diff --git "a/checkpoint-50000/trainer_state.json" "b/checkpoint-50000/trainer_state.json"
--- "a/checkpoint-50000/trainer_state.json"
+++ "b/checkpoint-50000/trainer_state.json"
@@ -1,5 +1,5 @@
 {
-  "best_metric": 3.397858142852783,
+  "best_metric": 3.4002845287323,
   "best_model_checkpoint": "/scratch/cl5625/exceptions/models/100M_low_2000_634/checkpoint-50000",
   "epoch": 5.390835579514825,
   "eval_steps": 1000,
@@ -10,7452 +10,7452 @@
   "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
     }
   ],