_rate": 0.00014944630329195898,
+      "loss": 3.2493,
       "step": 69700
     },
     {
       "epoch": 7.520215633423181,
-      "grad_norm": 0.8230692744255066,
-      "learning_rate": 0.00014913545601726928,
-      "loss": 3.2242,
+      "grad_norm": 0.8122262954711914,
+      "learning_rate": 0.00014912250404749055,
+      "loss": 3.2274,
       "step": 69750
     },
     {
       "epoch": 7.525606469002695,
-      "grad_norm": 0.7870705127716064,
-      "learning_rate": 0.00014881165677280085,
-      "loss": 3.2335,
+      "grad_norm": 0.7946466207504272,
+      "learning_rate": 0.0001487987048030221,
+      "loss": 3.2375,
       "step": 69800
     },
     {
       "epoch": 7.53099730458221,
-      "grad_norm": 0.7646998763084412,
-      "learning_rate": 0.00014848785752833243,
-      "loss": 3.219,
+      "grad_norm": 0.7666792869567871,
+      "learning_rate": 0.00014847490555855368,
+      "loss": 3.22,
       "step": 69850
     },
     {
       "epoch": 7.536388140161725,
-      "grad_norm": 0.8329770565032959,
-      "learning_rate": 0.00014816405828386398,
-      "loss": 3.239,
+      "grad_norm": 0.7951945066452026,
+      "learning_rate": 0.00014815110631408526,
+      "loss": 3.2422,
       "step": 69900
     },
     {
       "epoch": 7.54177897574124,
-      "grad_norm": 0.8003082275390625,
-      "learning_rate": 0.00014784025903939556,
-      "loss": 3.2322,
+      "grad_norm": 0.7725884914398193,
+      "learning_rate": 0.0001478273070696168,
+      "loss": 3.2336,
       "step": 69950
     },
     {
       "epoch": 7.547169811320755,
-      "grad_norm": 0.7998071312904358,
-      "learning_rate": 0.00014751645979492714,
-      "loss": 3.2254,
+      "grad_norm": 0.7850934267044067,
+      "learning_rate": 0.0001475035078251484,
+      "loss": 3.2271,
       "step": 70000
     },
     {
       "epoch": 7.547169811320755,
-      "eval_accuracy": 0.3892735261480237,
-      "eval_loss": 3.343639373779297,
-      "eval_runtime": 183.6901,
-      "eval_samples_per_second": 98.051,
-      "eval_steps_per_second": 6.13,
+      "eval_accuracy": 0.38896125774844487,
+      "eval_loss": 3.342238664627075,
+      "eval_runtime": 181.4447,
+      "eval_samples_per_second": 99.264,
+      "eval_steps_per_second": 6.206,
       "step": 70000
     },
     {
       "epoch": 7.55256064690027,
-      "grad_norm": 0.7884393930435181,
-      "learning_rate": 0.00014719266055045871,
-      "loss": 3.2285,
+      "grad_norm": 0.7838685512542725,
+      "learning_rate": 0.00014717970858067997,
+      "loss": 3.232,
       "step": 70050
     },
     {
       "epoch": 7.557951482479784,
-      "grad_norm": 0.8215367197990417,
-      "learning_rate": 0.00014686886130599027,
-      "loss": 3.238,
+      "grad_norm": 0.82815021276474,
+      "learning_rate": 0.00014685590933621154,
+      "loss": 3.2404,
       "step": 70100
     },
     {
       "epoch": 7.563342318059299,
-      "grad_norm": 0.794144332408905,
-      "learning_rate": 0.00014654506206152184,
-      "loss": 3.2335,
+      "grad_norm": 0.7995164394378662,
+      "learning_rate": 0.0001465321100917431,
+      "loss": 3.2346,
       "step": 70150
     },
     {
       "epoch": 7.568733153638814,
-      "grad_norm": 0.8532745838165283,
-      "learning_rate": 0.00014622126281705342,
-      "loss": 3.23,
+      "grad_norm": 0.8142074346542358,
+      "learning_rate": 0.00014620831084727467,
+      "loss": 3.2307,
       "step": 70200
     },
     {
       "epoch": 7.574123989218329,
-      "grad_norm": 0.8423836827278137,
-      "learning_rate": 0.000145897463572585,
-      "loss": 3.2318,
+      "grad_norm": 0.8016924262046814,
+      "learning_rate": 0.00014588451160280625,
+      "loss": 3.2374,
       "step": 70250
     },
     {
       "epoch": 7.579514824797844,
-      "grad_norm": 0.8306622505187988,
-      "learning_rate": 0.00014557366432811658,
-      "loss": 3.2334,
+      "grad_norm": 0.8279039263725281,
+      "learning_rate": 0.0001455607123583378,
+      "loss": 3.2357,
       "step": 70300
     },
     {
       "epoch": 7.584905660377358,
-      "grad_norm": 0.8016265630722046,
-      "learning_rate": 0.00014524986508364813,
-      "loss": 3.2495,
+      "grad_norm": 0.8122028112411499,
+      "learning_rate": 0.00014523691311386938,
+      "loss": 3.2493,
       "step": 70350
     },
     {
       "epoch": 7.590296495956873,
-      "grad_norm": 0.7804158329963684,
-      "learning_rate": 0.0001449260658391797,
-      "loss": 3.22,
+      "grad_norm": 0.8149353861808777,
+      "learning_rate": 0.00014491311386940096,
+      "loss": 3.2234,
       "step": 70400
     },
     {
       "epoch": 7.595687331536388,
-      "grad_norm": 0.8219479322433472,
-      "learning_rate": 0.00014460226659471128,
-      "loss": 3.2402,
+      "grad_norm": 0.8224563002586365,
+      "learning_rate": 0.00014458931462493253,
+      "loss": 3.2426,
       "step": 70450
     },
     {
       "epoch": 7.601078167115903,
-      "grad_norm": 0.8083851933479309,
-      "learning_rate": 0.00014427846735024283,
-      "loss": 3.2458,
+      "grad_norm": 0.8260905742645264,
+      "learning_rate": 0.00014426551538046408,
+      "loss": 3.2487,
       "step": 70500
     },
     {
       "epoch": 7.606469002695418,
-      "grad_norm": 0.7973158359527588,
-      "learning_rate": 0.0001439546681057744,
-      "loss": 3.2348,
+      "grad_norm": 0.8318034410476685,
+      "learning_rate": 0.00014394171613599566,
+      "loss": 3.2388,
       "step": 70550
     },
     {
       "epoch": 7.611859838274933,
-      "grad_norm": 0.8355487585067749,
-      "learning_rate": 0.000143630868861306,
-      "loss": 3.2238,
+      "grad_norm": 0.8205267786979675,
+      "learning_rate": 0.00014361791689152724,
+      "loss": 3.2263,
       "step": 70600
     },
     {
       "epoch": 7.617250673854447,
-      "grad_norm": 0.8130533695220947,
-      "learning_rate": 0.00014330706961683754,
-      "loss": 3.2209,
+      "grad_norm": 0.8046087622642517,
+      "learning_rate": 0.00014329411764705882,
+      "loss": 3.2256,
       "step": 70650
     },
     {
       "epoch": 7.622641509433962,
-      "grad_norm": 0.824119508266449,
-      "learning_rate": 0.00014298327037236912,
-      "loss": 3.2475,
+      "grad_norm": 0.8188928961753845,
+      "learning_rate": 0.0001429703184025904,
+      "loss": 3.2508,
       "step": 70700
     },
     {
       "epoch": 7.628032345013477,
-      "grad_norm": 0.8470996022224426,
-      "learning_rate": 0.0001426594711279007,
-      "loss": 3.2272,
+      "grad_norm": 0.7892650961875916,
+      "learning_rate": 0.00014264651915812194,
+      "loss": 3.231,
       "step": 70750
     },
     {
       "epoch": 7.633423180592992,
-      "grad_norm": 0.812836229801178,
-      "learning_rate": 0.00014233567188343224,
-      "loss": 3.2344,
+      "grad_norm": 0.8116347193717957,
+      "learning_rate": 0.00014232271991365352,
+      "loss": 3.238,
       "step": 70800
     },
     {
       "epoch": 7.638814016172507,
-      "grad_norm": 0.7536956667900085,
-      "learning_rate": 0.00014201187263896382,
-      "loss": 3.2221,
+      "grad_norm": 0.7771171927452087,
+      "learning_rate": 0.0001419989206691851,
+      "loss": 3.2253,
       "step": 70850
     },
     {
       "epoch": 7.644204851752022,
-      "grad_norm": 0.8352275490760803,
-      "learning_rate": 0.0001416880733944954,
-      "loss": 3.2347,
+      "grad_norm": 0.8153142333030701,
+      "learning_rate": 0.00014167512142471668,
+      "loss": 3.237,
       "step": 70900
     },
     {
       "epoch": 7.649595687331536,
-      "grad_norm": 0.7672180533409119,
-      "learning_rate": 0.00014136427415002695,
-      "loss": 3.2151,
+      "grad_norm": 0.810912549495697,
+      "learning_rate": 0.00014135132218024823,
+      "loss": 3.2179,
       "step": 70950
     },
     {
       "epoch": 7.654986522911051,
-      "grad_norm": 0.8054909706115723,
-      "learning_rate": 0.00014104047490555853,
-      "loss": 3.2215,
+      "grad_norm": 0.7712694406509399,
+      "learning_rate": 0.0001410275229357798,
+      "loss": 3.2237,
       "step": 71000
     },
     {
       "epoch": 7.654986522911051,
-      "eval_accuracy": 0.38935914462362153,
-      "eval_loss": 3.3389949798583984,
-      "eval_runtime": 183.3427,
-      "eval_samples_per_second": 98.237,
-      "eval_steps_per_second": 6.142,
+      "eval_accuracy": 0.3892282178938051,
+      "eval_loss": 3.338327407836914,
+      "eval_runtime": 181.43,
+      "eval_samples_per_second": 99.272,
+      "eval_steps_per_second": 6.206,
       "step": 71000
     },
     {
       "epoch": 7.660377358490566,
-      "grad_norm": 0.8233856558799744,
-      "learning_rate": 0.0001407166756610901,
-      "loss": 3.2494,
+      "grad_norm": 0.7952535152435303,
+      "learning_rate": 0.00014070372369131138,
+      "loss": 3.2515,
       "step": 71050
     },
     {
       "epoch": 7.665768194070081,
-      "grad_norm": 0.8189040422439575,
-      "learning_rate": 0.00014039287641662168,
-      "loss": 3.2215,
+      "grad_norm": 0.7871846556663513,
+      "learning_rate": 0.00014037992444684296,
+      "loss": 3.2253,
       "step": 71100
     },
     {
       "epoch": 7.671159029649596,
-      "grad_norm": 0.8202671408653259,
-      "learning_rate": 0.00014006907717215326,
-      "loss": 3.2173,
+      "grad_norm": 0.8093687295913696,
+      "learning_rate": 0.0001400561252023745,
+      "loss": 3.2189,
       "step": 71150
     },
     {
       "epoch": 7.67654986522911,
-      "grad_norm": 0.8153406977653503,
-      "learning_rate": 0.0001397452779276848,
-      "loss": 3.223,
+      "grad_norm": 0.8155460357666016,
+      "learning_rate": 0.0001397323259579061,
+      "loss": 3.2254,
       "step": 71200
     },
     {
       "epoch": 7.681940700808625,
-      "grad_norm": 0.8444876670837402,
-      "learning_rate": 0.0001394214786832164,
-      "loss": 3.2304,
+      "grad_norm": 0.8383728861808777,
+      "learning_rate": 0.00013940852671343767,
+      "loss": 3.2323,
       "step": 71250
     },
     {
       "epoch": 7.6873315363881405,
-      "grad_norm": 0.836453378200531,
-      "learning_rate": 0.00013909767943874797,
-      "loss": 3.2242,
+      "grad_norm": 0.8723028898239136,
+      "learning_rate": 0.00013908472746896924,
+      "loss": 3.2293,
       "step": 71300
     },
     {
       "epoch": 7.692722371967655,
-      "grad_norm": 0.7985456585884094,
-      "learning_rate": 0.00013877388019427954,
-      "loss": 3.2336,
+      "grad_norm": 0.8243257999420166,
+      "learning_rate": 0.0001387609282245008,
+      "loss": 3.237,
       "step": 71350
     },
     {
       "epoch": 7.69811320754717,
-      "grad_norm": 0.8078007102012634,
-      "learning_rate": 0.0001384500809498111,
-      "loss": 3.2358,
+      "grad_norm": 0.7792996764183044,
+      "learning_rate": 0.00013843712898003237,
+      "loss": 3.2376,
       "step": 71400
     },
     {
       "epoch": 7.703504043126685,
-      "grad_norm": 0.9007571935653687,
-      "learning_rate": 0.00013812628170534267,
-      "loss": 3.2288,
+      "grad_norm": 0.8691493272781372,
+      "learning_rate": 0.00013811332973556395,
+      "loss": 3.231,
       "step": 71450
     },
     {
       "epoch": 7.708894878706199,
-      "grad_norm": 0.7986894249916077,
-      "learning_rate": 0.00013780248246087425,
-      "loss": 3.2364,
+      "grad_norm": 0.7997547388076782,
+      "learning_rate": 0.0001377895304910955,
+      "loss": 3.2387,
       "step": 71500
     },
     {
       "epoch": 7.714285714285714,
-      "grad_norm": 0.8218685388565063,
-      "learning_rate": 0.00013747868321640583,
-      "loss": 3.2235,
+      "grad_norm": 0.8469955325126648,
+      "learning_rate": 0.00013746573124662708,
+      "loss": 3.2248,
       "step": 71550
     },
     {
       "epoch": 7.719676549865229,
-      "grad_norm": 0.8080571293830872,
-      "learning_rate": 0.00013715488397193738,
-      "loss": 3.2249,
+      "grad_norm": 0.821802020072937,
+      "learning_rate": 0.00013714193200215866,
+      "loss": 3.2306,
       "step": 71600
     },
     {
       "epoch": 7.725067385444744,
-      "grad_norm": 0.854092001914978,
-      "learning_rate": 0.00013683108472746896,
-      "loss": 3.2193,
+      "grad_norm": 0.8693572878837585,
+      "learning_rate": 0.00013682460874257958,
+      "loss": 3.2208,
       "step": 71650
     },
     {
       "epoch": 7.730458221024259,
-      "grad_norm": 0.7832936644554138,
-      "learning_rate": 0.00013650728548300053,
-      "loss": 3.2439,
+      "grad_norm": 0.8120995163917542,
+      "learning_rate": 0.00013650080949811116,
+      "loss": 3.2484,
       "step": 71700
     },
     {
       "epoch": 7.735849056603773,
-      "grad_norm": 0.779666006565094,
-      "learning_rate": 0.0001361834862385321,
-      "loss": 3.2301,
+      "grad_norm": 0.7727512121200562,
+      "learning_rate": 0.00013617701025364274,
+      "loss": 3.2328,
       "step": 71750
     },
     {
       "epoch": 7.741239892183288,
-      "grad_norm": 0.8854478597640991,
-      "learning_rate": 0.0001358596869940637,
-      "loss": 3.2211,
+      "grad_norm": 0.8336825966835022,
+      "learning_rate": 0.0001358532110091743,
+      "loss": 3.221,
       "step": 71800
     },
     {
       "epoch": 7.7466307277628035,
-      "grad_norm": 0.7972925901412964,
-      "learning_rate": 0.00013553588774959524,
-      "loss": 3.2189,
+      "grad_norm": 0.8126691579818726,
+      "learning_rate": 0.00013552941176470587,
+      "loss": 3.2201,
       "step": 71850
     },
     {
       "epoch": 7.752021563342318,
-      "grad_norm": 0.853445291519165,
-      "learning_rate": 0.00013521208850512682,
-      "loss": 3.2392,
+      "grad_norm": 0.8726671934127808,
+      "learning_rate": 0.00013520561252023744,
+      "loss": 3.241,
       "step": 71900
     },
     {
       "epoch": 7.757412398921833,
-      "grad_norm": 0.8733102083206177,
-      "learning_rate": 0.0001348882892606584,
-      "loss": 3.2315,
+      "grad_norm": 0.8359882235527039,
+      "learning_rate": 0.00013488181327576902,
+      "loss": 3.2334,
       "step": 71950
     },
     {
       "epoch": 7.762803234501348,
-      "grad_norm": 0.7951270937919617,
-      "learning_rate": 0.00013456449001618995,
-      "loss": 3.235,
+      "grad_norm": 0.8177676200866699,
+      "learning_rate": 0.00013455801403130057,
+      "loss": 3.2361,
       "step": 72000
     },
     {
       "epoch": 7.762803234501348,
-      "eval_accuracy": 0.39014046753989545,
-      "eval_loss": 3.3319358825683594,
-      "eval_runtime": 183.3781,
-      "eval_samples_per_second": 98.218,
-      "eval_steps_per_second": 6.14,
+      "eval_accuracy": 0.38985970847778284,
+      "eval_loss": 3.3334810733795166,
+      "eval_runtime": 184.1532,
+      "eval_samples_per_second": 97.804,
+      "eval_steps_per_second": 6.114,
       "step": 72000
     },
     {
       "epoch": 7.768194070080862,
-      "grad_norm": 0.7769864201545715,
-      "learning_rate": 0.00013424069077172152,
-      "loss": 3.2239,
+      "grad_norm": 0.8026477694511414,
+      "learning_rate": 0.00013423421478683215,
+      "loss": 3.2276,
       "step": 72050
     },
     {
       "epoch": 7.773584905660377,
-      "grad_norm": 0.7999830842018127,
-      "learning_rate": 0.0001339168915272531,
-      "loss": 3.2342,
+      "grad_norm": 0.8216882348060608,
+      "learning_rate": 0.00013391041554236373,
+      "loss": 3.2374,
       "step": 72100
     },
     {
       "epoch": 7.7789757412398925,
-      "grad_norm": 0.781497597694397,
-      "learning_rate": 0.00013359309228278465,
-      "loss": 3.2331,
+      "grad_norm": 0.7629038691520691,
+      "learning_rate": 0.00013358661629789528,
+      "loss": 3.2361,
       "step": 72150
     },
     {
       "epoch": 7.784366576819407,
-      "grad_norm": 0.8018577098846436,
-      "learning_rate": 0.00013326929303831623,
-      "loss": 3.2561,
+      "grad_norm": 0.7912164926528931,
+      "learning_rate": 0.00013326281705342685,
+      "loss": 3.2593,
       "step": 72200
     },
     {
       "epoch": 7.789757412398922,
-      "grad_norm": 0.8189263939857483,
-      "learning_rate": 0.0001329454937938478,
-      "loss": 3.211,
+      "grad_norm": 0.8290787935256958,
+      "learning_rate": 0.00013293901780895843,
+      "loss": 3.2125,
       "step": 72250
     },
     {
       "epoch": 7.795148247978437,
-      "grad_norm": 0.8497806191444397,
-      "learning_rate": 0.00013262169454937936,
-      "loss": 3.2363,
+      "grad_norm": 0.8224929571151733,
+      "learning_rate": 0.00013261521856449,
+      "loss": 3.2391,
       "step": 72300
     },
     {
       "epoch": 7.800539083557951,
-      "grad_norm": 0.8176953792572021,
-      "learning_rate": 0.00013229789530491093,
-      "loss": 3.2281,
+      "grad_norm": 0.7874560356140137,
+      "learning_rate": 0.00013229141932002156,
+      "loss": 3.229,
       "step": 72350
     },
     {
       "epoch": 7.8059299191374665,
-      "grad_norm": 0.8766255378723145,
-      "learning_rate": 0.0001319740960604425,
-      "loss": 3.2288,
+      "grad_norm": 0.8600299954414368,
+      "learning_rate": 0.00013196762007555314,
+      "loss": 3.2316,
       "step": 72400
     },
     {
       "epoch": 7.811320754716981,
-      "grad_norm": 0.8095853924751282,
-      "learning_rate": 0.0001316502968159741,
-      "loss": 3.2339,
+      "grad_norm": 0.7986945509910583,
+      "learning_rate": 0.00013164382083108472,
+      "loss": 3.2369,
       "step": 72450
     },
     {
       "epoch": 7.816711590296496,
-      "grad_norm": 0.8431499600410461,
-      "learning_rate": 0.00013132649757150564,
-      "loss": 3.2225,
+      "grad_norm": 0.8087888360023499,
+      "learning_rate": 0.0001313200215866163,
+      "loss": 3.2238,
       "step": 72500
     },
     {
       "epoch": 7.822102425876011,
-      "grad_norm": 0.8290701508522034,
-      "learning_rate": 0.00013100269832703722,
-      "loss": 3.2066,
+      "grad_norm": 0.801675021648407,
+      "learning_rate": 0.00013099622234214784,
+      "loss": 3.2099,
       "step": 72550
     },
     {
       "epoch": 7.827493261455525,
-      "grad_norm": 0.780984103679657,
-      "learning_rate": 0.0001306788990825688,
-      "loss": 3.2393,
+      "grad_norm": 0.7589983940124512,
+      "learning_rate": 0.00013067242309767942,
+      "loss": 3.2411,
       "step": 72600
     },
     {
       "epoch": 7.83288409703504,
-      "grad_norm": 0.795058012008667,
-      "learning_rate": 0.00013035509983810037,
-      "loss": 3.2264,
+      "grad_norm": 0.8077195286750793,
+      "learning_rate": 0.000130348623853211,
+      "loss": 3.2287,
       "step": 72650
     },
     {
       "epoch": 7.8382749326145555,
-      "grad_norm": 0.833223819732666,
-      "learning_rate": 0.00013003130059363192,
-      "loss": 3.2348,
+      "grad_norm": 0.8149362206459045,
+      "learning_rate": 0.00013002482460874258,
+      "loss": 3.2369,
       "step": 72700
     },
     {
       "epoch": 7.84366576819407,
-      "grad_norm": 0.8560569882392883,
-      "learning_rate": 0.0001297075013491635,
-      "loss": 3.229,
+      "grad_norm": 0.8104026317596436,
+      "learning_rate": 0.00012970102536427413,
+      "loss": 3.2328,
       "step": 72750
     },
     {
       "epoch": 7.849056603773585,
-      "grad_norm": 0.8031648397445679,
-      "learning_rate": 0.00012938370210469508,
-      "loss": 3.2249,
+      "grad_norm": 0.8013917207717896,
+      "learning_rate": 0.0001293772261198057,
+      "loss": 3.2285,
       "step": 72800
     },
     {
       "epoch": 7.8544474393531,
-      "grad_norm": 0.8022270202636719,
-      "learning_rate": 0.00012905990286022666,
-      "loss": 3.2204,
+      "grad_norm": 0.7973603010177612,
+      "learning_rate": 0.00012905342687533728,
+      "loss": 3.2218,
       "step": 72850
     },
     {
       "epoch": 7.859838274932614,
-      "grad_norm": 0.7824501991271973,
-      "learning_rate": 0.0001287361036157582,
-      "loss": 3.2205,
+      "grad_norm": 0.778017520904541,
+      "learning_rate": 0.00012872962763086886,
+      "loss": 3.2242,
       "step": 72900
     },
     {
       "epoch": 7.8652291105121295,
-      "grad_norm": 0.8288241028785706,
-      "learning_rate": 0.00012841230437128979,
-      "loss": 3.2178,
+      "grad_norm": 0.8143687844276428,
+      "learning_rate": 0.00012840582838640044,
+      "loss": 3.2183,
       "step": 72950
     },
     {
       "epoch": 7.870619946091644,
-      "grad_norm": 0.8087592124938965,
-      "learning_rate": 0.00012808850512682136,
-      "loss": 3.2186,
+      "grad_norm": 0.8002779483795166,
+      "learning_rate": 0.000128082029141932,
+      "loss": 3.2212,
       "step": 73000
     },
     {
       "epoch": 7.870619946091644,
-      "eval_accuracy": 0.3905230343578902,
-      "eval_loss": 3.329652786254883,
-      "eval_runtime": 183.3863,
-      "eval_samples_per_second": 98.213,
-      "eval_steps_per_second": 6.14,
+      "eval_accuracy": 0.3901797998852843,
+      "eval_loss": 3.3310370445251465,
+      "eval_runtime": 184.5237,
+      "eval_samples_per_second": 97.608,
+      "eval_steps_per_second": 6.102,
       "step": 73000
     },
     {
       "epoch": 7.876010781671159,
-      "grad_norm": 0.8783208727836609,
-      "learning_rate": 0.00012776470588235294,
-      "loss": 3.2303,
+      "grad_norm": 0.8779856562614441,
+      "learning_rate": 0.00012775822989746357,
+      "loss": 3.2333,
       "step": 73050
     },
     {
       "epoch": 7.881401617250674,
-      "grad_norm": 0.850925862789154,
-      "learning_rate": 0.0001274409066378845,
-      "loss": 3.2095,
+      "grad_norm": 0.8837502598762512,
+      "learning_rate": 0.00012743443065299514,
+      "loss": 3.2123,
       "step": 73100
     },
     {
       "epoch": 7.886792452830189,
-      "grad_norm": 0.7974444627761841,
-      "learning_rate": 0.00012711710739341607,
-      "loss": 3.2193,
+      "grad_norm": 0.7986597418785095,
+      "learning_rate": 0.00012711063140852672,
+      "loss": 3.2207,
       "step": 73150
     },
     {
       "epoch": 7.892183288409703,
-      "grad_norm": 0.8161489367485046,
-      "learning_rate": 0.00012679330814894765,
-      "loss": 3.2342,
+      "grad_norm": 0.8212838172912598,
+      "learning_rate": 0.00012678683216405827,
+      "loss": 3.2364,
       "step": 73200
     },
     {
       "epoch": 7.8975741239892185,
-      "grad_norm": 0.8146517872810364,
-      "learning_rate": 0.00012646950890447922,
-      "loss": 3.2266,
+      "grad_norm": 0.8679354786872864,
+      "learning_rate": 0.00012646303291958985,
+      "loss": 3.2281,
       "step": 73250
     },
     {
       "epoch": 7.902964959568733,
-      "grad_norm": 0.8209285140037537,
-      "learning_rate": 0.00012614570966001077,
-      "loss": 3.2198,
+      "grad_norm": 0.8051177859306335,
+      "learning_rate": 0.00012613923367512143,
+      "loss": 3.2223,
       "step": 73300
     },
     {
       "epoch": 7.908355795148248,
-      "grad_norm": 0.8260653614997864,
-      "learning_rate": 0.00012582191041554235,
-      "loss": 3.2057,
+      "grad_norm": 0.8221176862716675,
+      "learning_rate": 0.00012581543443065298,
+      "loss": 3.2083,
       "step": 73350
     },
     {
       "epoch": 7.913746630727763,
-      "grad_norm": 0.8421151638031006,
-      "learning_rate": 0.00012549811117107393,
-      "loss": 3.2516,
+      "grad_norm": 0.8385334610939026,
+      "learning_rate": 0.00012549163518618456,
+      "loss": 3.2529,
       "step": 73400
     },
     {
       "epoch": 7.919137466307277,
-      "grad_norm": 0.8532014489173889,
-      "learning_rate": 0.0001251743119266055,
-      "loss": 3.2325,
+      "grad_norm": 0.8031367659568787,
+      "learning_rate": 0.00012516783594171613,
+      "loss": 3.2334,
       "step": 73450
     },
     {
       "epoch": 7.9245283018867925,
-      "grad_norm": 0.7685270309448242,
-      "learning_rate": 0.00012485051268213706,
-      "loss": 3.2406,
+      "grad_norm": 0.8049946427345276,
+      "learning_rate": 0.00012484403669724768,
+      "loss": 3.245,
       "step": 73500
     },
     {
       "epoch": 7.929919137466308,
-      "grad_norm": 0.859190821647644,
-      "learning_rate": 0.00012452671343766864,
-      "loss": 3.2347,
+      "grad_norm": 0.8416993618011475,
+      "learning_rate": 0.00012452023745277926,
+      "loss": 3.2362,
       "step": 73550
     },
     {
       "epoch": 7.935309973045822,
-      "grad_norm": 0.8019168972969055,
-      "learning_rate": 0.00012420291419320021,
-      "loss": 3.2328,
+      "grad_norm": 0.8018704652786255,
+      "learning_rate": 0.00012419643820831084,
+      "loss": 3.2351,
       "step": 73600
     },
     {
       "epoch": 7.940700808625337,
-      "grad_norm": 0.849294126033783,
-      "learning_rate": 0.0001238791149487318,
-      "loss": 3.2427,
+      "grad_norm": 0.8655563592910767,
+      "learning_rate": 0.0001238726389638424,
+      "loss": 3.2429,
       "step": 73650
     },
     {
       "epoch": 7.946091644204852,
-      "grad_norm": 0.8092149496078491,
-      "learning_rate": 0.00012356179168915272,
-      "loss": 3.2481,
+      "grad_norm": 0.8544971942901611,
+      "learning_rate": 0.00012354883971937397,
+      "loss": 3.2505,
       "step": 73700
     },
     {
       "epoch": 7.951482479784366,
-      "grad_norm": 0.8331597447395325,
-      "learning_rate": 0.0001232379924446843,
-      "loss": 3.2047,
+      "grad_norm": 0.827673614025116,
+      "learning_rate": 0.00012322504047490555,
+      "loss": 3.2074,
       "step": 73750
     },
     {
       "epoch": 7.9568733153638815,
-      "grad_norm": 0.8321600556373596,
-      "learning_rate": 0.00012291419320021587,
-      "loss": 3.2282,
+      "grad_norm": 0.7904378175735474,
+      "learning_rate": 0.00012290124123043712,
+      "loss": 3.2291,
       "step": 73800
     },
     {
       "epoch": 7.962264150943396,
-      "grad_norm": 0.8063613772392273,
-      "learning_rate": 0.00012259039395574742,
-      "loss": 3.2137,
+      "grad_norm": 0.8064669370651245,
+      "learning_rate": 0.00012257744198596867,
+      "loss": 3.2144,
       "step": 73850
     },
     {
       "epoch": 7.967654986522911,
-      "grad_norm": 0.853704035282135,
-      "learning_rate": 0.000122266594711279,
-      "loss": 3.2196,
+      "grad_norm": 0.8689444065093994,
+      "learning_rate": 0.00012225364274150025,
+      "loss": 3.2219,
       "step": 73900
     },
     {
       "epoch": 7.973045822102426,
-      "grad_norm": 0.8520025014877319,
-      "learning_rate": 0.00012194279546681056,
-      "loss": 3.2307,
+      "grad_norm": 0.8743965029716492,
+      "learning_rate": 0.00012192984349703183,
+      "loss": 3.2345,
       "step": 73950
     },
     {
       "epoch": 7.97843665768194,
-      "grad_norm": 0.8468704223632812,
-      "learning_rate": 0.00012161899622234214,
-      "loss": 3.2275,
+      "grad_norm": 0.8444426655769348,
+      "learning_rate": 0.0001216060442525634,
+      "loss": 3.23,
       "step": 74000
     },
     {
       "epoch": 7.97843665768194,
-      "eval_accuracy": 0.3908582285167979,
-      "eval_loss": 3.326791286468506,
-      "eval_runtime": 183.707,
-      "eval_samples_per_second": 98.042,
-      "eval_steps_per_second": 6.129,
+      "eval_accuracy": 0.39047153288906616,
+      "eval_loss": 3.3288681507110596,
+      "eval_runtime": 185.9106,
+      "eval_samples_per_second": 96.88,
+      "eval_steps_per_second": 6.057,
       "step": 74000
     },
     {
       "epoch": 7.9838274932614555,
-      "grad_norm": 0.8065071105957031,
-      "learning_rate": 0.00012129519697787372,
-      "loss": 3.2299,
+      "grad_norm": 0.8021253347396851,
+      "learning_rate": 0.00012128224500809496,
+      "loss": 3.232,
       "step": 74050
     },
     {
       "epoch": 7.989218328840971,
-      "grad_norm": 0.8211476802825928,
-      "learning_rate": 0.00012097139773340527,
-      "loss": 3.2332,
+      "grad_norm": 0.8003281354904175,
+      "learning_rate": 0.00012095844576362653,
+      "loss": 3.2382,
       "step": 74100
     },
     {
       "epoch": 7.994609164420485,
-      "grad_norm": 0.7979119420051575,
-      "learning_rate": 0.00012064759848893685,
-      "loss": 3.2219,
+      "grad_norm": 0.8485066294670105,
+      "learning_rate": 0.00012063464651915811,
+      "loss": 3.2251,
       "step": 74150
     },
     {
       "epoch": 8.0,
-      "grad_norm": 1.7197800874710083,
-      "learning_rate": 0.00012032379924446843,
-      "loss": 3.2359,
+      "grad_norm": 1.7434502840042114,
+      "learning_rate": 0.00012031084727468969,
+      "loss": 3.2397,
       "step": 74200
     },
     {
       "epoch": 8.005390835579515,
-      "grad_norm": 0.8147539496421814,
-      "learning_rate": 0.00011999999999999999,
-      "loss": 3.1524,
+      "grad_norm": 0.8195734024047852,
+      "learning_rate": 0.00011998704803022124,
+      "loss": 3.1567,
       "step": 74250
     },
     {
       "epoch": 8.01078167115903,
-      "grad_norm": 0.8394148945808411,
-      "learning_rate": 0.00011967620075553155,
-      "loss": 3.1463,
+      "grad_norm": 0.8173742890357971,
+      "learning_rate": 0.00011966324878575282,
+      "loss": 3.1501,
       "step": 74300
     },
     {
       "epoch": 8.016172506738544,
-      "grad_norm": 0.8568943738937378,
-      "learning_rate": 0.00011935240151106313,
-      "loss": 3.1327,
+      "grad_norm": 0.8754720091819763,
+      "learning_rate": 0.0001193394495412844,
+      "loss": 3.1364,
       "step": 74350
     },
     {
       "epoch": 8.021563342318059,
-      "grad_norm": 0.8077348470687866,
-      "learning_rate": 0.00011902860226659471,
-      "loss": 3.1388,
+      "grad_norm": 0.8561992049217224,
+      "learning_rate": 0.00011901565029681597,
+      "loss": 3.1413,
       "step": 74400
     },
     {
       "epoch": 8.026954177897574,
-      "grad_norm": 0.785599410533905,
-      "learning_rate": 0.00011870480302212627,
-      "loss": 3.1547,
+      "grad_norm": 0.7828686237335205,
+      "learning_rate": 0.00011869185105234754,
+      "loss": 3.1596,
       "step": 74450
     },
     {
       "epoch": 8.032345013477089,
-      "grad_norm": 0.8443362712860107,
-      "learning_rate": 0.00011838100377765784,
-      "loss": 3.1424,
+      "grad_norm": 0.8400721549987793,
+      "learning_rate": 0.0001183680518078791,
+      "loss": 3.1453,
       "step": 74500
     },
     {
       "epoch": 8.037735849056604,
-      "grad_norm": 0.8324596881866455,
-      "learning_rate": 0.00011805720453318941,
-      "loss": 3.1601,
+      "grad_norm": 0.8275683522224426,
+      "learning_rate": 0.00011804425256341068,
+      "loss": 3.1622,
       "step": 74550
     },
     {
       "epoch": 8.04312668463612,
-      "grad_norm": 0.8503394722938538,
-      "learning_rate": 0.00011773340528872098,
-      "loss": 3.1528,
+      "grad_norm": 0.8730551600456238,
+      "learning_rate": 0.00011772045331894224,
+      "loss": 3.1548,
       "step": 74600
     },
     {
       "epoch": 8.048517520215633,
-      "grad_norm": 0.8025884628295898,
-      "learning_rate": 0.00011740960604425256,
-      "loss": 3.1453,
+      "grad_norm": 0.8044026494026184,
+      "learning_rate": 0.00011739665407447382,
+      "loss": 3.1501,
       "step": 74650
     },
     {
       "epoch": 8.053908355795148,
-      "grad_norm": 0.8412855267524719,
-      "learning_rate": 0.00011708580679978412,
-      "loss": 3.1539,
+      "grad_norm": 0.8507840037345886,
+      "learning_rate": 0.00011707285483000539,
+      "loss": 3.157,
       "step": 74700
     },
     {
       "epoch": 8.059299191374663,
-      "grad_norm": 0.8652845025062561,
-      "learning_rate": 0.00011676200755531568,
-      "loss": 3.1489,
+      "grad_norm": 0.8273726105690002,
+      "learning_rate": 0.00011674905558553695,
+      "loss": 3.1528,
       "step": 74750
     },
     {
       "epoch": 8.064690026954178,
-      "grad_norm": 0.8231407999992371,
-      "learning_rate": 0.00011643820831084726,
-      "loss": 3.156,
+      "grad_norm": 0.8017041087150574,
+      "learning_rate": 0.00011642525634106853,
+      "loss": 3.1589,
       "step": 74800
     },
     {
       "epoch": 8.070080862533693,
-      "grad_norm": 0.8152666687965393,
-      "learning_rate": 0.00011611440906637884,
-      "loss": 3.1646,
+      "grad_norm": 0.8189218044281006,
+      "learning_rate": 0.0001161014570966001,
+      "loss": 3.1665,
       "step": 74850
     },
     {
       "epoch": 8.075471698113208,
-      "grad_norm": 0.8229835033416748,
-      "learning_rate": 0.00011579060982191042,
-      "loss": 3.141,
+      "grad_norm": 0.8475618958473206,
+      "learning_rate": 0.00011577765785213166,
+      "loss": 3.1459,
       "step": 74900
     },
     {
       "epoch": 8.080862533692722,
-      "grad_norm": 0.8419477343559265,
-      "learning_rate": 0.00011546681057744197,
-      "loss": 3.1425,
+      "grad_norm": 0.8230682015419006,
+      "learning_rate": 0.00011545385860766323,
+      "loss": 3.1469,
       "step": 74950
     },
     {
       "epoch": 8.086253369272237,
-      "grad_norm": 0.8185001015663147,
-      "learning_rate": 0.00011514301133297355,
-      "loss": 3.1539,
+      "grad_norm": 0.821681559085846,
+      "learning_rate": 0.00011513005936319481,
+      "loss": 3.1565,
       "step": 75000
     },
     {
       "epoch": 8.086253369272237,
-      "eval_accuracy": 0.3905573686704396,
-      "eval_loss": 3.333664894104004,
-      "eval_runtime": 183.2216,
-      "eval_samples_per_second": 98.302,
-      "eval_steps_per_second": 6.146,
+      "eval_accuracy": 0.3902219572057563,
+      "eval_loss": 3.3358116149902344,
+      "eval_runtime": 185.4032,
+      "eval_samples_per_second": 97.145,
+      "eval_steps_per_second": 6.073,
       "step": 75000
     },
     {
       "epoch": 8.091644204851752,
-      "grad_norm": 0.8968559503555298,
-      "learning_rate": 0.00011481921208850512,
-      "loss": 3.1622,
+      "grad_norm": 0.8560944199562073,
+      "learning_rate": 0.00011480626011872639,
+      "loss": 3.166,
       "step": 75050
     },
     {
       "epoch": 8.097035040431267,
-      "grad_norm": 0.7851213812828064,
-      "learning_rate": 0.00011449541284403669,
-      "loss": 3.1479,
+      "grad_norm": 0.8111773133277893,
+      "learning_rate": 0.00011448893685914731,
+      "loss": 3.15,
       "step": 75100
     },
     {
       "epoch": 8.102425876010782,
-      "grad_norm": 0.8505614995956421,
-      "learning_rate": 0.00011417161359956825,
-      "loss": 3.1686,
+      "grad_norm": 0.8391156792640686,
+      "learning_rate": 0.00011416513761467889,
+      "loss": 3.1726,
       "step": 75150
     },
     {
       "epoch": 8.107816711590296,
-      "grad_norm": 0.8062684535980225,
-      "learning_rate": 0.00011384781435509983,
-      "loss": 3.1602,
+      "grad_norm": 0.86478191614151,
+      "learning_rate": 0.00011384133837021047,
+      "loss": 3.1637,
       "step": 75200
     },
     {
       "epoch": 8.11320754716981,
-      "grad_norm": 0.9041953682899475,
-      "learning_rate": 0.0001135240151106314,
-      "loss": 3.1506,
+      "grad_norm": 0.8701136112213135,
+      "learning_rate": 0.00011351753912574202,
+      "loss": 3.1545,
       "step": 75250
     },
     {
       "epoch": 8.118598382749326,
-      "grad_norm": 0.8515322208404541,
-      "learning_rate": 0.00011320021586616297,
-      "loss": 3.1608,
+      "grad_norm": 0.8543526530265808,
+      "learning_rate": 0.0001131937398812736,
+      "loss": 3.1621,
       "step": 75300
     },
     {
       "epoch": 8.123989218328841,
-      "grad_norm": 0.8603762984275818,
-      "learning_rate": 0.00011288289260658391,
-      "loss": 3.1645,
+      "grad_norm": 0.8470823764801025,
+      "learning_rate": 0.00011286994063680517,
+      "loss": 3.1673,
       "step": 75350
     },
     {
       "epoch": 8.129380053908356,
-      "grad_norm": 0.8565706014633179,
-      "learning_rate": 0.00011255909336211549,
-      "loss": 3.1632,
+      "grad_norm": 0.8628540635108948,
+      "learning_rate": 0.00011254614139233675,
+      "loss": 3.1667,
       "step": 75400
     },
     {
       "epoch": 8.134770889487871,
-      "grad_norm": 0.932990550994873,
-      "learning_rate": 0.00011223529411764705,
-      "loss": 3.167,
+      "grad_norm": 0.8851979970932007,
+      "learning_rate": 0.0001122223421478683,
+      "loss": 3.1701,
       "step": 75450
     },
     {
       "epoch": 8.140161725067385,
-      "grad_norm": 0.7992711663246155,
-      "learning_rate": 0.00011191149487317862,
-      "loss": 3.141,
+      "grad_norm": 0.8273195624351501,
+      "learning_rate": 0.00011189854290339988,
+      "loss": 3.1442,
       "step": 75500
     },
     {
       "epoch": 8.1455525606469,
-      "grad_norm": 0.835195004940033,
-      "learning_rate": 0.0001115876956287102,
-      "loss": 3.172,
+      "grad_norm": 0.8351007699966431,
+      "learning_rate": 0.00011157474365893146,
+      "loss": 3.177,
       "step": 75550
     },
     {
       "epoch": 8.150943396226415,
-      "grad_norm": 0.8241792917251587,
-      "learning_rate": 0.00011126389638424176,
-      "loss": 3.1497,
+      "grad_norm": 0.820804238319397,
+      "learning_rate": 0.00011125094441446302,
+      "loss": 3.1518,
       "step": 75600
     },
     {
       "epoch": 8.15633423180593,
-      "grad_norm": 0.7964764833450317,
-      "learning_rate": 0.00011094009713977334,
-      "loss": 3.1637,
+      "grad_norm": 0.8204725384712219,
+      "learning_rate": 0.00011092714516999459,
+      "loss": 3.167,
       "step": 75650
     },
     {
       "epoch": 8.161725067385445,
-      "grad_norm": 0.8399732708930969,
-      "learning_rate": 0.0001106162978953049,
-      "loss": 3.1866,
+      "grad_norm": 0.8204634189605713,
+      "learning_rate": 0.00011060334592552616,
+      "loss": 3.1887,
       "step": 75700
     },
     {
       "epoch": 8.167115902964959,
-      "grad_norm": 0.8350764513015747,
-      "learning_rate": 0.00011029249865083646,
-      "loss": 3.1631,
+      "grad_norm": 0.8703370094299316,
+      "learning_rate": 0.00011027954668105773,
+      "loss": 3.1652,
       "step": 75750
     },
     {
       "epoch": 8.172506738544474,
-      "grad_norm": 0.8351109623908997,
-      "learning_rate": 0.00010996869940636804,
-      "loss": 3.159,
+      "grad_norm": 0.8275737166404724,
+      "learning_rate": 0.0001099557474365893,
+      "loss": 3.1613,
       "step": 75800
     },
     {
       "epoch": 8.177897574123989,
-      "grad_norm": 0.8726935386657715,
-      "learning_rate": 0.00010964490016189962,
-      "loss": 3.1482,
+      "grad_norm": 0.8339884877204895,
+      "learning_rate": 0.00010963194819212088,
+      "loss": 3.1501,
       "step": 75850
     },
     {
       "epoch": 8.183288409703504,
-      "grad_norm": 0.8468180894851685,
-      "learning_rate": 0.00010932110091743117,
-      "loss": 3.1839,
+      "grad_norm": 0.8495912551879883,
+      "learning_rate": 0.00010930814894765243,
+      "loss": 3.1878,
       "step": 75900
     },
     {
       "epoch": 8.18867924528302,
-      "grad_norm": 0.8606348037719727,
-      "learning_rate": 0.00010899730167296275,
-      "loss": 3.1466,
+      "grad_norm": 0.8584531545639038,
+      "learning_rate": 0.00010898434970318401,
+      "loss": 3.1479,
       "step": 75950
     },
     {
       "epoch": 8.194070080862534,
-      "grad_norm": 0.9001939296722412,
-      "learning_rate": 0.00010867350242849432,
-      "loss": 3.1617,
+      "grad_norm": 0.8813877701759338,
+      "learning_rate": 0.00010866055045871559,
+      "loss": 3.1651,
       "step": 76000
     },
     {
       "epoch": 8.194070080862534,
-      "eval_accuracy": 0.3908328037410493,
-      "eval_loss": 3.3323988914489746,
-      "eval_runtime": 183.5683,
-      "eval_samples_per_second": 98.116,
-      "eval_steps_per_second": 6.134,
+      "eval_accuracy": 0.3907499015876469,
+      "eval_loss": 3.331759214401245,
+      "eval_runtime": 185.6787,
+      "eval_samples_per_second": 97.001,
+      "eval_steps_per_second": 6.064,
       "step": 76000
     },
     {
       "epoch": 8.199460916442048,
-      "grad_norm": 0.9116581082344055,
-      "learning_rate": 0.0001083497031840259,
-      "loss": 3.171,
+      "grad_norm": 0.8715637922286987,
+      "learning_rate": 0.00010833675121424717,
+      "loss": 3.1724,
       "step": 76050
     },
     {
       "epoch": 8.204851752021563,
-      "grad_norm": 0.8499873876571655,
-      "learning_rate": 0.00010802590393955745,
-      "loss": 3.1795,
+      "grad_norm": 0.8520619869232178,
+      "learning_rate": 0.00010801295196977872,
+      "loss": 3.1842,
       "step": 76100
     },
     {
       "epoch": 8.210242587601078,
-      "grad_norm": 0.8768589496612549,
-      "learning_rate": 0.00010770210469508903,
-      "loss": 3.1671,
+      "grad_norm": 0.873586893081665,
+      "learning_rate": 0.0001076891527253103,
+      "loss": 3.1691,
       "step": 76150
     },
     {
       "epoch": 8.215633423180593,
-      "grad_norm": 0.8752418756484985,
-      "learning_rate": 0.00010737830545062061,
-      "loss": 3.1726,
+      "grad_norm": 0.8279933929443359,
+      "learning_rate": 0.00010736535348084187,
+      "loss": 3.1746,
       "step": 76200
     },
     {
       "epoch": 8.221024258760108,
-      "grad_norm": 0.8474659323692322,
-      "learning_rate": 0.00010705450620615219,
-      "loss": 3.1591,
+      "grad_norm": 0.8118091225624084,
+      "learning_rate": 0.00010704155423637345,
+      "loss": 3.163,
       "step": 76250
     },
     {
       "epoch": 8.226415094339623,
-      "grad_norm": 0.8635431528091431,
-      "learning_rate": 0.00010673070696168375,
-      "loss": 3.1685,
+      "grad_norm": 0.8517959117889404,
+      "learning_rate": 0.000106717754991905,
+      "loss": 3.1711,
       "step": 76300
     },
     {
       "epoch": 8.231805929919137,
-      "grad_norm": 0.8334670662879944,
-      "learning_rate": 0.00010640690771721531,
-      "loss": 3.1738,
+      "grad_norm": 0.8346190452575684,
+      "learning_rate": 0.00010639395574743658,
+      "loss": 3.1773,
       "step": 76350
     },
     {
       "epoch": 8.237196765498652,
-      "grad_norm": 0.8614861369132996,
-      "learning_rate": 0.00010608310847274689,
-      "loss": 3.1823,
+      "grad_norm": 0.8859114050865173,
+      "learning_rate": 0.00010607015650296816,
+      "loss": 3.1837,
       "step": 76400
     },
     {
       "epoch": 8.242587601078167,
-      "grad_norm": 0.8414133191108704,
-      "learning_rate": 0.00010575930922827846,
-      "loss": 3.1497,
+      "grad_norm": 0.8647987246513367,
+      "learning_rate": 0.00010574635725849972,
+      "loss": 3.1541,
       "step": 76450
     },
     {
       "epoch": 8.247978436657682,
-      "grad_norm": 0.8237685561180115,
-      "learning_rate": 0.00010543550998381003,
-      "loss": 3.158,
+      "grad_norm": 0.8507528901100159,
+      "learning_rate": 0.00010542255801403128,
+      "loss": 3.16,
       "step": 76500
     },
     {
       "epoch": 8.253369272237197,
-      "grad_norm": 0.8597197532653809,
-      "learning_rate": 0.0001051117107393416,
-      "loss": 3.1511,
+      "grad_norm": 0.8596172332763672,
+      "learning_rate": 0.00010509875876956286,
+      "loss": 3.1544,
       "step": 76550
     },
     {
       "epoch": 8.25876010781671,
-      "grad_norm": 0.866402268409729,
-      "learning_rate": 0.00010478791149487316,
-      "loss": 3.1755,
+      "grad_norm": 0.8523345589637756,
+      "learning_rate": 0.00010477495952509443,
+      "loss": 3.1782,
       "step": 76600
     },
     {
       "epoch": 8.264150943396226,
-      "grad_norm": 0.8706876039505005,
-      "learning_rate": 0.00010446411225040474,
-      "loss": 3.1563,
+      "grad_norm": 0.8479462265968323,
+      "learning_rate": 0.000104451160280626,
+      "loss": 3.1591,
       "step": 76650
     },
     {
       "epoch": 8.269541778975741,
-      "grad_norm": 0.8851219415664673,
-      "learning_rate": 0.00010414031300593632,
-      "loss": 3.1876,
+      "grad_norm": 0.8411170244216919,
+      "learning_rate": 0.00010412736103615758,
+      "loss": 3.1911,
       "step": 76700
     },
     {
       "epoch": 8.274932614555256,
-      "grad_norm": 0.8189446330070496,
-      "learning_rate": 0.00010381651376146787,
-      "loss": 3.1711,
+      "grad_norm": 0.8051010966300964,
+      "learning_rate": 0.00010380356179168913,
+      "loss": 3.1744,
       "step": 76750
     },
     {
       "epoch": 8.280323450134771,
-      "grad_norm": 0.8582973480224609,
-      "learning_rate": 0.00010349271451699945,
-      "loss": 3.1521,
+      "grad_norm": 0.8394528031349182,
+      "learning_rate": 0.00010347976254722071,
+      "loss": 3.1579,
       "step": 76800
     },
     {
       "epoch": 8.285714285714286,
-      "grad_norm": 0.8564453721046448,
-      "learning_rate": 0.00010316891527253102,
-      "loss": 3.1961,
+      "grad_norm": 0.8649886250495911,
+      "learning_rate": 0.00010315596330275229,
+      "loss": 3.2007,
       "step": 76850
     },
     {
       "epoch": 8.2911051212938,
-      "grad_norm": 0.837863028049469,
-      "learning_rate": 0.0001028451160280626,
-      "loss": 3.1797,
+      "grad_norm": 0.8611036539077759,
+      "learning_rate": 0.00010283216405828387,
+      "loss": 3.1833,
       "step": 76900
     },
     {
       "epoch": 8.296495956873315,
-      "grad_norm": 0.8210705518722534,
-      "learning_rate": 0.00010252779276848353,
-      "loss": 3.1627,
+      "grad_norm": 0.8015654683113098,
+      "learning_rate": 0.00010250836481381542,
+      "loss": 3.1651,
       "step": 76950
     },
     {
       "epoch": 8.30188679245283,
-      "grad_norm": 0.8527716398239136,
-      "learning_rate": 0.0001022039935240151,
-      "loss": 3.1744,
+      "grad_norm": 0.8448264598846436,
+      "learning_rate": 0.000102184565569347,
+      "loss": 3.1772,
       "step": 77000
     },
     {
       "epoch": 8.30188679245283,
-      "eval_accuracy": 0.3909301567285311,
-      "eval_loss": 3.3277838230133057,
-      "eval_runtime": 183.3119,
-      "eval_samples_per_second": 98.253,
-      "eval_steps_per_second": 6.143,
+      "eval_accuracy": 0.39074827179432964,
+      "eval_loss": 3.3288376331329346,
+      "eval_runtime": 185.7821,
+      "eval_samples_per_second": 96.947,
+      "eval_steps_per_second": 6.061,
       "step": 77000
     },
     {
       "epoch": 8.307277628032345,
-      "grad_norm": 0.8103308081626892,
-      "learning_rate": 0.00010188019427954668,
-      "loss": 3.1687,
+      "grad_norm": 0.8100430369377136,
+      "learning_rate": 0.00010186076632487857,
+      "loss": 3.1737,
       "step": 77050
     },
     {
       "epoch": 8.31266846361186,
-      "grad_norm": 0.8414570093154907,
-      "learning_rate": 0.00010155639503507823,
-      "loss": 3.171,
+      "grad_norm": 0.8513318300247192,
+      "learning_rate": 0.00010153696708041014,
+      "loss": 3.1736,
       "step": 77100
     },
     {
       "epoch": 8.318059299191376,
-      "grad_norm": 0.8726758360862732,
-      "learning_rate": 0.00010123259579060981,
-      "loss": 3.1752,
+      "grad_norm": 0.8185202479362488,
+      "learning_rate": 0.0001012131678359417,
+      "loss": 3.1791,
       "step": 77150
     },
     {
       "epoch": 8.323450134770889,
-      "grad_norm": 0.8334712982177734,
-      "learning_rate": 0.00010090879654614139,
-      "loss": 3.1825,
+      "grad_norm": 0.813776969909668,
+      "learning_rate": 0.00010088936859147328,
+      "loss": 3.1857,
       "step": 77200
     },
     {
       "epoch": 8.328840970350404,
-      "grad_norm": 0.8159593939781189,
-      "learning_rate": 0.00010058499730167296,
-      "loss": 3.185,
+      "grad_norm": 0.8380706310272217,
+      "learning_rate": 0.00010056556934700485,
+      "loss": 3.1867,
       "step": 77250
     },
     {
       "epoch": 8.33423180592992,
-      "grad_norm": 0.8698613047599792,
-      "learning_rate": 0.00010026119805720452,
-      "loss": 3.1679,
+      "grad_norm": 0.8581069111824036,
+      "learning_rate": 0.00010024177010253642,
+      "loss": 3.1724,
       "step": 77300
     },
     {
       "epoch": 8.339622641509434,
-      "grad_norm": 0.854110598564148,
-      "learning_rate": 9.993739881273609e-05,
-      "loss": 3.1685,
+      "grad_norm": 0.8658299446105957,
+      "learning_rate": 9.9917970858068e-05,
+      "loss": 3.171,
       "step": 77350
     },
     {
       "epoch": 8.34501347708895,
-      "grad_norm": 0.8709824085235596,
-      "learning_rate": 9.961359956826767e-05,
-      "loss": 3.1878,
+      "grad_norm": 0.8810209035873413,
+      "learning_rate": 9.959417161359956e-05,
+      "loss": 3.1893,
       "step": 77400
     },
     {
       "epoch": 8.350404312668463,
-      "grad_norm": 0.8535655736923218,
-      "learning_rate": 9.928980032379923e-05,
-      "loss": 3.1643,
+      "grad_norm": 0.8300270438194275,
+      "learning_rate": 9.927037236913112e-05,
+      "loss": 3.1684,
       "step": 77450
     },
     {
       "epoch": 8.355795148247978,
-      "grad_norm": 0.9109821915626526,
-      "learning_rate": 9.896600107933081e-05,
-      "loss": 3.1706,
+      "grad_norm": 0.868481457233429,
+      "learning_rate": 9.89465731246627e-05,
+      "loss": 3.1741,
       "step": 77500
     },
     {
       "epoch": 8.361185983827493,
-      "grad_norm": 0.8124513030052185,
-      "learning_rate": 9.864220183486238e-05,
-      "loss": 3.1827,
+      "grad_norm": 0.8139988780021667,
+      "learning_rate": 9.862277388019428e-05,
+      "loss": 3.1854,
       "step": 77550
     },
     {
       "epoch": 8.366576819407008,
-      "grad_norm": 0.8645600080490112,
-      "learning_rate": 9.831840259039394e-05,
-      "loss": 3.1798,
+      "grad_norm": 0.8499587178230286,
+      "learning_rate": 9.829897463572583e-05,
+      "loss": 3.1841,
       "step": 77600
     },
     {
       "epoch": 8.371967654986523,
-      "grad_norm": 0.8976259231567383,
-      "learning_rate": 9.799460334592552e-05,
-      "loss": 3.1675,
+      "grad_norm": 0.8566383719444275,
+      "learning_rate": 9.797517539125741e-05,
+      "loss": 3.1716,
       "step": 77650
     },
     {
       "epoch": 8.377358490566039,
-      "grad_norm": 0.8319971561431885,
-      "learning_rate": 9.76708041014571e-05,
-      "loss": 3.1847,
+      "grad_norm": 0.8233956098556519,
+      "learning_rate": 9.765137614678899e-05,
+      "loss": 3.189,
       "step": 77700
     },
     {
       "epoch": 8.382749326145552,
-      "grad_norm": 0.8768755793571472,
-      "learning_rate": 9.734700485698865e-05,
-      "loss": 3.1646,
+      "grad_norm": 0.8583993315696716,
+      "learning_rate": 9.732757690232056e-05,
+      "loss": 3.1685,
       "step": 77750
     },
     {
       "epoch": 8.388140161725067,
-      "grad_norm": 0.8494071960449219,
-      "learning_rate": 9.702320561252022e-05,
-      "loss": 3.1844,
+      "grad_norm": 0.8790739178657532,
+      "learning_rate": 9.700377765785211e-05,
+      "loss": 3.189,
       "step": 77800
     },
     {
       "epoch": 8.393530997304582,
-      "grad_norm": 0.8520126938819885,
-      "learning_rate": 9.66994063680518e-05,
-      "loss": 3.173,
+      "grad_norm": 0.8590960502624512,
+      "learning_rate": 9.667997841338369e-05,
+      "loss": 3.175,
       "step": 77850
     },
     {
       "epoch": 8.398921832884097,
-      "grad_norm": 0.8483304381370544,
-      "learning_rate": 9.637560712358338e-05,
-      "loss": 3.176,
+      "grad_norm": 0.8205679655075073,
+      "learning_rate": 9.635617916891527e-05,
+      "loss": 3.178,
       "step": 77900
     },
     {
       "epoch": 8.404312668463612,
-      "grad_norm": 0.8193878531455994,
-      "learning_rate": 9.605180787911493e-05,
-      "loss": 3.1709,
+      "grad_norm": 0.8271093964576721,
+      "learning_rate": 9.603237992444683e-05,
+      "loss": 3.1735,
       "step": 77950
     },
     {
       "epoch": 8.409703504043126,
-      "grad_norm": 0.8840237259864807,
-      "learning_rate": 9.572800863464651e-05,
-      "loss": 3.1638,
+      "grad_norm": 0.8990002274513245,
+      "learning_rate": 9.57085806799784e-05,
+      "loss": 3.1663,
       "step": 78000
     },
     {
       "epoch": 8.409703504043126,
-      "eval_accuracy": 0.3914478877389671,
-      "eval_loss": 3.3253719806671143,
-      "eval_runtime": 183.4973,
-      "eval_samples_per_second": 98.154,
-      "eval_steps_per_second": 6.136,
+      "eval_accuracy": 0.39128012768018156,
+      "eval_loss": 3.3256309032440186,
+      "eval_runtime": 186.2016,
+      "eval_samples_per_second": 96.729,
+      "eval_steps_per_second": 6.047,
       "step": 78000
     },
     {
       "epoch": 8.415094339622641,
-      "grad_norm": 0.881489098072052,
-      "learning_rate": 9.540420939017809e-05,
-      "loss": 3.1724,
+      "grad_norm": 0.8596180081367493,
+      "learning_rate": 9.538478143550998e-05,
+      "loss": 3.1759,
       "step": 78050
     },
     {
       "epoch": 8.420485175202156,
-      "grad_norm": 0.86197429895401,
-      "learning_rate": 9.508041014570966e-05,
-      "loss": 3.1732,
+      "grad_norm": 0.872639000415802,
+      "learning_rate": 9.506098219104154e-05,
+      "loss": 3.1767,
       "step": 78100
     },
     {
       "epoch": 8.425876010781671,
-      "grad_norm": 0.9012764692306519,
-      "learning_rate": 9.475661090124121e-05,
-      "loss": 3.1727,
+      "grad_norm": 0.9674401879310608,
+      "learning_rate": 9.473718294657312e-05,
+      "loss": 3.1765,
       "step": 78150
     },
     {
       "epoch": 8.431266846361186,
-      "grad_norm": 0.8909586071968079,
-      "learning_rate": 9.443281165677279e-05,
-      "loss": 3.1593,
+      "grad_norm": 0.908412516117096,
+      "learning_rate": 9.44133837021047e-05,
+      "loss": 3.1642,
       "step": 78200
     },
     {
       "epoch": 8.436657681940702,
-      "grad_norm": 0.848753035068512,
-      "learning_rate": 9.410901241230437e-05,
-      "loss": 3.1753,
+      "grad_norm": 0.8564751744270325,
+      "learning_rate": 9.408958445763625e-05,
+      "loss": 3.1789,
       "step": 78250
     },
     {
       "epoch": 8.442048517520215,
-      "grad_norm": 0.9008297324180603,
-      "learning_rate": 9.378521316783593e-05,
-      "loss": 3.1694,
+      "grad_norm": 0.8696043491363525,
+      "learning_rate": 9.376578521316782e-05,
+      "loss": 3.1713,
       "step": 78300
     },
     {
       "epoch": 8.44743935309973,
-      "grad_norm": 0.8660398125648499,
-      "learning_rate": 9.346141392336751e-05,
-      "loss": 3.1938,
+      "grad_norm": 0.8673651814460754,
+      "learning_rate": 9.34419859686994e-05,
+      "loss": 3.196,
       "step": 78350
     },
     {
       "epoch": 8.452830188679245,
-      "grad_norm": 0.8556224703788757,
-      "learning_rate": 9.313761467889907e-05,
-      "loss": 3.1827,
+      "grad_norm": 0.8563075065612793,
+      "learning_rate": 9.311818672423098e-05,
+      "loss": 3.1858,
       "step": 78400
     },
     {
       "epoch": 8.45822102425876,
-      "grad_norm": 0.8545512557029724,
-      "learning_rate": 9.281381543443064e-05,
-      "loss": 3.1655,
+      "grad_norm": 0.8462834358215332,
+      "learning_rate": 9.279438747976253e-05,
+      "loss": 3.1672,
       "step": 78450
     },
     {
       "epoch": 8.463611859838275,
-      "grad_norm": 0.8092154264450073,
-      "learning_rate": 9.249001618996222e-05,
-      "loss": 3.1973,
+      "grad_norm": 0.8198238611221313,
+      "learning_rate": 9.247058823529411e-05,
+      "loss": 3.1991,
       "step": 78500
     },
     {
       "epoch": 8.46900269541779,
-      "grad_norm": 0.8078595399856567,
-      "learning_rate": 9.21662169454938e-05,
-      "loss": 3.1669,
+      "grad_norm": 0.8206710815429688,
+      "learning_rate": 9.214678899082568e-05,
+      "loss": 3.1697,
       "step": 78550
     },
     {
       "epoch": 8.474393530997304,
-      "grad_norm": 0.7993897199630737,
-      "learning_rate": 9.184241770102534e-05,
-      "loss": 3.1705,
+      "grad_norm": 0.7890952825546265,
+      "learning_rate": 9.182298974635726e-05,
+      "loss": 3.1729,
       "step": 78600
     },
     {
       "epoch": 8.479784366576819,
-      "grad_norm": 0.8611373901367188,
-      "learning_rate": 9.151861845655692e-05,
-      "loss": 3.1709,
+      "grad_norm": 0.8805410265922546,
+      "learning_rate": 9.149919050188881e-05,
+      "loss": 3.1739,
       "step": 78650
     },
     {
       "epoch": 8.485175202156334,
-      "grad_norm": 0.8634024262428284,
-      "learning_rate": 9.11948192120885e-05,
-      "loss": 3.1975,
+      "grad_norm": 0.8415430188179016,
+      "learning_rate": 9.117539125742039e-05,
+      "loss": 3.2004,
       "step": 78700
     },
     {
       "epoch": 8.49056603773585,
-      "grad_norm": 0.8928923010826111,
-      "learning_rate": 9.087101996762008e-05,
-      "loss": 3.1808,
+      "grad_norm": 0.844524085521698,
+      "learning_rate": 9.085159201295197e-05,
+      "loss": 3.1834,
       "step": 78750
     },
     {
       "epoch": 8.495956873315365,
-      "grad_norm": 0.8743278384208679,
-      "learning_rate": 9.054722072315163e-05,
-      "loss": 3.173,
+      "grad_norm": 0.8768396377563477,
+      "learning_rate": 9.052779276848353e-05,
+      "loss": 3.177,
       "step": 78800
     },
     {
       "epoch": 8.501347708894878,
-      "grad_norm": 0.8760882019996643,
-      "learning_rate": 9.02234214786832e-05,
-      "loss": 3.1766,
+      "grad_norm": 0.8678855299949646,
+      "learning_rate": 9.02039935240151e-05,
+      "loss": 3.1768,
       "step": 78850
     },
     {
       "epoch": 8.506738544474393,
-      "grad_norm": 0.8380445241928101,
-      "learning_rate": 8.989962223421478e-05,
-      "loss": 3.1711,
+      "grad_norm": 0.8700560331344604,
+      "learning_rate": 8.988019427954667e-05,
+      "loss": 3.1752,
       "step": 78900
     },
     {
       "epoch": 8.512129380053908,
-      "grad_norm": 0.8691101670265198,
-      "learning_rate": 8.958229897463571e-05,
-      "loss": 3.1905,
+      "grad_norm": 0.9049444198608398,
+      "learning_rate": 8.955639503507824e-05,
+      "loss": 3.1943,
       "step": 78950
     },
     {
       "epoch": 8.517520215633423,
-      "grad_norm": 0.8391762971878052,
-      "learning_rate": 8.925849973016729e-05,
-      "loss": 3.1771,
+      "grad_norm": 0.8347573280334473,
+      "learning_rate": 8.923259579060982e-05,
+      "loss": 3.1808,
       "step": 79000
     },
     {
       "epoch": 8.517520215633423,
-      "eval_accuracy": 0.39183414875514755,
-      "eval_loss": 3.3210854530334473,
-      "eval_runtime": 183.4514,
-      "eval_samples_per_second": 98.179,
-      "eval_steps_per_second": 6.138,
+      "eval_accuracy": 0.39176059075009717,
+      "eval_loss": 3.3227598667144775,
+      "eval_runtime": 185.7539,
+      "eval_samples_per_second": 96.962,
+      "eval_steps_per_second": 6.062,
       "step": 79000
     },
     {
       "epoch": 8.522911051212938,
-      "grad_norm": 0.8594598174095154,
-      "learning_rate": 8.893470048569886e-05,
-      "loss": 3.1846,
+      "grad_norm": 0.8408942222595215,
+      "learning_rate": 8.89087965461414e-05,
+      "loss": 3.1877,
       "step": 79050
     },
     {
       "epoch": 8.528301886792454,
-      "grad_norm": 0.8403089642524719,
-      "learning_rate": 8.861090124123044e-05,
-      "loss": 3.1923,
+      "grad_norm": 0.8584231734275818,
+      "learning_rate": 8.859147328656233e-05,
+      "loss": 3.1959,
       "step": 79100
     },
     {
       "epoch": 8.533692722371967,
-      "grad_norm": 0.815895676612854,
-      "learning_rate": 8.828710199676199e-05,
-      "loss": 3.1661,
+      "grad_norm": 0.8307530879974365,
+      "learning_rate": 8.82676740420939e-05,
+      "loss": 3.1698,
       "step": 79150
     },
     {
       "epoch": 8.539083557951482,
-      "grad_norm": 0.8398244380950928,
-      "learning_rate": 8.796330275229357e-05,
-      "loss": 3.1722,
+      "grad_norm": 0.8899841904640198,
+      "learning_rate": 8.794387479762546e-05,
+      "loss": 3.1746,
       "step": 79200
     },
     {
       "epoch": 8.544474393530997,
-      "grad_norm": 0.8309531807899475,
-      "learning_rate": 8.763950350782515e-05,
-      "loss": 3.163,
+      "grad_norm": 0.8379618525505066,
+      "learning_rate": 8.762007555315704e-05,
+      "loss": 3.165,
       "step": 79250
     },
     {
       "epoch": 8.549865229110512,
-      "grad_norm": 0.8906266093254089,
-      "learning_rate": 8.731570426335671e-05,
-      "loss": 3.1818,
+      "grad_norm": 0.8717741966247559,
+      "learning_rate": 8.72962763086886e-05,
+      "loss": 3.1838,
       "step": 79300
     },
     {
       "epoch": 8.555256064690028,
-      "grad_norm": 0.8349162340164185,
-      "learning_rate": 8.699190501888828e-05,
-      "loss": 3.1804,
+      "grad_norm": 0.8427934050559998,
+      "learning_rate": 8.697247706422018e-05,
+      "loss": 3.1847,
       "step": 79350
     },
     {
       "epoch": 8.560646900269543,
-      "grad_norm": 0.8070085048675537,
-      "learning_rate": 8.666810577441985e-05,
-      "loss": 3.1515,
+      "grad_norm": 0.8378936648368835,
+      "learning_rate": 8.664867781975174e-05,
+      "loss": 3.1555,
       "step": 79400
     },
     {
       "epoch": 8.566037735849056,
-      "grad_norm": 0.8301887512207031,
-      "learning_rate": 8.634430652995142e-05,
-      "loss": 3.1848,
+      "grad_norm": 0.9016141295433044,
+      "learning_rate": 8.632487857528331e-05,
+      "loss": 3.1902,
       "step": 79450
     },
     {
       "epoch": 8.571428571428571,
-      "grad_norm": 0.8491299748420715,
-      "learning_rate": 8.6020507285483e-05,
-      "loss": 3.1771,
+      "grad_norm": 0.8445428609848022,
+      "learning_rate": 8.600107933081489e-05,
+      "loss": 3.179,
       "step": 79500
     },
     {
       "epoch": 8.576819407008086,
-      "grad_norm": 0.8146814703941345,
-      "learning_rate": 8.569670804101456e-05,
-      "loss": 3.1755,
+      "grad_norm": 0.8253672122955322,
+      "learning_rate": 8.567728008634646e-05,
+      "loss": 3.1776,
       "step": 79550
     },
     {
       "epoch": 8.582210242587601,
-      "grad_norm": 0.8444635272026062,
-      "learning_rate": 8.537290879654612e-05,
-      "loss": 3.1763,
+      "grad_norm": 0.8554487824440002,
+      "learning_rate": 8.535348084187804e-05,
+      "loss": 3.1801,
       "step": 79600
     },
     {
       "epoch": 8.587601078167117,
-      "grad_norm": 0.8767122030258179,
-      "learning_rate": 8.50491095520777e-05,
-      "loss": 3.1571,
+      "grad_norm": 0.9377881288528442,
+      "learning_rate": 8.502968159740959e-05,
+      "loss": 3.1602,
       "step": 79650
     },
     {
       "epoch": 8.59299191374663,
-      "grad_norm": 0.8696253895759583,
-      "learning_rate": 8.472531030760928e-05,
-      "loss": 3.1813,
+      "grad_norm": 0.8728728890419006,
+      "learning_rate": 8.470588235294117e-05,
+      "loss": 3.1841,
       "step": 79700
     },
     {
       "epoch": 8.598382749326145,
-      "grad_norm": 0.8822606205940247,
-      "learning_rate": 8.440151106314086e-05,
-      "loss": 3.189,
+      "grad_norm": 0.9526692628860474,
+      "learning_rate": 8.438208310847275e-05,
+      "loss": 3.1915,
       "step": 79750
     },
     {
       "epoch": 8.60377358490566,
-      "grad_norm": 0.8302372694015503,
-      "learning_rate": 8.407771181867241e-05,
-      "loss": 3.1693,
+      "grad_norm": 0.8466939926147461,
+      "learning_rate": 8.405828386400431e-05,
+      "loss": 3.1745,
       "step": 79800
     },
     {
       "epoch": 8.609164420485175,
-      "grad_norm": 0.8339834213256836,
-      "learning_rate": 8.375391257420398e-05,
-      "loss": 3.1542,
+      "grad_norm": 0.8755860328674316,
+      "learning_rate": 8.373448461953587e-05,
+      "loss": 3.158,
       "step": 79850
     },
     {
       "epoch": 8.61455525606469,
-      "grad_norm": 0.862395703792572,
-      "learning_rate": 8.343011332973556e-05,
-      "loss": 3.1626,
+      "grad_norm": 0.8443488478660583,
+      "learning_rate": 8.341068537506745e-05,
+      "loss": 3.1656,
       "step": 79900
     },
     {
       "epoch": 8.619946091644206,
-      "grad_norm": 0.846082329750061,
-      "learning_rate": 8.310631408526714e-05,
-      "loss": 3.1861,
+      "grad_norm": 0.8386605978012085,
+      "learning_rate": 8.308688613059902e-05,
+      "loss": 3.1887,
       "step": 79950
     },
     {
       "epoch": 8.625336927223719,
-      "grad_norm": 0.8844338059425354,
-      "learning_rate": 8.278251484079869e-05,
-      "loss": 3.1753,
+      "grad_norm": 0.8474873900413513,
+      "learning_rate": 8.27630868861306e-05,
+      "loss": 3.1781,
       "step": 80000
     },
     {
       "epoch": 8.625336927223719,
-      "eval_accuracy": 0.3923148291308388,
-      "eval_loss": 3.3195295333862305,
-      "eval_runtime": 183.5025,
-      "eval_samples_per_second": 98.151,
-      "eval_steps_per_second": 6.136,
+      "eval_accuracy": 0.39203722100247285,
+      "eval_loss": 3.3204987049102783,
+      "eval_runtime": 184.7605,
+      "eval_samples_per_second": 97.483,
+      "eval_steps_per_second": 6.094,
       "step": 80000
     },
     {
       "epoch": 8.630727762803234,
-      "grad_norm": 0.8043167591094971,
-      "learning_rate": 8.245871559633027e-05,
-      "loss": 3.1842,
+      "grad_norm": 0.8463869094848633,
+      "learning_rate": 8.243928764166216e-05,
+      "loss": 3.1866,
       "step": 80050
     },
     {
       "epoch": 8.63611859838275,
-      "grad_norm": 0.8519346714019775,
-      "learning_rate": 8.213491635186185e-05,
-      "loss": 3.1895,
+      "grad_norm": 0.8860599398612976,
+      "learning_rate": 8.211548839719374e-05,
+      "loss": 3.1933,
       "step": 80100
     },
     {
       "epoch": 8.641509433962264,
-      "grad_norm": 0.8410979509353638,
-      "learning_rate": 8.181111710739341e-05,
-      "loss": 3.1753,
+      "grad_norm": 0.8397736549377441,
+      "learning_rate": 8.17916891527253e-05,
+      "loss": 3.1792,
       "step": 80150
     },
     {
       "epoch": 8.64690026954178,
-      "grad_norm": 0.859257161617279,
-      "learning_rate": 8.148731786292497e-05,
-      "loss": 3.1692,
+      "grad_norm": 0.8628084063529968,
+      "learning_rate": 8.146788990825688e-05,
+      "loss": 3.1729,
       "step": 80200
     },
     {
       "epoch": 8.652291105121293,
-      "grad_norm": 0.8353541493415833,
-      "learning_rate": 8.116351861845655e-05,
-      "loss": 3.179,
+      "grad_norm": 0.8329259753227234,
+      "learning_rate": 8.114409066378844e-05,
+      "loss": 3.1815,
       "step": 80250
     },
     {
       "epoch": 8.657681940700808,
-      "grad_norm": 1.4765746593475342,
-      "learning_rate": 8.083971937398812e-05,
-      "loss": 3.1822,
+      "grad_norm": 0.8436859846115112,
+      "learning_rate": 8.082029141932e-05,
+      "loss": 3.1861,
       "step": 80300
     },
     {
       "epoch": 8.663072776280323,
-      "grad_norm": 0.8867174386978149,
-      "learning_rate": 8.05159201295197e-05,
-      "loss": 3.1748,
+      "grad_norm": 0.888627827167511,
+      "learning_rate": 8.049649217485158e-05,
+      "loss": 3.1779,
       "step": 80350
     },
     {
       "epoch": 8.668463611859838,
-      "grad_norm": 0.8695386052131653,
-      "learning_rate": 8.019212088505126e-05,
-      "loss": 3.1764,
+      "grad_norm": 0.8507674932479858,
+      "learning_rate": 8.017916891527252e-05,
+      "loss": 3.1788,
       "step": 80400
     },
     {
       "epoch": 8.673854447439354,
-      "grad_norm": 0.8609994649887085,
-      "learning_rate": 7.986832164058282e-05,
-      "loss": 3.1638,
+      "grad_norm": 0.8327531814575195,
+      "learning_rate": 7.985536967080409e-05,
+      "loss": 3.1664,
       "step": 80450
     },
     {
       "epoch": 8.679245283018869,
-      "grad_norm": 0.8649408221244812,
-      "learning_rate": 7.95445223961144e-05,
-      "loss": 3.1631,
+      "grad_norm": 0.8763095140457153,
+      "learning_rate": 7.953157042633566e-05,
+      "loss": 3.165,
       "step": 80500
     },
     {
       "epoch": 8.684636118598382,
-      "grad_norm": 0.8425276279449463,
-      "learning_rate": 7.922072315164598e-05,
-      "loss": 3.1902,
+      "grad_norm": 0.8629357218742371,
+      "learning_rate": 7.920777118186724e-05,
+      "loss": 3.1933,
       "step": 80550
     },
     {
       "epoch": 8.690026954177897,
-      "grad_norm": 0.8469082117080688,
-      "learning_rate": 7.889692390717755e-05,
-      "loss": 3.1604,
+      "grad_norm": 0.8500385880470276,
+      "learning_rate": 7.888397193739879e-05,
+      "loss": 3.1638,
       "step": 80600
     },
     {
       "epoch": 8.695417789757412,
-      "grad_norm": 0.9055061936378479,
-      "learning_rate": 7.85731246627091e-05,
-      "loss": 3.1877,
+      "grad_norm": 0.8588795065879822,
+      "learning_rate": 7.856017269293037e-05,
+      "loss": 3.1913,
       "step": 80650
     },
     {
       "epoch": 8.700808625336927,
-      "grad_norm": 0.8626482486724854,
-      "learning_rate": 7.824932541824068e-05,
-      "loss": 3.1765,
+      "grad_norm": 0.8750964999198914,
+      "learning_rate": 7.823637344846195e-05,
+      "loss": 3.1817,
       "step": 80700
     },
     {
       "epoch": 8.706199460916443,
-      "grad_norm": 0.913098931312561,
-      "learning_rate": 7.792552617377226e-05,
-      "loss": 3.1799,
+      "grad_norm": 0.8942277431488037,
+      "learning_rate": 7.791257420399353e-05,
+      "loss": 3.1816,
       "step": 80750
     },
     {
       "epoch": 8.711590296495956,
-      "grad_norm": 0.8522776961326599,
-      "learning_rate": 7.760172692930382e-05,
-      "loss": 3.1608,
+      "grad_norm": 0.851742684841156,
+      "learning_rate": 7.758877495952508e-05,
+      "loss": 3.165,
       "step": 80800
     },
     {
       "epoch": 8.716981132075471,
-      "grad_norm": 0.8847063183784485,
-      "learning_rate": 7.727792768483539e-05,
-      "loss": 3.1656,
+      "grad_norm": 0.8816221356391907,
+      "learning_rate": 7.726497571505665e-05,
+      "loss": 3.1686,
       "step": 80850
     },
     {
       "epoch": 8.722371967654986,
-      "grad_norm": 0.9037518501281738,
-      "learning_rate": 7.695412844036697e-05,
-      "loss": 3.1761,
+      "grad_norm": 0.8947892785072327,
+      "learning_rate": 7.694117647058823e-05,
+      "loss": 3.1791,
       "step": 80900
     },
     {
       "epoch": 8.727762803234501,
-      "grad_norm": 0.8497751951217651,
-      "learning_rate": 7.663032919589854e-05,
-      "loss": 3.1929,
+      "grad_norm": 0.837598979473114,
+      "learning_rate": 7.661737722611981e-05,
+      "loss": 3.1961,
       "step": 80950
     },
     {
       "epoch": 8.733153638814017,
-      "grad_norm": 0.8641258478164673,
-      "learning_rate": 7.630652995143011e-05,
-      "loss": 3.1883,
+      "grad_norm": 0.866131603717804,
+      "learning_rate": 7.629357798165137e-05,
+      "loss": 3.1919,
       "step": 81000
     },
     {
       "epoch": 8.733153638814017,
-      "eval_accuracy": 0.39242739352261463,
-      "eval_loss": 3.3154139518737793,
-      "eval_runtime": 183.473,
-      "eval_samples_per_second": 98.167,
-      "eval_steps_per_second": 6.137,
+      "eval_accuracy": 0.39230363788339395,
+      "eval_loss": 3.315007448196411,
+      "eval_runtime": 181.4331,
+      "eval_samples_per_second": 99.271,
+      "eval_steps_per_second": 6.206,
       "step": 81000
     },
     {
       "epoch": 8.738544474393532,
-      "grad_norm": 0.8507005572319031,
-      "learning_rate": 7.598273070696167e-05,
-      "loss": 3.1759,
+      "grad_norm": 0.853217363357544,
+      "learning_rate": 7.596977873718294e-05,
+      "loss": 3.1784,
       "step": 81050
     },
     {
       "epoch": 8.743935309973045,
-      "grad_norm": 0.8656426072120667,
-      "learning_rate": 7.565893146249325e-05,
-      "loss": 3.1855,
+      "grad_norm": 0.8749951124191284,
+      "learning_rate": 7.564597949271451e-05,
+      "loss": 3.1906,
       "step": 81100
     },
     {
       "epoch": 8.74932614555256,
-      "grad_norm": 0.8609006404876709,
-      "learning_rate": 7.533513221802481e-05,
-      "loss": 3.1698,
+      "grad_norm": 0.7965062856674194,
+      "learning_rate": 7.532218024824608e-05,
+      "loss": 3.1732,
       "step": 81150
     },
     {
       "epoch": 8.754716981132075,
-      "grad_norm": 0.973294198513031,
-      "learning_rate": 7.501133297355639e-05,
-      "loss": 3.1679,
+      "grad_norm": 0.9589442014694214,
+      "learning_rate": 7.499838100377764e-05,
+      "loss": 3.1704,
       "step": 81200
     },
     {
       "epoch": 8.76010781671159,
-      "grad_norm": 0.8869118094444275,
-      "learning_rate": 7.468753372908796e-05,
-      "loss": 3.1771,
+      "grad_norm": 0.8942793011665344,
+      "learning_rate": 7.467458175930922e-05,
+      "loss": 3.1799,
       "step": 81250
     },
     {
       "epoch": 8.765498652291106,
-      "grad_norm": 0.8876540064811707,
-      "learning_rate": 7.436373448461953e-05,
-      "loss": 3.1754,
+      "grad_norm": 0.9064013957977295,
+      "learning_rate": 7.435078251484078e-05,
+      "loss": 3.1788,
       "step": 81300
     },
     {
       "epoch": 8.77088948787062,
-      "grad_norm": 0.8496720790863037,
-      "learning_rate": 7.40399352401511e-05,
-      "loss": 3.1767,
+      "grad_norm": 0.8686546683311462,
+      "learning_rate": 7.402698327037236e-05,
+      "loss": 3.1807,
       "step": 81350
     },
     {
       "epoch": 8.776280323450134,
-      "grad_norm": 0.8450832366943359,
-      "learning_rate": 7.371613599568268e-05,
-      "loss": 3.1846,
+      "grad_norm": 0.8572303056716919,
+      "learning_rate": 7.370318402590393e-05,
+      "loss": 3.1875,
       "step": 81400
     },
     {
       "epoch": 8.78167115902965,
-      "grad_norm": 0.8679110407829285,
-      "learning_rate": 7.339233675121424e-05,
-      "loss": 3.1839,
+      "grad_norm": 0.8310827016830444,
+      "learning_rate": 7.33793847814355e-05,
+      "loss": 3.1878,
       "step": 81450
     },
     {
       "epoch": 8.787061994609164,
-      "grad_norm": 0.9450589418411255,
-      "learning_rate": 7.306853750674582e-05,
-      "loss": 3.2063,
+      "grad_norm": 0.8690220713615417,
+      "learning_rate": 7.305558553696707e-05,
+      "loss": 3.2094,
       "step": 81500
     },
     {
       "epoch": 8.79245283018868,
-      "grad_norm": 0.8842014670372009,
-      "learning_rate": 7.274473826227738e-05,
-      "loss": 3.1691,
+      "grad_norm": 0.9176035523414612,
+      "learning_rate": 7.273178629249865e-05,
+      "loss": 3.1711,
       "step": 81550
     },
     {
       "epoch": 8.797843665768195,
-      "grad_norm": 0.8416998386383057,
-      "learning_rate": 7.242093901780896e-05,
-      "loss": 3.1774,
+      "grad_norm": 0.8315758109092712,
+      "learning_rate": 7.240798704803021e-05,
+      "loss": 3.1799,
       "step": 81600
     },
     {
       "epoch": 8.80323450134771,
-      "grad_norm": 0.9002299904823303,
-      "learning_rate": 7.209713977334052e-05,
-      "loss": 3.1794,
+      "grad_norm": 0.9258739948272705,
+      "learning_rate": 7.208418780356179e-05,
+      "loss": 3.1832,
       "step": 81650
     },
     {
       "epoch": 8.808625336927223,
-      "grad_norm": 0.8381384015083313,
-      "learning_rate": 7.17733405288721e-05,
-      "loss": 3.1905,
+      "grad_norm": 0.798367440700531,
+      "learning_rate": 7.176038855909337e-05,
+      "loss": 3.1916,
       "step": 81700
     },
     {
       "epoch": 8.814016172506738,
-      "grad_norm": 0.8705641031265259,
-      "learning_rate": 7.144954128440366e-05,
-      "loss": 3.1742,
+      "grad_norm": 0.8513667583465576,
+      "learning_rate": 7.143658931462493e-05,
+      "loss": 3.1784,
       "step": 81750
     },
     {
       "epoch": 8.819407008086253,
-      "grad_norm": 0.8659651279449463,
-      "learning_rate": 7.112574203993523e-05,
-      "loss": 3.1648,
+      "grad_norm": 0.8541272878646851,
+      "learning_rate": 7.11127900701565e-05,
+      "loss": 3.1694,
       "step": 81800
     },
     {
       "epoch": 8.824797843665769,
-      "grad_norm": 0.8624113202095032,
-      "learning_rate": 7.08019427954668e-05,
-      "loss": 3.181,
+      "grad_norm": 0.8517417907714844,
+      "learning_rate": 7.078899082568807e-05,
+      "loss": 3.1829,
       "step": 81850
     },
     {
       "epoch": 8.830188679245284,
-      "grad_norm": 0.8544336557388306,
-      "learning_rate": 7.047814355099837e-05,
-      "loss": 3.2008,
+      "grad_norm": 0.9081867933273315,
+      "learning_rate": 7.046519158121964e-05,
+      "loss": 3.2028,
       "step": 81900
     },
     {
       "epoch": 8.835579514824797,
-      "grad_norm": 0.8490225076675415,
-      "learning_rate": 7.015434430652993e-05,
-      "loss": 3.1846,
+      "grad_norm": 0.8268014192581177,
+      "learning_rate": 7.014139233675121e-05,
+      "loss": 3.1877,
       "step": 81950
     },
     {
       "epoch": 8.840970350404312,
-      "grad_norm": 0.8430733680725098,
-      "learning_rate": 6.983054506206151e-05,
-      "loss": 3.173,
+      "grad_norm": 0.8537060618400574,
+      "learning_rate": 6.981759309228278e-05,
+      "loss": 3.1769,
       "step": 82000
     },
     {
       "epoch": 8.840970350404312,
-      "eval_accuracy": 0.3929860866717567,
-      "eval_loss": 3.311267137527466,
-      "eval_runtime": 183.6755,
-      "eval_samples_per_second": 98.059,
-      "eval_steps_per_second": 6.13,
+      "eval_accuracy": 0.3927071747087369,
+      "eval_loss": 3.3120408058166504,
+      "eval_runtime": 181.6575,
+      "eval_samples_per_second": 99.148,
+      "eval_steps_per_second": 6.198,
       "step": 82000
     },
     {
       "epoch": 8.846361185983827,
-      "grad_norm": 0.8979526162147522,
-      "learning_rate": 6.950674581759309e-05,
-      "loss": 3.1733,
+      "grad_norm": 0.868050217628479,
+      "learning_rate": 6.950026983270372e-05,
+      "loss": 3.1774,
       "step": 82050
     },
     {
       "epoch": 8.851752021563343,
-      "grad_norm": 0.8584031462669373,
-      "learning_rate": 6.918294657312465e-05,
-      "loss": 3.1726,
+      "grad_norm": 0.8678497076034546,
+      "learning_rate": 6.91764705882353e-05,
+      "loss": 3.1762,
       "step": 82100
     },
     {
       "epoch": 8.857142857142858,
-      "grad_norm": 0.8862038850784302,
-      "learning_rate": 6.885914732865623e-05,
-      "loss": 3.169,
+      "grad_norm": 0.8837611675262451,
+      "learning_rate": 6.885267134376686e-05,
+      "loss": 3.1722,
       "step": 82150
     },
     {
       "epoch": 8.862533692722373,
-      "grad_norm": 0.8636807799339294,
-      "learning_rate": 6.85353480841878e-05,
-      "loss": 3.1808,
+      "grad_norm": 0.8401360511779785,
+      "learning_rate": 6.852887209929844e-05,
+      "loss": 3.1831,
       "step": 82200
     },
     {
       "epoch": 8.867924528301886,
-      "grad_norm": 0.8127702474594116,
-      "learning_rate": 6.821154883971937e-05,
-      "loss": 3.182,
+      "grad_norm": 0.8119093179702759,
+      "learning_rate": 6.820507285483e-05,
+      "loss": 3.1866,
       "step": 82250
     },
     {
       "epoch": 8.873315363881401,
-      "grad_norm": 0.8639582991600037,
-      "learning_rate": 6.788774959525094e-05,
-      "loss": 3.1617,
+      "grad_norm": 0.8857676982879639,
+      "learning_rate": 6.788127361036156e-05,
+      "loss": 3.1632,
       "step": 82300
     },
     {
       "epoch": 8.878706199460916,
-      "grad_norm": 0.8684182167053223,
-      "learning_rate": 6.756395035078252e-05,
-      "loss": 3.1501,
+      "grad_norm": 0.8609165549278259,
+      "learning_rate": 6.755747436589314e-05,
+      "loss": 3.1546,
       "step": 82350
     },
     {
       "epoch": 8.884097035040432,
-      "grad_norm": 0.8748733401298523,
-      "learning_rate": 6.724015110631408e-05,
-      "loss": 3.1779,
+      "grad_norm": 0.8864471912384033,
+      "learning_rate": 6.72336751214247e-05,
+      "loss": 3.1831,
       "step": 82400
     },
     {
       "epoch": 8.889487870619947,
-      "grad_norm": 0.8723418116569519,
-      "learning_rate": 6.691635186184566e-05,
-      "loss": 3.1651,
+      "grad_norm": 0.8692185878753662,
+      "learning_rate": 6.690987587695628e-05,
+      "loss": 3.1663,
       "step": 82450
     },
     {
       "epoch": 8.89487870619946,
-      "grad_norm": 0.8358088135719299,
-      "learning_rate": 6.659255261737722e-05,
-      "loss": 3.1762,
+      "grad_norm": 0.8401042222976685,
+      "learning_rate": 6.658607663248785e-05,
+      "loss": 3.1795,
       "step": 82500
     },
     {
       "epoch": 8.900269541778975,
-      "grad_norm": 0.8295577168464661,
-      "learning_rate": 6.626875337290879e-05,
-      "loss": 3.1649,
+      "grad_norm": 0.8570231199264526,
+      "learning_rate": 6.626227738801942e-05,
+      "loss": 3.1683,
       "step": 82550
     },
     {
       "epoch": 8.90566037735849,
-      "grad_norm": 0.8575540781021118,
-      "learning_rate": 6.594495412844036e-05,
-      "loss": 3.1891,
+      "grad_norm": 0.8442992568016052,
+      "learning_rate": 6.593847814355099e-05,
+      "loss": 3.192,
       "step": 82600
     },
     {
       "epoch": 8.911051212938006,
-      "grad_norm": 0.8286822438240051,
-      "learning_rate": 6.562115488397193e-05,
-      "loss": 3.1915,
+      "grad_norm": 0.8482553958892822,
+      "learning_rate": 6.561467889908257e-05,
+      "loss": 3.1944,
       "step": 82650
     },
     {
       "epoch": 8.91644204851752,
-      "grad_norm": 0.8888596296310425,
-      "learning_rate": 6.52973556395035e-05,
-      "loss": 3.1872,
+      "grad_norm": 0.8677952885627747,
+      "learning_rate": 6.529087965461413e-05,
+      "loss": 3.191,
       "step": 82700
     },
     {
       "epoch": 8.921832884097036,
-      "grad_norm": 0.863433301448822,
-      "learning_rate": 6.497355639503507e-05,
-      "loss": 3.1827,
+      "grad_norm": 0.8685084581375122,
+      "learning_rate": 6.496708041014571e-05,
+      "loss": 3.1852,
       "step": 82750
     },
     {
       "epoch": 8.92722371967655,
-      "grad_norm": 0.8844254612922668,
-      "learning_rate": 6.464975715056663e-05,
-      "loss": 3.181,
+      "grad_norm": 0.8863459229469299,
+      "learning_rate": 6.464328116567727e-05,
+      "loss": 3.1846,
       "step": 82800
     },
     {
       "epoch": 8.932614555256064,
-      "grad_norm": 0.8509985208511353,
-      "learning_rate": 6.432595790609821e-05,
-      "loss": 3.1813,
+      "grad_norm": 0.8586819171905518,
+      "learning_rate": 6.431948192120885e-05,
+      "loss": 3.1853,
       "step": 82850
     },
     {
       "epoch": 8.93800539083558,
-      "grad_norm": 0.896156907081604,
-      "learning_rate": 6.400215866162979e-05,
-      "loss": 3.177,
+      "grad_norm": 0.8691402077674866,
+      "learning_rate": 6.399568267674041e-05,
+      "loss": 3.1809,
       "step": 82900
     },
     {
       "epoch": 8.943396226415095,
-      "grad_norm": 0.8738552927970886,
-      "learning_rate": 6.367835941716135e-05,
-      "loss": 3.1741,
+      "grad_norm": 0.8704937696456909,
+      "learning_rate": 6.367188343227199e-05,
+      "loss": 3.1783,
       "step": 82950
     },
     {
       "epoch": 8.94878706199461,
-      "grad_norm": 0.8401568531990051,
-      "learning_rate": 6.335456017269293e-05,
-      "loss": 3.1879,
+      "grad_norm": 0.8600329756736755,
+      "learning_rate": 6.334808418780356e-05,
+      "loss": 3.192,
       "step": 83000
     },
     {
       "epoch": 8.94878706199461,
-      "eval_accuracy": 0.3931521082843372,
-      "eval_loss": 3.3100833892822266,
-      "eval_runtime": 183.3659,
-      "eval_samples_per_second": 98.224,
-      "eval_steps_per_second": 6.141,
+      "eval_accuracy": 0.39299238853924995,
+      "eval_loss": 3.309659719467163,
+      "eval_runtime": 181.3969,
+      "eval_samples_per_second": 99.291,
+      "eval_steps_per_second": 6.207,
       "step": 83000
     },
     {
       "epoch": 8.954177897574123,
-      "grad_norm": 0.8782355785369873,
-      "learning_rate": 6.30307609282245e-05,
-      "loss": 3.1927,
+      "grad_norm": 0.8590782880783081,
+      "learning_rate": 6.302428494333512e-05,
+      "loss": 3.1965,
       "step": 83050
     },
     {
       "epoch": 8.959568733153638,
-      "grad_norm": 0.8759222030639648,
-      "learning_rate": 6.270696168375607e-05,
-      "loss": 3.1784,
+      "grad_norm": 0.9092225432395935,
+      "learning_rate": 6.27004856988667e-05,
+      "loss": 3.1809,
       "step": 83100
     },
     {
       "epoch": 8.964959568733153,
-      "grad_norm": 0.8785428404808044,
-      "learning_rate": 6.238316243928764e-05,
-      "loss": 3.1803,
+      "grad_norm": 0.879080593585968,
+      "learning_rate": 6.237668645439826e-05,
+      "loss": 3.1819,
       "step": 83150
     },
     {
       "epoch": 8.970350404312669,
-      "grad_norm": 0.8433955311775208,
-      "learning_rate": 6.205936319481921e-05,
-      "loss": 3.188,
+      "grad_norm": 0.8661657571792603,
+      "learning_rate": 6.205288720992984e-05,
+      "loss": 3.1894,
       "step": 83200
     },
     {
       "epoch": 8.975741239892184,
-      "grad_norm": 0.877407968044281,
-      "learning_rate": 6.173556395035078e-05,
-      "loss": 3.1738,
+      "grad_norm": 0.8800835013389587,
+      "learning_rate": 6.17290879654614e-05,
+      "loss": 3.1743,
       "step": 83250
     },
     {
       "epoch": 8.981132075471699,
-      "grad_norm": 0.8445099592208862,
-      "learning_rate": 6.141176470588236e-05,
-      "loss": 3.1865,
+      "grad_norm": 0.8630102276802063,
+      "learning_rate": 6.140528872099298e-05,
+      "loss": 3.1885,
       "step": 83300
     },
     {
       "epoch": 8.986522911051212,
-      "grad_norm": 0.8456767797470093,
-      "learning_rate": 6.108796546141392e-05,
-      "loss": 3.1833,
+      "grad_norm": 0.86602383852005,
+      "learning_rate": 6.108148947652455e-05,
+      "loss": 3.1854,
       "step": 83350
     },
     {
       "epoch": 8.991913746630727,
-      "grad_norm": 0.8464460968971252,
-      "learning_rate": 6.076416621694549e-05,
-      "loss": 3.1835,
+      "grad_norm": 0.8288998603820801,
+      "learning_rate": 6.075769023205612e-05,
+      "loss": 3.1878,
       "step": 83400
     },
     {
       "epoch": 8.997304582210242,
-      "grad_norm": 0.8834063410758972,
-      "learning_rate": 6.0440366972477055e-05,
-      "loss": 3.1796,
+      "grad_norm": 0.8891361355781555,
+      "learning_rate": 6.043389098758769e-05,
+      "loss": 3.183,
       "step": 83450
     },
     {
       "epoch": 9.002695417789758,
-      "grad_norm": 0.8756682276725769,
-      "learning_rate": 6.011656772800863e-05,
-      "loss": 3.1574,
+      "grad_norm": 0.8949942588806152,
+      "learning_rate": 6.0110091743119265e-05,
+      "loss": 3.1602,
       "step": 83500
     },
     {
       "epoch": 9.008086253369273,
-      "grad_norm": 0.8651514053344727,
-      "learning_rate": 5.9792768483540197e-05,
-      "loss": 3.1089,
+      "grad_norm": 0.8898561000823975,
+      "learning_rate": 5.978629249865083e-05,
+      "loss": 3.113,
       "step": 83550
     },
     {
       "epoch": 9.013477088948788,
-      "grad_norm": 0.8796454668045044,
-      "learning_rate": 5.946896923907177e-05,
-      "loss": 3.1068,
+      "grad_norm": 0.9109175801277161,
+      "learning_rate": 5.94624932541824e-05,
+      "loss": 3.1086,
       "step": 83600
     },
     {
       "epoch": 9.018867924528301,
-      "grad_norm": 0.9122756123542786,
-      "learning_rate": 5.9145169994603345e-05,
-      "loss": 3.1101,
+      "grad_norm": 0.8608222603797913,
+      "learning_rate": 5.913869400971397e-05,
+      "loss": 3.1122,
       "step": 83650
     },
     {
       "epoch": 9.024258760107816,
-      "grad_norm": 0.9356156587600708,
-      "learning_rate": 5.882137075013491e-05,
-      "loss": 3.1069,
+      "grad_norm": 1.012410283088684,
+      "learning_rate": 5.881489476524554e-05,
+      "loss": 3.1111,
       "step": 83700
     },
     {
       "epoch": 9.029649595687331,
-      "grad_norm": 0.8637686371803284,
-      "learning_rate": 5.849757150566649e-05,
-      "loss": 3.1241,
+      "grad_norm": 0.8533174395561218,
+      "learning_rate": 5.849109552077711e-05,
+      "loss": 3.1264,
       "step": 83750
     },
     {
       "epoch": 9.035040431266847,
-      "grad_norm": 0.830084502696991,
-      "learning_rate": 5.817377226119805e-05,
-      "loss": 3.1141,
+      "grad_norm": 0.8322952389717102,
+      "learning_rate": 5.8167296276308683e-05,
+      "loss": 3.1197,
       "step": 83800
     },
     {
       "epoch": 9.040431266846362,
-      "grad_norm": 0.8634269833564758,
-      "learning_rate": 5.784997301672963e-05,
-      "loss": 3.1249,
+      "grad_norm": 0.8747698664665222,
+      "learning_rate": 5.784349703184026e-05,
+      "loss": 3.1287,
       "step": 83850
     },
     {
       "epoch": 9.045822102425875,
-      "grad_norm": 0.8700315356254578,
-      "learning_rate": 5.752617377226119e-05,
-      "loss": 3.1224,
+      "grad_norm": 0.8423470258712769,
+      "learning_rate": 5.7519697787371825e-05,
+      "loss": 3.1262,
       "step": 83900
     },
     {
       "epoch": 9.05121293800539,
-      "grad_norm": 0.8730193376541138,
-      "learning_rate": 5.7202374527792764e-05,
-      "loss": 3.1196,
+      "grad_norm": 0.8985799551010132,
+      "learning_rate": 5.7195898542903396e-05,
+      "loss": 3.1226,
       "step": 83950
     },
     {
       "epoch": 9.056603773584905,
-      "grad_norm": 0.9231434464454651,
-      "learning_rate": 5.6878575283324335e-05,
-      "loss": 3.1161,
+      "grad_norm": 0.8890034556388855,
+      "learning_rate": 5.687209929843497e-05,
+      "loss": 3.1208,
       "step": 84000
     },
     {
       "epoch": 9.056603773584905,
-      "eval_accuracy": 0.39329487817892544,
-      "eval_loss": 3.3126213550567627,
-      "eval_runtime": 183.6807,
-      "eval_samples_per_second": 98.056,
-      "eval_steps_per_second": 6.13,
+      "eval_accuracy": 0.3929635955239791,
+      "eval_loss": 3.3126463890075684,
+      "eval_runtime": 181.5662,
+      "eval_samples_per_second": 99.198,
+      "eval_steps_per_second": 6.202,
       "step": 84000
     },
     {
       "epoch": 9.06199460916442,
-      "grad_norm": 0.8769927024841309,
-      "learning_rate": 5.6554776038855905e-05,
-      "loss": 3.1255,
+      "grad_norm": 0.88873690366745,
+      "learning_rate": 5.654830005396654e-05,
+      "loss": 3.129,
       "step": 84050
     },
     {
       "epoch": 9.067385444743936,
-      "grad_norm": 0.8235280513763428,
-      "learning_rate": 5.623097679438747e-05,
-      "loss": 3.1235,
+      "grad_norm": 0.8265731334686279,
+      "learning_rate": 5.62245008094981e-05,
+      "loss": 3.1267,
       "step": 84100
     },
     {
       "epoch": 9.07277628032345,
-      "grad_norm": 0.8591448664665222,
-      "learning_rate": 5.590717754991905e-05,
-      "loss": 3.1205,
+      "grad_norm": 0.854137122631073,
+      "learning_rate": 5.590070156502968e-05,
+      "loss": 3.1223,
       "step": 84150
     },
     {
       "epoch": 9.078167115902964,
-      "grad_norm": 0.8690507411956787,
-      "learning_rate": 5.558337830545061e-05,
-      "loss": 3.1238,
+      "grad_norm": 0.8830826282501221,
+      "learning_rate": 5.5576902320561244e-05,
+      "loss": 3.1284,
       "step": 84200
     },
     {
       "epoch": 9.08355795148248,
-      "grad_norm": 0.841168224811554,
-      "learning_rate": 5.525957906098219e-05,
-      "loss": 3.1343,
+      "grad_norm": 0.8506742119789124,
+      "learning_rate": 5.525310307609282e-05,
+      "loss": 3.1379,
       "step": 84250
     },
     {
       "epoch": 9.088948787061994,
-      "grad_norm": 0.8726950883865356,
-      "learning_rate": 5.493577981651375e-05,
-      "loss": 3.1138,
+      "grad_norm": 0.8546109199523926,
+      "learning_rate": 5.4929303831624386e-05,
+      "loss": 3.1166,
       "step": 84300
     },
     {
       "epoch": 9.09433962264151,
-      "grad_norm": 0.898984968662262,
-      "learning_rate": 5.461845655693469e-05,
-      "loss": 3.131,
+      "grad_norm": 0.8783098459243774,
+      "learning_rate": 5.460550458715596e-05,
+      "loss": 3.1345,
       "step": 84350
     },
     {
       "epoch": 9.099730458221025,
-      "grad_norm": 0.871024489402771,
-      "learning_rate": 5.429465731246627e-05,
-      "loss": 3.1095,
+      "grad_norm": 0.8363707661628723,
+      "learning_rate": 5.428170534268753e-05,
+      "loss": 3.1139,
       "step": 84400
     },
     {
       "epoch": 9.10512129380054,
-      "grad_norm": 0.9210131764411926,
-      "learning_rate": 5.3970858067997833e-05,
-      "loss": 3.1129,
+      "grad_norm": 0.8737838864326477,
+      "learning_rate": 5.39579060982191e-05,
+      "loss": 3.1151,
       "step": 84450
     },
     {
       "epoch": 9.110512129380053,
-      "grad_norm": 0.8362647294998169,
-      "learning_rate": 5.364705882352941e-05,
-      "loss": 3.1118,
+      "grad_norm": 0.8333865404129028,
+      "learning_rate": 5.363410685375067e-05,
+      "loss": 3.1147,
       "step": 84500
     },
     {
       "epoch": 9.115902964959568,
-      "grad_norm": 0.9286665320396423,
-      "learning_rate": 5.3323259579060975e-05,
-      "loss": 3.1208,
+      "grad_norm": 0.89090496301651,
+      "learning_rate": 5.331030760928224e-05,
+      "loss": 3.1249,
       "step": 84550
     },
     {
       "epoch": 9.121293800539084,
-      "grad_norm": 0.8982177972793579,
-      "learning_rate": 5.299946033459255e-05,
-      "loss": 3.1197,
+      "grad_norm": 0.9061359763145447,
+      "learning_rate": 5.2986508364813804e-05,
+      "loss": 3.1238,
       "step": 84600
     },
     {
       "epoch": 9.126684636118599,
-      "grad_norm": 0.8783411979675293,
-      "learning_rate": 5.267566109012412e-05,
-      "loss": 3.1145,
+      "grad_norm": 0.8999995589256287,
+      "learning_rate": 5.266270912034538e-05,
+      "loss": 3.1182,
       "step": 84650
     },
     {
       "epoch": 9.132075471698114,
-      "grad_norm": 0.8987107276916504,
-      "learning_rate": 5.235186184565569e-05,
-      "loss": 3.1187,
+      "grad_norm": 0.8917686343193054,
+      "learning_rate": 5.233890987587695e-05,
+      "loss": 3.1206,
       "step": 84700
     },
     {
       "epoch": 9.137466307277627,
-      "grad_norm": 0.872234582901001,
-      "learning_rate": 5.202806260118726e-05,
-      "loss": 3.128,
+      "grad_norm": 0.849173367023468,
+      "learning_rate": 5.2015110631408524e-05,
+      "loss": 3.1327,
       "step": 84750
     },
     {
       "epoch": 9.142857142857142,
-      "grad_norm": 0.8869773745536804,
-      "learning_rate": 5.170426335671883e-05,
-      "loss": 3.132,
+      "grad_norm": 0.8533074855804443,
+      "learning_rate": 5.1691311386940094e-05,
+      "loss": 3.135,
       "step": 84800
     },
     {
       "epoch": 9.148247978436657,
-      "grad_norm": 0.8714069724082947,
-      "learning_rate": 5.1380464112250394e-05,
-      "loss": 3.1384,
+      "grad_norm": 0.8497921228408813,
+      "learning_rate": 5.136751214247166e-05,
+      "loss": 3.1404,
       "step": 84850
     },
     {
       "epoch": 9.153638814016173,
-      "grad_norm": 0.8658605217933655,
-      "learning_rate": 5.105666486778197e-05,
-      "loss": 3.1152,
+      "grad_norm": 0.8530235290527344,
+      "learning_rate": 5.1043712898003236e-05,
+      "loss": 3.1165,
       "step": 84900
     },
     {
       "epoch": 9.159029649595688,
-      "grad_norm": 0.8711130619049072,
-      "learning_rate": 5.0732865623313536e-05,
-      "loss": 3.1266,
+      "grad_norm": 0.8733996748924255,
+      "learning_rate": 5.07199136535348e-05,
+      "loss": 3.1311,
       "step": 84950
     },
     {
       "epoch": 9.164420485175203,
-      "grad_norm": 0.8554506301879883,
-      "learning_rate": 5.040906637884511e-05,
-      "loss": 3.1182,
+      "grad_norm": 0.8780140280723572,
+      "learning_rate": 5.039611440906638e-05,
+      "loss": 3.1214,
       "step": 85000
     },
     {
       "epoch": 9.164420485175203,
-      "eval_accuracy": 0.3934888235836743,
-      "eval_loss": 3.310671806335449,
-      "eval_runtime": 183.8724,
-      "eval_samples_per_second": 97.954,
-      "eval_steps_per_second": 6.124,
+      "eval_accuracy": 0.39322088556232376,
+      "eval_loss": 3.3106119632720947,
+      "eval_runtime": 181.5249,
+      "eval_samples_per_second": 99.221,
+      "eval_steps_per_second": 6.203,
       "step": 85000
     },
     {
       "epoch": 9.169811320754716,
-      "grad_norm": 0.8684996962547302,
-      "learning_rate": 5.0085267134376684e-05,
-      "loss": 3.1199,
+      "grad_norm": 0.8870514631271362,
+      "learning_rate": 5.007231516459794e-05,
+      "loss": 3.124,
       "step": 85050
     },
     {
       "epoch": 9.175202156334231,
-      "grad_norm": 0.8756259679794312,
-      "learning_rate": 4.9761467889908255e-05,
-      "loss": 3.1533,
+      "grad_norm": 0.8743305802345276,
+      "learning_rate": 4.974851592012952e-05,
+      "loss": 3.1566,
       "step": 85100
     },
     {
       "epoch": 9.180592991913747,
-      "grad_norm": 0.8757433295249939,
-      "learning_rate": 4.9437668645439826e-05,
-      "loss": 3.1191,
+      "grad_norm": 0.8916103839874268,
+      "learning_rate": 4.9424716675661084e-05,
+      "loss": 3.1225,
       "step": 85150
     },
     {
       "epoch": 9.185983827493262,
-      "grad_norm": 0.857020378112793,
-      "learning_rate": 4.911386940097139e-05,
-      "loss": 3.1121,
+      "grad_norm": 0.9013537764549255,
+      "learning_rate": 4.9100917431192655e-05,
+      "loss": 3.1149,
       "step": 85200
     },
     {
       "epoch": 9.191374663072777,
-      "grad_norm": 0.9065777659416199,
-      "learning_rate": 4.879007015650297e-05,
-      "loss": 3.1354,
+      "grad_norm": 0.9145349264144897,
+      "learning_rate": 4.8777118186724226e-05,
+      "loss": 3.1385,
       "step": 85250
     },
     {
       "epoch": 9.19676549865229,
-      "grad_norm": 0.848362147808075,
-      "learning_rate": 4.846627091203453e-05,
-      "loss": 3.1066,
+      "grad_norm": 0.8603264093399048,
+      "learning_rate": 4.8453318942255797e-05,
+      "loss": 3.1109,
       "step": 85300
     },
     {
       "epoch": 9.202156334231805,
-      "grad_norm": 0.8962082862854004,
-      "learning_rate": 4.814247166756611e-05,
-      "loss": 3.1158,
+      "grad_norm": 0.9225505590438843,
+      "learning_rate": 4.812951969778736e-05,
+      "loss": 3.1189,
       "step": 85350
     },
     {
       "epoch": 9.20754716981132,
-      "grad_norm": 0.8896428942680359,
-      "learning_rate": 4.7818672423097674e-05,
-      "loss": 3.1332,
+      "grad_norm": 0.8797885179519653,
+      "learning_rate": 4.780572045331894e-05,
+      "loss": 3.1375,
       "step": 85400
     },
     {
       "epoch": 9.212938005390836,
-      "grad_norm": 0.8223966360092163,
-      "learning_rate": 4.7494873178629244e-05,
-      "loss": 3.097,
+      "grad_norm": 0.8161735534667969,
+      "learning_rate": 4.748192120885051e-05,
+      "loss": 3.099,
       "step": 85450
     },
     {
       "epoch": 9.21832884097035,
-      "grad_norm": 0.867064893245697,
-      "learning_rate": 4.7171073934160815e-05,
-      "loss": 3.1308,
+      "grad_norm": 0.8942599892616272,
+      "learning_rate": 4.715812196438208e-05,
+      "loss": 3.1356,
       "step": 85500
     },
     {
       "epoch": 9.223719676549866,
-      "grad_norm": 0.9328610301017761,
-      "learning_rate": 4.6847274689692386e-05,
-      "loss": 3.1116,
+      "grad_norm": 0.9018413424491882,
+      "learning_rate": 4.683432271991365e-05,
+      "loss": 3.1138,
       "step": 85550
     },
     {
       "epoch": 9.22911051212938,
-      "grad_norm": 0.8840168714523315,
-      "learning_rate": 4.652347544522396e-05,
-      "loss": 3.134,
+      "grad_norm": 0.8639441728591919,
+      "learning_rate": 4.651052347544522e-05,
+      "loss": 3.1362,
       "step": 85600
     },
     {
       "epoch": 9.234501347708894,
-      "grad_norm": 0.8884910345077515,
-      "learning_rate": 4.619967620075553e-05,
-      "loss": 3.1195,
+      "grad_norm": 0.8743155598640442,
+      "learning_rate": 4.618672423097679e-05,
+      "loss": 3.123,
       "step": 85650
     },
     {
       "epoch": 9.23989218328841,
-      "grad_norm": 0.9012236595153809,
-      "learning_rate": 4.587587695628709e-05,
-      "loss": 3.1036,
+      "grad_norm": 0.8649117350578308,
+      "learning_rate": 4.586292498650836e-05,
+      "loss": 3.1065,
       "step": 85700
     },
     {
       "epoch": 9.245283018867925,
-      "grad_norm": 0.8753321170806885,
-      "learning_rate": 4.555207771181867e-05,
-      "loss": 3.1286,
+      "grad_norm": 0.842155396938324,
+      "learning_rate": 4.5539125742039935e-05,
+      "loss": 3.1306,
       "step": 85750
     },
     {
       "epoch": 9.25067385444744,
-      "grad_norm": 0.8457467555999756,
-      "learning_rate": 4.522827846735024e-05,
-      "loss": 3.1347,
+      "grad_norm": 0.8524735569953918,
+      "learning_rate": 4.52153264975715e-05,
+      "loss": 3.1376,
       "step": 85800
     },
     {
       "epoch": 9.256064690026955,
-      "grad_norm": 0.8559665083885193,
-      "learning_rate": 4.490447922288181e-05,
-      "loss": 3.134,
+      "grad_norm": 0.8670454621315002,
+      "learning_rate": 4.4891527253103076e-05,
+      "loss": 3.1375,
       "step": 85850
     },
     {
       "epoch": 9.261455525606468,
-      "grad_norm": 0.8733518123626709,
-      "learning_rate": 4.458067997841338e-05,
-      "loss": 3.0981,
+      "grad_norm": 0.8947070240974426,
+      "learning_rate": 4.456772800863464e-05,
+      "loss": 3.1014,
       "step": 85900
     },
     {
       "epoch": 9.266846361185983,
-      "grad_norm": 0.8558553457260132,
-      "learning_rate": 4.4256880733944947e-05,
-      "loss": 3.134,
+      "grad_norm": 0.8871631622314453,
+      "learning_rate": 4.424392876416621e-05,
+      "loss": 3.1376,
       "step": 85950
     },
     {
       "epoch": 9.272237196765499,
-      "grad_norm": 0.8802152872085571,
-      "learning_rate": 4.3933081489476524e-05,
-      "loss": 3.1283,
+      "grad_norm": 0.8541000485420227,
+      "learning_rate": 4.392012951969778e-05,
+      "loss": 3.1308,
       "step": 86000
     },
     {
       "epoch": 9.272237196765499,
-      "eval_accuracy": 0.39366853546011943,
-      "eval_loss": 3.3087074756622314,
-      "eval_runtime": 183.4292,
-      "eval_samples_per_second": 98.19,
-      "eval_steps_per_second": 6.139,
+      "eval_accuracy": 0.3933883196624459,
+      "eval_loss": 3.309619188308716,
+      "eval_runtime": 181.3013,
+      "eval_samples_per_second": 99.343,
+      "eval_steps_per_second": 6.211,
       "step": 86000
     },
     {
       "epoch": 9.277628032345014,
-      "grad_norm": 0.8693817257881165,
-      "learning_rate": 4.360928224500809e-05,
-      "loss": 3.1253,
+      "grad_norm": 0.8713949918746948,
+      "learning_rate": 4.359633027522935e-05,
+      "loss": 3.1312,
       "step": 86050
     },
     {
       "epoch": 9.283018867924529,
-      "grad_norm": 0.929667592048645,
-      "learning_rate": 4.3285483000539666e-05,
-      "loss": 3.124,
+      "grad_norm": 0.8741675019264221,
+      "learning_rate": 4.3272531030760924e-05,
+      "loss": 3.1287,
       "step": 86100
     },
     {
       "epoch": 9.288409703504042,
-      "grad_norm": 0.9104236960411072,
-      "learning_rate": 4.296168375607123e-05,
-      "loss": 3.1201,
+      "grad_norm": 0.8772149085998535,
+      "learning_rate": 4.2948731786292495e-05,
+      "loss": 3.1253,
       "step": 86150
     },
     {
       "epoch": 9.293800539083557,
-      "grad_norm": 0.8924854397773743,
-      "learning_rate": 4.263788451160281e-05,
-      "loss": 3.134,
+      "grad_norm": 0.9387300610542297,
+      "learning_rate": 4.262493254182407e-05,
+      "loss": 3.1367,
       "step": 86200
     },
     {
       "epoch": 9.299191374663073,
-      "grad_norm": 0.8825605511665344,
-      "learning_rate": 4.231408526713437e-05,
-      "loss": 3.1373,
+      "grad_norm": 0.8727113604545593,
+      "learning_rate": 4.230113329735564e-05,
+      "loss": 3.14,
       "step": 86250
     },
     {
       "epoch": 9.304582210242588,
-      "grad_norm": 0.8415103554725647,
-      "learning_rate": 4.199028602266594e-05,
-      "loss": 3.1284,
+      "grad_norm": 0.8436615467071533,
+      "learning_rate": 4.197733405288721e-05,
+      "loss": 3.1318,
       "step": 86300
     },
     {
       "epoch": 9.309973045822103,
-      "grad_norm": 0.8934675455093384,
-      "learning_rate": 4.1666486778197514e-05,
-      "loss": 3.0965,
+      "grad_norm": 0.8674768805503845,
+      "learning_rate": 4.165353480841878e-05,
+      "loss": 3.0986,
       "step": 86350
     },
     {
       "epoch": 9.315363881401618,
-      "grad_norm": 0.8981188535690308,
-      "learning_rate": 4.1342687533729085e-05,
-      "loss": 3.1221,
+      "grad_norm": 0.8669427037239075,
+      "learning_rate": 4.132973556395035e-05,
+      "loss": 3.1257,
       "step": 86400
     },
     {
       "epoch": 9.320754716981131,
-      "grad_norm": 0.8392598032951355,
-      "learning_rate": 4.101888828926065e-05,
-      "loss": 3.1186,
+      "grad_norm": 0.8331587910652161,
+      "learning_rate": 4.1005936319481913e-05,
+      "loss": 3.1204,
       "step": 86450
     },
     {
       "epoch": 9.326145552560646,
-      "grad_norm": 0.8929069638252258,
-      "learning_rate": 4.0695089044792226e-05,
-      "loss": 3.1407,
+      "grad_norm": 0.8975145220756531,
+      "learning_rate": 4.068213707501349e-05,
+      "loss": 3.1432,
       "step": 86500
     },
     {
       "epoch": 9.331536388140162,
-      "grad_norm": 0.9197957515716553,
-      "learning_rate": 4.037128980032379e-05,
-      "loss": 3.1202,
+      "grad_norm": 0.9537723660469055,
+      "learning_rate": 4.0358337830545055e-05,
+      "loss": 3.1236,
       "step": 86550
     },
     {
       "epoch": 9.336927223719677,
-      "grad_norm": 0.8734220266342163,
-      "learning_rate": 4.004749055585537e-05,
-      "loss": 3.1092,
+      "grad_norm": 0.8615782260894775,
+      "learning_rate": 4.003453858607663e-05,
+      "loss": 3.1125,
       "step": 86600
     },
     {
       "epoch": 9.342318059299192,
-      "grad_norm": 0.8806682229042053,
-      "learning_rate": 3.972369131138694e-05,
-      "loss": 3.1167,
+      "grad_norm": 0.8585619330406189,
+      "learning_rate": 3.97107393416082e-05,
+      "loss": 3.1209,
       "step": 86650
     },
     {
       "epoch": 9.347708894878707,
-      "grad_norm": 0.8593670129776001,
-      "learning_rate": 3.93998920669185e-05,
-      "loss": 3.1351,
+      "grad_norm": 0.8578895330429077,
+      "learning_rate": 3.9386940097139775e-05,
+      "loss": 3.1384,
       "step": 86700
     },
     {
       "epoch": 9.35309973045822,
-      "grad_norm": 0.8641849160194397,
-      "learning_rate": 3.907609282245008e-05,
-      "loss": 3.1207,
+      "grad_norm": 0.9198311567306519,
+      "learning_rate": 3.906314085267134e-05,
+      "loss": 3.1243,
       "step": 86750
     },
     {
       "epoch": 9.358490566037736,
-      "grad_norm": 0.8967059254646301,
-      "learning_rate": 3.8752293577981645e-05,
-      "loss": 3.1447,
+      "grad_norm": 0.8892605304718018,
+      "learning_rate": 3.874581759309228e-05,
+      "loss": 3.1497,
       "step": 86800
     },
     {
       "epoch": 9.36388140161725,
-      "grad_norm": 0.851588249206543,
-      "learning_rate": 3.842849433351322e-05,
-      "loss": 3.1141,
+      "grad_norm": 0.8520983457565308,
+      "learning_rate": 3.8422018348623855e-05,
+      "loss": 3.118,
       "step": 86850
     },
     {
       "epoch": 9.369272237196766,
-      "grad_norm": 0.8949471116065979,
-      "learning_rate": 3.810469508904479e-05,
-      "loss": 3.1332,
+      "grad_norm": 0.8694999814033508,
+      "learning_rate": 3.809821910415542e-05,
+      "loss": 3.1382,
       "step": 86900
     },
     {
       "epoch": 9.374663072776281,
-      "grad_norm": 0.8716377019882202,
-      "learning_rate": 3.7780895844576364e-05,
-      "loss": 3.1285,
+      "grad_norm": 0.8894853591918945,
+      "learning_rate": 3.7774419859687e-05,
+      "loss": 3.1312,
       "step": 86950
     },
     {
       "epoch": 9.380053908355794,
-      "grad_norm": 0.8617969155311584,
-      "learning_rate": 3.745709660010793e-05,
-      "loss": 3.1145,
+      "grad_norm": 0.9015088081359863,
+      "learning_rate": 3.745062061521856e-05,
+      "loss": 3.1173,
       "step": 87000
     },
     {
       "epoch": 9.380053908355794,
-      "eval_accuracy": 0.39399134318981627,
-      "eval_loss": 3.307685613632202,
-      "eval_runtime": 183.5956,
-      "eval_samples_per_second": 98.101,
-      "eval_steps_per_second": 6.133,
+      "eval_accuracy": 0.3937577394810151,
+      "eval_loss": 3.30816388130188,
+      "eval_runtime": 181.5933,
+      "eval_samples_per_second": 99.183,
+      "eval_steps_per_second": 6.201,
       "step": 87000
     },
     {
       "epoch": 9.38544474393531,
-      "grad_norm": 0.8978739976882935,
-      "learning_rate": 3.71332973556395e-05,
-      "loss": 3.1249,
+      "grad_norm": 0.8614080548286438,
+      "learning_rate": 3.712682137075013e-05,
+      "loss": 3.1268,
       "step": 87050
     },
     {
       "epoch": 9.390835579514825,
-      "grad_norm": 0.857925295829773,
-      "learning_rate": 3.680949811117107e-05,
-      "loss": 3.1099,
+      "grad_norm": 0.8563096523284912,
+      "learning_rate": 3.68030221262817e-05,
+      "loss": 3.1158,
       "step": 87100
     },
     {
       "epoch": 9.39622641509434,
-      "grad_norm": 0.8732088208198547,
-      "learning_rate": 3.648569886670264e-05,
-      "loss": 3.1194,
+      "grad_norm": 0.8736945986747742,
+      "learning_rate": 3.6479222881813274e-05,
+      "loss": 3.1228,
       "step": 87150
     },
     {
       "epoch": 9.401617250673855,
-      "grad_norm": 0.9072591066360474,
-      "learning_rate": 3.616189962223421e-05,
-      "loss": 3.1287,
+      "grad_norm": 0.8687424063682556,
+      "learning_rate": 3.6155423637344844e-05,
+      "loss": 3.1344,
       "step": 87200
     },
     {
       "epoch": 9.40700808625337,
-      "grad_norm": 0.8680362105369568,
-      "learning_rate": 3.583810037776578e-05,
-      "loss": 3.1274,
+      "grad_norm": 0.9159214496612549,
+      "learning_rate": 3.5831624392876415e-05,
+      "loss": 3.1308,
       "step": 87250
     },
     {
       "epoch": 9.412398921832883,
-      "grad_norm": 0.8892818093299866,
-      "learning_rate": 3.5514301133297354e-05,
-      "loss": 3.1368,
+      "grad_norm": 0.9170340895652771,
+      "learning_rate": 3.5507825148407986e-05,
+      "loss": 3.1414,
       "step": 87300
     },
     {
       "epoch": 9.417789757412399,
-      "grad_norm": 0.9270879626274109,
-      "learning_rate": 3.519697787371829e-05,
-      "loss": 3.1019,
+      "grad_norm": 0.9083608984947205,
+      "learning_rate": 3.518402590393956e-05,
+      "loss": 3.1053,
       "step": 87350
     },
     {
       "epoch": 9.423180592991914,
-      "grad_norm": 0.9087534546852112,
-      "learning_rate": 3.487317862924986e-05,
-      "loss": 3.1312,
+      "grad_norm": 0.8961578607559204,
+      "learning_rate": 3.486022665947113e-05,
+      "loss": 3.1325,
       "step": 87400
     },
     {
       "epoch": 9.428571428571429,
-      "grad_norm": 0.8477288484573364,
-      "learning_rate": 3.4549379384781434e-05,
-      "loss": 3.1269,
+      "grad_norm": 0.8620476126670837,
+      "learning_rate": 3.45364274150027e-05,
+      "loss": 3.1312,
       "step": 87450
     },
     {
       "epoch": 9.433962264150944,
-      "grad_norm": 0.8833394646644592,
-      "learning_rate": 3.4225580140313e-05,
-      "loss": 3.1109,
+      "grad_norm": 0.869385838508606,
+      "learning_rate": 3.421262817053426e-05,
+      "loss": 3.1146,
       "step": 87500
     },
     {
       "epoch": 9.439353099730457,
-      "grad_norm": 0.834549605846405,
-      "learning_rate": 3.390178089584457e-05,
-      "loss": 3.1338,
+      "grad_norm": 0.8649206161499023,
+      "learning_rate": 3.3888828926065834e-05,
+      "loss": 3.1382,
       "step": 87550
     },
     {
       "epoch": 9.444743935309972,
-      "grad_norm": 0.8486028909683228,
-      "learning_rate": 3.357798165137615e-05,
-      "loss": 3.1179,
+      "grad_norm": 0.8611971735954285,
+      "learning_rate": 3.3565029681597405e-05,
+      "loss": 3.1212,
       "step": 87600
     },
     {
       "epoch": 9.450134770889488,
-      "grad_norm": 0.8409470915794373,
-      "learning_rate": 3.325418240690772e-05,
-      "loss": 3.1084,
+      "grad_norm": 0.8563823699951172,
+      "learning_rate": 3.3241230437128976e-05,
+      "loss": 3.1109,
       "step": 87650
     },
     {
       "epoch": 9.455525606469003,
-      "grad_norm": 0.9353410005569458,
-      "learning_rate": 3.293038316243929e-05,
-      "loss": 3.1225,
+      "grad_norm": 0.8646349906921387,
+      "learning_rate": 3.2917431192660547e-05,
+      "loss": 3.1266,
       "step": 87700
     },
     {
       "epoch": 9.460916442048518,
-      "grad_norm": 0.886806309223175,
-      "learning_rate": 3.260658391797086e-05,
-      "loss": 3.1205,
+      "grad_norm": 0.8988960981369019,
+      "learning_rate": 3.259363194819212e-05,
+      "loss": 3.122,
       "step": 87750
     },
     {
       "epoch": 9.466307277628033,
-      "grad_norm": 0.8259122371673584,
-      "learning_rate": 3.2282784673502424e-05,
-      "loss": 3.121,
+      "grad_norm": 0.8426046967506409,
+      "learning_rate": 3.226983270372369e-05,
+      "loss": 3.1241,
       "step": 87800
     },
     {
       "epoch": 9.471698113207546,
-      "grad_norm": 0.8698675036430359,
-      "learning_rate": 3.1958985429033995e-05,
-      "loss": 3.1229,
+      "grad_norm": 0.8700035810470581,
+      "learning_rate": 3.194603345925526e-05,
+      "loss": 3.126,
       "step": 87850
     },
     {
       "epoch": 9.477088948787062,
-      "grad_norm": 0.8667991757392883,
-      "learning_rate": 3.1635186184565565e-05,
-      "loss": 3.1285,
+      "grad_norm": 0.8410253524780273,
+      "learning_rate": 3.162223421478683e-05,
+      "loss": 3.135,
       "step": 87900
     },
     {
       "epoch": 9.482479784366577,
-      "grad_norm": 0.9097034931182861,
-      "learning_rate": 3.1311386940097136e-05,
-      "loss": 3.1274,
+      "grad_norm": 0.8620033860206604,
+      "learning_rate": 3.12984349703184e-05,
+      "loss": 3.1307,
       "step": 87950
     },
     {
       "epoch": 9.487870619946092,
-      "grad_norm": 0.9387816786766052,
-      "learning_rate": 3.098758769562871e-05,
-      "loss": 3.1251,
+      "grad_norm": 0.9386901259422302,
+      "learning_rate": 3.097463572584997e-05,
+      "loss": 3.127,
       "step": 88000
     },
     {
       "epoch": 9.487870619946092,
-      "eval_accuracy": 0.3944075924030335,
-      "eval_loss": 3.3036043643951416,
-      "eval_runtime": 183.5422,
-      "eval_samples_per_second": 98.13,
-      "eval_steps_per_second": 6.135,
+      "eval_accuracy": 0.3941805078675013,
+      "eval_loss": 3.3037567138671875,
+      "eval_runtime": 181.5433,
+      "eval_samples_per_second": 99.21,
+      "eval_steps_per_second": 6.202,
       "step": 88000
     },
     {
       "epoch": 9.493261455525607,
-      "grad_norm": 0.8956674933433533,
-      "learning_rate": 3.066378845116028e-05,
-      "loss": 3.1346,
+      "grad_norm": 0.907065212726593,
+      "learning_rate": 3.065083648138154e-05,
+      "loss": 3.1374,
       "step": 88050
     },
     {
       "epoch": 9.498652291105122,
-      "grad_norm": 0.9043859839439392,
-      "learning_rate": 3.033998920669185e-05,
-      "loss": 3.13,
+      "grad_norm": 0.8494638204574585,
+      "learning_rate": 3.0327037236913114e-05,
+      "loss": 3.1335,
       "step": 88100
     },
     {
       "epoch": 9.504043126684635,
-      "grad_norm": 0.8860067129135132,
-      "learning_rate": 3.0016189962223416e-05,
-      "loss": 3.1347,
+      "grad_norm": 0.8722262382507324,
+      "learning_rate": 3.000323799244468e-05,
+      "loss": 3.1368,
       "step": 88150
     },
     {
       "epoch": 9.50943396226415,
-      "grad_norm": 0.8624042272567749,
-      "learning_rate": 2.9692390717754987e-05,
-      "loss": 3.1483,
+      "grad_norm": 0.8585368394851685,
+      "learning_rate": 2.9679438747976252e-05,
+      "loss": 3.1504,
       "step": 88200
     },
     {
       "epoch": 9.514824797843666,
-      "grad_norm": 0.9003132581710815,
-      "learning_rate": 2.9368591473286558e-05,
-      "loss": 3.1226,
+      "grad_norm": 0.8924391865730286,
+      "learning_rate": 2.9355639503507823e-05,
+      "loss": 3.1253,
       "step": 88250
     },
     {
       "epoch": 9.520215633423181,
-      "grad_norm": 0.884479284286499,
-      "learning_rate": 2.904479222881813e-05,
-      "loss": 3.1117,
+      "grad_norm": 0.8812455534934998,
+      "learning_rate": 2.9031840259039394e-05,
+      "loss": 3.1166,
       "step": 88300
     },
     {
       "epoch": 9.525606469002696,
-      "grad_norm": 0.9293926954269409,
-      "learning_rate": 2.87209929843497e-05,
-      "loss": 3.1243,
+      "grad_norm": 0.8853342533111572,
+      "learning_rate": 2.8708041014570965e-05,
+      "loss": 3.128,
       "step": 88350
     },
     {
       "epoch": 9.530997304582211,
-      "grad_norm": 0.9077318906784058,
-      "learning_rate": 2.8397193739881274e-05,
-      "loss": 3.1164,
+      "grad_norm": 0.8742244839668274,
+      "learning_rate": 2.8384241770102532e-05,
+      "loss": 3.1204,
       "step": 88400
     },
     {
       "epoch": 9.536388140161725,
-      "grad_norm": 0.8652298450469971,
-      "learning_rate": 2.8073394495412842e-05,
-      "loss": 3.1231,
+      "grad_norm": 0.8439992070198059,
+      "learning_rate": 2.8060442525634103e-05,
+      "loss": 3.1246,
       "step": 88450
     },
     {
       "epoch": 9.54177897574124,
-      "grad_norm": 0.8628005385398865,
-      "learning_rate": 2.7749595250944413e-05,
-      "loss": 3.1198,
+      "grad_norm": 0.8583610653877258,
+      "learning_rate": 2.7736643281165674e-05,
+      "loss": 3.1232,
       "step": 88500
     },
     {
       "epoch": 9.547169811320755,
-      "grad_norm": 0.8354774117469788,
-      "learning_rate": 2.7425796006475984e-05,
-      "loss": 3.1267,
+      "grad_norm": 0.8362330794334412,
+      "learning_rate": 2.7412844036697245e-05,
+      "loss": 3.1289,
       "step": 88550
     },
     {
       "epoch": 9.55256064690027,
-      "grad_norm": 0.8727371692657471,
-      "learning_rate": 2.7101996762007554e-05,
-      "loss": 3.1216,
+      "grad_norm": 0.8831988573074341,
+      "learning_rate": 2.7089044792228816e-05,
+      "loss": 3.1265,
       "step": 88600
     },
     {
       "epoch": 9.557951482479785,
-      "grad_norm": 0.8466863036155701,
-      "learning_rate": 2.6778197517539125e-05,
-      "loss": 3.1325,
+      "grad_norm": 0.860260009765625,
+      "learning_rate": 2.6765245547760383e-05,
+      "loss": 3.1356,
       "step": 88650
     },
     {
       "epoch": 9.563342318059298,
-      "grad_norm": 0.8633363842964172,
-      "learning_rate": 2.6454398273070693e-05,
-      "loss": 3.1185,
+      "grad_norm": 0.8877055644989014,
+      "learning_rate": 2.6441446303291954e-05,
+      "loss": 3.1221,
       "step": 88700
     },
     {
       "epoch": 9.568733153638814,
-      "grad_norm": 0.8681420087814331,
-      "learning_rate": 2.6130599028602264e-05,
-      "loss": 3.1405,
+      "grad_norm": 0.8568811416625977,
+      "learning_rate": 2.6117647058823525e-05,
+      "loss": 3.1421,
       "step": 88750
     },
     {
       "epoch": 9.574123989218329,
-      "grad_norm": 0.9068518877029419,
-      "learning_rate": 2.5806799784133835e-05,
-      "loss": 3.1347,
+      "grad_norm": 0.8610482811927795,
+      "learning_rate": 2.57938478143551e-05,
+      "loss": 3.1381,
       "step": 88800
     },
     {
       "epoch": 9.579514824797844,
-      "grad_norm": 0.8506572842597961,
-      "learning_rate": 2.5483000539665406e-05,
-      "loss": 3.1212,
+      "grad_norm": 0.8704871535301208,
+      "learning_rate": 2.547004856988667e-05,
+      "loss": 3.1244,
       "step": 88850
     },
     {
       "epoch": 9.584905660377359,
-      "grad_norm": 0.8176720142364502,
-      "learning_rate": 2.5159201295196976e-05,
-      "loss": 3.1273,
+      "grad_norm": 0.8440133333206177,
+      "learning_rate": 2.514624932541824e-05,
+      "loss": 3.1297,
       "step": 88900
     },
     {
       "epoch": 9.590296495956874,
-      "grad_norm": 0.9253360629081726,
-      "learning_rate": 2.4835402050728544e-05,
-      "loss": 3.1155,
+      "grad_norm": 0.8958004713058472,
+      "learning_rate": 2.482245008094981e-05,
+      "loss": 3.1197,
       "step": 88950
     },
     {
       "epoch": 9.595687331536388,
-      "grad_norm": 0.8876266479492188,
-      "learning_rate": 2.4511602806260115e-05,
-      "loss": 3.1404,
+      "grad_norm": 0.9002637267112732,
+      "learning_rate": 2.449865083648138e-05,
+      "loss": 3.1444,
       "step": 89000
     },
     {
       "epoch": 9.595687331536388,
-      "eval_accuracy": 0.39452298176989253,
-      "eval_loss": 3.302065372467041,
-      "eval_runtime": 183.5463,
-      "eval_samples_per_second": 98.128,
-      "eval_steps_per_second": 6.135,
+      "eval_accuracy": 0.39441508945229276,
+      "eval_loss": 3.30206036567688,
+      "eval_runtime": 181.6419,
+      "eval_samples_per_second": 99.157,
+      "eval_steps_per_second": 6.199,
       "step": 89000
     },
     {
       "epoch": 9.601078167115903,
-      "grad_norm": 0.8948820233345032,
-      "learning_rate": 2.4187803561791686e-05,
-      "loss": 3.1014,
+      "grad_norm": 0.8430779576301575,
+      "learning_rate": 2.417485159201295e-05,
+      "loss": 3.1061,
       "step": 89050
     },
     {
       "epoch": 9.606469002695418,
-      "grad_norm": 0.8849186897277832,
-      "learning_rate": 2.3864004317323257e-05,
-      "loss": 3.1244,
+      "grad_norm": 0.8381374478340149,
+      "learning_rate": 2.385105234754452e-05,
+      "loss": 3.1284,
       "step": 89100
     },
     {
       "epoch": 9.611859838274933,
-      "grad_norm": 0.8476836085319519,
-      "learning_rate": 2.354020507285483e-05,
-      "loss": 3.1156,
+      "grad_norm": 0.8521980047225952,
+      "learning_rate": 2.3527253103076092e-05,
+      "loss": 3.1183,
       "step": 89150
     },
     {
       "epoch": 9.617250673854448,
-      "grad_norm": 0.8894110321998596,
-      "learning_rate": 2.3216405828386402e-05,
-      "loss": 3.1006,
+      "grad_norm": 0.8843023777008057,
+      "learning_rate": 2.320345385860766e-05,
+      "loss": 3.1035,
       "step": 89200
     },
     {
       "epoch": 9.622641509433961,
-      "grad_norm": 0.8986078500747681,
-      "learning_rate": 2.289260658391797e-05,
-      "loss": 3.1276,
+      "grad_norm": 0.8870645761489868,
+      "learning_rate": 2.287965461413923e-05,
+      "loss": 3.1296,
       "step": 89250
     },
     {
       "epoch": 9.628032345013477,
-      "grad_norm": 0.9119869470596313,
-      "learning_rate": 2.256880733944954e-05,
-      "loss": 3.1514,
+      "grad_norm": 0.8845605850219727,
+      "learning_rate": 2.25558553696708e-05,
+      "loss": 3.1532,
       "step": 89300
     },
     {
       "epoch": 9.633423180592992,
-      "grad_norm": 0.8732916712760925,
-      "learning_rate": 2.224500809498111e-05,
-      "loss": 3.1433,
+      "grad_norm": 0.8621330261230469,
+      "learning_rate": 2.2232056125202372e-05,
+      "loss": 3.145,
       "step": 89350
     },
     {
       "epoch": 9.638814016172507,
-      "grad_norm": 0.8498047590255737,
-      "learning_rate": 2.1921208850512682e-05,
-      "loss": 3.1164,
+      "grad_norm": 0.8898807764053345,
+      "learning_rate": 2.1908256880733943e-05,
+      "loss": 3.1184,
       "step": 89400
     },
     {
       "epoch": 9.644204851752022,
-      "grad_norm": 0.8872281312942505,
-      "learning_rate": 2.1597409606044253e-05,
-      "loss": 3.1294,
+      "grad_norm": 0.8443333506584167,
+      "learning_rate": 2.158445763626551e-05,
+      "loss": 3.1347,
       "step": 89450
     },
     {
       "epoch": 9.649595687331537,
-      "grad_norm": 0.8698834776878357,
-      "learning_rate": 2.127361036157582e-05,
-      "loss": 3.1374,
+      "grad_norm": 0.8922162652015686,
+      "learning_rate": 2.126065839179708e-05,
+      "loss": 3.1402,
       "step": 89500
     },
     {
       "epoch": 9.65498652291105,
-      "grad_norm": 0.8387486338615417,
-      "learning_rate": 2.094981111710739e-05,
-      "loss": 3.1255,
+      "grad_norm": 0.8212519884109497,
+      "learning_rate": 2.0936859147328653e-05,
+      "loss": 3.13,
       "step": 89550
     },
     {
       "epoch": 9.660377358490566,
-      "grad_norm": 0.8434497714042664,
-      "learning_rate": 2.0626011872638962e-05,
-      "loss": 3.1261,
+      "grad_norm": 0.8509665131568909,
+      "learning_rate": 2.0613059902860227e-05,
+      "loss": 3.1285,
       "step": 89600
     },
     {
       "epoch": 9.66576819407008,
-      "grad_norm": 0.8968543410301208,
-      "learning_rate": 2.0302212628170533e-05,
-      "loss": 3.1309,
+      "grad_norm": 0.8982486128807068,
+      "learning_rate": 2.0289260658391798e-05,
+      "loss": 3.133,
       "step": 89650
     },
     {
       "epoch": 9.671159029649596,
-      "grad_norm": 0.9121111035346985,
-      "learning_rate": 1.99784133837021e-05,
-      "loss": 3.1151,
+      "grad_norm": 0.9122908115386963,
+      "learning_rate": 1.996546141392337e-05,
+      "loss": 3.1196,
       "step": 89700
     },
     {
       "epoch": 9.676549865229111,
-      "grad_norm": 0.9186899065971375,
-      "learning_rate": 1.965461413923367e-05,
-      "loss": 3.1392,
+      "grad_norm": 0.9822547435760498,
+      "learning_rate": 1.9641662169454936e-05,
+      "loss": 3.1431,
       "step": 89750
     },
     {
       "epoch": 9.681940700808624,
-      "grad_norm": 0.8917949199676514,
-      "learning_rate": 1.9330814894765242e-05,
-      "loss": 3.1306,
+      "grad_norm": 0.8897708058357239,
+      "learning_rate": 1.9317862924986507e-05,
+      "loss": 3.133,
       "step": 89800
     },
     {
       "epoch": 9.68733153638814,
-      "grad_norm": 0.8518571257591248,
-      "learning_rate": 1.9007015650296813e-05,
-      "loss": 3.1275,
+      "grad_norm": 0.8570009469985962,
+      "learning_rate": 1.8994063680518078e-05,
+      "loss": 3.1298,
       "step": 89850
     },
     {
       "epoch": 9.692722371967655,
-      "grad_norm": 0.8881104588508606,
-      "learning_rate": 1.8683216405828384e-05,
-      "loss": 3.1199,
+      "grad_norm": 0.8820420503616333,
+      "learning_rate": 1.8676740420939016e-05,
+      "loss": 3.1211,
       "step": 89900
     },
     {
       "epoch": 9.69811320754717,
-      "grad_norm": 0.8823761343955994,
-      "learning_rate": 1.8359417161359955e-05,
-      "loss": 3.1253,
+      "grad_norm": 0.8794217705726624,
+      "learning_rate": 1.8352941176470587e-05,
+      "loss": 3.1268,
       "step": 89950
     },
     {
       "epoch": 9.703504043126685,
-      "grad_norm": 0.8819106817245483,
-      "learning_rate": 1.8035617916891526e-05,
-      "loss": 3.1261,
+      "grad_norm": 0.8845636248588562,
+      "learning_rate": 1.8029141932002158e-05,
+      "loss": 3.1292,
       "step": 90000
     },
     {
       "epoch": 9.703504043126685,
-      "eval_accuracy": 0.3946354375087805,
-      "eval_loss": 3.299715518951416,
-      "eval_runtime": 183.4384,
-      "eval_samples_per_second": 98.186,
-      "eval_steps_per_second": 6.138,
+      "eval_accuracy": 0.39450353290297374,
+      "eval_loss": 3.300025701522827,
+      "eval_runtime": 181.3342,
+      "eval_samples_per_second": 99.325,
+      "eval_steps_per_second": 6.21,
       "step": 90000
     },
     {
       "epoch": 9.7088948787062,
-      "grad_norm": 0.8763957023620605,
-      "learning_rate": 1.7711818672423097e-05,
-      "loss": 3.1182,
+      "grad_norm": 0.8940553665161133,
+      "learning_rate": 1.770534268753373e-05,
+      "loss": 3.1232,
       "step": 90050
     },
     {
       "epoch": 9.714285714285714,
-      "grad_norm": 0.8612681031227112,
-      "learning_rate": 1.7388019427954664e-05,
-      "loss": 3.1227,
+      "grad_norm": 0.9007864594459534,
+      "learning_rate": 1.7381543443065297e-05,
+      "loss": 3.1254,
       "step": 90100
     },
     {
       "epoch": 9.719676549865229,
-      "grad_norm": 0.8781333565711975,
-      "learning_rate": 1.706422018348624e-05,
-      "loss": 3.1294,
+      "grad_norm": 0.9010265469551086,
+      "learning_rate": 1.7057744198596867e-05,
+      "loss": 3.1325,
       "step": 90150
     },
     {
       "epoch": 9.725067385444744,
-      "grad_norm": 0.865515410900116,
-      "learning_rate": 1.674042093901781e-05,
-      "loss": 3.1297,
+      "grad_norm": 0.8837758302688599,
+      "learning_rate": 1.673394495412844e-05,
+      "loss": 3.1332,
       "step": 90200
     },
     {
       "epoch": 9.730458221024259,
-      "grad_norm": 0.8704222440719604,
-      "learning_rate": 1.6423097679438745e-05,
-      "loss": 3.1347,
+      "grad_norm": 0.881119966506958,
+      "learning_rate": 1.641014570966001e-05,
+      "loss": 3.1379,
       "step": 90250
     },
     {
       "epoch": 9.735849056603774,
-      "grad_norm": 0.8565595746040344,
-      "learning_rate": 1.6099298434970315e-05,
-      "loss": 3.117,
+      "grad_norm": 0.8747886419296265,
+      "learning_rate": 1.608634646519158e-05,
+      "loss": 3.1203,
       "step": 90300
     },
     {
       "epoch": 9.74123989218329,
-      "grad_norm": 0.8548470735549927,
-      "learning_rate": 1.577549919050189e-05,
-      "loss": 3.1338,
+      "grad_norm": 0.8446343541145325,
+      "learning_rate": 1.576254722072315e-05,
+      "loss": 3.1353,
       "step": 90350
     },
     {
       "epoch": 9.746630727762803,
-      "grad_norm": 0.8468580842018127,
-      "learning_rate": 1.5451699946033457e-05,
-      "loss": 3.1301,
+      "grad_norm": 0.8257328867912292,
+      "learning_rate": 1.5438747976254722e-05,
+      "loss": 3.1347,
       "step": 90400
     },
     {
       "epoch": 9.752021563342318,
-      "grad_norm": 0.9051481485366821,
-      "learning_rate": 1.5127900701565028e-05,
-      "loss": 3.1297,
+      "grad_norm": 0.8903543949127197,
+      "learning_rate": 1.5114948731786291e-05,
+      "loss": 3.1316,
       "step": 90450
     },
     {
       "epoch": 9.757412398921833,
-      "grad_norm": 0.8959400057792664,
-      "learning_rate": 1.4804101457096599e-05,
-      "loss": 3.1377,
+      "grad_norm": 0.8656268119812012,
+      "learning_rate": 1.4791149487317862e-05,
+      "loss": 3.1412,
       "step": 90500
     },
     {
       "epoch": 9.762803234501348,
-      "grad_norm": 0.9014003872871399,
-      "learning_rate": 1.448030221262817e-05,
-      "loss": 3.1243,
+      "grad_norm": 0.8959280848503113,
+      "learning_rate": 1.4467350242849431e-05,
+      "loss": 3.1269,
       "step": 90550
     },
     {
       "epoch": 9.768194070080863,
-      "grad_norm": 0.8818705677986145,
-      "learning_rate": 1.4156502968159739e-05,
-      "loss": 3.1397,
+      "grad_norm": 0.8403332829475403,
+      "learning_rate": 1.4143550998381002e-05,
+      "loss": 3.1427,
       "step": 90600
     },
     {
       "epoch": 9.773584905660378,
-      "grad_norm": 0.8922802209854126,
-      "learning_rate": 1.383270372369131e-05,
-      "loss": 3.1233,
+      "grad_norm": 0.8411329984664917,
+      "learning_rate": 1.3819751753912573e-05,
+      "loss": 3.127,
       "step": 90650
     },
     {
       "epoch": 9.778975741239892,
-      "grad_norm": 0.9080004096031189,
-      "learning_rate": 1.3508904479222879e-05,
-      "loss": 3.1246,
+      "grad_norm": 0.8733254075050354,
+      "learning_rate": 1.3495952509444142e-05,
+      "loss": 3.1284,
       "step": 90700
     },
     {
       "epoch": 9.784366576819407,
-      "grad_norm": 0.9191465973854065,
-      "learning_rate": 1.3185105234754452e-05,
-      "loss": 3.1084,
+      "grad_norm": 0.8979040384292603,
+      "learning_rate": 1.3172153264975715e-05,
+      "loss": 3.1116,
       "step": 90750
     },
     {
       "epoch": 9.789757412398922,
-      "grad_norm": 0.9019896984100342,
-      "learning_rate": 1.2861305990286023e-05,
-      "loss": 3.1193,
+      "grad_norm": 0.9167905449867249,
+      "learning_rate": 1.2848354020507286e-05,
+      "loss": 3.1235,
       "step": 90800
     },
     {
       "epoch": 9.795148247978437,
-      "grad_norm": 0.8658702373504639,
-      "learning_rate": 1.2537506745817592e-05,
-      "loss": 3.1277,
+      "grad_norm": 0.8761138319969177,
+      "learning_rate": 1.2524554776038855e-05,
+      "loss": 3.1317,
       "step": 90850
     },
     {
       "epoch": 9.800539083557952,
-      "grad_norm": 0.8594561219215393,
-      "learning_rate": 1.2213707501349163e-05,
-      "loss": 3.1363,
+      "grad_norm": 0.8663395047187805,
+      "learning_rate": 1.2200755531570426e-05,
+      "loss": 3.1379,
       "step": 90900
     },
     {
       "epoch": 9.805929919137466,
-      "grad_norm": 0.8785389065742493,
-      "learning_rate": 1.1889908256880732e-05,
-      "loss": 3.1238,
+      "grad_norm": 0.9014497399330139,
+      "learning_rate": 1.1876956287101995e-05,
+      "loss": 3.1271,
       "step": 90950
     },
     {
       "epoch": 9.81132075471698,
-      "grad_norm": 0.8855112195014954,
-      "learning_rate": 1.1566109012412303e-05,
-      "loss": 3.1272,
+      "grad_norm": 0.8891124129295349,
+      "learning_rate": 1.1553157042633566e-05,
+      "loss": 3.1312,
       "step": 91000
     },
     {
       "epoch": 9.81132075471698,
-      "eval_accuracy": 0.3948550249950536,
-      "eval_loss": 3.297788381576538,
-      "eval_runtime": 183.5523,
-      "eval_samples_per_second": 98.125,
-      "eval_steps_per_second": 6.134,
+      "eval_accuracy": 0.39473767987621394,
+      "eval_loss": 3.2979013919830322,
+      "eval_runtime": 181.3991,
+      "eval_samples_per_second": 99.289,
+      "eval_steps_per_second": 6.207,
       "step": 91000
     },
     {
       "epoch": 9.816711590296496,
-      "grad_norm": 0.868584394454956,
-      "learning_rate": 1.1242309767943874e-05,
-      "loss": 3.1144,
+      "grad_norm": 0.869036853313446,
+      "learning_rate": 1.1229357798165137e-05,
+      "loss": 3.1186,
       "step": 91050
     },
     {
       "epoch": 9.822102425876011,
-      "grad_norm": 0.8530396819114685,
-      "learning_rate": 1.0918510523475443e-05,
-      "loss": 3.1235,
+      "grad_norm": 0.8199326395988464,
+      "learning_rate": 1.0905558553696706e-05,
+      "loss": 3.1268,
       "step": 91100
     },
     {
       "epoch": 9.827493261455526,
-      "grad_norm": 0.8684161901473999,
-      "learning_rate": 1.0594711279007015e-05,
-      "loss": 3.1208,
+      "grad_norm": 0.843533992767334,
+      "learning_rate": 1.0581759309228278e-05,
+      "loss": 3.1246,
       "step": 91150
     },
     {
       "epoch": 9.832884097035041,
-      "grad_norm": 0.8899792432785034,
-      "learning_rate": 1.0270912034538586e-05,
-      "loss": 3.1213,
+      "grad_norm": 0.854178249835968,
+      "learning_rate": 1.025796006475985e-05,
+      "loss": 3.1262,
       "step": 91200
     },
     {
       "epoch": 9.838274932614555,
-      "grad_norm": 0.8613511919975281,
-      "learning_rate": 9.947112790070156e-06,
-      "loss": 3.1172,
+      "grad_norm": 0.8823135495185852,
+      "learning_rate": 9.934160820291419e-06,
+      "loss": 3.1192,
       "step": 91250
     },
     {
       "epoch": 9.84366576819407,
-      "grad_norm": 0.8709462881088257,
-      "learning_rate": 9.623313545601726e-06,
-      "loss": 3.1474,
+      "grad_norm": 0.8559948205947876,
+      "learning_rate": 9.61036157582299e-06,
+      "loss": 3.1506,
       "step": 91300
     },
     {
       "epoch": 9.849056603773585,
-      "grad_norm": 0.896946132183075,
-      "learning_rate": 9.299514301133296e-06,
-      "loss": 3.1258,
+      "grad_norm": 0.8760740160942078,
+      "learning_rate": 9.286562331354559e-06,
+      "loss": 3.1271,
       "step": 91350
     },
     {
       "epoch": 9.8544474393531,
-      "grad_norm": 0.8559094071388245,
-      "learning_rate": 8.975715056664867e-06,
-      "loss": 3.1215,
+      "grad_norm": 0.8908193707466125,
+      "learning_rate": 8.96276308688613e-06,
+      "loss": 3.1251,
       "step": 91400
     },
     {
       "epoch": 9.859838274932615,
-      "grad_norm": 0.8676149249076843,
-      "learning_rate": 8.651915812196437e-06,
-      "loss": 3.1358,
+      "grad_norm": 0.8712329268455505,
+      "learning_rate": 8.6389638424177e-06,
+      "loss": 3.1383,
       "step": 91450
     },
     {
       "epoch": 9.865229110512129,
-      "grad_norm": 0.8665869235992432,
-      "learning_rate": 8.328116567728008e-06,
-      "loss": 3.0967,
+      "grad_norm": 0.858143150806427,
+      "learning_rate": 8.315164597949271e-06,
+      "loss": 3.0999,
       "step": 91500
     },
     {
       "epoch": 9.870619946091644,
-      "grad_norm": 0.8311969637870789,
-      "learning_rate": 8.004317323259577e-06,
-      "loss": 3.0991,
+      "grad_norm": 0.8412820100784302,
+      "learning_rate": 7.99136535348084e-06,
+      "loss": 3.1016,
       "step": 91550
     },
     {
       "epoch": 9.876010781671159,
-      "grad_norm": 0.8719154000282288,
-      "learning_rate": 7.680518078791148e-06,
-      "loss": 3.12,
+      "grad_norm": 0.897463858127594,
+      "learning_rate": 7.667566109012411e-06,
+      "loss": 3.1231,
       "step": 91600
     },
     {
       "epoch": 9.881401617250674,
-      "grad_norm": 0.8440600633621216,
-      "learning_rate": 7.356718834322719e-06,
-      "loss": 3.1188,
+      "grad_norm": 0.8523684144020081,
+      "learning_rate": 7.343766864543982e-06,
+      "loss": 3.1223,
       "step": 91650
     },
     {
       "epoch": 9.88679245283019,
-      "grad_norm": 0.8644077777862549,
-      "learning_rate": 7.03291958985429e-06,
-      "loss": 3.1107,
+      "grad_norm": 0.8708697557449341,
+      "learning_rate": 7.019967620075552e-06,
+      "loss": 3.1144,
       "step": 91700
     },
     {
       "epoch": 9.892183288409704,
-      "grad_norm": 0.8675840497016907,
-      "learning_rate": 6.70912034538586e-06,
-      "loss": 3.1316,
+      "grad_norm": 0.8736843466758728,
+      "learning_rate": 6.696168375607123e-06,
+      "loss": 3.1342,
       "step": 91750
     },
     {
       "epoch": 9.897574123989218,
-      "grad_norm": 0.873511552810669,
-      "learning_rate": 6.38532110091743e-06,
-      "loss": 3.1241,
+      "grad_norm": 0.8864836692810059,
+      "learning_rate": 6.372369131138693e-06,
+      "loss": 3.1278,
       "step": 91800
     },
     {
       "epoch": 9.902964959568733,
-      "grad_norm": 0.9048805236816406,
-      "learning_rate": 6.061521856449001e-06,
-      "loss": 3.1268,
+      "grad_norm": 0.9168736338615417,
+      "learning_rate": 6.048569886670264e-06,
+      "loss": 3.1294,
       "step": 91850
     },
     {
       "epoch": 9.908355795148248,
-      "grad_norm": 0.8429412245750427,
-      "learning_rate": 5.737722611980571e-06,
-      "loss": 3.1232,
+      "grad_norm": 0.8574074506759644,
+      "learning_rate": 5.724770642201834e-06,
+      "loss": 3.1259,
       "step": 91900
     },
     {
       "epoch": 9.913746630727763,
-      "grad_norm": 0.8978517651557922,
-      "learning_rate": 5.413923367512142e-06,
-      "loss": 3.1066,
+      "grad_norm": 0.8834133148193359,
+      "learning_rate": 5.400971397733404e-06,
+      "loss": 3.1099,
       "step": 91950
     },
     {
       "epoch": 9.919137466307278,
-      "grad_norm": 0.8608512282371521,
-      "learning_rate": 5.090124123043712e-06,
-      "loss": 3.0993,
+      "grad_norm": 0.8338649272918701,
+      "learning_rate": 5.077172153264975e-06,
+      "loss": 3.1057,
       "step": 92000
     },
     {
       "epoch": 9.919137466307278,
-      "eval_accuracy": 0.39501604857479466,
-      "eval_loss": 3.2966372966766357,
-      "eval_runtime": 183.5216,
-      "eval_samples_per_second": 98.141,
-      "eval_steps_per_second": 6.136,
+      "eval_accuracy": 0.3949261926365721,
+      "eval_loss": 3.2965760231018066,
+      "eval_runtime": 181.4796,
+      "eval_samples_per_second": 99.245,
+      "eval_steps_per_second": 6.205,
       "step": 92000
     },
     {
       "epoch": 9.924528301886792,
-      "grad_norm": 0.8535104990005493,
-      "learning_rate": 4.766324878575283e-06,
-      "loss": 3.1249,
+      "grad_norm": 0.869513750076294,
+      "learning_rate": 4.753372908796546e-06,
+      "loss": 3.1257,
       "step": 92050
     },
     {
       "epoch": 9.929919137466307,
-      "grad_norm": 0.9014983177185059,
-      "learning_rate": 4.442525634106853e-06,
-      "loss": 3.1135,
+      "grad_norm": 0.8724406361579895,
+      "learning_rate": 4.429573664328116e-06,
+      "loss": 3.1165,
       "step": 92100
     },
     {
       "epoch": 9.935309973045822,
-      "grad_norm": 0.8669118285179138,
-      "learning_rate": 4.118726389638424e-06,
-      "loss": 3.1419,
+      "grad_norm": 0.826997697353363,
+      "learning_rate": 4.105774419859686e-06,
+      "loss": 3.1427,
       "step": 92150
     },
     {
       "epoch": 9.940700808625337,
-      "grad_norm": 0.8581469058990479,
-      "learning_rate": 3.7949271451699944e-06,
-      "loss": 3.1323,
+      "grad_norm": 0.8791482448577881,
+      "learning_rate": 3.781975175391257e-06,
+      "loss": 3.1351,
       "step": 92200
     },
     {
       "epoch": 9.946091644204852,
-      "grad_norm": 0.8778931498527527,
-      "learning_rate": 3.4711279007015644e-06,
-      "loss": 3.124,
+      "grad_norm": 0.9015523791313171,
+      "learning_rate": 3.4581759309228275e-06,
+      "loss": 3.1276,
       "step": 92250
     },
     {
       "epoch": 9.951482479784367,
-      "grad_norm": 0.9205542206764221,
-      "learning_rate": 3.1473286562331353e-06,
-      "loss": 3.1097,
+      "grad_norm": 0.9038318991661072,
+      "learning_rate": 3.134376686454398e-06,
+      "loss": 3.1148,
       "step": 92300
     },
     {
       "epoch": 9.95687331536388,
-      "grad_norm": 0.9244022965431213,
-      "learning_rate": 2.8235294117647054e-06,
-      "loss": 3.1215,
+      "grad_norm": 0.9109078049659729,
+      "learning_rate": 2.8105774419859684e-06,
+      "loss": 3.1241,
       "step": 92350
     },
     {
       "epoch": 9.962264150943396,
-      "grad_norm": 0.858595073223114,
-      "learning_rate": 2.4997301672962763e-06,
-      "loss": 3.1055,
+      "grad_norm": 0.8301215767860413,
+      "learning_rate": 2.486778197517539e-06,
+      "loss": 3.1077,
       "step": 92400
     },
     {
       "epoch": 9.967654986522911,
-      "grad_norm": 0.8582491874694824,
-      "learning_rate": 2.1759309228278467e-06,
-      "loss": 3.1251,
+      "grad_norm": 0.8766610622406006,
+      "learning_rate": 2.1629789530491094e-06,
+      "loss": 3.1269,
       "step": 92450
     },
     {
       "epoch": 9.973045822102426,
-      "grad_norm": 0.8598997592926025,
-      "learning_rate": 1.852131678359417e-06,
-      "loss": 3.12,
+      "grad_norm": 0.8508982062339783,
+      "learning_rate": 1.8391797085806798e-06,
+      "loss": 3.1222,
       "step": 92500
     },
     {
       "epoch": 9.978436657681941,
-      "grad_norm": 0.8376339077949524,
-      "learning_rate": 1.5283324338909875e-06,
-      "loss": 3.1245,
+      "grad_norm": 0.8604896068572998,
+      "learning_rate": 1.5153804641122503e-06,
+      "loss": 3.1275,
       "step": 92550
     },
     {
       "epoch": 9.983827493261456,
-      "grad_norm": 0.8226578831672668,
-      "learning_rate": 1.204533189422558e-06,
-      "loss": 3.1198,
+      "grad_norm": 0.8426108360290527,
+      "learning_rate": 1.1915812196438208e-06,
+      "loss": 3.1217,
       "step": 92600
     },
     {
       "epoch": 9.98921832884097,
-      "grad_norm": 0.8843859434127808,
-      "learning_rate": 8.807339449541284e-07,
-      "loss": 3.1088,
+      "grad_norm": 0.8589921593666077,
+      "learning_rate": 8.677819751753911e-07,
+      "loss": 3.1115,
       "step": 92650
     },
     {
       "epoch": 9.994609164420485,
-      "grad_norm": 0.886563777923584,
-      "learning_rate": 5.569347004856989e-07,
-      "loss": 3.131,
+      "grad_norm": 0.8870005011558533,
+      "learning_rate": 5.439827307069617e-07,
+      "loss": 3.1345,
       "step": 92700
     },
     {
       "epoch": 10.0,
-      "grad_norm": 1.917934536933899,
-      "learning_rate": 2.3313545601726927e-07,
-      "loss": 3.1285,
+      "grad_norm": 1.8870831727981567,
+      "learning_rate": 2.201834862385321e-07,
+      "loss": 3.1305,
       "step": 92750
     }
   ],