diff --git "a/checkpoint-90000/trainer_state.json" "b/checkpoint-90000/trainer_state.json"
--- "a/checkpoint-90000/trainer_state.json"
+++ "b/checkpoint-90000/trainer_state.json"
@@ -1,5 +1,5 @@
 {
-  "best_metric": 3.3018898963928223,
+  "best_metric": 3.3053202629089355,
   "best_model_checkpoint": "/scratch/cl5625/exceptions/models/100M_low_2000_8397/checkpoint-90000",
   "epoch": 9.703504043126685,
   "eval_steps": 1000,
@@ -10,13412 +10,13412 @@
   "log_history": [
     {
       "epoch": 0.005390835579514825,
-      "grad_norm": 1.5813435316085815,
+      "grad_norm": 2.569659471511841,
       "learning_rate": 0.0003,
-      "loss": 8.6318,
+      "loss": 8.6331,
       "step": 50
     },
     {
       "epoch": 0.01078167115902965,
-      "grad_norm": 1.0082982778549194,
+      "grad_norm": 3.8161051273345947,
       "learning_rate": 0.0006,
-      "loss": 6.8697,
+      "loss": 6.8745,
       "step": 100
     },
     {
       "epoch": 0.016172506738544475,
-      "grad_norm": 2.0049550533294678,
+      "grad_norm": 4.1664886474609375,
       "learning_rate": 0.0005996762007555315,
-      "loss": 6.4573,
+      "loss": 6.4574,
       "step": 150
     },
     {
       "epoch": 0.0215633423180593,
-      "grad_norm": 2.044867515563965,
+      "grad_norm": 1.2024744749069214,
       "learning_rate": 0.000599352401511063,
-      "loss": 6.239,
+      "loss": 6.2266,
       "step": 200
     },
     {
       "epoch": 0.026954177897574125,
-      "grad_norm": 1.7265796661376953,
+      "grad_norm": 2.2366983890533447,
       "learning_rate": 0.0005990286022665946,
-      "loss": 6.0844,
+      "loss": 6.0998,
       "step": 250
     },
     {
       "epoch": 0.03234501347708895,
-      "grad_norm": 3.774665594100952,
+      "grad_norm": 3.547074317932129,
       "learning_rate": 0.0005987048030221263,
-      "loss": 5.9817,
+      "loss": 5.9913,
       "step": 300
     },
     {
       "epoch": 0.03773584905660377,
-      "grad_norm": 1.1093494892120361,
+      "grad_norm": 1.4036786556243896,
       "learning_rate": 0.0005983810037776578,
-      "loss": 5.8816,
+      "loss": 5.8857,
       "step": 350
     },
     {
       "epoch": 0.0431266846361186,
-      "grad_norm": 1.401320457458496,
+      "grad_norm": 1.0421242713928223,
       "learning_rate": 0.0005980572045331894,
-      "loss": 5.8369,
+      "loss": 5.8352,
       "step": 400
     },
     {
       "epoch": 0.04851752021563342,
-      "grad_norm": 1.4539614915847778,
+      "grad_norm": 0.9555350542068481,
       "learning_rate": 0.0005977334052887209,
-      "loss": 5.756,
+      "loss": 5.7695,
       "step": 450
     },
     {
       "epoch": 0.05390835579514825,
-      "grad_norm": 1.237868309020996,
+      "grad_norm": 1.0334396362304688,
       "learning_rate": 0.0005974096060442526,
-      "loss": 5.6613,
+      "loss": 5.6656,
       "step": 500
     },
     {
       "epoch": 0.05929919137466307,
-      "grad_norm": 1.7472739219665527,
+      "grad_norm": 2.126458168029785,
       "learning_rate": 0.0005970858067997841,
-      "loss": 5.5746,
+      "loss": 5.5837,
       "step": 550
     },
     {
       "epoch": 0.0646900269541779,
-      "grad_norm": 1.1309614181518555,
+      "grad_norm": 1.156680941581726,
       "learning_rate": 0.0005967620075553157,
-      "loss": 5.5044,
+      "loss": 5.5045,
       "step": 600
     },
     {
       "epoch": 0.07008086253369272,
-      "grad_norm": 1.2399049997329712,
+      "grad_norm": 1.2812466621398926,
       "learning_rate": 0.0005964382083108472,
-      "loss": 5.4345,
+      "loss": 5.4387,
       "step": 650
     },
     {
       "epoch": 0.07547169811320754,
-      "grad_norm": 1.7309972047805786,
+      "grad_norm": 1.8948819637298584,
       "learning_rate": 0.0005961144090663788,
-      "loss": 5.386,
+      "loss": 5.3888,
       "step": 700
     },
     {
       "epoch": 0.08086253369272237,
-      "grad_norm": 1.1939959526062012,
+      "grad_norm": 1.0907261371612549,
       "learning_rate": 0.0005957906098219104,
-      "loss": 5.3088,
+      "loss": 5.3155,
       "step": 750
     },
     {
       "epoch": 0.0862533692722372,
-      "grad_norm": 1.6205377578735352,
+      "grad_norm": 1.7458995580673218,
       "learning_rate": 0.0005954668105774419,
-      "loss": 5.2622,
+      "loss": 5.2655,
       "step": 800
     },
     {
       "epoch": 0.09164420485175202,
-      "grad_norm": 1.3945573568344116,
+      "grad_norm": 1.611615538597107,
       "learning_rate": 0.0005951430113329735,
-      "loss": 5.2251,
+      "loss": 5.2284,
       "step": 850
     },
     {
       "epoch": 0.09703504043126684,
-      "grad_norm": 1.3046308755874634,
+      "grad_norm": 1.189422607421875,
       "learning_rate": 0.0005948192120885051,
-      "loss": 5.1654,
+      "loss": 5.1714,
       "step": 900
     },
     {
       "epoch": 0.10242587601078167,
-      "grad_norm": 1.2131174802780151,
+      "grad_norm": 1.2515290975570679,
       "learning_rate": 0.0005944954128440366,
-      "loss": 5.1448,
+      "loss": 5.1451,
       "step": 950
     },
     {
       "epoch": 0.1078167115902965,
-      "grad_norm": 0.9675126075744629,
+      "grad_norm": 1.3560065031051636,
       "learning_rate": 0.0005941716135995682,
-      "loss": 5.1035,
+      "loss": 5.1027,
       "step": 1000
     },
     {
       "epoch": 0.1078167115902965,
-      "eval_accuracy": 0.22749035406825205,
-      "eval_loss": 5.018069267272949,
-      "eval_runtime": 184.6816,
-      "eval_samples_per_second": 97.525,
-      "eval_steps_per_second": 6.097,
+      "eval_accuracy": 0.22700869581657046,
+      "eval_loss": 5.025312423706055,
+      "eval_runtime": 180.9369,
+      "eval_samples_per_second": 99.543,
+      "eval_steps_per_second": 6.223,
       "step": 1000
     },
     {
       "epoch": 0.11320754716981132,
-      "grad_norm": 1.0149263143539429,
+      "grad_norm": 1.2824947834014893,
       "learning_rate": 0.0005938478143550997,
-      "loss": 5.0486,
+      "loss": 5.0527,
       "step": 1050
     },
     {
       "epoch": 0.11859838274932614,
-      "grad_norm": 1.3123433589935303,
+      "grad_norm": 1.0902217626571655,
       "learning_rate": 0.0005935240151106314,
-      "loss": 5.0207,
+      "loss": 5.0291,
       "step": 1100
     },
     {
       "epoch": 0.12398921832884097,
-      "grad_norm": 0.9050843715667725,
+      "grad_norm": 1.123762607574463,
       "learning_rate": 0.0005932002158661629,
-      "loss": 4.9928,
+      "loss": 4.9996,
       "step": 1150
     },
     {
       "epoch": 0.1293800539083558,
-      "grad_norm": 1.274807095527649,
+      "grad_norm": 1.1804653406143188,
       "learning_rate": 0.0005928764166216945,
-      "loss": 4.9535,
+      "loss": 4.9594,
       "step": 1200
     },
     {
       "epoch": 0.1347708894878706,
-      "grad_norm": 1.00777006149292,
+      "grad_norm": 1.2184960842132568,
       "learning_rate": 0.000592552617377226,
-      "loss": 4.9137,
+      "loss": 4.9153,
       "step": 1250
     },
     {
       "epoch": 0.14016172506738545,
-      "grad_norm": 0.8887598514556885,
+      "grad_norm": 0.9934311509132385,
       "learning_rate": 0.0005922288181327577,
-      "loss": 4.8818,
+      "loss": 4.8838,
       "step": 1300
     },
     {
       "epoch": 0.14555256064690028,
-      "grad_norm": 0.9829823970794678,
+      "grad_norm": 0.9794490933418274,
       "learning_rate": 0.0005919050188882893,
-      "loss": 4.8648,
+      "loss": 4.8729,
       "step": 1350
     },
     {
       "epoch": 0.1509433962264151,
-      "grad_norm": 0.9974054098129272,
+      "grad_norm": 1.2818341255187988,
       "learning_rate": 0.0005915812196438207,
-      "loss": 4.8498,
+      "loss": 4.8554,
       "step": 1400
     },
     {
       "epoch": 0.15633423180592992,
-      "grad_norm": 1.0419261455535889,
+      "grad_norm": 1.0256102085113525,
       "learning_rate": 0.0005912574203993524,
-      "loss": 4.8005,
+      "loss": 4.8053,
       "step": 1450
     },
     {
       "epoch": 0.16172506738544473,
-      "grad_norm": 1.1737334728240967,
+      "grad_norm": 1.174377679824829,
       "learning_rate": 0.0005909336211548839,
-      "loss": 4.8138,
+      "loss": 4.8233,
       "step": 1500
     },
     {
       "epoch": 0.16711590296495957,
-      "grad_norm": 1.0961012840270996,
+      "grad_norm": 0.9294725060462952,
       "learning_rate": 0.0005906098219104155,
-      "loss": 4.7659,
+      "loss": 4.7745,
       "step": 1550
     },
     {
       "epoch": 0.1725067385444744,
-      "grad_norm": 1.1323438882827759,
+      "grad_norm": 0.9997650980949402,
       "learning_rate": 0.000590286022665947,
-      "loss": 4.7663,
+      "loss": 4.7731,
       "step": 1600
     },
     {
       "epoch": 0.1778975741239892,
-      "grad_norm": 0.8736843466758728,
+      "grad_norm": 0.8246453404426575,
       "learning_rate": 0.0005899622234214787,
-      "loss": 4.7157,
+      "loss": 4.7323,
       "step": 1650
     },
     {
       "epoch": 0.18328840970350405,
-      "grad_norm": 0.7891756892204285,
+      "grad_norm": 1.0215401649475098,
       "learning_rate": 0.0005896384241770102,
-      "loss": 4.7099,
+      "loss": 4.7145,
       "step": 1700
     },
     {
       "epoch": 0.18867924528301888,
-      "grad_norm": 1.0397777557373047,
+      "grad_norm": 1.0960590839385986,
       "learning_rate": 0.0005893146249325418,
-      "loss": 4.6729,
+      "loss": 4.678,
       "step": 1750
     },
     {
       "epoch": 0.1940700808625337,
-      "grad_norm": 1.151475429534912,
+      "grad_norm": 0.818347692489624,
       "learning_rate": 0.0005889908256880733,
-      "loss": 4.6804,
+      "loss": 4.6877,
       "step": 1800
     },
     {
       "epoch": 0.19946091644204852,
-      "grad_norm": 1.5584900379180908,
+      "grad_norm": 1.6717796325683594,
       "learning_rate": 0.0005886670264436049,
-      "loss": 4.6534,
+      "loss": 4.6592,
       "step": 1850
     },
     {
       "epoch": 0.20485175202156333,
-      "grad_norm": 0.7592549920082092,
+      "grad_norm": 0.8092234134674072,
       "learning_rate": 0.0005883432271991365,
-      "loss": 4.6368,
+      "loss": 4.6447,
       "step": 1900
     },
     {
       "epoch": 0.21024258760107817,
-      "grad_norm": 1.132956862449646,
+      "grad_norm": 0.9391996264457703,
       "learning_rate": 0.0005880194279546681,
-      "loss": 4.6073,
+      "loss": 4.6133,
       "step": 1950
     },
     {
       "epoch": 0.215633423180593,
-      "grad_norm": 0.904139518737793,
+      "grad_norm": 0.9681597352027893,
       "learning_rate": 0.0005876956287101996,
-      "loss": 4.5922,
+      "loss": 4.5966,
       "step": 2000
     },
     {
       "epoch": 0.215633423180593,
-      "eval_accuracy": 0.26994136329603313,
-      "eval_loss": 4.514337062835693,
-      "eval_runtime": 183.9245,
-      "eval_samples_per_second": 97.926,
-      "eval_steps_per_second": 6.122,
+      "eval_accuracy": 0.2693512694623127,
+      "eval_loss": 4.519053936004639,
+      "eval_runtime": 180.5608,
+      "eval_samples_per_second": 99.75,
+      "eval_steps_per_second": 6.236,
       "step": 2000
     },
     {
       "epoch": 0.2210242587601078,
-      "grad_norm": 0.8630588054656982,
+      "grad_norm": 1.3102365732192993,
       "learning_rate": 0.0005873718294657312,
-      "loss": 4.5736,
+      "loss": 4.5777,
       "step": 2050
     },
     {
       "epoch": 0.22641509433962265,
-      "grad_norm": 0.751631498336792,
+      "grad_norm": 0.8627723455429077,
       "learning_rate": 0.0005870480302212628,
-      "loss": 4.5577,
+      "loss": 4.563,
       "step": 2100
     },
     {
       "epoch": 0.23180592991913745,
-      "grad_norm": 0.9096190333366394,
+      "grad_norm": 0.9546903371810913,
       "learning_rate": 0.0005867242309767943,
-      "loss": 4.5327,
+      "loss": 4.5345,
       "step": 2150
     },
     {
       "epoch": 0.2371967654986523,
-      "grad_norm": 0.9275792241096497,
+      "grad_norm": 0.9777777194976807,
       "learning_rate": 0.0005864004317323259,
-      "loss": 4.513,
+      "loss": 4.5162,
       "step": 2200
     },
     {
       "epoch": 0.24258760107816713,
-      "grad_norm": 0.8501319885253906,
+      "grad_norm": 0.7585428357124329,
       "learning_rate": 0.0005860766324878575,
-      "loss": 4.4957,
+      "loss": 4.5015,
       "step": 2250
     },
     {
       "epoch": 0.24797843665768193,
-      "grad_norm": 0.8335550427436829,
+      "grad_norm": 0.80382239818573,
       "learning_rate": 0.000585752833243389,
-      "loss": 4.505,
+      "loss": 4.51,
       "step": 2300
     },
     {
       "epoch": 0.25336927223719674,
-      "grad_norm": 0.7996338605880737,
+      "grad_norm": 0.8545572757720947,
       "learning_rate": 0.0005854290339989206,
-      "loss": 4.4554,
+      "loss": 4.4646,
       "step": 2350
     },
     {
       "epoch": 0.2587601078167116,
-      "grad_norm": 1.0249171257019043,
+      "grad_norm": 0.9123208522796631,
       "learning_rate": 0.0005851052347544521,
-      "loss": 4.4629,
+      "loss": 4.4689,
       "step": 2400
     },
     {
       "epoch": 0.2641509433962264,
-      "grad_norm": 0.9536342620849609,
+      "grad_norm": 1.0219218730926514,
       "learning_rate": 0.0005847814355099838,
-      "loss": 4.4412,
+      "loss": 4.4448,
       "step": 2450
     },
     {
       "epoch": 0.2695417789757412,
-      "grad_norm": 1.2684301137924194,
+      "grad_norm": 1.1517032384872437,
       "learning_rate": 0.0005844576362655154,
-      "loss": 4.4274,
+      "loss": 4.4302,
       "step": 2500
     },
     {
       "epoch": 0.2749326145552561,
-      "grad_norm": 0.8013315796852112,
+      "grad_norm": 0.6815633177757263,
       "learning_rate": 0.0005841338370210469,
-      "loss": 4.4099,
+      "loss": 4.4157,
       "step": 2550
     },
     {
       "epoch": 0.2803234501347709,
-      "grad_norm": 0.8502827882766724,
+      "grad_norm": 0.8250149488449097,
       "learning_rate": 0.0005838100377765785,
-      "loss": 4.3802,
+      "loss": 4.3869,
       "step": 2600
     },
     {
       "epoch": 0.2857142857142857,
-      "grad_norm": 0.8461282849311829,
+      "grad_norm": 0.8895910382270813,
       "learning_rate": 0.0005834862385321101,
-      "loss": 4.3947,
+      "loss": 4.3982,
       "step": 2650
     },
     {
       "epoch": 0.29110512129380056,
-      "grad_norm": 1.0242016315460205,
+      "grad_norm": 0.8321819305419922,
       "learning_rate": 0.0005831624392876417,
-      "loss": 4.3976,
+      "loss": 4.4025,
       "step": 2700
     },
     {
       "epoch": 0.29649595687331537,
-      "grad_norm": 0.8631992936134338,
+      "grad_norm": 0.9302955865859985,
       "learning_rate": 0.0005828386400431731,
-      "loss": 4.3521,
+      "loss": 4.3627,
       "step": 2750
     },
     {
       "epoch": 0.3018867924528302,
-      "grad_norm": 0.7447966933250427,
+      "grad_norm": 0.6355003118515015,
       "learning_rate": 0.0005825148407987048,
-      "loss": 4.3323,
+      "loss": 4.3397,
       "step": 2800
     },
     {
       "epoch": 0.30727762803234504,
-      "grad_norm": 0.7234294414520264,
+      "grad_norm": 0.819584310054779,
       "learning_rate": 0.0005821910415542363,
-      "loss": 4.3364,
+      "loss": 4.3423,
       "step": 2850
     },
     {
       "epoch": 0.31266846361185985,
-      "grad_norm": 0.8805587887763977,
+      "grad_norm": 0.8713323473930359,
       "learning_rate": 0.0005818672423097679,
-      "loss": 4.3307,
+      "loss": 4.3326,
       "step": 2900
     },
     {
       "epoch": 0.31805929919137466,
-      "grad_norm": 0.8215044140815735,
+      "grad_norm": 0.9243606328964233,
       "learning_rate": 0.0005815434430652994,
-      "loss": 4.3141,
+      "loss": 4.3229,
       "step": 2950
     },
     {
       "epoch": 0.32345013477088946,
-      "grad_norm": 0.7359305620193481,
+      "grad_norm": 0.7551540732383728,
       "learning_rate": 0.0005812196438208311,
-      "loss": 4.311,
+      "loss": 4.3193,
       "step": 3000
     },
     {
       "epoch": 0.32345013477088946,
-      "eval_accuracy": 0.298171882566655,
-      "eval_loss": 4.2451090812683105,
-      "eval_runtime": 183.7967,
-      "eval_samples_per_second": 97.994,
-      "eval_steps_per_second": 6.126,
+      "eval_accuracy": 0.29709339400220847,
+      "eval_loss": 4.250678539276123,
+      "eval_runtime": 180.5928,
+      "eval_samples_per_second": 99.733,
+      "eval_steps_per_second": 6.235,
       "step": 3000
     },
     {
       "epoch": 0.3288409703504043,
-      "grad_norm": 0.770511269569397,
+      "grad_norm": 0.7257675528526306,
       "learning_rate": 0.0005808958445763626,
-      "loss": 4.3027,
+      "loss": 4.3107,
       "step": 3050
     },
     {
       "epoch": 0.33423180592991913,
-      "grad_norm": 0.677880048751831,
+      "grad_norm": 0.735853374004364,
       "learning_rate": 0.0005805720453318942,
-      "loss": 4.3083,
+      "loss": 4.3168,
       "step": 3100
     },
     {
       "epoch": 0.33962264150943394,
-      "grad_norm": 0.6858367919921875,
+      "grad_norm": 0.6535053253173828,
       "learning_rate": 0.0005802482460874257,
-      "loss": 4.2657,
+      "loss": 4.2723,
       "step": 3150
     },
     {
       "epoch": 0.3450134770889488,
-      "grad_norm": 0.6485952138900757,
+      "grad_norm": 0.6910092830657959,
       "learning_rate": 0.0005799244468429573,
-      "loss": 4.261,
+      "loss": 4.2703,
       "step": 3200
     },
     {
       "epoch": 0.3504043126684636,
-      "grad_norm": 0.706744372844696,
+      "grad_norm": 0.8999305367469788,
       "learning_rate": 0.0005796006475984889,
-      "loss": 4.2609,
+      "loss": 4.2706,
       "step": 3250
     },
     {
       "epoch": 0.3557951482479784,
-      "grad_norm": 0.7495291233062744,
+      "grad_norm": 0.8477680683135986,
       "learning_rate": 0.0005792768483540205,
-      "loss": 4.2458,
+      "loss": 4.2529,
       "step": 3300
     },
     {
       "epoch": 0.3611859838274933,
-      "grad_norm": 0.6684705018997192,
+      "grad_norm": 0.6384448409080505,
       "learning_rate": 0.000578953049109552,
-      "loss": 4.2625,
+      "loss": 4.2699,
       "step": 3350
     },
     {
       "epoch": 0.3665768194070081,
-      "grad_norm": 0.6015046834945679,
+      "grad_norm": 0.8196437954902649,
       "learning_rate": 0.0005786292498650836,
-      "loss": 4.2344,
+      "loss": 4.2396,
       "step": 3400
     },
     {
       "epoch": 0.3719676549865229,
-      "grad_norm": 0.6746869087219238,
+      "grad_norm": 0.727353572845459,
       "learning_rate": 0.0005783054506206152,
-      "loss": 4.2115,
+      "loss": 4.2167,
       "step": 3450
     },
     {
       "epoch": 0.37735849056603776,
-      "grad_norm": 0.6800693273544312,
+      "grad_norm": 0.740081787109375,
       "learning_rate": 0.0005779816513761467,
-      "loss": 4.2238,
+      "loss": 4.2268,
       "step": 3500
     },
     {
       "epoch": 0.38274932614555257,
-      "grad_norm": 0.8763014674186707,
+      "grad_norm": 0.7981793284416199,
       "learning_rate": 0.0005776578521316782,
-      "loss": 4.2176,
+      "loss": 4.2238,
       "step": 3550
     },
     {
       "epoch": 0.3881401617250674,
-      "grad_norm": 0.6071361899375916,
+      "grad_norm": 0.622768759727478,
       "learning_rate": 0.0005773340528872099,
-      "loss": 4.2172,
+      "loss": 4.2219,
       "step": 3600
     },
     {
       "epoch": 0.3935309973045822,
-      "grad_norm": 0.8424677848815918,
+      "grad_norm": 0.7501055598258972,
       "learning_rate": 0.0005770102536427414,
-      "loss": 4.1909,
+      "loss": 4.1952,
       "step": 3650
     },
     {
       "epoch": 0.39892183288409705,
-      "grad_norm": 0.7024722099304199,
+      "grad_norm": 0.7931749224662781,
       "learning_rate": 0.000576686454398273,
-      "loss": 4.193,
+      "loss": 4.201,
       "step": 3700
     },
     {
       "epoch": 0.40431266846361186,
-      "grad_norm": 0.6678298711776733,
+      "grad_norm": 0.7232137322425842,
       "learning_rate": 0.0005763626551538045,
-      "loss": 4.1954,
+      "loss": 4.2026,
       "step": 3750
     },
     {
       "epoch": 0.40970350404312667,
-      "grad_norm": 0.591792643070221,
+      "grad_norm": 0.662381112575531,
       "learning_rate": 0.0005760388559093362,
-      "loss": 4.1897,
+      "loss": 4.1963,
       "step": 3800
     },
     {
       "epoch": 0.41509433962264153,
-      "grad_norm": 0.7370854020118713,
+      "grad_norm": 0.6325182318687439,
       "learning_rate": 0.0005757150566648678,
-      "loss": 4.1703,
+      "loss": 4.1772,
       "step": 3850
     },
     {
       "epoch": 0.42048517520215634,
-      "grad_norm": 0.8055949807167053,
+      "grad_norm": 0.7051071524620056,
       "learning_rate": 0.0005753912574203993,
-      "loss": 4.1666,
+      "loss": 4.1742,
       "step": 3900
     },
     {
       "epoch": 0.42587601078167114,
-      "grad_norm": 0.6214135885238647,
+      "grad_norm": 0.7175087928771973,
       "learning_rate": 0.0005750674581759309,
-      "loss": 4.1586,
+      "loss": 4.1655,
       "step": 3950
     },
     {
       "epoch": 0.431266846361186,
-      "grad_norm": 0.7452210783958435,
+      "grad_norm": 0.7469754815101624,
       "learning_rate": 0.0005747436589314624,
-      "loss": 4.159,
+      "loss": 4.1641,
       "step": 4000
     },
     {
       "epoch": 0.431266846361186,
-      "eval_accuracy": 0.31298409615085193,
-      "eval_loss": 4.08732271194458,
-      "eval_runtime": 183.8347,
-      "eval_samples_per_second": 97.974,
-      "eval_steps_per_second": 6.125,
+      "eval_accuracy": 0.3123611891450115,
+      "eval_loss": 4.092988014221191,
+      "eval_runtime": 180.6843,
+      "eval_samples_per_second": 99.682,
+      "eval_steps_per_second": 6.232,
       "step": 4000
     },
     {
       "epoch": 0.4366576819407008,
-      "grad_norm": 0.5454065203666687,
+      "grad_norm": 0.601862370967865,
       "learning_rate": 0.0005744198596869941,
-      "loss": 4.1509,
+      "loss": 4.1573,
       "step": 4050
     },
     {
       "epoch": 0.4420485175202156,
-      "grad_norm": 0.6480352282524109,
+      "grad_norm": 0.6490374803543091,
       "learning_rate": 0.0005740960604425255,
-      "loss": 4.1515,
+      "loss": 4.1595,
       "step": 4100
     },
     {
       "epoch": 0.4474393530997305,
-      "grad_norm": 0.7491263151168823,
+      "grad_norm": 0.7015263438224792,
       "learning_rate": 0.0005737722611980572,
-      "loss": 4.1543,
+      "loss": 4.1601,
       "step": 4150
     },
     {
       "epoch": 0.4528301886792453,
-      "grad_norm": 0.5414714217185974,
+      "grad_norm": 0.6764467358589172,
       "learning_rate": 0.0005734484619535887,
-      "loss": 4.1397,
+      "loss": 4.1445,
       "step": 4200
     },
     {
       "epoch": 0.4582210242587601,
-      "grad_norm": 0.735364556312561,
+      "grad_norm": 0.6843761801719666,
       "learning_rate": 0.0005731246627091203,
-      "loss": 4.1231,
+      "loss": 4.1272,
       "step": 4250
     },
     {
       "epoch": 0.4636118598382749,
-      "grad_norm": 0.7785477042198181,
+      "grad_norm": 0.8332685828208923,
       "learning_rate": 0.0005728008634646518,
-      "loss": 4.1242,
+      "loss": 4.1243,
       "step": 4300
     },
     {
       "epoch": 0.46900269541778977,
-      "grad_norm": 0.6108700633049011,
+      "grad_norm": 0.5701176524162292,
       "learning_rate": 0.0005724770642201835,
-      "loss": 4.118,
+      "loss": 4.1218,
       "step": 4350
     },
     {
       "epoch": 0.4743935309973046,
-      "grad_norm": 0.6415695548057556,
+      "grad_norm": 0.6857277154922485,
       "learning_rate": 0.000572153264975715,
-      "loss": 4.1178,
+      "loss": 4.1232,
       "step": 4400
     },
     {
       "epoch": 0.4797843665768194,
-      "grad_norm": 0.8770371675491333,
+      "grad_norm": 0.7610508799552917,
       "learning_rate": 0.0005718294657312466,
-      "loss": 4.1028,
+      "loss": 4.1083,
       "step": 4450
     },
     {
       "epoch": 0.48517520215633425,
-      "grad_norm": 0.7115625143051147,
+      "grad_norm": 0.6872192025184631,
       "learning_rate": 0.0005715056664867781,
-      "loss": 4.1109,
+      "loss": 4.1167,
       "step": 4500
     },
     {
       "epoch": 0.49056603773584906,
-      "grad_norm": 0.660191535949707,
+      "grad_norm": 0.6687753796577454,
       "learning_rate": 0.0005711818672423097,
-      "loss": 4.1011,
+      "loss": 4.1084,
       "step": 4550
     },
     {
       "epoch": 0.49595687331536387,
-      "grad_norm": 0.6192442178726196,
+      "grad_norm": 0.7406129837036133,
       "learning_rate": 0.0005708580679978413,
-      "loss": 4.0857,
+      "loss": 4.0924,
       "step": 4600
     },
     {
       "epoch": 0.5013477088948787,
-      "grad_norm": 0.6783645153045654,
+      "grad_norm": 0.6267918944358826,
       "learning_rate": 0.0005705342687533729,
-      "loss": 4.078,
+      "loss": 4.0872,
       "step": 4650
     },
     {
       "epoch": 0.5067385444743935,
-      "grad_norm": 0.6928367018699646,
+      "grad_norm": 0.7284395694732666,
       "learning_rate": 0.0005702104695089044,
-      "loss": 4.0956,
+      "loss": 4.0995,
       "step": 4700
     },
     {
       "epoch": 0.5121293800539084,
-      "grad_norm": 0.6265180110931396,
+      "grad_norm": 0.6475328803062439,
       "learning_rate": 0.000569886670264436,
-      "loss": 4.0822,
+      "loss": 4.0883,
       "step": 4750
     },
     {
       "epoch": 0.5175202156334232,
-      "grad_norm": 0.7510149478912354,
+      "grad_norm": 0.7573050260543823,
       "learning_rate": 0.0005695628710199675,
-      "loss": 4.0755,
+      "loss": 4.0838,
       "step": 4800
     },
     {
       "epoch": 0.522911051212938,
-      "grad_norm": 0.7354061603546143,
+      "grad_norm": 0.6486386060714722,
       "learning_rate": 0.0005692390717754991,
-      "loss": 4.0655,
+      "loss": 4.0727,
       "step": 4850
     },
     {
       "epoch": 0.5283018867924528,
-      "grad_norm": 0.6540889739990234,
+      "grad_norm": 0.6475781798362732,
       "learning_rate": 0.0005689152725310306,
-      "loss": 4.055,
+      "loss": 4.0552,
       "step": 4900
     },
     {
       "epoch": 0.5336927223719676,
-      "grad_norm": 0.674555778503418,
+      "grad_norm": 0.6808075904846191,
       "learning_rate": 0.0005685914732865623,
-      "loss": 4.0696,
+      "loss": 4.074,
       "step": 4950
     },
     {
       "epoch": 0.5390835579514824,
-      "grad_norm": 0.4996623992919922,
+      "grad_norm": 0.5275306701660156,
       "learning_rate": 0.0005682676740420939,
-      "loss": 4.0472,
+      "loss": 4.0541,
       "step": 5000
     },
     {
       "epoch": 0.5390835579514824,
-      "eval_accuracy": 0.3215982057496718,
-      "eval_loss": 3.9894087314605713,
-      "eval_runtime": 183.8499,
-      "eval_samples_per_second": 97.966,
-      "eval_steps_per_second": 6.125,
+      "eval_accuracy": 0.3209118454573477,
+      "eval_loss": 3.9960134029388428,
+      "eval_runtime": 180.5882,
+      "eval_samples_per_second": 99.735,
+      "eval_steps_per_second": 6.235,
       "step": 5000
     },
     {
       "epoch": 0.5444743935309974,
-      "grad_norm": 0.7127335071563721,
+      "grad_norm": 0.650256335735321,
       "learning_rate": 0.0005679438747976254,
-      "loss": 4.0568,
+      "loss": 4.0632,
       "step": 5050
     },
     {
       "epoch": 0.5498652291105122,
-      "grad_norm": 0.6820212602615356,
+      "grad_norm": 0.6265420317649841,
       "learning_rate": 0.000567620075553157,
-      "loss": 4.0516,
+      "loss": 4.0563,
       "step": 5100
     },
     {
       "epoch": 0.555256064690027,
-      "grad_norm": 0.7278046011924744,
+      "grad_norm": 0.697827935218811,
       "learning_rate": 0.0005672962763086886,
-      "loss": 4.0498,
+      "loss": 4.0546,
       "step": 5150
     },
     {
       "epoch": 0.5606469002695418,
-      "grad_norm": 0.624863862991333,
+      "grad_norm": 0.5988860726356506,
       "learning_rate": 0.0005669724770642202,
-      "loss": 4.0671,
+      "loss": 4.0726,
       "step": 5200
     },
     {
       "epoch": 0.5660377358490566,
-      "grad_norm": 0.6197010278701782,
+      "grad_norm": 0.6732010245323181,
       "learning_rate": 0.0005666486778197517,
-      "loss": 4.041,
+      "loss": 4.0505,
       "step": 5250
     },
     {
       "epoch": 0.5714285714285714,
-      "grad_norm": 0.7419002056121826,
+      "grad_norm": 0.7412548661231995,
       "learning_rate": 0.0005663248785752833,
-      "loss": 4.0326,
+      "loss": 4.04,
       "step": 5300
     },
     {
       "epoch": 0.5768194070080862,
-      "grad_norm": 0.6294003129005432,
+      "grad_norm": 0.6741536259651184,
       "learning_rate": 0.0005660010793308148,
-      "loss": 4.0263,
+      "loss": 4.0345,
       "step": 5350
     },
     {
       "epoch": 0.5822102425876011,
-      "grad_norm": 0.5749978423118591,
+      "grad_norm": 0.6361691355705261,
       "learning_rate": 0.0005656772800863465,
-      "loss": 4.0359,
+      "loss": 4.0424,
       "step": 5400
     },
     {
       "epoch": 0.5876010781671159,
-      "grad_norm": 0.5415307879447937,
+      "grad_norm": 0.5312321186065674,
       "learning_rate": 0.0005653534808418779,
-      "loss": 4.0382,
+      "loss": 4.0439,
       "step": 5450
     },
     {
       "epoch": 0.5929919137466307,
-      "grad_norm": 0.6727957725524902,
+      "grad_norm": 0.6982681751251221,
       "learning_rate": 0.0005650296815974096,
-      "loss": 4.0181,
+      "loss": 4.0263,
       "step": 5500
     },
     {
       "epoch": 0.5983827493261455,
-      "grad_norm": 0.6418060660362244,
+      "grad_norm": 0.6990839838981628,
       "learning_rate": 0.0005647058823529411,
-      "loss": 4.0069,
+      "loss": 4.014,
       "step": 5550
     },
     {
       "epoch": 0.6037735849056604,
-      "grad_norm": 0.7099882960319519,
+      "grad_norm": 0.6587233543395996,
       "learning_rate": 0.0005643820831084727,
-      "loss": 4.0226,
+      "loss": 4.0308,
       "step": 5600
     },
     {
       "epoch": 0.6091644204851752,
-      "grad_norm": 0.6498609185218811,
+      "grad_norm": 0.5876255035400391,
       "learning_rate": 0.0005640582838640042,
-      "loss": 4.0046,
+      "loss": 4.0114,
       "step": 5650
     },
     {
       "epoch": 0.6145552560646901,
-      "grad_norm": 0.5713142156600952,
+      "grad_norm": 0.5635818839073181,
       "learning_rate": 0.0005637344846195358,
-      "loss": 4.0043,
+      "loss": 4.0089,
       "step": 5700
     },
     {
       "epoch": 0.6199460916442049,
-      "grad_norm": 0.6617730259895325,
+      "grad_norm": 0.6282250285148621,
       "learning_rate": 0.0005634106853750674,
-      "loss": 3.979,
+      "loss": 3.9827,
       "step": 5750
     },
     {
       "epoch": 0.6253369272237197,
-      "grad_norm": 0.6460627317428589,
+      "grad_norm": 0.667834997177124,
       "learning_rate": 0.000563086886130599,
-      "loss": 3.9846,
+      "loss": 3.9909,
       "step": 5800
     },
     {
       "epoch": 0.6307277628032345,
-      "grad_norm": 0.6782198548316956,
+      "grad_norm": 0.6544371247291565,
       "learning_rate": 0.0005627630868861305,
-      "loss": 3.9947,
+      "loss": 4.003,
       "step": 5850
     },
     {
       "epoch": 0.6361185983827493,
-      "grad_norm": 0.5800774097442627,
+      "grad_norm": 0.6035220623016357,
       "learning_rate": 0.0005624392876416621,
-      "loss": 3.9922,
+      "loss": 4.0019,
       "step": 5900
     },
     {
       "epoch": 0.6415094339622641,
-      "grad_norm": 0.676412045955658,
+      "grad_norm": 0.6140830516815186,
       "learning_rate": 0.0005621154883971937,
-      "loss": 3.9791,
+      "loss": 3.9873,
       "step": 5950
     },
     {
       "epoch": 0.6469002695417789,
-      "grad_norm": 0.583685040473938,
+      "grad_norm": 0.6172138452529907,
       "learning_rate": 0.0005617916891527253,
-      "loss": 3.9931,
+      "loss": 4.0006,
       "step": 6000
     },
     {
       "epoch": 0.6469002695417789,
-      "eval_accuracy": 0.32843508006251015,
-      "eval_loss": 3.9174036979675293,
-      "eval_runtime": 183.9133,
-      "eval_samples_per_second": 97.932,
-      "eval_steps_per_second": 6.122,
+      "eval_accuracy": 0.32769602311959606,
+      "eval_loss": 3.924186944961548,
+      "eval_runtime": 180.7415,
+      "eval_samples_per_second": 99.651,
+      "eval_steps_per_second": 6.23,
       "step": 6000
     },
     {
       "epoch": 0.6522911051212938,
-      "grad_norm": 0.5780828595161438,
+      "grad_norm": 0.617624044418335,
       "learning_rate": 0.0005614678899082568,
-      "loss": 3.9663,
+      "loss": 3.9719,
       "step": 6050
     },
     {
       "epoch": 0.6576819407008087,
-      "grad_norm": 0.5652017593383789,
+      "grad_norm": 0.6407833099365234,
       "learning_rate": 0.0005611440906637884,
-      "loss": 3.9971,
+      "loss": 4.0037,
       "step": 6100
     },
     {
       "epoch": 0.6630727762803235,
-      "grad_norm": 0.8341758847236633,
+      "grad_norm": 0.6433732509613037,
       "learning_rate": 0.00056082029141932,
-      "loss": 3.9793,
+      "loss": 3.9888,
       "step": 6150
     },
     {
       "epoch": 0.6684636118598383,
-      "grad_norm": 0.578648567199707,
+      "grad_norm": 0.5665285587310791,
       "learning_rate": 0.0005604964921748515,
-      "loss": 3.9499,
+      "loss": 3.9583,
       "step": 6200
     },
     {
       "epoch": 0.6738544474393531,
-      "grad_norm": 0.59043949842453,
+      "grad_norm": 0.6015879511833191,
       "learning_rate": 0.000560172692930383,
-      "loss": 3.9715,
+      "loss": 3.9779,
       "step": 6250
     },
     {
       "epoch": 0.6792452830188679,
-      "grad_norm": 0.6629766225814819,
+      "grad_norm": 0.7052006721496582,
       "learning_rate": 0.0005598488936859147,
-      "loss": 3.9749,
+      "loss": 3.9822,
       "step": 6300
     },
     {
       "epoch": 0.6846361185983828,
-      "grad_norm": 0.5541216135025024,
+      "grad_norm": 0.554301381111145,
       "learning_rate": 0.0005595250944414463,
-      "loss": 3.9608,
+      "loss": 3.9679,
       "step": 6350
     },
     {
       "epoch": 0.6900269541778976,
-      "grad_norm": 0.6767361164093018,
+      "grad_norm": 0.6193522810935974,
       "learning_rate": 0.0005592012951969778,
-      "loss": 3.9694,
+      "loss": 3.9763,
       "step": 6400
     },
     {
       "epoch": 0.6954177897574124,
-      "grad_norm": 0.6981496810913086,
-      "learning_rate": 0.0005588774959525094,
-      "loss": 3.9655,
+      "grad_norm": 0.6429438591003418,
+      "learning_rate": 0.0005588839719373988,
+      "loss": 3.9716,
       "step": 6450
     },
     {
       "epoch": 0.7008086253369272,
-      "grad_norm": 0.5768890380859375,
-      "learning_rate": 0.000558553696708041,
-      "loss": 3.96,
+      "grad_norm": 0.6176355481147766,
+      "learning_rate": 0.0005585601726929303,
+      "loss": 3.9647,
       "step": 6500
     },
     {
       "epoch": 0.706199460916442,
-      "grad_norm": 0.6919478178024292,
-      "learning_rate": 0.0005582298974635726,
-      "loss": 3.958,
+      "grad_norm": 0.6647421717643738,
+      "learning_rate": 0.0005582363734484619,
+      "loss": 3.9638,
       "step": 6550
     },
     {
       "epoch": 0.7115902964959568,
-      "grad_norm": 0.5978683829307556,
-      "learning_rate": 0.0005579060982191041,
-      "loss": 3.9587,
+      "grad_norm": 0.5974478721618652,
+      "learning_rate": 0.0005579125742039935,
+      "loss": 3.967,
       "step": 6600
     },
     {
       "epoch": 0.7169811320754716,
-      "grad_norm": 0.6285462379455566,
-      "learning_rate": 0.0005575822989746357,
-      "loss": 3.9703,
+      "grad_norm": 0.6024782657623291,
+      "learning_rate": 0.0005575887749595251,
+      "loss": 3.9752,
       "step": 6650
     },
     {
       "epoch": 0.7223719676549866,
-      "grad_norm": 0.7304993867874146,
-      "learning_rate": 0.0005572584997301672,
-      "loss": 3.9348,
+      "grad_norm": 0.6122707724571228,
+      "learning_rate": 0.0005572649757150566,
+      "loss": 3.9388,
       "step": 6700
     },
     {
       "epoch": 0.7277628032345014,
-      "grad_norm": 0.5624628663063049,
-      "learning_rate": 0.0005569347004856989,
-      "loss": 3.9602,
+      "grad_norm": 0.5399841666221619,
+      "learning_rate": 0.0005569411764705882,
+      "loss": 3.9672,
       "step": 6750
     },
     {
       "epoch": 0.7331536388140162,
-      "grad_norm": 0.6064887642860413,
-      "learning_rate": 0.0005566109012412303,
-      "loss": 3.9331,
+      "grad_norm": 0.6701462268829346,
+      "learning_rate": 0.0005566173772261198,
+      "loss": 3.9395,
       "step": 6800
     },
     {
       "epoch": 0.738544474393531,
-      "grad_norm": 0.5876811146736145,
-      "learning_rate": 0.000556287101996762,
-      "loss": 3.9323,
+      "grad_norm": 0.6382229328155518,
+      "learning_rate": 0.0005562935779816513,
+      "loss": 3.9407,
       "step": 6850
     },
     {
       "epoch": 0.7439353099730458,
-      "grad_norm": 0.5871797204017639,
-      "learning_rate": 0.0005559633027522935,
-      "loss": 3.9329,
+      "grad_norm": 0.5983497500419617,
+      "learning_rate": 0.0005559697787371828,
+      "loss": 3.9411,
       "step": 6900
     },
     {
       "epoch": 0.7493261455525606,
-      "grad_norm": 0.6020776629447937,
-      "learning_rate": 0.0005556395035078251,
-      "loss": 3.9362,
+      "grad_norm": 0.6194353699684143,
+      "learning_rate": 0.0005556459794927145,
+      "loss": 3.9455,
       "step": 6950
     },
     {
       "epoch": 0.7547169811320755,
-      "grad_norm": 0.6208156943321228,
-      "learning_rate": 0.0005553157042633566,
-      "loss": 3.9191,
+      "grad_norm": 0.6101710796356201,
+      "learning_rate": 0.000555322180248246,
+      "loss": 3.9272,
       "step": 7000
     },
     {
       "epoch": 0.7547169811320755,
-      "eval_accuracy": 0.33344278300899177,
-      "eval_loss": 3.8618791103363037,
-      "eval_runtime": 183.7498,
-      "eval_samples_per_second": 98.019,
-      "eval_steps_per_second": 6.128,
+      "eval_accuracy": 0.3329229875936873,
+      "eval_loss": 3.865872859954834,
+      "eval_runtime": 180.181,
+      "eval_samples_per_second": 99.961,
+      "eval_steps_per_second": 6.249,
       "step": 7000
     },
     {
       "epoch": 0.7601078167115903,
-      "grad_norm": 0.6798801422119141,
-      "learning_rate": 0.0005549919050188882,
-      "loss": 3.9331,
+      "grad_norm": 0.6827537417411804,
+      "learning_rate": 0.0005549983810037776,
+      "loss": 3.9395,
       "step": 7050
     },
     {
       "epoch": 0.7654986522911051,
-      "grad_norm": 1.4628221988677979,
-      "learning_rate": 0.0005546681057744198,
-      "loss": 3.906,
+      "grad_norm": 0.9468194842338562,
+      "learning_rate": 0.0005546745817593091,
+      "loss": 3.9127,
       "step": 7100
     },
     {
       "epoch": 0.77088948787062,
-      "grad_norm": 0.6019647121429443,
-      "learning_rate": 0.0005543443065299514,
-      "loss": 3.9351,
+      "grad_norm": 0.6613170504570007,
+      "learning_rate": 0.0005543507825148408,
+      "loss": 3.9409,
       "step": 7150
     },
     {
       "epoch": 0.7762803234501348,
-      "grad_norm": 0.5555400252342224,
-      "learning_rate": 0.000554020507285483,
-      "loss": 3.9058,
+      "grad_norm": 0.5522786378860474,
+      "learning_rate": 0.0005540269832703723,
+      "loss": 3.9139,
       "step": 7200
     },
     {
       "epoch": 0.7816711590296496,
-      "grad_norm": 0.6362798810005188,
-      "learning_rate": 0.0005536967080410145,
-      "loss": 3.8967,
+      "grad_norm": 0.6424623727798462,
+      "learning_rate": 0.0005537031840259039,
+      "loss": 3.9059,
       "step": 7250
     },
     {
       "epoch": 0.7870619946091644,
-      "grad_norm": 0.5722786784172058,
-      "learning_rate": 0.0005533729087965462,
-      "loss": 3.9111,
+      "grad_norm": 0.5609923005104065,
+      "learning_rate": 0.0005533793847814354,
+      "loss": 3.9183,
       "step": 7300
     },
     {
       "epoch": 0.7924528301886793,
-      "grad_norm": 0.580388069152832,
-      "learning_rate": 0.0005530491095520777,
-      "loss": 3.9099,
+      "grad_norm": 0.6663058996200562,
+      "learning_rate": 0.000553055585536967,
+      "loss": 3.9182,
       "step": 7350
     },
     {
       "epoch": 0.7978436657681941,
-      "grad_norm": 0.5334943532943726,
-      "learning_rate": 0.0005527253103076093,
-      "loss": 3.9072,
+      "grad_norm": 0.5433474183082581,
+      "learning_rate": 0.0005527317862924987,
+      "loss": 3.9149,
       "step": 7400
     },
     {
       "epoch": 0.8032345013477089,
-      "grad_norm": 0.6169307827949524,
-      "learning_rate": 0.0005524015110631408,
-      "loss": 3.8832,
+      "grad_norm": 0.566183865070343,
+      "learning_rate": 0.0005524079870480301,
+      "loss": 3.8916,
       "step": 7450
     },
     {
       "epoch": 0.8086253369272237,
-      "grad_norm": 0.6007163524627686,
-      "learning_rate": 0.0005520777118186724,
-      "loss": 3.9024,
+      "grad_norm": 0.6637818813323975,
+      "learning_rate": 0.0005520841878035618,
+      "loss": 3.9096,
       "step": 7500
     },
     {
       "epoch": 0.8140161725067385,
-      "grad_norm": 0.5735055804252625,
-      "learning_rate": 0.0005517539125742039,
-      "loss": 3.8979,
+      "grad_norm": 0.6299164295196533,
+      "learning_rate": 0.0005517603885590933,
+      "loss": 3.9037,
       "step": 7550
     },
     {
       "epoch": 0.8194070080862533,
-      "grad_norm": 0.5585691928863525,
+      "grad_norm": 0.6113953590393066,
       "learning_rate": 0.0005514365893146249,
-      "loss": 3.8902,
+      "loss": 3.8978,
       "step": 7600
     },
     {
       "epoch": 0.8247978436657682,
-      "grad_norm": 0.5613856315612793,
+      "grad_norm": 0.5476465225219727,
       "learning_rate": 0.0005511127900701564,
-      "loss": 3.8952,
+      "loss": 3.9038,
       "step": 7650
     },
     {
       "epoch": 0.8301886792452831,
-      "grad_norm": 0.5988004803657532,
+      "grad_norm": 0.611580491065979,
       "learning_rate": 0.000550788990825688,
-      "loss": 3.8894,
+      "loss": 3.8975,
       "step": 7700
     },
     {
       "epoch": 0.8355795148247979,
-      "grad_norm": 0.5915306210517883,
+      "grad_norm": 0.626191258430481,
       "learning_rate": 0.0005504651915812196,
-      "loss": 3.8922,
+      "loss": 3.8981,
       "step": 7750
     },
     {
       "epoch": 0.8409703504043127,
-      "grad_norm": 0.7122664451599121,
+      "grad_norm": 0.5997198224067688,
       "learning_rate": 0.0005501413923367512,
-      "loss": 3.8766,
+      "loss": 3.8835,
       "step": 7800
     },
     {
       "epoch": 0.8463611859838275,
-      "grad_norm": 0.5410688519477844,
+      "grad_norm": 0.5855820775032043,
       "learning_rate": 0.0005498175930922827,
-      "loss": 3.8794,
+      "loss": 3.8885,
       "step": 7850
     },
     {
       "epoch": 0.8517520215633423,
-      "grad_norm": 0.5719017386436462,
+      "grad_norm": 0.6042124629020691,
       "learning_rate": 0.0005494937938478143,
-      "loss": 3.8727,
+      "loss": 3.8831,
       "step": 7900
     },
     {
       "epoch": 0.8571428571428571,
-      "grad_norm": 0.570475697517395,
+      "grad_norm": 0.6467875838279724,
       "learning_rate": 0.0005491699946033459,
-      "loss": 3.9063,
+      "loss": 3.9133,
       "step": 7950
     },
     {
       "epoch": 0.862533692722372,
-      "grad_norm": 0.5836682915687561,
+      "grad_norm": 0.580238938331604,
       "learning_rate": 0.0005488461953588775,
-      "loss": 3.8542,
+      "loss": 3.8604,
       "step": 8000
     },
     {
       "epoch": 0.862533692722372,
-      "eval_accuracy": 0.33794155582895036,
-      "eval_loss": 3.816067695617676,
-      "eval_runtime": 183.5803,
-      "eval_samples_per_second": 98.11,
-      "eval_steps_per_second": 6.134,
+      "eval_accuracy": 0.33701235633236093,
+      "eval_loss": 3.823117733001709,
+      "eval_runtime": 179.933,
+      "eval_samples_per_second": 100.098,
+      "eval_steps_per_second": 6.258,
       "step": 8000
     },
     {
       "epoch": 0.8679245283018868,
-      "grad_norm": 0.6498636603355408,
+      "grad_norm": 0.6544236540794373,
       "learning_rate": 0.000548522396114409,
-      "loss": 3.895,
+      "loss": 3.9033,
       "step": 8050
     },
     {
       "epoch": 0.8733153638814016,
-      "grad_norm": 0.6250914931297302,
+      "grad_norm": 0.6407549977302551,
       "learning_rate": 0.0005481985968699406,
-      "loss": 3.8735,
+      "loss": 3.8815,
       "step": 8100
     },
     {
       "epoch": 0.8787061994609164,
-      "grad_norm": 0.5511185526847839,
+      "grad_norm": 0.5456429719924927,
       "learning_rate": 0.0005478747976254721,
-      "loss": 3.8758,
+      "loss": 3.8833,
       "step": 8150
     },
     {
       "epoch": 0.8840970350404312,
-      "grad_norm": 0.6440311074256897,
+      "grad_norm": 0.6943271160125732,
       "learning_rate": 0.0005475509983810037,
-      "loss": 3.869,
+      "loss": 3.8788,
       "step": 8200
     },
     {
       "epoch": 0.889487870619946,
-      "grad_norm": 0.5398238301277161,
+      "grad_norm": 0.6314564347267151,
       "learning_rate": 0.0005472271991365352,
-      "loss": 3.866,
+      "loss": 3.876,
       "step": 8250
     },
     {
       "epoch": 0.894878706199461,
-      "grad_norm": 0.5822187066078186,
+      "grad_norm": 0.6067231893539429,
       "learning_rate": 0.0005469033998920669,
-      "loss": 3.8864,
+      "loss": 3.893,
       "step": 8300
     },
     {
       "epoch": 0.9002695417789758,
-      "grad_norm": 0.5405433773994446,
+      "grad_norm": 0.5689924359321594,
       "learning_rate": 0.0005465796006475984,
-      "loss": 3.857,
+      "loss": 3.8639,
       "step": 8350
     },
     {
       "epoch": 0.9056603773584906,
-      "grad_norm": 0.5735152959823608,
+      "grad_norm": 0.5926893353462219,
       "learning_rate": 0.00054625580140313,
-      "loss": 3.8628,
+      "loss": 3.8692,
       "step": 8400
     },
     {
       "epoch": 0.9110512129380054,
-      "grad_norm": 0.5623797178268433,
+      "grad_norm": 0.5659659504890442,
       "learning_rate": 0.0005459320021586615,
-      "loss": 3.858,
+      "loss": 3.8642,
       "step": 8450
     },
     {
       "epoch": 0.9164420485175202,
-      "grad_norm": 0.5893810391426086,
+      "grad_norm": 0.617048442363739,
       "learning_rate": 0.0005456082029141932,
-      "loss": 3.8638,
+      "loss": 3.8696,
       "step": 8500
     },
     {
       "epoch": 0.921832884097035,
-      "grad_norm": 0.6306450366973877,
-      "learning_rate": 0.0005452844036697248,
-      "loss": 3.8565,
+      "grad_norm": 0.5619012713432312,
+      "learning_rate": 0.000545290879654614,
+      "loss": 3.8651,
       "step": 8550
     },
     {
       "epoch": 0.9272237196765498,
-      "grad_norm": 0.5499542951583862,
-      "learning_rate": 0.0005449606044252563,
-      "loss": 3.8639,
+      "grad_norm": 0.551312267780304,
+      "learning_rate": 0.0005449670804101457,
+      "loss": 3.8736,
       "step": 8600
     },
     {
       "epoch": 0.9326145552560647,
-      "grad_norm": 0.563671886920929,
-      "learning_rate": 0.0005446368051807879,
-      "loss": 3.844,
+      "grad_norm": 0.5884478092193604,
+      "learning_rate": 0.0005446432811656773,
+      "loss": 3.8492,
       "step": 8650
     },
     {
       "epoch": 0.9380053908355795,
-      "grad_norm": 0.5789870619773865,
-      "learning_rate": 0.0005443130059363194,
-      "loss": 3.8492,
+      "grad_norm": 0.6045072078704834,
+      "learning_rate": 0.0005443194819212088,
+      "loss": 3.857,
       "step": 8700
     },
     {
       "epoch": 0.9433962264150944,
-      "grad_norm": 0.5480877757072449,
-      "learning_rate": 0.0005439892066918511,
-      "loss": 3.8458,
+      "grad_norm": 0.5507612824440002,
+      "learning_rate": 0.0005439956826767404,
+      "loss": 3.8551,
       "step": 8750
     },
     {
       "epoch": 0.9487870619946092,
-      "grad_norm": 0.5352542996406555,
-      "learning_rate": 0.0005436654074473825,
-      "loss": 3.8487,
+      "grad_norm": 0.48222315311431885,
+      "learning_rate": 0.000543671883432272,
+      "loss": 3.8567,
       "step": 8800
     },
     {
       "epoch": 0.954177897574124,
-      "grad_norm": 0.5493354797363281,
-      "learning_rate": 0.0005433416082029142,
-      "loss": 3.8454,
+      "grad_norm": 0.6002428531646729,
+      "learning_rate": 0.0005433480841878035,
+      "loss": 3.8561,
       "step": 8850
     },
     {
       "epoch": 0.9595687331536388,
-      "grad_norm": 0.5753820538520813,
-      "learning_rate": 0.0005430178089584457,
-      "loss": 3.8372,
+      "grad_norm": 0.5420414805412292,
+      "learning_rate": 0.000543024284943335,
+      "loss": 3.847,
       "step": 8900
     },
     {
       "epoch": 0.9649595687331537,
-      "grad_norm": 0.5328227877616882,
-      "learning_rate": 0.0005426940097139773,
-      "loss": 3.8276,
+      "grad_norm": 0.565650224685669,
+      "learning_rate": 0.0005427004856988667,
+      "loss": 3.8364,
       "step": 8950
     },
     {
       "epoch": 0.9703504043126685,
-      "grad_norm": 0.5667588114738464,
-      "learning_rate": 0.0005423702104695088,
-      "loss": 3.8531,
+      "grad_norm": 0.5593795776367188,
+      "learning_rate": 0.0005423766864543982,
+      "loss": 3.8606,
       "step": 9000
     },
     {
       "epoch": 0.9703504043126685,
-      "eval_accuracy": 0.34165379039401994,
-      "eval_loss": 3.7784414291381836,
-      "eval_runtime": 184.2031,
-      "eval_samples_per_second": 97.778,
-      "eval_steps_per_second": 6.113,
+      "eval_accuracy": 0.34049826693211294,
+      "eval_loss": 3.784738779067993,
+      "eval_runtime": 180.014,
+      "eval_samples_per_second": 100.053,
+      "eval_steps_per_second": 6.255,
       "step": 9000
     },
     {
       "epoch": 0.9757412398921833,
-      "grad_norm": 0.6010326743125916,
-      "learning_rate": 0.0005420464112250404,
-      "loss": 3.8678,
+      "grad_norm": 0.6755319237709045,
+      "learning_rate": 0.0005420528872099298,
+      "loss": 3.8761,
       "step": 9050
     },
     {
       "epoch": 0.9811320754716981,
-      "grad_norm": 0.5293521285057068,
-      "learning_rate": 0.000541722611980572,
-      "loss": 3.8241,
+      "grad_norm": 0.5751345753669739,
+      "learning_rate": 0.0005417290879654613,
+      "loss": 3.8319,
       "step": 9100
     },
     {
       "epoch": 0.9865229110512129,
-      "grad_norm": 0.5834252238273621,
-      "learning_rate": 0.0005413988127361036,
-      "loss": 3.8378,
+      "grad_norm": 0.5585288405418396,
+      "learning_rate": 0.000541405288720993,
+      "loss": 3.8452,
       "step": 9150
     },
     {
       "epoch": 0.9919137466307277,
-      "grad_norm": 0.5734151005744934,
-      "learning_rate": 0.0005410750134916351,
-      "loss": 3.8295,
+      "grad_norm": 0.6097003817558289,
+      "learning_rate": 0.0005410814894765245,
+      "loss": 3.8362,
       "step": 9200
     },
     {
       "epoch": 0.9973045822102425,
-      "grad_norm": 0.5477326512336731,
-      "learning_rate": 0.0005407512142471667,
-      "loss": 3.8451,
+      "grad_norm": 0.5463394522666931,
+      "learning_rate": 0.0005407576902320561,
+      "loss": 3.852,
       "step": 9250
     },
     {
       "epoch": 1.0026954177897573,
-      "grad_norm": 0.5776128172874451,
-      "learning_rate": 0.0005404274150026983,
-      "loss": 3.7938,
+      "grad_norm": 0.5733648538589478,
+      "learning_rate": 0.0005404338909875876,
+      "loss": 3.8027,
       "step": 9300
     },
     {
       "epoch": 1.0080862533692723,
-      "grad_norm": 0.5947125554084778,
-      "learning_rate": 0.0005401036157582299,
-      "loss": 3.7605,
+      "grad_norm": 0.5949172973632812,
+      "learning_rate": 0.0005401100917431192,
+      "loss": 3.7663,
       "step": 9350
     },
     {
       "epoch": 1.013477088948787,
-      "grad_norm": 0.6069636940956116,
-      "learning_rate": 0.0005397798165137614,
-      "loss": 3.7525,
+      "grad_norm": 0.5749472379684448,
+      "learning_rate": 0.0005397862924986508,
+      "loss": 3.7635,
       "step": 9400
     },
     {
       "epoch": 1.0188679245283019,
-      "grad_norm": 0.5255810618400574,
-      "learning_rate": 0.000539456017269293,
-      "loss": 3.7525,
+      "grad_norm": 0.5567948818206787,
+      "learning_rate": 0.0005394624932541824,
+      "loss": 3.7604,
       "step": 9450
     },
     {
       "epoch": 1.0242587601078168,
-      "grad_norm": 0.546853244304657,
-      "learning_rate": 0.0005391322180248245,
-      "loss": 3.7576,
+      "grad_norm": 0.5213462710380554,
+      "learning_rate": 0.0005391386940097139,
+      "loss": 3.7657,
       "step": 9500
     },
     {
       "epoch": 1.0296495956873315,
-      "grad_norm": 0.5615027546882629,
-      "learning_rate": 0.0005388084187803561,
-      "loss": 3.7679,
+      "grad_norm": 0.5804442167282104,
+      "learning_rate": 0.0005388148947652455,
+      "loss": 3.7763,
       "step": 9550
     },
     {
       "epoch": 1.0350404312668464,
-      "grad_norm": 0.579430341720581,
-      "learning_rate": 0.0005384846195358876,
-      "loss": 3.7661,
+      "grad_norm": 0.5572142601013184,
+      "learning_rate": 0.000538491095520777,
+      "loss": 3.7733,
       "step": 9600
     },
     {
       "epoch": 1.0404312668463611,
-      "grad_norm": 0.6099395751953125,
-      "learning_rate": 0.0005381608202914193,
-      "loss": 3.7707,
+      "grad_norm": 0.6099660396575928,
+      "learning_rate": 0.0005381672962763086,
+      "loss": 3.779,
       "step": 9650
     },
     {
       "epoch": 1.045822102425876,
-      "grad_norm": 0.5670583248138428,
-      "learning_rate": 0.0005378370210469509,
-      "loss": 3.7544,
+      "grad_norm": 0.6277936697006226,
+      "learning_rate": 0.0005378434970318403,
+      "loss": 3.7623,
       "step": 9700
     },
     {
       "epoch": 1.0512129380053907,
-      "grad_norm": 0.5511836409568787,
-      "learning_rate": 0.0005375132218024824,
-      "loss": 3.7525,
+      "grad_norm": 0.5859461426734924,
+      "learning_rate": 0.0005375196977873718,
+      "loss": 3.7597,
       "step": 9750
     },
     {
       "epoch": 1.0566037735849056,
-      "grad_norm": 0.5288443565368652,
-      "learning_rate": 0.000537189422558014,
-      "loss": 3.7805,
+      "grad_norm": 0.5517092943191528,
+      "learning_rate": 0.0005371958985429034,
+      "loss": 3.7876,
       "step": 9800
     },
     {
       "epoch": 1.0619946091644206,
-      "grad_norm": 0.5281191468238831,
-      "learning_rate": 0.0005368656233135455,
-      "loss": 3.7818,
+      "grad_norm": 0.5141564011573792,
+      "learning_rate": 0.0005368720992984349,
+      "loss": 3.7866,
       "step": 9850
     },
     {
       "epoch": 1.0673854447439353,
-      "grad_norm": 0.5530230402946472,
+      "grad_norm": 0.6039884090423584,
       "learning_rate": 0.0005365483000539665,
-      "loss": 3.7735,
+      "loss": 3.7795,
       "step": 9900
     },
     {
       "epoch": 1.0727762803234502,
-      "grad_norm": 0.6312490701675415,
+      "grad_norm": 0.5600705742835999,
       "learning_rate": 0.0005362245008094981,
-      "loss": 3.7825,
+      "loss": 3.7911,
       "step": 9950
     },
     {
       "epoch": 1.0781671159029649,
-      "grad_norm": 0.5409021377563477,
+      "grad_norm": 0.5785020589828491,
       "learning_rate": 0.0005359007015650297,
-      "loss": 3.7613,
+      "loss": 3.7669,
       "step": 10000
     },
     {
       "epoch": 1.0781671159029649,
-      "eval_accuracy": 0.3451733833889944,
-      "eval_loss": 3.7442734241485596,
-      "eval_runtime": 183.7504,
-      "eval_samples_per_second": 98.019,
-      "eval_steps_per_second": 6.128,
+      "eval_accuracy": 0.3439608171609848,
+      "eval_loss": 3.753610610961914,
+      "eval_runtime": 180.7337,
+      "eval_samples_per_second": 99.655,
+      "eval_steps_per_second": 6.23,
       "step": 10000
     },
     {
       "epoch": 1.0835579514824798,
-      "grad_norm": 0.5455986857414246,
+      "grad_norm": 0.549750804901123,
       "learning_rate": 0.0005355769023205612,
-      "loss": 3.7552,
+      "loss": 3.7626,
       "step": 10050
     },
     {
       "epoch": 1.0889487870619945,
-      "grad_norm": 0.5736761093139648,
+      "grad_norm": 0.6019790768623352,
       "learning_rate": 0.0005352531030760928,
-      "loss": 3.7591,
+      "loss": 3.7648,
       "step": 10100
     },
     {
       "epoch": 1.0943396226415094,
-      "grad_norm": 0.5683985948562622,
+      "grad_norm": 0.5909546613693237,
       "learning_rate": 0.0005349293038316244,
-      "loss": 3.7521,
+      "loss": 3.7604,
       "step": 10150
     },
     {
       "epoch": 1.0997304582210243,
-      "grad_norm": 0.5910122394561768,
+      "grad_norm": 0.619135320186615,
       "learning_rate": 0.0005346055045871559,
-      "loss": 3.7653,
+      "loss": 3.7714,
       "step": 10200
     },
     {
       "epoch": 1.105121293800539,
-      "grad_norm": 0.6161583065986633,
+      "grad_norm": 0.6440742015838623,
       "learning_rate": 0.0005342817053426874,
-      "loss": 3.753,
+      "loss": 3.76,
       "step": 10250
     },
     {
       "epoch": 1.110512129380054,
-      "grad_norm": 0.5718348026275635,
+      "grad_norm": 0.5757265686988831,
       "learning_rate": 0.0005339579060982191,
-      "loss": 3.7615,
+      "loss": 3.7664,
       "step": 10300
     },
     {
       "epoch": 1.1159029649595686,
-      "grad_norm": 0.576435387134552,
+      "grad_norm": 0.5689772963523865,
       "learning_rate": 0.0005336341068537506,
-      "loss": 3.7567,
+      "loss": 3.7639,
       "step": 10350
     },
     {
       "epoch": 1.1212938005390836,
-      "grad_norm": 0.5745424628257751,
+      "grad_norm": 0.549622118473053,
       "learning_rate": 0.0005333103076092822,
-      "loss": 3.7612,
+      "loss": 3.7678,
       "step": 10400
     },
     {
       "epoch": 1.1266846361185983,
-      "grad_norm": 0.5627392530441284,
+      "grad_norm": 0.5835170149803162,
       "learning_rate": 0.0005329865083648137,
-      "loss": 3.757,
+      "loss": 3.7666,
       "step": 10450
     },
     {
       "epoch": 1.1320754716981132,
-      "grad_norm": 0.6117731332778931,
+      "grad_norm": 0.6249995827674866,
       "learning_rate": 0.0005326627091203454,
-      "loss": 3.7597,
+      "loss": 3.7676,
       "step": 10500
     },
     {
       "epoch": 1.137466307277628,
-      "grad_norm": 0.6170853972434998,
+      "grad_norm": 0.572984516620636,
       "learning_rate": 0.0005323389098758769,
-      "loss": 3.7656,
+      "loss": 3.7717,
       "step": 10550
     },
     {
       "epoch": 1.1428571428571428,
-      "grad_norm": 0.5776093006134033,
+      "grad_norm": 0.5635685324668884,
       "learning_rate": 0.0005320151106314085,
-      "loss": 3.7554,
+      "loss": 3.7622,
       "step": 10600
     },
     {
       "epoch": 1.1482479784366577,
-      "grad_norm": 0.6169910430908203,
+      "grad_norm": 0.6093471050262451,
       "learning_rate": 0.00053169131138694,
-      "loss": 3.7413,
+      "loss": 3.7474,
       "step": 10650
     },
     {
       "epoch": 1.1536388140161726,
-      "grad_norm": 0.5566491484642029,
+      "grad_norm": 0.6207146048545837,
       "learning_rate": 0.0005313675121424716,
-      "loss": 3.7528,
+      "loss": 3.7593,
       "step": 10700
     },
     {
       "epoch": 1.1590296495956873,
-      "grad_norm": 0.5667610168457031,
+      "grad_norm": 0.5490242838859558,
       "learning_rate": 0.0005310437128980032,
-      "loss": 3.7331,
+      "loss": 3.74,
       "step": 10750
     },
     {
       "epoch": 1.1644204851752022,
-      "grad_norm": 0.5442368984222412,
-      "learning_rate": 0.0005307199136535348,
-      "loss": 3.7391,
+      "grad_norm": 0.5845029950141907,
+      "learning_rate": 0.0005307263896384242,
+      "loss": 3.7447,
       "step": 10800
     },
     {
       "epoch": 1.169811320754717,
-      "grad_norm": 0.5643858313560486,
-      "learning_rate": 0.0005303961144090663,
-      "loss": 3.7309,
+      "grad_norm": 0.5138887763023376,
+      "learning_rate": 0.0005304025903939556,
+      "loss": 3.7355,
       "step": 10850
     },
     {
       "epoch": 1.1752021563342319,
-      "grad_norm": 0.6005975604057312,
-      "learning_rate": 0.0005300723151645979,
-      "loss": 3.7448,
+      "grad_norm": 0.6029220223426819,
+      "learning_rate": 0.0005300787911494873,
+      "loss": 3.7495,
       "step": 10900
     },
     {
       "epoch": 1.1805929919137466,
-      "grad_norm": 0.5089237093925476,
-      "learning_rate": 0.0005297485159201295,
-      "loss": 3.7585,
+      "grad_norm": 0.5275118350982666,
+      "learning_rate": 0.0005297549919050189,
+      "loss": 3.7646,
       "step": 10950
     },
     {
       "epoch": 1.1859838274932615,
-      "grad_norm": 0.5960420370101929,
-      "learning_rate": 0.000529424716675661,
-      "loss": 3.7685,
+      "grad_norm": 0.5724518895149231,
+      "learning_rate": 0.0005294311926605504,
+      "loss": 3.7755,
       "step": 11000
     },
     {
       "epoch": 1.1859838274932615,
-      "eval_accuracy": 0.34775128180528075,
-      "eval_loss": 3.720940589904785,
-      "eval_runtime": 183.5903,
-      "eval_samples_per_second": 98.104,
-      "eval_steps_per_second": 6.133,
+      "eval_accuracy": 0.3467635184021593,
+      "eval_loss": 3.7266314029693604,
+      "eval_runtime": 181.2827,
+      "eval_samples_per_second": 99.353,
+      "eval_steps_per_second": 6.211,
       "step": 11000
     },
     {
       "epoch": 1.1913746630727764,
-      "grad_norm": 0.575962245464325,
-      "learning_rate": 0.0005291009174311926,
-      "loss": 3.755,
+      "grad_norm": 0.583351731300354,
+      "learning_rate": 0.000529107393416082,
+      "loss": 3.7621,
       "step": 11050
     },
     {
       "epoch": 1.196765498652291,
-      "grad_norm": 0.523404598236084,
-      "learning_rate": 0.0005287771181867242,
-      "loss": 3.7415,
+      "grad_norm": 0.550574541091919,
+      "learning_rate": 0.0005287835941716135,
+      "loss": 3.7457,
       "step": 11100
     },
     {
       "epoch": 1.202156334231806,
-      "grad_norm": 0.6159588694572449,
-      "learning_rate": 0.0005284533189422558,
-      "loss": 3.7365,
+      "grad_norm": 0.5491907000541687,
+      "learning_rate": 0.0005284597949271452,
+      "loss": 3.7462,
       "step": 11150
     },
     {
       "epoch": 1.2075471698113207,
-      "grad_norm": 0.49879810214042664,
-      "learning_rate": 0.0005281295196977873,
-      "loss": 3.7591,
+      "grad_norm": 0.5312425494194031,
+      "learning_rate": 0.0005281359956826767,
+      "loss": 3.7675,
       "step": 11200
     },
     {
       "epoch": 1.2129380053908356,
-      "grad_norm": 0.5774774551391602,
-      "learning_rate": 0.0005278057204533189,
-      "loss": 3.7483,
+      "grad_norm": 0.6146191954612732,
+      "learning_rate": 0.0005278121964382083,
+      "loss": 3.7549,
       "step": 11250
     },
     {
       "epoch": 1.2183288409703503,
-      "grad_norm": 0.5057991743087769,
-      "learning_rate": 0.0005274819212088505,
-      "loss": 3.732,
+      "grad_norm": 0.5598134398460388,
+      "learning_rate": 0.0005274883971937398,
+      "loss": 3.7391,
       "step": 11300
     },
     {
       "epoch": 1.2237196765498652,
-      "grad_norm": 0.5313456654548645,
-      "learning_rate": 0.0005271581219643821,
-      "loss": 3.7282,
+      "grad_norm": 0.5811231732368469,
+      "learning_rate": 0.0005271645979492714,
+      "loss": 3.7362,
       "step": 11350
     },
     {
       "epoch": 1.2291105121293802,
-      "grad_norm": 0.7374264597892761,
-      "learning_rate": 0.0005268343227199136,
-      "loss": 3.7426,
+      "grad_norm": 0.6216686367988586,
+      "learning_rate": 0.000526840798704803,
+      "loss": 3.7485,
       "step": 11400
     },
     {
       "epoch": 1.2345013477088949,
-      "grad_norm": 0.5699669718742371,
-      "learning_rate": 0.0005265105234754452,
-      "loss": 3.7535,
+      "grad_norm": 0.5307666063308716,
+      "learning_rate": 0.0005265169994603346,
+      "loss": 3.7607,
       "step": 11450
     },
     {
       "epoch": 1.2398921832884098,
-      "grad_norm": 0.567206084728241,
-      "learning_rate": 0.0005261867242309767,
-      "loss": 3.7431,
+      "grad_norm": 0.5830156207084656,
+      "learning_rate": 0.0005261932002158661,
+      "loss": 3.7492,
       "step": 11500
     },
     {
       "epoch": 1.2452830188679245,
-      "grad_norm": 0.5753964781761169,
-      "learning_rate": 0.0005258629249865083,
-      "loss": 3.7371,
+      "grad_norm": 0.5295611619949341,
+      "learning_rate": 0.0005258694009713977,
+      "loss": 3.7436,
       "step": 11550
     },
     {
       "epoch": 1.2506738544474394,
-      "grad_norm": 0.5284551978111267,
-      "learning_rate": 0.0005255391257420398,
-      "loss": 3.7509,
+      "grad_norm": 0.5932489633560181,
+      "learning_rate": 0.0005255456017269292,
+      "loss": 3.7553,
       "step": 11600
     },
     {
       "epoch": 1.256064690026954,
-      "grad_norm": 0.5363852381706238,
-      "learning_rate": 0.0005252153264975715,
-      "loss": 3.7284,
+      "grad_norm": 0.5623101592063904,
+      "learning_rate": 0.0005252218024824608,
+      "loss": 3.7376,
       "step": 11650
     },
     {
       "epoch": 1.261455525606469,
-      "grad_norm": 0.5310810804367065,
-      "learning_rate": 0.000524891527253103,
-      "loss": 3.7414,
+      "grad_norm": 0.5590541958808899,
+      "learning_rate": 0.0005248980032379924,
+      "loss": 3.7477,
       "step": 11700
     },
     {
       "epoch": 1.266846361185984,
-      "grad_norm": 0.5255619287490845,
-      "learning_rate": 0.0005245677280086346,
-      "loss": 3.7409,
+      "grad_norm": 0.5333404541015625,
+      "learning_rate": 0.000524574203993524,
+      "loss": 3.7474,
       "step": 11750
     },
     {
       "epoch": 1.2722371967654986,
-      "grad_norm": 0.6119208931922913,
-      "learning_rate": 0.0005242439287641661,
-      "loss": 3.7545,
+      "grad_norm": 0.5815286636352539,
+      "learning_rate": 0.0005242504047490555,
+      "loss": 3.7594,
       "step": 11800
     },
     {
       "epoch": 1.2776280323450135,
-      "grad_norm": 0.6038151979446411,
-      "learning_rate": 0.0005239201295196978,
-      "loss": 3.7357,
+      "grad_norm": 0.5916560888290405,
+      "learning_rate": 0.0005239266055045871,
+      "loss": 3.7433,
       "step": 11850
     },
     {
       "epoch": 1.2830188679245282,
-      "grad_norm": 0.5870354175567627,
-      "learning_rate": 0.0005235963302752293,
-      "loss": 3.7318,
+      "grad_norm": 0.5896475315093994,
+      "learning_rate": 0.0005236028062601186,
+      "loss": 3.7393,
       "step": 11900
     },
     {
       "epoch": 1.2884097035040432,
-      "grad_norm": 0.6072551012039185,
-      "learning_rate": 0.0005232725310307609,
-      "loss": 3.7151,
+      "grad_norm": 0.582275927066803,
+      "learning_rate": 0.0005232790070156503,
+      "loss": 3.7214,
       "step": 11950
     },
     {
       "epoch": 1.2938005390835579,
-      "grad_norm": 0.5396766662597656,
-      "learning_rate": 0.0005229487317862924,
-      "loss": 3.7287,
+      "grad_norm": 0.5797357559204102,
+      "learning_rate": 0.0005229552077711818,
+      "loss": 3.7364,
       "step": 12000
     },
     {
       "epoch": 1.2938005390835579,
-      "eval_accuracy": 0.3499499816430946,
-      "eval_loss": 3.6942965984344482,
-      "eval_runtime": 183.9882,
-      "eval_samples_per_second": 97.892,
-      "eval_steps_per_second": 6.12,
+      "eval_accuracy": 0.3489121292586907,
+      "eval_loss": 3.6995816230773926,
+      "eval_runtime": 181.5844,
+      "eval_samples_per_second": 99.188,
+      "eval_steps_per_second": 6.201,
       "step": 12000
     },
     {
       "epoch": 1.2991913746630728,
-      "grad_norm": 0.5627809166908264,
-      "learning_rate": 0.000522624932541824,
-      "loss": 3.7433,
+      "grad_norm": 0.5879301428794861,
+      "learning_rate": 0.0005226314085267134,
+      "loss": 3.7513,
       "step": 12050
     },
     {
       "epoch": 1.3045822102425877,
-      "grad_norm": 0.5790122151374817,
-      "learning_rate": 0.0005223011332973557,
-      "loss": 3.7371,
+      "grad_norm": 0.6262207627296448,
+      "learning_rate": 0.000522307609282245,
+      "loss": 3.7451,
       "step": 12100
     },
     {
       "epoch": 1.3099730458221024,
-      "grad_norm": 0.5617051720619202,
-      "learning_rate": 0.0005219773340528872,
-      "loss": 3.725,
+      "grad_norm": 0.5379753112792969,
+      "learning_rate": 0.0005219838100377766,
+      "loss": 3.7323,
       "step": 12150
     },
     {
       "epoch": 1.3153638814016173,
-      "grad_norm": 0.5151467323303223,
-      "learning_rate": 0.0005216535348084188,
-      "loss": 3.7435,
+      "grad_norm": 0.5410135984420776,
+      "learning_rate": 0.000521660010793308,
+      "loss": 3.749,
       "step": 12200
     },
     {
       "epoch": 1.320754716981132,
-      "grad_norm": 0.571425199508667,
-      "learning_rate": 0.0005213297355639503,
-      "loss": 3.7168,
+      "grad_norm": 0.603247880935669,
+      "learning_rate": 0.0005213362115488396,
+      "loss": 3.7249,
       "step": 12250
     },
     {
       "epoch": 1.326145552560647,
-      "grad_norm": 0.554076611995697,
-      "learning_rate": 0.0005210059363194819,
-      "loss": 3.7248,
+      "grad_norm": 0.6170111298561096,
+      "learning_rate": 0.0005210124123043713,
+      "loss": 3.7319,
       "step": 12300
     },
     {
       "epoch": 1.3315363881401616,
-      "grad_norm": 0.5426397919654846,
-      "learning_rate": 0.0005206821370750134,
-      "loss": 3.733,
+      "grad_norm": 0.5713621973991394,
+      "learning_rate": 0.0005206886130599028,
+      "loss": 3.7408,
       "step": 12350
     },
     {
       "epoch": 1.3369272237196765,
-      "grad_norm": 0.5451375246047974,
-      "learning_rate": 0.000520358337830545,
-      "loss": 3.7304,
+      "grad_norm": 0.6131263971328735,
+      "learning_rate": 0.0005203648138154344,
+      "loss": 3.7376,
       "step": 12400
     },
     {
       "epoch": 1.3423180592991915,
-      "grad_norm": 0.5756955146789551,
-      "learning_rate": 0.0005200345385860766,
-      "loss": 3.7188,
+      "grad_norm": 0.6258272528648376,
+      "learning_rate": 0.0005200410145709659,
+      "loss": 3.7251,
       "step": 12450
     },
     {
       "epoch": 1.3477088948787062,
-      "grad_norm": 0.5725782513618469,
-      "learning_rate": 0.0005197107393416082,
-      "loss": 3.7372,
+      "grad_norm": 0.6006142497062683,
+      "learning_rate": 0.0005197172153264976,
+      "loss": 3.7444,
       "step": 12500
     },
     {
       "epoch": 1.353099730458221,
-      "grad_norm": 0.5362164974212646,
-      "learning_rate": 0.0005193869400971397,
-      "loss": 3.7099,
+      "grad_norm": 0.5389991998672485,
+      "learning_rate": 0.0005193934160820291,
+      "loss": 3.7174,
       "step": 12550
     },
     {
       "epoch": 1.3584905660377358,
-      "grad_norm": 0.5761519074440002,
-      "learning_rate": 0.0005190631408526713,
-      "loss": 3.7107,
+      "grad_norm": 0.576200008392334,
+      "learning_rate": 0.0005190696168375607,
+      "loss": 3.7154,
       "step": 12600
     },
     {
       "epoch": 1.3638814016172507,
-      "grad_norm": 0.6134148240089417,
-      "learning_rate": 0.0005187393416082029,
-      "loss": 3.7015,
+      "grad_norm": 0.5497236251831055,
+      "learning_rate": 0.0005187458175930922,
+      "loss": 3.7082,
       "step": 12650
     },
     {
       "epoch": 1.3692722371967654,
-      "grad_norm": 0.556781530380249,
+      "grad_norm": 0.5433369874954224,
       "learning_rate": 0.0005184220183486238,
-      "loss": 3.7163,
+      "loss": 3.7215,
       "step": 12700
     },
     {
       "epoch": 1.3746630727762803,
-      "grad_norm": 0.534551739692688,
+      "grad_norm": 0.5957639217376709,
       "learning_rate": 0.0005180982191041554,
-      "loss": 3.7077,
+      "loss": 3.7126,
       "step": 12750
     },
     {
       "epoch": 1.3800539083557952,
-      "grad_norm": 0.5834203362464905,
+      "grad_norm": 0.5817251801490784,
       "learning_rate": 0.000517774419859687,
-      "loss": 3.7058,
+      "loss": 3.7119,
       "step": 12800
     },
     {
       "epoch": 1.38544474393531,
-      "grad_norm": 0.5853343605995178,
-      "learning_rate": 0.0005174506206152185,
-      "loss": 3.7172,
+      "grad_norm": 0.6018874645233154,
+      "learning_rate": 0.0005174570966001078,
+      "loss": 3.7225,
       "step": 12850
     },
     {
       "epoch": 1.3908355795148248,
-      "grad_norm": 0.5919023156166077,
-      "learning_rate": 0.0005171268213707501,
-      "loss": 3.7155,
+      "grad_norm": 0.6323360800743103,
+      "learning_rate": 0.0005171332973556395,
+      "loss": 3.7222,
       "step": 12900
     },
     {
       "epoch": 1.3962264150943398,
-      "grad_norm": 0.5478957295417786,
-      "learning_rate": 0.0005168030221262816,
-      "loss": 3.724,
+      "grad_norm": 0.5343092083930969,
+      "learning_rate": 0.000516809498111171,
+      "loss": 3.7295,
       "step": 12950
     },
     {
       "epoch": 1.4016172506738545,
-      "grad_norm": 0.5379932522773743,
-      "learning_rate": 0.0005164792228818132,
-      "loss": 3.7051,
+      "grad_norm": 0.598622739315033,
+      "learning_rate": 0.0005164856988667026,
+      "loss": 3.7135,
       "step": 13000
     },
     {
       "epoch": 1.4016172506738545,
-      "eval_accuracy": 0.3524520403436865,
-      "eval_loss": 3.672461986541748,
-      "eval_runtime": 183.6622,
-      "eval_samples_per_second": 98.066,
-      "eval_steps_per_second": 6.131,
+      "eval_accuracy": 0.35126739790784517,
+      "eval_loss": 3.6761436462402344,
+      "eval_runtime": 181.5215,
+      "eval_samples_per_second": 99.222,
+      "eval_steps_per_second": 6.203,
       "step": 13000
     },
     {
       "epoch": 1.4070080862533692,
-      "grad_norm": 0.5632694363594055,
-      "learning_rate": 0.0005161554236373448,
-      "loss": 3.7037,
+      "grad_norm": 0.5752383470535278,
+      "learning_rate": 0.0005161618996222341,
+      "loss": 3.7112,
       "step": 13050
     },
     {
       "epoch": 1.412398921832884,
-      "grad_norm": 0.5602201223373413,
-      "learning_rate": 0.0005158316243928764,
-      "loss": 3.698,
+      "grad_norm": 0.6265646815299988,
+      "learning_rate": 0.0005158381003777657,
+      "loss": 3.7052,
       "step": 13100
     },
     {
       "epoch": 1.417789757412399,
-      "grad_norm": 0.6008055806159973,
-      "learning_rate": 0.0005155078251484079,
-      "loss": 3.7036,
+      "grad_norm": 0.5487644672393799,
+      "learning_rate": 0.0005155143011332973,
+      "loss": 3.7102,
       "step": 13150
     },
     {
       "epoch": 1.4231805929919137,
-      "grad_norm": 0.5000778436660767,
-      "learning_rate": 0.0005151840259039395,
-      "loss": 3.7021,
+      "grad_norm": 0.5608962774276733,
+      "learning_rate": 0.0005151905018888289,
+      "loss": 3.7072,
       "step": 13200
     },
     {
       "epoch": 1.4285714285714286,
-      "grad_norm": 0.5265121459960938,
-      "learning_rate": 0.000514860226659471,
-      "loss": 3.7143,
+      "grad_norm": 0.5424461960792542,
+      "learning_rate": 0.0005148667026443604,
+      "loss": 3.7203,
       "step": 13250
     },
     {
       "epoch": 1.4339622641509435,
-      "grad_norm": 0.5470620393753052,
-      "learning_rate": 0.0005145364274150027,
-      "loss": 3.6872,
+      "grad_norm": 0.5759827494621277,
+      "learning_rate": 0.000514542903399892,
+      "loss": 3.6927,
       "step": 13300
     },
     {
       "epoch": 1.4393530997304582,
-      "grad_norm": 0.5715850591659546,
-      "learning_rate": 0.0005142126281705343,
-      "loss": 3.7053,
+      "grad_norm": 0.5633515119552612,
+      "learning_rate": 0.0005142191041554237,
+      "loss": 3.7129,
       "step": 13350
     },
     {
       "epoch": 1.444743935309973,
-      "grad_norm": 0.5808437466621399,
-      "learning_rate": 0.0005138888289260658,
-      "loss": 3.7218,
+      "grad_norm": 0.6104333996772766,
+      "learning_rate": 0.0005138953049109552,
+      "loss": 3.7287,
       "step": 13400
     },
     {
       "epoch": 1.4501347708894878,
-      "grad_norm": 0.5598849058151245,
-      "learning_rate": 0.0005135650296815974,
-      "loss": 3.7246,
+      "grad_norm": 0.5509889125823975,
+      "learning_rate": 0.0005135715056664868,
+      "loss": 3.7316,
       "step": 13450
     },
     {
       "epoch": 1.4555256064690028,
-      "grad_norm": 0.6316524744033813,
-      "learning_rate": 0.0005132412304371289,
-      "loss": 3.7159,
+      "grad_norm": 0.5871342420578003,
+      "learning_rate": 0.0005132477064220183,
+      "loss": 3.7219,
       "step": 13500
     },
     {
       "epoch": 1.4609164420485174,
-      "grad_norm": 0.6849909424781799,
-      "learning_rate": 0.0005129174311926605,
-      "loss": 3.7046,
+      "grad_norm": 0.5736863017082214,
+      "learning_rate": 0.0005129239071775499,
+      "loss": 3.7097,
       "step": 13550
     },
     {
       "epoch": 1.4663072776280324,
-      "grad_norm": 0.6671061515808105,
-      "learning_rate": 0.000512593631948192,
-      "loss": 3.7215,
+      "grad_norm": 0.6301795244216919,
+      "learning_rate": 0.0005126001079330814,
+      "loss": 3.7272,
       "step": 13600
     },
     {
       "epoch": 1.4716981132075473,
-      "grad_norm": 0.5725294947624207,
-      "learning_rate": 0.0005122698327037237,
-      "loss": 3.7105,
+      "grad_norm": 0.6035773754119873,
+      "learning_rate": 0.000512276308688613,
+      "loss": 3.7163,
       "step": 13650
     },
     {
       "epoch": 1.477088948787062,
-      "grad_norm": 0.6049585342407227,
-      "learning_rate": 0.0005119460334592552,
-      "loss": 3.6986,
+      "grad_norm": 0.6564353704452515,
+      "learning_rate": 0.0005119525094441446,
+      "loss": 3.7031,
       "step": 13700
     },
     {
       "epoch": 1.482479784366577,
-      "grad_norm": 0.582939624786377,
-      "learning_rate": 0.0005116222342147868,
-      "loss": 3.699,
+      "grad_norm": 0.5601286292076111,
+      "learning_rate": 0.0005116287101996762,
+      "loss": 3.7065,
       "step": 13750
     },
     {
       "epoch": 1.4878706199460916,
-      "grad_norm": 0.510508120059967,
-      "learning_rate": 0.0005112984349703183,
-      "loss": 3.7006,
+      "grad_norm": 0.5683754086494446,
+      "learning_rate": 0.0005113049109552077,
+      "loss": 3.7097,
       "step": 13800
     },
     {
       "epoch": 1.4932614555256065,
-      "grad_norm": 0.5556900501251221,
-      "learning_rate": 0.00051097463572585,
-      "loss": 3.7061,
+      "grad_norm": 0.5673937797546387,
+      "learning_rate": 0.0005109811117107393,
+      "loss": 3.7118,
       "step": 13850
     },
     {
       "epoch": 1.4986522911051212,
-      "grad_norm": 0.6585386395454407,
-      "learning_rate": 0.0005106508364813815,
-      "loss": 3.6874,
+      "grad_norm": 0.6348940134048462,
+      "learning_rate": 0.0005106573124662708,
+      "loss": 3.6928,
       "step": 13900
     },
     {
       "epoch": 1.5040431266846361,
-      "grad_norm": 0.5924018025398254,
-      "learning_rate": 0.0005103270372369131,
-      "loss": 3.688,
+      "grad_norm": 0.5598605871200562,
+      "learning_rate": 0.0005103335132218025,
+      "loss": 3.6943,
       "step": 13950
     },
     {
       "epoch": 1.509433962264151,
-      "grad_norm": 0.5659307241439819,
-      "learning_rate": 0.0005100032379924446,
-      "loss": 3.7032,
+      "grad_norm": 0.6554436683654785,
+      "learning_rate": 0.000510009713977334,
+      "loss": 3.7085,
       "step": 14000
     },
     {
       "epoch": 1.509433962264151,
-      "eval_accuracy": 0.3539072284701858,
-      "eval_loss": 3.6559207439422607,
-      "eval_runtime": 183.7429,
-      "eval_samples_per_second": 98.023,
-      "eval_steps_per_second": 6.128,
+      "eval_accuracy": 0.3533183298182313,
+      "eval_loss": 3.659724473953247,
+      "eval_runtime": 184.2123,
+      "eval_samples_per_second": 97.773,
+      "eval_steps_per_second": 6.113,
       "step": 14000
     },
     {
       "epoch": 1.5148247978436657,
-      "grad_norm": 0.5375577807426453,
-      "learning_rate": 0.0005096794387479762,
-      "loss": 3.7013,
+      "grad_norm": 0.5855810046195984,
+      "learning_rate": 0.0005096859147328656,
+      "loss": 3.7067,
       "step": 14050
     },
     {
       "epoch": 1.5202156334231804,
-      "grad_norm": 0.631818950176239,
-      "learning_rate": 0.0005093556395035078,
-      "loss": 3.6999,
+      "grad_norm": 0.5719590187072754,
+      "learning_rate": 0.0005093621154883971,
+      "loss": 3.7044,
       "step": 14100
     },
     {
       "epoch": 1.5256064690026954,
-      "grad_norm": 0.56558758020401,
-      "learning_rate": 0.0005090318402590394,
-      "loss": 3.6931,
+      "grad_norm": 0.5538426637649536,
+      "learning_rate": 0.0005090383162439288,
+      "loss": 3.6983,
       "step": 14150
     },
     {
       "epoch": 1.5309973045822103,
-      "grad_norm": 0.5545461773872375,
-      "learning_rate": 0.0005087080410145709,
-      "loss": 3.6975,
+      "grad_norm": 0.5432640910148621,
+      "learning_rate": 0.0005087145169994602,
+      "loss": 3.7045,
       "step": 14200
     },
     {
       "epoch": 1.536388140161725,
-      "grad_norm": 0.5795785784721375,
-      "learning_rate": 0.0005083842417701025,
-      "loss": 3.6976,
+      "grad_norm": 0.6232120394706726,
+      "learning_rate": 0.0005083907177549918,
+      "loss": 3.7069,
       "step": 14250
     },
     {
       "epoch": 1.54177897574124,
-      "grad_norm": 0.5446183681488037,
-      "learning_rate": 0.000508060442525634,
-      "loss": 3.6944,
+      "grad_norm": 0.5137596726417542,
+      "learning_rate": 0.0005080669185105234,
+      "loss": 3.7,
       "step": 14300
     },
     {
       "epoch": 1.5471698113207548,
-      "grad_norm": 0.5753372311592102,
-      "learning_rate": 0.0005077366432811656,
-      "loss": 3.6866,
+      "grad_norm": 0.5768091678619385,
+      "learning_rate": 0.000507743119266055,
+      "loss": 3.6902,
       "step": 14350
     },
     {
       "epoch": 1.5525606469002695,
-      "grad_norm": 0.5649062991142273,
-      "learning_rate": 0.0005074128440366971,
-      "loss": 3.6891,
+      "grad_norm": 0.596432626247406,
+      "learning_rate": 0.0005074193200215865,
+      "loss": 3.6943,
       "step": 14400
     },
     {
       "epoch": 1.5579514824797842,
-      "grad_norm": 0.5094801783561707,
-      "learning_rate": 0.0005070890447922288,
-      "loss": 3.6906,
+      "grad_norm": 0.5589677095413208,
+      "learning_rate": 0.0005070955207771181,
+      "loss": 3.6982,
       "step": 14450
     },
     {
       "epoch": 1.5633423180592994,
-      "grad_norm": 0.5297936797142029,
-      "learning_rate": 0.0005067652455477604,
-      "loss": 3.6888,
+      "grad_norm": 0.5615023374557495,
+      "learning_rate": 0.0005067717215326498,
+      "loss": 3.6963,
       "step": 14500
     },
     {
       "epoch": 1.568733153638814,
-      "grad_norm": 0.537732720375061,
-      "learning_rate": 0.0005064414463032919,
-      "loss": 3.6874,
+      "grad_norm": 0.5714856386184692,
+      "learning_rate": 0.0005064479222881813,
+      "loss": 3.6955,
       "step": 14550
     },
     {
       "epoch": 1.5741239892183287,
-      "grad_norm": 0.5864264369010925,
-      "learning_rate": 0.0005061176470588235,
-      "loss": 3.6742,
+      "grad_norm": 0.5951200127601624,
+      "learning_rate": 0.0005061241230437129,
+      "loss": 3.6794,
       "step": 14600
     },
     {
       "epoch": 1.5795148247978437,
-      "grad_norm": 0.5446892380714417,
-      "learning_rate": 0.0005057938478143551,
-      "loss": 3.6769,
+      "grad_norm": 0.5872084498405457,
+      "learning_rate": 0.0005058003237992444,
+      "loss": 3.683,
       "step": 14650
     },
     {
       "epoch": 1.5849056603773586,
-      "grad_norm": 0.5335270762443542,
-      "learning_rate": 0.0005054700485698867,
-      "loss": 3.6846,
+      "grad_norm": 0.6111684441566467,
+      "learning_rate": 0.000505476524554776,
+      "loss": 3.6911,
       "step": 14700
     },
     {
       "epoch": 1.5902964959568733,
-      "grad_norm": 0.5628997087478638,
+      "grad_norm": 0.5691280961036682,
       "learning_rate": 0.0005051527253103076,
-      "loss": 3.664,
+      "loss": 3.6702,
       "step": 14750
     },
     {
       "epoch": 1.595687331536388,
-      "grad_norm": 0.5562835931777954,
+      "grad_norm": 0.5549219846725464,
       "learning_rate": 0.0005048289260658392,
-      "loss": 3.6578,
+      "loss": 3.6621,
       "step": 14800
     },
     {
       "epoch": 1.6010781671159031,
-      "grad_norm": 0.5934857726097107,
+      "grad_norm": 0.596741795539856,
       "learning_rate": 0.0005045051268213707,
-      "loss": 3.6825,
+      "loss": 3.6877,
       "step": 14850
     },
     {
       "epoch": 1.6064690026954178,
-      "grad_norm": 0.5580174922943115,
+      "grad_norm": 0.5717527270317078,
       "learning_rate": 0.0005041813275769023,
-      "loss": 3.6859,
+      "loss": 3.6919,
       "step": 14900
     },
     {
       "epoch": 1.6118598382749325,
-      "grad_norm": 0.5808007121086121,
+      "grad_norm": 0.5621931552886963,
       "learning_rate": 0.0005038575283324338,
-      "loss": 3.6708,
+      "loss": 3.678,
       "step": 14950
     },
     {
       "epoch": 1.6172506738544474,
-      "grad_norm": 0.504628598690033,
+      "grad_norm": 0.5136688947677612,
       "learning_rate": 0.0005035337290879654,
-      "loss": 3.6641,
+      "loss": 3.6698,
       "step": 15000
     },
     {
       "epoch": 1.6172506738544474,
-      "eval_accuracy": 0.3560253078652413,
-      "eval_loss": 3.6341910362243652,
-      "eval_runtime": 183.7265,
-      "eval_samples_per_second": 98.032,
-      "eval_steps_per_second": 6.129,
+      "eval_accuracy": 0.35519280943880677,
+      "eval_loss": 3.6415674686431885,
+      "eval_runtime": 184.4321,
+      "eval_samples_per_second": 97.657,
+      "eval_steps_per_second": 6.105,
       "step": 15000
     },
     {
       "epoch": 1.6226415094339623,
-      "grad_norm": 0.5475382208824158,
+      "grad_norm": 0.5894722938537598,
       "learning_rate": 0.000503209929843497,
-      "loss": 3.6619,
+      "loss": 3.6679,
       "step": 15050
     },
     {
       "epoch": 1.628032345013477,
-      "grad_norm": 0.561851978302002,
+      "grad_norm": 0.5535485148429871,
       "learning_rate": 0.0005028861305990286,
-      "loss": 3.6737,
+      "loss": 3.6792,
       "step": 15100
     },
     {
       "epoch": 1.633423180592992,
-      "grad_norm": 0.5445704460144043,
-      "learning_rate": 0.0005025623313545601,
-      "loss": 3.6936,
+      "grad_norm": 0.551805853843689,
+      "learning_rate": 0.0005025688073394495,
+      "loss": 3.6981,
       "step": 15150
     },
     {
       "epoch": 1.6388140161725069,
-      "grad_norm": 0.5345889329910278,
-      "learning_rate": 0.0005022385321100917,
-      "loss": 3.6804,
+      "grad_norm": 0.5345020890235901,
+      "learning_rate": 0.0005022450080949811,
+      "loss": 3.6856,
       "step": 15200
     },
     {
       "epoch": 1.6442048517520216,
-      "grad_norm": 0.4933413863182068,
-      "learning_rate": 0.0005019147328656232,
-      "loss": 3.6721,
+      "grad_norm": 0.5428183078765869,
+      "learning_rate": 0.0005019212088505126,
+      "loss": 3.6795,
       "step": 15250
     },
     {
       "epoch": 1.6495956873315363,
-      "grad_norm": 0.5393181443214417,
-      "learning_rate": 0.0005015909336211549,
-      "loss": 3.6716,
+      "grad_norm": 0.6240091323852539,
+      "learning_rate": 0.0005015974096060442,
+      "loss": 3.6773,
       "step": 15300
     },
     {
       "epoch": 1.6549865229110512,
-      "grad_norm": 0.5043426752090454,
-      "learning_rate": 0.0005012671343766864,
-      "loss": 3.6568,
+      "grad_norm": 0.5113633871078491,
+      "learning_rate": 0.0005012736103615758,
+      "loss": 3.663,
       "step": 15350
     },
     {
       "epoch": 1.6603773584905661,
-      "grad_norm": 0.5442577004432678,
-      "learning_rate": 0.000500943335132218,
-      "loss": 3.675,
+      "grad_norm": 0.5642730593681335,
+      "learning_rate": 0.0005009498111171074,
+      "loss": 3.6806,
       "step": 15400
     },
     {
       "epoch": 1.6657681940700808,
-      "grad_norm": 0.5932238101959229,
-      "learning_rate": 0.0005006195358877495,
-      "loss": 3.6634,
+      "grad_norm": 0.6118944883346558,
+      "learning_rate": 0.0005006260118726389,
+      "loss": 3.6703,
       "step": 15450
     },
     {
       "epoch": 1.6711590296495957,
-      "grad_norm": 0.5122867226600647,
-      "learning_rate": 0.0005002957366432812,
-      "loss": 3.6645,
+      "grad_norm": 0.5733397006988525,
+      "learning_rate": 0.0005003022126281705,
+      "loss": 3.6719,
       "step": 15500
     },
     {
       "epoch": 1.6765498652291106,
-      "grad_norm": 0.5891242027282715,
-      "learning_rate": 0.0004999719373988127,
-      "loss": 3.651,
+      "grad_norm": 0.5732635855674744,
+      "learning_rate": 0.000499978413383702,
+      "loss": 3.656,
       "step": 15550
     },
     {
       "epoch": 1.6819407008086253,
-      "grad_norm": 0.570210874080658,
-      "learning_rate": 0.0004996481381543442,
-      "loss": 3.6675,
+      "grad_norm": 0.579186201095581,
+      "learning_rate": 0.0004996546141392336,
+      "loss": 3.6748,
       "step": 15600
     },
     {
       "epoch": 1.68733153638814,
-      "grad_norm": 0.5468628406524658,
-      "learning_rate": 0.0004993243389098758,
-      "loss": 3.6734,
+      "grad_norm": 0.5598515868186951,
+      "learning_rate": 0.0004993308148947651,
+      "loss": 3.6822,
       "step": 15650
     },
     {
       "epoch": 1.692722371967655,
-      "grad_norm": 0.6194903254508972,
-      "learning_rate": 0.0004990005396654074,
-      "loss": 3.6724,
+      "grad_norm": 0.6149080991744995,
+      "learning_rate": 0.0004990070156502968,
+      "loss": 3.6771,
       "step": 15700
     },
     {
       "epoch": 1.6981132075471699,
-      "grad_norm": 0.5616596937179565,
-      "learning_rate": 0.000498676740420939,
-      "loss": 3.6535,
+      "grad_norm": 0.5728682279586792,
+      "learning_rate": 0.0004986832164058284,
+      "loss": 3.6572,
       "step": 15750
     },
     {
       "epoch": 1.7035040431266846,
-      "grad_norm": 0.5665090084075928,
-      "learning_rate": 0.0004983529411764705,
-      "loss": 3.6628,
+      "grad_norm": 0.5917532444000244,
+      "learning_rate": 0.0004983594171613599,
+      "loss": 3.6691,
       "step": 15800
     },
     {
       "epoch": 1.7088948787061995,
-      "grad_norm": 0.5654871463775635,
-      "learning_rate": 0.0004980291419320022,
-      "loss": 3.6803,
+      "grad_norm": 0.5908268094062805,
+      "learning_rate": 0.0004980356179168915,
+      "loss": 3.6856,
       "step": 15850
     },
     {
       "epoch": 1.7142857142857144,
-      "grad_norm": 0.5348817110061646,
-      "learning_rate": 0.0004977053426875337,
-      "loss": 3.6793,
+      "grad_norm": 0.566153347492218,
+      "learning_rate": 0.000497711818672423,
+      "loss": 3.6855,
       "step": 15900
     },
     {
       "epoch": 1.719676549865229,
-      "grad_norm": 0.556675136089325,
-      "learning_rate": 0.0004973815434430653,
-      "loss": 3.6665,
+      "grad_norm": 0.518864095211029,
+      "learning_rate": 0.0004973880194279547,
+      "loss": 3.6739,
       "step": 15950
     },
     {
       "epoch": 1.7250673854447438,
-      "grad_norm": 0.5713870525360107,
-      "learning_rate": 0.0004970577441985968,
-      "loss": 3.6638,
+      "grad_norm": 0.5470095276832581,
+      "learning_rate": 0.0004970642201834862,
+      "loss": 3.6694,
       "step": 16000
     },
     {
       "epoch": 1.7250673854447438,
-      "eval_accuracy": 0.35771866312183,
-      "eval_loss": 3.618290662765503,
-      "eval_runtime": 183.7921,
-      "eval_samples_per_second": 97.997,
-      "eval_steps_per_second": 6.126,
+      "eval_accuracy": 0.3571103156029567,
+      "eval_loss": 3.623772621154785,
+      "eval_runtime": 184.5354,
+      "eval_samples_per_second": 97.602,
+      "eval_steps_per_second": 6.102,
       "step": 16000
     },
     {
       "epoch": 1.7304582210242587,
-      "grad_norm": 0.5948731899261475,
-      "learning_rate": 0.0004967339449541284,
-      "loss": 3.6498,
+      "grad_norm": 0.599932074546814,
+      "learning_rate": 0.0004967404209390178,
+      "loss": 3.6553,
       "step": 16050
     },
     {
       "epoch": 1.7358490566037736,
-      "grad_norm": 0.5521835684776306,
-      "learning_rate": 0.00049641014570966,
-      "loss": 3.6514,
+      "grad_norm": 0.5855822563171387,
+      "learning_rate": 0.0004964166216945493,
+      "loss": 3.6578,
       "step": 16100
     },
     {
       "epoch": 1.7412398921832883,
-      "grad_norm": 0.6063022613525391,
-      "learning_rate": 0.0004960863464651916,
-      "loss": 3.6574,
+      "grad_norm": 0.6412010192871094,
+      "learning_rate": 0.000496092822450081,
+      "loss": 3.6669,
       "step": 16150
     },
     {
       "epoch": 1.7466307277628033,
-      "grad_norm": 0.6571112871170044,
-      "learning_rate": 0.0004957625472207231,
-      "loss": 3.6673,
+      "grad_norm": 0.6825807690620422,
+      "learning_rate": 0.0004957690232056125,
+      "loss": 3.6734,
       "step": 16200
     },
     {
       "epoch": 1.7520215633423182,
-      "grad_norm": 0.5960469841957092,
-      "learning_rate": 0.0004954387479762547,
-      "loss": 3.65,
+      "grad_norm": 0.6272662878036499,
+      "learning_rate": 0.0004954452239611441,
+      "loss": 3.6553,
       "step": 16250
     },
     {
       "epoch": 1.7574123989218329,
-      "grad_norm": 0.518771231174469,
-      "learning_rate": 0.0004951149487317862,
-      "loss": 3.6426,
+      "grad_norm": 0.570270836353302,
+      "learning_rate": 0.0004951214247166756,
+      "loss": 3.648,
       "step": 16300
     },
     {
       "epoch": 1.7628032345013476,
-      "grad_norm": 0.5418721437454224,
-      "learning_rate": 0.0004947911494873178,
-      "loss": 3.6663,
+      "grad_norm": 0.5867336988449097,
+      "learning_rate": 0.0004947976254722072,
+      "loss": 3.6742,
       "step": 16350
     },
     {
       "epoch": 1.7681940700808625,
-      "grad_norm": 0.4927101135253906,
-      "learning_rate": 0.0004944673502428493,
-      "loss": 3.646,
+      "grad_norm": 0.5367298722267151,
+      "learning_rate": 0.0004944738262277387,
+      "loss": 3.6513,
       "step": 16400
     },
     {
       "epoch": 1.7735849056603774,
-      "grad_norm": 0.5151129364967346,
-      "learning_rate": 0.000494143550998381,
-      "loss": 3.6555,
+      "grad_norm": 0.5539150238037109,
+      "learning_rate": 0.0004941500269832703,
+      "loss": 3.662,
       "step": 16450
     },
     {
       "epoch": 1.778975741239892,
-      "grad_norm": 0.5475391149520874,
-      "learning_rate": 0.0004938197517539125,
-      "loss": 3.6402,
+      "grad_norm": 0.5436108112335205,
+      "learning_rate": 0.0004938262277388019,
+      "loss": 3.6472,
       "step": 16500
     },
     {
       "epoch": 1.784366576819407,
-      "grad_norm": 0.5400313138961792,
-      "learning_rate": 0.0004934959525094441,
-      "loss": 3.6485,
+      "grad_norm": 0.5677002668380737,
+      "learning_rate": 0.0004935024284943335,
+      "loss": 3.6546,
       "step": 16550
     },
     {
       "epoch": 1.789757412398922,
-      "grad_norm": 0.5347439646720886,
-      "learning_rate": 0.0004931721532649756,
-      "loss": 3.6607,
+      "grad_norm": 0.4927968680858612,
+      "learning_rate": 0.000493178629249865,
+      "loss": 3.6675,
       "step": 16600
     },
     {
       "epoch": 1.7951482479784366,
-      "grad_norm": 0.5058249235153198,
-      "learning_rate": 0.0004928483540205073,
-      "loss": 3.6536,
+      "grad_norm": 0.5354840755462646,
+      "learning_rate": 0.0004928548300053966,
+      "loss": 3.6587,
       "step": 16650
     },
     {
       "epoch": 1.8005390835579513,
-      "grad_norm": 0.5410025119781494,
-      "learning_rate": 0.0004925245547760388,
-      "loss": 3.6574,
+      "grad_norm": 0.6506155133247375,
+      "learning_rate": 0.0004925310307609282,
+      "loss": 3.6647,
       "step": 16700
     },
     {
       "epoch": 1.8059299191374663,
-      "grad_norm": 0.5334374904632568,
+      "grad_norm": 0.5581393241882324,
       "learning_rate": 0.0004922072315164598,
-      "loss": 3.6566,
+      "loss": 3.6619,
       "step": 16750
     },
     {
       "epoch": 1.8113207547169812,
-      "grad_norm": 0.5640711784362793,
+      "grad_norm": 0.6036659479141235,
       "learning_rate": 0.0004918834322719913,
-      "loss": 3.6645,
+      "loss": 3.6719,
       "step": 16800
     },
     {
       "epoch": 1.8167115902964959,
-      "grad_norm": 0.5393502116203308,
+      "grad_norm": 0.5904647707939148,
       "learning_rate": 0.0004915596330275229,
-      "loss": 3.6718,
+      "loss": 3.6787,
       "step": 16850
     },
     {
       "epoch": 1.8221024258760108,
-      "grad_norm": 0.5351586937904358,
+      "grad_norm": 0.5560145974159241,
       "learning_rate": 0.0004912358337830544,
-      "loss": 3.6445,
+      "loss": 3.6519,
       "step": 16900
     },
     {
       "epoch": 1.8274932614555257,
-      "grad_norm": 0.5266234874725342,
+      "grad_norm": 0.5826396942138672,
       "learning_rate": 0.000490912034538586,
-      "loss": 3.6559,
+      "loss": 3.6626,
       "step": 16950
     },
     {
       "epoch": 1.8328840970350404,
-      "grad_norm": 0.5003353357315063,
+      "grad_norm": 0.542641818523407,
       "learning_rate": 0.0004905882352941175,
-      "loss": 3.6368,
+      "loss": 3.6432,
       "step": 17000
     },
     {
       "epoch": 1.8328840970350404,
-      "eval_accuracy": 0.3592136182052694,
-      "eval_loss": 3.602701187133789,
-      "eval_runtime": 183.6818,
-      "eval_samples_per_second": 98.055,
-      "eval_steps_per_second": 6.13,
+      "eval_accuracy": 0.3584828188818292,
+      "eval_loss": 3.6078975200653076,
+      "eval_runtime": 184.2782,
+      "eval_samples_per_second": 97.738,
+      "eval_steps_per_second": 6.11,
       "step": 17000
     },
     {
       "epoch": 1.838274932614555,
-      "grad_norm": 0.5567882657051086,
+      "grad_norm": 0.5734180212020874,
       "learning_rate": 0.0004902644360496492,
-      "loss": 3.6407,
+      "loss": 3.6482,
       "step": 17050
     },
     {
       "epoch": 1.8436657681940702,
-      "grad_norm": 0.5742266178131104,
+      "grad_norm": 0.5473641157150269,
       "learning_rate": 0.0004899406368051808,
-      "loss": 3.6327,
+      "loss": 3.6388,
       "step": 17100
     },
     {
       "epoch": 1.849056603773585,
-      "grad_norm": 0.5474602580070496,
-      "learning_rate": 0.0004896168375607123,
-      "loss": 3.6396,
+      "grad_norm": 0.5937027335166931,
+      "learning_rate": 0.0004896233135456017,
+      "loss": 3.6462,
       "step": 17150
     },
     {
       "epoch": 1.8544474393530996,
-      "grad_norm": 0.5230289697647095,
-      "learning_rate": 0.0004892930383162439,
-      "loss": 3.6459,
+      "grad_norm": 0.548816442489624,
+      "learning_rate": 0.0004892995143011333,
+      "loss": 3.6523,
       "step": 17200
     },
     {
       "epoch": 1.8598382749326146,
-      "grad_norm": 0.4841998517513275,
-      "learning_rate": 0.0004889692390717754,
-      "loss": 3.6515,
+      "grad_norm": 0.5214555263519287,
+      "learning_rate": 0.0004889757150566648,
+      "loss": 3.6586,
       "step": 17250
     },
     {
       "epoch": 1.8652291105121295,
-      "grad_norm": 0.5501573085784912,
-      "learning_rate": 0.0004886454398273071,
-      "loss": 3.6547,
+      "grad_norm": 0.5880135297775269,
+      "learning_rate": 0.0004886519158121964,
+      "loss": 3.662,
       "step": 17300
     },
     {
       "epoch": 1.8706199460916442,
-      "grad_norm": 0.543190598487854,
-      "learning_rate": 0.0004883216405828386,
-      "loss": 3.6554,
+      "grad_norm": 0.58387690782547,
+      "learning_rate": 0.000488328116567728,
+      "loss": 3.6608,
       "step": 17350
     },
     {
       "epoch": 1.8760107816711589,
-      "grad_norm": 0.5470271706581116,
-      "learning_rate": 0.00048799784133837017,
-      "loss": 3.6487,
+      "grad_norm": 0.543831467628479,
+      "learning_rate": 0.0004880043173232595,
+      "loss": 3.6545,
       "step": 17400
     },
     {
       "epoch": 1.881401617250674,
-      "grad_norm": 0.5897737145423889,
-      "learning_rate": 0.0004876740420939017,
-      "loss": 3.6389,
+      "grad_norm": 0.5606027245521545,
+      "learning_rate": 0.0004876805180787911,
+      "loss": 3.6429,
       "step": 17450
     },
     {
       "epoch": 1.8867924528301887,
-      "grad_norm": 0.5379341244697571,
-      "learning_rate": 0.0004873502428494333,
-      "loss": 3.6411,
+      "grad_norm": 0.5298070311546326,
+      "learning_rate": 0.0004873567188343227,
+      "loss": 3.6465,
       "step": 17500
     },
     {
       "epoch": 1.8921832884097034,
-      "grad_norm": 0.5637924671173096,
-      "learning_rate": 0.0004870264436049649,
-      "loss": 3.6377,
+      "grad_norm": 0.6168541312217712,
+      "learning_rate": 0.0004870329195898542,
+      "loss": 3.6431,
       "step": 17550
     },
     {
       "epoch": 1.8975741239892183,
-      "grad_norm": 0.5570976734161377,
-      "learning_rate": 0.00048670264436049643,
-      "loss": 3.6509,
+      "grad_norm": 0.5477967262268066,
+      "learning_rate": 0.00048670912034538583,
+      "loss": 3.6576,
       "step": 17600
     },
     {
       "epoch": 1.9029649595687332,
-      "grad_norm": 0.6049391031265259,
-      "learning_rate": 0.00048637884511602803,
-      "loss": 3.6425,
+      "grad_norm": 0.6176745295524597,
+      "learning_rate": 0.0004863853211009174,
+      "loss": 3.6485,
       "step": 17650
     },
     {
       "epoch": 1.908355795148248,
-      "grad_norm": 0.5467279553413391,
-      "learning_rate": 0.0004860550458715596,
-      "loss": 3.6374,
+      "grad_norm": 0.5221447348594666,
+      "learning_rate": 0.000486061521856449,
+      "loss": 3.6423,
       "step": 17700
     },
     {
       "epoch": 1.9137466307277629,
-      "grad_norm": 0.5799625515937805,
-      "learning_rate": 0.0004857312466270912,
-      "loss": 3.6579,
+      "grad_norm": 0.5973334312438965,
+      "learning_rate": 0.00048573772261198054,
+      "loss": 3.6639,
       "step": 17750
     },
     {
       "epoch": 1.9191374663072778,
-      "grad_norm": 0.5544312000274658,
-      "learning_rate": 0.00048540744738262274,
-      "loss": 3.6415,
+      "grad_norm": 0.586326539516449,
+      "learning_rate": 0.00048541392336751214,
+      "loss": 3.648,
       "step": 17800
     },
     {
       "epoch": 1.9245283018867925,
-      "grad_norm": 0.577680766582489,
-      "learning_rate": 0.00048508364813815434,
-      "loss": 3.6342,
+      "grad_norm": 0.6071659922599792,
+      "learning_rate": 0.0004850901241230437,
+      "loss": 3.6391,
       "step": 17850
     },
     {
       "epoch": 1.9299191374663072,
-      "grad_norm": 0.5421233177185059,
-      "learning_rate": 0.00048475984889368584,
-      "loss": 3.6444,
+      "grad_norm": 0.5929883718490601,
+      "learning_rate": 0.0004847663248785753,
+      "loss": 3.6502,
       "step": 17900
     },
     {
       "epoch": 1.935309973045822,
-      "grad_norm": 0.5379252433776855,
-      "learning_rate": 0.0004844360496492175,
-      "loss": 3.6261,
+      "grad_norm": 0.5701756477355957,
+      "learning_rate": 0.0004844425256341068,
+      "loss": 3.6313,
       "step": 17950
     },
     {
       "epoch": 1.940700808625337,
-      "grad_norm": 0.5494712591171265,
-      "learning_rate": 0.000484112250404749,
-      "loss": 3.6291,
+      "grad_norm": 0.5965040922164917,
+      "learning_rate": 0.00048411872638963834,
+      "loss": 3.6354,
       "step": 18000
     },
     {
       "epoch": 1.940700808625337,
-      "eval_accuracy": 0.3600362292189129,
-      "eval_loss": 3.5892419815063477,
-      "eval_runtime": 184.0541,
-      "eval_samples_per_second": 97.857,
-      "eval_steps_per_second": 6.118,
+      "eval_accuracy": 0.3595456614304305,
+      "eval_loss": 3.592846632003784,
+      "eval_runtime": 183.8363,
+      "eval_samples_per_second": 97.973,
+      "eval_steps_per_second": 6.125,
       "step": 18000
     },
     {
       "epoch": 1.9460916442048517,
-      "grad_norm": 0.5733945369720459,
-      "learning_rate": 0.00048378845116028055,
-      "loss": 3.6336,
+      "grad_norm": 0.5628731846809387,
+      "learning_rate": 0.00048379492714516995,
+      "loss": 3.6392,
       "step": 18050
     },
     {
       "epoch": 1.9514824797843666,
-      "grad_norm": 0.567290186882019,
-      "learning_rate": 0.00048346465191581215,
-      "loss": 3.633,
+      "grad_norm": 0.553044319152832,
+      "learning_rate": 0.0004834711279007015,
+      "loss": 3.6394,
       "step": 18100
     },
     {
       "epoch": 1.9568733153638815,
-      "grad_norm": 0.5201252698898315,
-      "learning_rate": 0.0004831408526713437,
-      "loss": 3.6308,
+      "grad_norm": 0.5722833871841431,
+      "learning_rate": 0.0004831473286562331,
+      "loss": 3.6365,
       "step": 18150
     },
     {
       "epoch": 1.9622641509433962,
-      "grad_norm": 0.5754126310348511,
-      "learning_rate": 0.0004828170534268753,
-      "loss": 3.6186,
+      "grad_norm": 0.5541419386863708,
+      "learning_rate": 0.00048282352941176465,
+      "loss": 3.626,
       "step": 18200
     },
     {
       "epoch": 1.967654986522911,
-      "grad_norm": 0.5907006859779358,
-      "learning_rate": 0.00048249325418240686,
-      "loss": 3.6398,
+      "grad_norm": 0.6036373376846313,
+      "learning_rate": 0.00048249973016729626,
+      "loss": 3.6459,
       "step": 18250
     },
     {
       "epoch": 1.9730458221024259,
-      "grad_norm": 0.6127999424934387,
-      "learning_rate": 0.00048216945493793846,
-      "loss": 3.6509,
+      "grad_norm": 0.5856481194496155,
+      "learning_rate": 0.0004821759309228278,
+      "loss": 3.6561,
       "step": 18300
     },
     {
       "epoch": 1.9784366576819408,
-      "grad_norm": 0.5465409159660339,
-      "learning_rate": 0.00048184565569347,
-      "loss": 3.6337,
+      "grad_norm": 0.5856950283050537,
+      "learning_rate": 0.00048185213167835936,
+      "loss": 3.6416,
       "step": 18350
     },
     {
       "epoch": 1.9838274932614555,
-      "grad_norm": 0.6550977230072021,
-      "learning_rate": 0.0004815218564490016,
-      "loss": 3.636,
+      "grad_norm": 0.6323913335800171,
+      "learning_rate": 0.00048152833243389096,
+      "loss": 3.643,
       "step": 18400
     },
     {
       "epoch": 1.9892183288409704,
-      "grad_norm": 0.5740667581558228,
-      "learning_rate": 0.00048119805720453317,
-      "loss": 3.6441,
+      "grad_norm": 0.5707237124443054,
+      "learning_rate": 0.0004812045331894225,
+      "loss": 3.6505,
       "step": 18450
     },
     {
       "epoch": 1.9946091644204853,
-      "grad_norm": 0.5781065225601196,
-      "learning_rate": 0.0004808742579600647,
-      "loss": 3.6359,
+      "grad_norm": 0.5647567510604858,
+      "learning_rate": 0.0004808807339449541,
+      "loss": 3.6416,
       "step": 18500
     },
     {
       "epoch": 2.0,
-      "grad_norm": 1.1652530431747437,
-      "learning_rate": 0.0004805504587155963,
-      "loss": 3.6409,
+      "grad_norm": 1.2244846820831299,
+      "learning_rate": 0.0004805569347004856,
+      "loss": 3.6479,
       "step": 18550
     },
     {
       "epoch": 2.0053908355795147,
-      "grad_norm": 0.5303277373313904,
-      "learning_rate": 0.0004802266594711278,
-      "loss": 3.5688,
+      "grad_norm": 0.5471111536026001,
+      "learning_rate": 0.0004802331354560173,
+      "loss": 3.5743,
       "step": 18600
     },
     {
       "epoch": 2.01078167115903,
-      "grad_norm": 0.5572181344032288,
-      "learning_rate": 0.0004799028602266594,
-      "loss": 3.5383,
+      "grad_norm": 0.5748924016952515,
+      "learning_rate": 0.00047990933621154877,
+      "loss": 3.544,
       "step": 18650
     },
     {
       "epoch": 2.0161725067385445,
-      "grad_norm": 0.577308177947998,
-      "learning_rate": 0.000479579060982191,
-      "loss": 3.5517,
+      "grad_norm": 0.5614541172981262,
+      "learning_rate": 0.0004795855369670804,
+      "loss": 3.5572,
       "step": 18700
     },
     {
       "epoch": 2.0215633423180592,
-      "grad_norm": 0.5191822648048401,
-      "learning_rate": 0.0004792552617377226,
-      "loss": 3.5297,
+      "grad_norm": 0.5483747720718384,
+      "learning_rate": 0.0004792617377226119,
+      "loss": 3.5367,
       "step": 18750
     },
     {
       "epoch": 2.026954177897574,
-      "grad_norm": 0.5460506081581116,
-      "learning_rate": 0.00047893146249325413,
-      "loss": 3.5495,
+      "grad_norm": 0.6159261465072632,
+      "learning_rate": 0.0004789379384781435,
+      "loss": 3.5543,
       "step": 18800
     },
     {
       "epoch": 2.032345013477089,
-      "grad_norm": 0.5392601490020752,
+      "grad_norm": 0.5394082069396973,
       "learning_rate": 0.0004786141392336751,
-      "loss": 3.5583,
+      "loss": 3.5623,
       "step": 18850
     },
     {
       "epoch": 2.0377358490566038,
-      "grad_norm": 0.5567241311073303,
+      "grad_norm": 0.5845710039138794,
       "learning_rate": 0.00047829033998920663,
-      "loss": 3.5447,
+      "loss": 3.5513,
       "step": 18900
     },
     {
       "epoch": 2.0431266846361185,
-      "grad_norm": 0.5814675092697144,
+      "grad_norm": 0.5727204084396362,
       "learning_rate": 0.00047796654074473824,
-      "loss": 3.539,
+      "loss": 3.5448,
       "step": 18950
     },
     {
       "epoch": 2.0485175202156336,
-      "grad_norm": 0.6219574213027954,
+      "grad_norm": 0.5626639723777771,
       "learning_rate": 0.0004776427415002698,
-      "loss": 3.5668,
+      "loss": 3.5732,
       "step": 19000
     },
     {
       "epoch": 2.0485175202156336,
-      "eval_accuracy": 0.3623585760430596,
-      "eval_loss": 3.5776376724243164,
-      "eval_runtime": 183.6098,
-      "eval_samples_per_second": 98.094,
-      "eval_steps_per_second": 6.133,
+      "eval_accuracy": 0.3612944296598046,
+      "eval_loss": 3.585456371307373,
+      "eval_runtime": 184.0545,
+      "eval_samples_per_second": 97.857,
+      "eval_steps_per_second": 6.118,
       "step": 19000
     },
     {
       "epoch": 2.0539083557951483,
-      "grad_norm": 0.5591341257095337,
+      "grad_norm": 0.5868081450462341,
       "learning_rate": 0.0004773189422558014,
-      "loss": 3.5422,
+      "loss": 3.5489,
       "step": 19050
     },
     {
       "epoch": 2.059299191374663,
-      "grad_norm": 0.5593909025192261,
+      "grad_norm": 0.5627261996269226,
       "learning_rate": 0.00047699514301133294,
-      "loss": 3.566,
+      "loss": 3.5708,
       "step": 19100
     },
     {
       "epoch": 2.0646900269541777,
-      "grad_norm": 0.5888485908508301,
+      "grad_norm": 0.5930964350700378,
       "learning_rate": 0.00047667134376686455,
-      "loss": 3.5615,
+      "loss": 3.5674,
       "step": 19150
     },
     {
       "epoch": 2.070080862533693,
-      "grad_norm": 0.5862798094749451,
-      "learning_rate": 0.0004763475445223961,
-      "loss": 3.5691,
+      "grad_norm": 0.5827620625495911,
+      "learning_rate": 0.0004763540205072854,
+      "loss": 3.5755,
       "step": 19200
     },
     {
       "epoch": 2.0754716981132075,
-      "grad_norm": 0.5456637740135193,
-      "learning_rate": 0.0004760237452779276,
-      "loss": 3.5466,
+      "grad_norm": 0.5772703886032104,
+      "learning_rate": 0.00047603022126281705,
+      "loss": 3.5545,
       "step": 19250
     },
     {
       "epoch": 2.0808625336927222,
-      "grad_norm": 0.5669352412223816,
-      "learning_rate": 0.0004756999460334592,
-      "loss": 3.544,
+      "grad_norm": 0.5826517939567566,
+      "learning_rate": 0.00047570642201834855,
+      "loss": 3.5496,
       "step": 19300
     },
     {
       "epoch": 2.0862533692722374,
-      "grad_norm": 0.6276587247848511,
-      "learning_rate": 0.00047537614678899075,
-      "loss": 3.5469,
+      "grad_norm": 0.6076366901397705,
+      "learning_rate": 0.00047538262277388015,
+      "loss": 3.5522,
       "step": 19350
     },
     {
       "epoch": 2.091644204851752,
-      "grad_norm": 0.5461747646331787,
-      "learning_rate": 0.00047505234754452235,
-      "loss": 3.5435,
+      "grad_norm": 0.5733623504638672,
+      "learning_rate": 0.0004750588235294117,
+      "loss": 3.5483,
       "step": 19400
     },
     {
       "epoch": 2.0970350404312668,
-      "grad_norm": 0.7402114868164062,
-      "learning_rate": 0.0004747285483000539,
-      "loss": 3.5519,
+      "grad_norm": 0.5831547379493713,
+      "learning_rate": 0.0004747350242849433,
+      "loss": 3.5584,
       "step": 19450
     },
     {
       "epoch": 2.1024258760107815,
-      "grad_norm": 0.6080474853515625,
-      "learning_rate": 0.0004744047490555855,
-      "loss": 3.5453,
+      "grad_norm": 0.5739231109619141,
+      "learning_rate": 0.00047441122504047486,
+      "loss": 3.5497,
       "step": 19500
     },
     {
       "epoch": 2.1078167115902966,
-      "grad_norm": 0.5156586170196533,
-      "learning_rate": 0.00047408094981111706,
-      "loss": 3.5511,
+      "grad_norm": 0.5826650261878967,
+      "learning_rate": 0.0004740874257960064,
+      "loss": 3.5573,
       "step": 19550
     },
     {
       "epoch": 2.1132075471698113,
-      "grad_norm": 0.6247926354408264,
-      "learning_rate": 0.00047375715056664866,
-      "loss": 3.548,
+      "grad_norm": 0.6077868342399597,
+      "learning_rate": 0.000473763626551538,
+      "loss": 3.5546,
       "step": 19600
     },
     {
       "epoch": 2.118598382749326,
-      "grad_norm": 0.5468766093254089,
-      "learning_rate": 0.0004734333513221802,
-      "loss": 3.5704,
+      "grad_norm": 0.5813013911247253,
+      "learning_rate": 0.00047343982730706956,
+      "loss": 3.5754,
       "step": 19650
     },
     {
       "epoch": 2.123989218328841,
-      "grad_norm": 0.5999529957771301,
-      "learning_rate": 0.00047310955207771177,
-      "loss": 3.5414,
+      "grad_norm": 0.5970227718353271,
+      "learning_rate": 0.00047311602806260117,
+      "loss": 3.5468,
       "step": 19700
     },
     {
       "epoch": 2.129380053908356,
-      "grad_norm": 0.5577241778373718,
-      "learning_rate": 0.00047278575283324337,
-      "loss": 3.561,
+      "grad_norm": 0.5493763089179993,
+      "learning_rate": 0.0004727922288181327,
+      "loss": 3.5675,
       "step": 19750
     },
     {
       "epoch": 2.1347708894878705,
-      "grad_norm": 0.5632361769676208,
-      "learning_rate": 0.0004724619535887749,
-      "loss": 3.5539,
+      "grad_norm": 0.5389633774757385,
+      "learning_rate": 0.0004724684295736643,
+      "loss": 3.5607,
       "step": 19800
     },
     {
       "epoch": 2.1401617250673857,
-      "grad_norm": 0.5894713401794434,
-      "learning_rate": 0.0004721381543443065,
-      "loss": 3.5463,
+      "grad_norm": 0.5832297801971436,
+      "learning_rate": 0.0004721446303291959,
+      "loss": 3.5535,
       "step": 19850
     },
     {
       "epoch": 2.1455525606469004,
-      "grad_norm": 0.5929577946662903,
-      "learning_rate": 0.000471814355099838,
-      "loss": 3.5588,
+      "grad_norm": 0.5914731025695801,
+      "learning_rate": 0.0004718208310847275,
+      "loss": 3.5636,
       "step": 19900
     },
     {
       "epoch": 2.150943396226415,
-      "grad_norm": 0.5735053420066833,
-      "learning_rate": 0.0004714905558553697,
-      "loss": 3.5507,
+      "grad_norm": 0.5819589495658875,
+      "learning_rate": 0.000471497031840259,
+      "loss": 3.556,
       "step": 19950
     },
     {
       "epoch": 2.1563342318059298,
-      "grad_norm": 0.572102427482605,
-      "learning_rate": 0.0004711667566109012,
-      "loss": 3.5744,
+      "grad_norm": 0.6233876347541809,
+      "learning_rate": 0.0004711732325957905,
+      "loss": 3.5788,
       "step": 20000
     },
     {
       "epoch": 2.1563342318059298,
-      "eval_accuracy": 0.3623309782095547,
-      "eval_loss": 3.5741710662841797,
-      "eval_runtime": 184.0834,
-      "eval_samples_per_second": 97.842,
-      "eval_steps_per_second": 6.117,
+      "eval_accuracy": 0.36208520537731836,
+      "eval_loss": 3.5757808685302734,
+      "eval_runtime": 184.0991,
+      "eval_samples_per_second": 97.833,
+      "eval_steps_per_second": 6.116,
       "step": 20000
     },
     {
       "epoch": 2.161725067385445,
-      "grad_norm": 0.5751939415931702,
-      "learning_rate": 0.00047084295736643273,
-      "loss": 3.5742,
+      "grad_norm": 0.5925044417381287,
+      "learning_rate": 0.00047084943335132213,
+      "loss": 3.5787,
       "step": 20050
     },
     {
       "epoch": 2.1671159029649596,
-      "grad_norm": 0.5660809874534607,
-      "learning_rate": 0.00047051915812196433,
-      "loss": 3.5769,
+      "grad_norm": 0.6043457984924316,
+      "learning_rate": 0.0004705256341068537,
+      "loss": 3.5818,
       "step": 20100
     },
     {
       "epoch": 2.1725067385444743,
-      "grad_norm": 0.5410208702087402,
-      "learning_rate": 0.0004701953588774959,
-      "loss": 3.57,
+      "grad_norm": 0.5603416562080383,
+      "learning_rate": 0.0004702018348623853,
+      "loss": 3.574,
       "step": 20150
     },
     {
       "epoch": 2.177897574123989,
-      "grad_norm": 0.5733333826065063,
-      "learning_rate": 0.0004698715596330275,
-      "loss": 3.5658,
+      "grad_norm": 0.5870110392570496,
+      "learning_rate": 0.00046987803561791684,
+      "loss": 3.5711,
       "step": 20200
     },
     {
       "epoch": 2.183288409703504,
-      "grad_norm": 0.6063035130500793,
-      "learning_rate": 0.00046954776038855904,
-      "loss": 3.5663,
+      "grad_norm": 0.5642842054367065,
+      "learning_rate": 0.00046955423637344844,
+      "loss": 3.5728,
       "step": 20250
     },
     {
       "epoch": 2.188679245283019,
-      "grad_norm": 0.5596222281455994,
-      "learning_rate": 0.00046922396114409064,
-      "loss": 3.5569,
+      "grad_norm": 0.5840718746185303,
+      "learning_rate": 0.00046923043712898,
+      "loss": 3.5639,
       "step": 20300
     },
     {
       "epoch": 2.1940700808625335,
-      "grad_norm": 0.5413583517074585,
-      "learning_rate": 0.0004689001618996222,
-      "loss": 3.5729,
+      "grad_norm": 0.5522575974464417,
+      "learning_rate": 0.0004689066378845116,
+      "loss": 3.5772,
       "step": 20350
     },
     {
       "epoch": 2.1994609164420487,
-      "grad_norm": 0.550186276435852,
-      "learning_rate": 0.0004685763626551538,
-      "loss": 3.5729,
+      "grad_norm": 0.5565154552459717,
+      "learning_rate": 0.00046858283864004315,
+      "loss": 3.58,
       "step": 20400
     },
     {
       "epoch": 2.2048517520215634,
-      "grad_norm": 0.6196731925010681,
-      "learning_rate": 0.00046825256341068535,
-      "loss": 3.5705,
+      "grad_norm": 0.5971863269805908,
+      "learning_rate": 0.0004682590393955747,
+      "loss": 3.5758,
       "step": 20450
     },
     {
       "epoch": 2.210242587601078,
-      "grad_norm": 0.5624878406524658,
-      "learning_rate": 0.0004679287641662169,
-      "loss": 3.5411,
+      "grad_norm": 0.5942404866218567,
+      "learning_rate": 0.0004679352401511063,
+      "loss": 3.548,
       "step": 20500
     },
     {
       "epoch": 2.215633423180593,
-      "grad_norm": 0.55267733335495,
-      "learning_rate": 0.0004676049649217485,
-      "loss": 3.5695,
+      "grad_norm": 0.6009751558303833,
+      "learning_rate": 0.0004676114409066378,
+      "loss": 3.5747,
       "step": 20550
     },
     {
       "epoch": 2.221024258760108,
-      "grad_norm": 0.6480850577354431,
-      "learning_rate": 0.00046728116567728,
-      "loss": 3.5455,
+      "grad_norm": 0.6824178695678711,
+      "learning_rate": 0.00046728764166216946,
+      "loss": 3.5531,
       "step": 20600
     },
     {
       "epoch": 2.2264150943396226,
-      "grad_norm": 0.561083197593689,
-      "learning_rate": 0.0004669573664328116,
-      "loss": 3.5533,
+      "grad_norm": 0.6136130094528198,
+      "learning_rate": 0.00046696384241770095,
+      "loss": 3.5571,
       "step": 20650
     },
     {
       "epoch": 2.2318059299191373,
-      "grad_norm": 0.5583511590957642,
-      "learning_rate": 0.00046663356718834316,
-      "loss": 3.5504,
+      "grad_norm": 0.6039118766784668,
+      "learning_rate": 0.00046664004317323256,
+      "loss": 3.5562,
       "step": 20700
     },
     {
       "epoch": 2.2371967654986524,
-      "grad_norm": 0.6693071126937866,
-      "learning_rate": 0.00046630976794387476,
-      "loss": 3.5581,
+      "grad_norm": 0.5857826471328735,
+      "learning_rate": 0.0004663162439287641,
+      "loss": 3.5619,
       "step": 20750
     },
     {
       "epoch": 2.242587601078167,
-      "grad_norm": 0.59278404712677,
-      "learning_rate": 0.0004659859686994063,
-      "loss": 3.5522,
+      "grad_norm": 0.5661748051643372,
+      "learning_rate": 0.00046599244468429566,
+      "loss": 3.5565,
       "step": 20800
     },
     {
       "epoch": 2.247978436657682,
-      "grad_norm": 0.5818943977355957,
-      "learning_rate": 0.0004656621694549379,
-      "loss": 3.5674,
+      "grad_norm": 0.5799033641815186,
+      "learning_rate": 0.00046566864543982726,
+      "loss": 3.5734,
       "step": 20850
     },
     {
       "epoch": 2.2533692722371965,
-      "grad_norm": 0.5376104116439819,
-      "learning_rate": 0.00046533837021046947,
-      "loss": 3.5601,
+      "grad_norm": 0.561801016330719,
+      "learning_rate": 0.0004653448461953588,
+      "loss": 3.5667,
       "step": 20900
     },
     {
       "epoch": 2.2587601078167117,
-      "grad_norm": 0.5341477990150452,
-      "learning_rate": 0.000465014570966001,
-      "loss": 3.5495,
+      "grad_norm": 0.5338796973228455,
+      "learning_rate": 0.0004650210469508904,
+      "loss": 3.5561,
       "step": 20950
     },
     {
       "epoch": 2.2641509433962264,
-      "grad_norm": 0.5756168961524963,
-      "learning_rate": 0.0004646907717215326,
-      "loss": 3.5621,
+      "grad_norm": 0.6155247092247009,
+      "learning_rate": 0.00046469724770642197,
+      "loss": 3.5674,
       "step": 21000
     },
     {
       "epoch": 2.2641509433962264,
-      "eval_accuracy": 0.3640750743647527,
-      "eval_loss": 3.559407949447632,
-      "eval_runtime": 183.7723,
-      "eval_samples_per_second": 98.007,
-      "eval_steps_per_second": 6.127,
+      "eval_accuracy": 0.36337882665963755,
+      "eval_loss": 3.5647754669189453,
+      "eval_runtime": 183.4598,
+      "eval_samples_per_second": 98.174,
+      "eval_steps_per_second": 6.138,
       "step": 21000
     },
     {
       "epoch": 2.269541778975741,
-      "grad_norm": 0.5583627223968506,
+      "grad_norm": 0.5377642512321472,
       "learning_rate": 0.0004643734484619536,
-      "loss": 3.5553,
+      "loss": 3.5615,
       "step": 21050
     },
     {
       "epoch": 2.274932614555256,
-      "grad_norm": 0.6221109628677368,
+      "grad_norm": 0.597271740436554,
       "learning_rate": 0.0004640496492174851,
-      "loss": 3.5521,
+      "loss": 3.5596,
       "step": 21100
     },
     {
       "epoch": 2.280323450134771,
-      "grad_norm": 0.5557717680931091,
+      "grad_norm": 0.5945741534233093,
       "learning_rate": 0.00046372584997301673,
-      "loss": 3.5767,
+      "loss": 3.5798,
       "step": 21150
     },
     {
       "epoch": 2.2857142857142856,
-      "grad_norm": 0.5979974269866943,
+      "grad_norm": 0.6722349524497986,
       "learning_rate": 0.0004634020507285483,
-      "loss": 3.5602,
+      "loss": 3.5668,
       "step": 21200
     },
     {
       "epoch": 2.2911051212938007,
-      "grad_norm": 0.6088611483573914,
-      "learning_rate": 0.0004630782514840798,
-      "loss": 3.5701,
+      "grad_norm": 0.5914362668991089,
+      "learning_rate": 0.00046308472746896923,
+      "loss": 3.5769,
       "step": 21250
     },
     {
       "epoch": 2.2964959568733154,
-      "grad_norm": 0.561540961265564,
-      "learning_rate": 0.0004627544522396114,
-      "loss": 3.5368,
+      "grad_norm": 0.6281972527503967,
+      "learning_rate": 0.00046276092822450073,
+      "loss": 3.5434,
       "step": 21300
     },
     {
       "epoch": 2.30188679245283,
-      "grad_norm": 0.6222971677780151,
-      "learning_rate": 0.00046243065299514293,
-      "loss": 3.559,
+      "grad_norm": 0.6473223567008972,
+      "learning_rate": 0.00046243712898003233,
+      "loss": 3.5623,
       "step": 21350
     },
     {
       "epoch": 2.3072776280323453,
-      "grad_norm": 0.5412083864212036,
-      "learning_rate": 0.00046210685375067454,
-      "loss": 3.5516,
+      "grad_norm": 0.6009921431541443,
+      "learning_rate": 0.0004621133297355639,
+      "loss": 3.5574,
       "step": 21400
     },
     {
       "epoch": 2.31266846361186,
-      "grad_norm": 0.5675464868545532,
-      "learning_rate": 0.0004617830545062061,
-      "loss": 3.5741,
+      "grad_norm": 0.5805341601371765,
+      "learning_rate": 0.0004617895304910955,
+      "loss": 3.5807,
       "step": 21450
     },
     {
       "epoch": 2.3180592991913747,
-      "grad_norm": 0.5713619589805603,
-      "learning_rate": 0.0004614592552617377,
-      "loss": 3.5468,
+      "grad_norm": 0.5670769810676575,
+      "learning_rate": 0.00046146573124662704,
+      "loss": 3.552,
       "step": 21500
     },
     {
       "epoch": 2.3234501347708894,
-      "grad_norm": 0.5480716824531555,
-      "learning_rate": 0.00046113545601726924,
-      "loss": 3.5626,
+      "grad_norm": 0.6296991109848022,
+      "learning_rate": 0.00046114193200215864,
+      "loss": 3.5674,
       "step": 21550
     },
     {
       "epoch": 2.3288409703504045,
-      "grad_norm": 0.6332092881202698,
-      "learning_rate": 0.00046081165677280085,
-      "loss": 3.5526,
+      "grad_norm": 0.5793233513832092,
+      "learning_rate": 0.0004608181327576902,
+      "loss": 3.5594,
       "step": 21600
     },
     {
       "epoch": 2.334231805929919,
-      "grad_norm": 0.6008698344230652,
-      "learning_rate": 0.0004604878575283324,
-      "loss": 3.5542,
+      "grad_norm": 0.6445789933204651,
+      "learning_rate": 0.00046049433351322175,
+      "loss": 3.5611,
       "step": 21650
     },
     {
       "epoch": 2.339622641509434,
-      "grad_norm": 0.6072392463684082,
-      "learning_rate": 0.00046016405828386395,
-      "loss": 3.5722,
+      "grad_norm": 0.6258645057678223,
+      "learning_rate": 0.00046017053426875335,
+      "loss": 3.5772,
       "step": 21700
     },
     {
       "epoch": 2.3450134770889486,
-      "grad_norm": 0.5895165801048279,
-      "learning_rate": 0.00045984025903939555,
-      "loss": 3.5609,
+      "grad_norm": 0.598357081413269,
+      "learning_rate": 0.0004598467350242849,
+      "loss": 3.5667,
       "step": 21750
     },
     {
       "epoch": 2.3504043126684637,
-      "grad_norm": 0.6073868274688721,
-      "learning_rate": 0.0004595164597949271,
-      "loss": 3.5427,
+      "grad_norm": 0.5611604452133179,
+      "learning_rate": 0.0004595229357798165,
+      "loss": 3.549,
       "step": 21800
     },
     {
       "epoch": 2.3557951482479784,
-      "grad_norm": 0.5244640111923218,
-      "learning_rate": 0.0004591926605504587,
-      "loss": 3.5541,
+      "grad_norm": 0.6011940836906433,
+      "learning_rate": 0.00045919913653534806,
+      "loss": 3.5591,
       "step": 21850
     },
     {
       "epoch": 2.361185983827493,
-      "grad_norm": 0.5883101224899292,
-      "learning_rate": 0.0004588688613059902,
-      "loss": 3.5479,
+      "grad_norm": 0.6146547794342041,
+      "learning_rate": 0.00045887533729087966,
+      "loss": 3.5532,
       "step": 21900
     },
     {
       "epoch": 2.3665768194070083,
-      "grad_norm": 0.5923961997032166,
-      "learning_rate": 0.00045854506206152186,
-      "loss": 3.5506,
+      "grad_norm": 0.5715474486351013,
+      "learning_rate": 0.00045855153804641116,
+      "loss": 3.5564,
       "step": 21950
     },
     {
       "epoch": 2.371967654986523,
-      "grad_norm": 0.5549956560134888,
-      "learning_rate": 0.00045822126281705336,
-      "loss": 3.5697,
+      "grad_norm": 0.5572586059570312,
+      "learning_rate": 0.0004582277388019427,
+      "loss": 3.5759,
       "step": 22000
     },
     {
       "epoch": 2.371967654986523,
-      "eval_accuracy": 0.3651554100282921,
-      "eval_loss": 3.5489094257354736,
-      "eval_runtime": 183.5303,
-      "eval_samples_per_second": 98.136,
-      "eval_steps_per_second": 6.135,
+      "eval_accuracy": 0.3641635178154337,
+      "eval_loss": 3.5540733337402344,
+      "eval_runtime": 183.5927,
+      "eval_samples_per_second": 98.103,
+      "eval_steps_per_second": 6.133,
       "step": 22000
     },
     {
       "epoch": 2.3773584905660377,
-      "grad_norm": 0.5528952479362488,
-      "learning_rate": 0.00045789746357258497,
-      "loss": 3.5474,
+      "grad_norm": 0.5874884128570557,
+      "learning_rate": 0.0004579039395574743,
+      "loss": 3.5523,
       "step": 22050
     },
     {
       "epoch": 2.382749326145553,
-      "grad_norm": 0.6946583390235901,
-      "learning_rate": 0.0004575736643281165,
-      "loss": 3.5487,
+      "grad_norm": 0.6247643828392029,
+      "learning_rate": 0.00045758014031300586,
+      "loss": 3.5557,
       "step": 22100
     },
     {
       "epoch": 2.3881401617250675,
-      "grad_norm": 0.5659127235412598,
-      "learning_rate": 0.00045724986508364807,
-      "loss": 3.5513,
+      "grad_norm": 0.5906260013580322,
+      "learning_rate": 0.00045725634106853747,
+      "loss": 3.5578,
       "step": 22150
     },
     {
       "epoch": 2.393530997304582,
-      "grad_norm": 0.5336350798606873,
-      "learning_rate": 0.00045692606583917967,
-      "loss": 3.5623,
+      "grad_norm": 0.567096471786499,
+      "learning_rate": 0.000456932541824069,
+      "loss": 3.5689,
       "step": 22200
     },
     {
       "epoch": 2.398921832884097,
-      "grad_norm": 0.5657646656036377,
-      "learning_rate": 0.0004566022665947112,
-      "loss": 3.5658,
+      "grad_norm": 0.5838461518287659,
+      "learning_rate": 0.0004566087425796006,
+      "loss": 3.5735,
       "step": 22250
     },
     {
       "epoch": 2.404312668463612,
-      "grad_norm": 0.5422862768173218,
+      "grad_norm": 0.603792667388916,
       "learning_rate": 0.0004562849433351322,
-      "loss": 3.5705,
+      "loss": 3.5755,
       "step": 22300
     },
     {
       "epoch": 2.4097035040431267,
-      "grad_norm": 0.5550121068954468,
+      "grad_norm": 0.6152181029319763,
       "learning_rate": 0.0004559611440906638,
-      "loss": 3.5507,
+      "loss": 3.5544,
       "step": 22350
     },
     {
       "epoch": 2.4150943396226414,
-      "grad_norm": 0.5688802003860474,
+      "grad_norm": 0.5923592448234558,
       "learning_rate": 0.00045563734484619533,
-      "loss": 3.5711,
+      "loss": 3.5777,
       "step": 22400
     },
     {
       "epoch": 2.420485175202156,
-      "grad_norm": 0.5675523281097412,
+      "grad_norm": 0.5686638951301575,
       "learning_rate": 0.0004553135456017269,
-      "loss": 3.5616,
+      "loss": 3.5681,
       "step": 22450
     },
     {
       "epoch": 2.4258760107816713,
-      "grad_norm": 0.5429262518882751,
+      "grad_norm": 0.5507339835166931,
       "learning_rate": 0.0004549897463572585,
-      "loss": 3.5557,
+      "loss": 3.5626,
       "step": 22500
     },
     {
       "epoch": 2.431266846361186,
-      "grad_norm": 0.6132704615592957,
+      "grad_norm": 0.5886709690093994,
       "learning_rate": 0.00045466594711279,
-      "loss": 3.5504,
+      "loss": 3.5558,
       "step": 22550
     },
     {
       "epoch": 2.4366576819407006,
-      "grad_norm": 0.5489976406097412,
+      "grad_norm": 0.561502993106842,
       "learning_rate": 0.00045434214786832164,
-      "loss": 3.5518,
+      "loss": 3.5584,
       "step": 22600
     },
     {
       "epoch": 2.442048517520216,
-      "grad_norm": 0.5603680610656738,
+      "grad_norm": 0.5907917618751526,
       "learning_rate": 0.00045401834862385314,
-      "loss": 3.5373,
+      "loss": 3.5444,
       "step": 22650
     },
     {
       "epoch": 2.4474393530997305,
-      "grad_norm": 0.5937761068344116,
+      "grad_norm": 0.5945461988449097,
       "learning_rate": 0.00045369454937938474,
-      "loss": 3.5433,
+      "loss": 3.5476,
       "step": 22700
     },
     {
       "epoch": 2.452830188679245,
-      "grad_norm": 0.5349526405334473,
+      "grad_norm": 0.5842607021331787,
       "learning_rate": 0.0004533707501349163,
-      "loss": 3.5468,
+      "loss": 3.5525,
       "step": 22750
     },
     {
       "epoch": 2.4582210242587603,
-      "grad_norm": 0.5487583875656128,
+      "grad_norm": 0.5414595603942871,
       "learning_rate": 0.0004530469508904479,
-      "loss": 3.5434,
+      "loss": 3.5478,
       "step": 22800
     },
     {
       "epoch": 2.463611859838275,
-      "grad_norm": 0.6188079714775085,
+      "grad_norm": 0.6275479197502136,
       "learning_rate": 0.00045272315164597945,
-      "loss": 3.5518,
+      "loss": 3.5578,
       "step": 22850
     },
     {
       "epoch": 2.4690026954177897,
-      "grad_norm": 0.6644785404205322,
+      "grad_norm": 0.6523239612579346,
       "learning_rate": 0.000452399352401511,
-      "loss": 3.5607,
+      "loss": 3.5642,
       "step": 22900
     },
     {
       "epoch": 2.4743935309973044,
-      "grad_norm": 0.579646646976471,
+      "grad_norm": 0.6264610290527344,
       "learning_rate": 0.0004520755531570426,
-      "loss": 3.5622,
+      "loss": 3.568,
       "step": 22950
     },
     {
       "epoch": 2.4797843665768196,
-      "grad_norm": 0.5342908501625061,
+      "grad_norm": 0.5830268859863281,
       "learning_rate": 0.00045175175391257415,
-      "loss": 3.537,
+      "loss": 3.5439,
       "step": 23000
     },
     {
       "epoch": 2.4797843665768196,
-      "eval_accuracy": 0.36602680618856426,
-      "eval_loss": 3.5380289554595947,
-      "eval_runtime": 183.7529,
-      "eval_samples_per_second": 98.017,
-      "eval_steps_per_second": 6.128,
+      "eval_accuracy": 0.3653592428458321,
+      "eval_loss": 3.544100046157837,
+      "eval_runtime": 183.4913,
+      "eval_samples_per_second": 98.157,
+      "eval_steps_per_second": 6.137,
       "step": 23000
     },
     {
       "epoch": 2.4851752021563343,
-      "grad_norm": 0.570767343044281,
+      "grad_norm": 0.5847001075744629,
       "learning_rate": 0.00045142795466810576,
-      "loss": 3.5417,
+      "loss": 3.5484,
       "step": 23050
     },
     {
       "epoch": 2.490566037735849,
-      "grad_norm": 0.5754830837249756,
+      "grad_norm": 0.5852646827697754,
       "learning_rate": 0.0004511041554236373,
-      "loss": 3.5566,
+      "loss": 3.5642,
       "step": 23100
     },
     {
       "epoch": 2.4959568733153636,
-      "grad_norm": 0.5624446868896484,
+      "grad_norm": 0.6043870449066162,
       "learning_rate": 0.0004507803561791689,
-      "loss": 3.5579,
+      "loss": 3.5632,
       "step": 23150
     },
     {
       "epoch": 2.501347708894879,
-      "grad_norm": 0.6118834614753723,
+      "grad_norm": 0.6264826059341431,
       "learning_rate": 0.00045045655693470046,
-      "loss": 3.5519,
+      "loss": 3.5549,
       "step": 23200
     },
     {
       "epoch": 2.5067385444743935,
-      "grad_norm": 0.5770419836044312,
-      "learning_rate": 0.00045013275769023207,
-      "loss": 3.5446,
+      "grad_norm": 0.5479467511177063,
+      "learning_rate": 0.0004501392336751214,
+      "loss": 3.5491,
       "step": 23250
     },
     {
       "epoch": 2.512129380053908,
-      "grad_norm": 0.5695980787277222,
-      "learning_rate": 0.00044980895844576356,
-      "loss": 3.544,
+      "grad_norm": 0.6958391070365906,
+      "learning_rate": 0.0004498154344306529,
+      "loss": 3.5503,
       "step": 23300
     },
     {
       "epoch": 2.5175202156334233,
-      "grad_norm": 0.6304803490638733,
-      "learning_rate": 0.0004494851592012951,
-      "loss": 3.5394,
+      "grad_norm": 0.65898197889328,
+      "learning_rate": 0.0004494916351861845,
+      "loss": 3.5468,
       "step": 23350
     },
     {
       "epoch": 2.522911051212938,
-      "grad_norm": 0.5704509019851685,
-      "learning_rate": 0.0004491613599568267,
-      "loss": 3.5319,
+      "grad_norm": 0.5873842835426331,
+      "learning_rate": 0.00044916783594171607,
+      "loss": 3.5384,
       "step": 23400
     },
     {
       "epoch": 2.5283018867924527,
-      "grad_norm": 0.5579874515533447,
-      "learning_rate": 0.00044883756071235827,
-      "loss": 3.5454,
+      "grad_norm": 0.5899979472160339,
+      "learning_rate": 0.00044884403669724767,
+      "loss": 3.5509,
       "step": 23450
     },
     {
       "epoch": 2.533692722371968,
-      "grad_norm": 0.5578489899635315,
-      "learning_rate": 0.0004485137614678899,
-      "loss": 3.5406,
+      "grad_norm": 0.5558457970619202,
+      "learning_rate": 0.0004485202374527792,
+      "loss": 3.5475,
       "step": 23500
     },
     {
       "epoch": 2.5390835579514826,
-      "grad_norm": 0.5366498231887817,
-      "learning_rate": 0.0004481899622234214,
-      "loss": 3.5523,
+      "grad_norm": 0.5665252804756165,
+      "learning_rate": 0.00044819643820831083,
+      "loss": 3.5582,
       "step": 23550
     },
     {
       "epoch": 2.5444743935309972,
-      "grad_norm": 0.5812051892280579,
-      "learning_rate": 0.00044786616297895303,
-      "loss": 3.5521,
+      "grad_norm": 0.6234211921691895,
+      "learning_rate": 0.0004478726389638424,
+      "loss": 3.5564,
       "step": 23600
     },
     {
       "epoch": 2.5498652291105124,
-      "grad_norm": 0.577279806137085,
-      "learning_rate": 0.0004475423637344846,
-      "loss": 3.5546,
+      "grad_norm": 0.60724276304245,
+      "learning_rate": 0.00044754883971937393,
+      "loss": 3.5612,
       "step": 23650
     },
     {
       "epoch": 2.555256064690027,
-      "grad_norm": 0.5395628213882446,
-      "learning_rate": 0.00044721856449001613,
-      "loss": 3.5627,
+      "grad_norm": 0.587846577167511,
+      "learning_rate": 0.00044722504047490553,
+      "loss": 3.569,
       "step": 23700
     },
     {
       "epoch": 2.560646900269542,
-      "grad_norm": 0.5482959151268005,
-      "learning_rate": 0.00044689476524554774,
-      "loss": 3.5664,
+      "grad_norm": 0.5777546763420105,
+      "learning_rate": 0.0004469012412304371,
+      "loss": 3.5738,
       "step": 23750
     },
     {
       "epoch": 2.5660377358490565,
-      "grad_norm": 0.6769027709960938,
-      "learning_rate": 0.0004465709660010793,
-      "loss": 3.5453,
+      "grad_norm": 0.7845169305801392,
+      "learning_rate": 0.0004465774419859687,
+      "loss": 3.5513,
       "step": 23800
     },
     {
       "epoch": 2.571428571428571,
-      "grad_norm": 0.5909700989723206,
-      "learning_rate": 0.0004462471667566109,
-      "loss": 3.5611,
+      "grad_norm": 0.5615964531898499,
+      "learning_rate": 0.00044625364274150024,
+      "loss": 3.5635,
       "step": 23850
     },
     {
       "epoch": 2.5768194070080863,
-      "grad_norm": 0.6435213088989258,
-      "learning_rate": 0.00044592336751214244,
-      "loss": 3.5433,
+      "grad_norm": 0.576488733291626,
+      "learning_rate": 0.00044592984349703184,
+      "loss": 3.5471,
       "step": 23900
     },
     {
       "epoch": 2.582210242587601,
-      "grad_norm": 0.5850532650947571,
-      "learning_rate": 0.00044559956826767405,
-      "loss": 3.5489,
+      "grad_norm": 0.5760179758071899,
+      "learning_rate": 0.00044560604425256334,
+      "loss": 3.5553,
       "step": 23950
     },
     {
       "epoch": 2.5876010781671157,
-      "grad_norm": 0.663982093334198,
-      "learning_rate": 0.00044527576902320554,
-      "loss": 3.5319,
+      "grad_norm": 0.6169016361236572,
+      "learning_rate": 0.000445282245008095,
+      "loss": 3.5368,
       "step": 24000
     },
     {
       "epoch": 2.5876010781671157,
-      "eval_accuracy": 0.36721959759099165,
-      "eval_loss": 3.5314929485321045,
-      "eval_runtime": 183.664,
-      "eval_samples_per_second": 98.065,
-      "eval_steps_per_second": 6.131,
+      "eval_accuracy": 0.36643175550144885,
+      "eval_loss": 3.535325288772583,
+      "eval_runtime": 183.5952,
+      "eval_samples_per_second": 98.102,
+      "eval_steps_per_second": 6.133,
       "step": 24000
     },
     {
       "epoch": 2.592991913746631,
-      "grad_norm": 0.5680175423622131,
-      "learning_rate": 0.00044495196977873715,
-      "loss": 3.5434,
+      "grad_norm": 0.5668704509735107,
+      "learning_rate": 0.0004449584457636265,
+      "loss": 3.5499,
       "step": 24050
     },
     {
       "epoch": 2.5983827493261455,
-      "grad_norm": 0.6317371129989624,
-      "learning_rate": 0.0004446281705342687,
-      "loss": 3.538,
+      "grad_norm": 0.7200941443443298,
+      "learning_rate": 0.00044463464651915805,
+      "loss": 3.5444,
       "step": 24100
     },
     {
       "epoch": 2.6037735849056602,
-      "grad_norm": 0.5761317610740662,
-      "learning_rate": 0.00044430437128980025,
-      "loss": 3.5633,
+      "grad_norm": 0.6006847023963928,
+      "learning_rate": 0.00044431084727468965,
+      "loss": 3.5704,
       "step": 24150
     },
     {
       "epoch": 2.6091644204851754,
-      "grad_norm": 0.5837670564651489,
-      "learning_rate": 0.00044398057204533185,
-      "loss": 3.5289,
+      "grad_norm": 0.6405774354934692,
+      "learning_rate": 0.0004439870480302212,
+      "loss": 3.536,
       "step": 24200
     },
     {
       "epoch": 2.61455525606469,
-      "grad_norm": 0.5544631481170654,
-      "learning_rate": 0.0004436567728008634,
-      "loss": 3.54,
+      "grad_norm": 0.5956117510795593,
+      "learning_rate": 0.0004436632487857528,
+      "loss": 3.546,
       "step": 24250
     },
     {
       "epoch": 2.6199460916442048,
-      "grad_norm": 0.5587742924690247,
-      "learning_rate": 0.000443332973556395,
-      "loss": 3.5479,
+      "grad_norm": 0.6144042015075684,
+      "learning_rate": 0.00044333944954128436,
+      "loss": 3.554,
       "step": 24300
     },
     {
       "epoch": 2.62533692722372,
-      "grad_norm": 0.5495783090591431,
-      "learning_rate": 0.00044300917431192656,
-      "loss": 3.5642,
+      "grad_norm": 0.5479573607444763,
+      "learning_rate": 0.00044301565029681596,
+      "loss": 3.5706,
       "step": 24350
     },
     {
       "epoch": 2.6307277628032346,
-      "grad_norm": 0.5933278203010559,
-      "learning_rate": 0.00044268537506745816,
-      "loss": 3.526,
+      "grad_norm": 0.566356360912323,
+      "learning_rate": 0.0004426918510523475,
+      "loss": 3.5317,
       "step": 24400
     },
     {
       "epoch": 2.6361185983827493,
-      "grad_norm": 0.5730457305908203,
-      "learning_rate": 0.0004423615758229897,
-      "loss": 3.546,
+      "grad_norm": 0.5739872455596924,
+      "learning_rate": 0.0004423680518078791,
+      "loss": 3.55,
       "step": 24450
     },
     {
       "epoch": 2.641509433962264,
-      "grad_norm": 0.5949668288230896,
-      "learning_rate": 0.0004420377765785213,
-      "loss": 3.5588,
+      "grad_norm": 0.5683674812316895,
+      "learning_rate": 0.00044204425256341067,
+      "loss": 3.5647,
       "step": 24500
     },
     {
       "epoch": 2.6469002695417787,
-      "grad_norm": 0.5814875364303589,
-      "learning_rate": 0.00044171397733405287,
-      "loss": 3.5262,
+      "grad_norm": 0.5911427736282349,
+      "learning_rate": 0.0004417204533189422,
+      "loss": 3.5319,
       "step": 24550
     },
     {
       "epoch": 2.652291105121294,
-      "grad_norm": 0.5717437267303467,
-      "learning_rate": 0.00044139017808958437,
-      "loss": 3.5318,
+      "grad_norm": 0.6214153170585632,
+      "learning_rate": 0.0004413966540744738,
+      "loss": 3.5386,
       "step": 24600
     },
     {
       "epoch": 2.6576819407008085,
-      "grad_norm": 0.5098128914833069,
-      "learning_rate": 0.00044106637884511597,
-      "loss": 3.5355,
+      "grad_norm": 0.5593351125717163,
+      "learning_rate": 0.0004410728548300053,
+      "loss": 3.5407,
       "step": 24650
     },
     {
       "epoch": 2.6630727762803232,
-      "grad_norm": 0.5785273909568787,
-      "learning_rate": 0.0004407425796006475,
-      "loss": 3.5384,
+      "grad_norm": 0.614547073841095,
+      "learning_rate": 0.0004407490555855369,
+      "loss": 3.5452,
       "step": 24700
     },
     {
       "epoch": 2.6684636118598384,
-      "grad_norm": 0.5735189318656921,
+      "grad_norm": 0.5920159220695496,
       "learning_rate": 0.0004404252563410685,
-      "loss": 3.5379,
+      "loss": 3.5438,
       "step": 24750
     },
     {
       "epoch": 2.673854447439353,
-      "grad_norm": 0.5718660354614258,
+      "grad_norm": 0.595122218132019,
       "learning_rate": 0.0004401014570966001,
-      "loss": 3.5167,
+      "loss": 3.5215,
       "step": 24800
     },
     {
       "epoch": 2.6792452830188678,
-      "grad_norm": 0.5400899052619934,
+      "grad_norm": 0.5354313850402832,
       "learning_rate": 0.00043977765785213163,
-      "loss": 3.5144,
+      "loss": 3.5196,
       "step": 24850
     },
     {
       "epoch": 2.684636118598383,
-      "grad_norm": 0.5528421998023987,
+      "grad_norm": 0.5641005635261536,
       "learning_rate": 0.0004394538586076632,
-      "loss": 3.5536,
+      "loss": 3.5589,
       "step": 24900
     },
     {
       "epoch": 2.6900269541778976,
-      "grad_norm": 0.5778202414512634,
+      "grad_norm": 0.5943961143493652,
       "learning_rate": 0.0004391300593631948,
-      "loss": 3.536,
+      "loss": 3.5434,
       "step": 24950
     },
     {
       "epoch": 2.6954177897574123,
-      "grad_norm": 0.5615628957748413,
+      "grad_norm": 0.5887384414672852,
       "learning_rate": 0.00043880626011872634,
-      "loss": 3.5265,
+      "loss": 3.5327,
       "step": 25000
     },
     {
       "epoch": 2.6954177897574123,
-      "eval_accuracy": 0.36784598148924214,
-      "eval_loss": 3.5187320709228516,
-      "eval_runtime": 183.9267,
-      "eval_samples_per_second": 97.925,
-      "eval_steps_per_second": 6.122,
+      "eval_accuracy": 0.36708769298518484,
+      "eval_loss": 3.524667263031006,
+      "eval_runtime": 183.5052,
+      "eval_samples_per_second": 98.15,
+      "eval_steps_per_second": 6.136,
       "step": 25000
     },
     {
       "epoch": 2.7008086253369274,
-      "grad_norm": 0.5722801685333252,
+      "grad_norm": 0.5729919672012329,
       "learning_rate": 0.00043848246087425794,
-      "loss": 3.5369,
+      "loss": 3.5417,
       "step": 25050
     },
     {
       "epoch": 2.706199460916442,
-      "grad_norm": 0.5277543067932129,
+      "grad_norm": 0.5481346249580383,
       "learning_rate": 0.0004381586616297895,
-      "loss": 3.5222,
+      "loss": 3.5267,
       "step": 25100
     },
     {
       "epoch": 2.711590296495957,
-      "grad_norm": 0.5880123972892761,
+      "grad_norm": 0.605682909488678,
       "learning_rate": 0.0004378348623853211,
-      "loss": 3.5126,
+      "loss": 3.5183,
       "step": 25150
     },
     {
       "epoch": 2.7169811320754715,
-      "grad_norm": 0.5826494097709656,
+      "grad_norm": 0.5646705031394958,
       "learning_rate": 0.00043751106314085265,
-      "loss": 3.5171,
+      "loss": 3.5248,
       "step": 25200
     },
     {
       "epoch": 2.7223719676549867,
-      "grad_norm": 0.5825526118278503,
+      "grad_norm": 0.6011978387832642,
       "learning_rate": 0.00043718726389638425,
-      "loss": 3.5533,
+      "loss": 3.5592,
       "step": 25250
     },
     {
       "epoch": 2.7277628032345014,
-      "grad_norm": 0.5669980049133301,
+      "grad_norm": 0.6028566360473633,
       "learning_rate": 0.00043686346465191575,
-      "loss": 3.5104,
+      "loss": 3.5153,
       "step": 25300
     },
     {
       "epoch": 2.733153638814016,
-      "grad_norm": 0.6107063889503479,
+      "grad_norm": 0.6496447324752808,
       "learning_rate": 0.0004365396654074473,
-      "loss": 3.5351,
+      "loss": 3.5403,
       "step": 25350
     },
     {
       "epoch": 2.7385444743935308,
-      "grad_norm": 0.6292389035224915,
+      "grad_norm": 0.6164609789848328,
       "learning_rate": 0.0004362158661629789,
-      "loss": 3.5373,
+      "loss": 3.5431,
       "step": 25400
     },
     {
       "epoch": 2.743935309973046,
-      "grad_norm": 0.5547637343406677,
+      "grad_norm": 0.5738456845283508,
       "learning_rate": 0.00043589206691851045,
-      "loss": 3.5098,
+      "loss": 3.5146,
       "step": 25450
     },
     {
       "epoch": 2.7493261455525606,
-      "grad_norm": 0.5594532489776611,
+      "grad_norm": 0.5638880729675293,
       "learning_rate": 0.00043556826767404206,
-      "loss": 3.5693,
+      "loss": 3.5758,
       "step": 25500
     },
     {
       "epoch": 2.7547169811320753,
-      "grad_norm": 0.5568592548370361,
+      "grad_norm": 0.564645528793335,
       "learning_rate": 0.0004352444684295736,
-      "loss": 3.5577,
+      "loss": 3.5627,
       "step": 25550
     },
     {
       "epoch": 2.7601078167115904,
-      "grad_norm": 0.5955683588981628,
+      "grad_norm": 0.614765465259552,
       "learning_rate": 0.0004349206691851052,
-      "loss": 3.5345,
+      "loss": 3.5395,
       "step": 25600
     },
     {
       "epoch": 2.765498652291105,
-      "grad_norm": 0.6182993054389954,
-      "learning_rate": 0.00043459686994063676,
-      "loss": 3.5233,
+      "grad_norm": 0.6745593547821045,
+      "learning_rate": 0.0004346033459255261,
+      "loss": 3.5291,
       "step": 25650
     },
     {
       "epoch": 2.77088948787062,
-      "grad_norm": 0.5633643865585327,
-      "learning_rate": 0.00043427307069616837,
-      "loss": 3.5408,
+      "grad_norm": 0.5785972476005554,
+      "learning_rate": 0.0004342795466810577,
+      "loss": 3.5474,
       "step": 25700
     },
     {
       "epoch": 2.776280323450135,
-      "grad_norm": 0.5655704140663147,
-      "learning_rate": 0.0004339492714516999,
-      "loss": 3.5331,
+      "grad_norm": 0.625646710395813,
+      "learning_rate": 0.00043395574743658927,
+      "loss": 3.5367,
       "step": 25750
     },
     {
       "epoch": 2.7816711590296497,
-      "grad_norm": 0.560209333896637,
-      "learning_rate": 0.00043362547220723147,
-      "loss": 3.5319,
+      "grad_norm": 0.5450944900512695,
+      "learning_rate": 0.00043363194819212087,
+      "loss": 3.5392,
       "step": 25800
     },
     {
       "epoch": 2.7870619946091644,
-      "grad_norm": 0.5956665873527527,
-      "learning_rate": 0.0004333016729627631,
-      "loss": 3.5201,
+      "grad_norm": 0.6140152812004089,
+      "learning_rate": 0.0004333081489476524,
+      "loss": 3.5272,
       "step": 25850
     },
     {
       "epoch": 2.7924528301886795,
-      "grad_norm": 0.586514949798584,
-      "learning_rate": 0.0004329778737182946,
-      "loss": 3.5297,
+      "grad_norm": 0.6200935244560242,
+      "learning_rate": 0.000432984349703184,
+      "loss": 3.5358,
       "step": 25900
     },
     {
       "epoch": 2.797843665768194,
-      "grad_norm": 0.5690777897834778,
-      "learning_rate": 0.00043265407447382623,
-      "loss": 3.5196,
+      "grad_norm": 0.5898178815841675,
+      "learning_rate": 0.0004326605504587155,
+      "loss": 3.5242,
       "step": 25950
     },
     {
       "epoch": 2.803234501347709,
-      "grad_norm": 0.5629533529281616,
-      "learning_rate": 0.0004323302752293577,
-      "loss": 3.5313,
+      "grad_norm": 0.5562497973442078,
+      "learning_rate": 0.0004323367512142472,
+      "loss": 3.5385,
       "step": 26000
     },
     {
       "epoch": 2.803234501347709,
-      "eval_accuracy": 0.36869542976617575,
-      "eval_loss": 3.510869026184082,
-      "eval_runtime": 183.6283,
-      "eval_samples_per_second": 98.084,
-      "eval_steps_per_second": 6.132,
+      "eval_accuracy": 0.36819138901960435,
+      "eval_loss": 3.515011787414551,
+      "eval_runtime": 183.3626,
+      "eval_samples_per_second": 98.226,
+      "eval_steps_per_second": 6.141,
       "step": 26000
     },
     {
       "epoch": 2.8086253369272236,
-      "grad_norm": 0.5801641345024109,
-      "learning_rate": 0.00043200647598488933,
-      "loss": 3.5313,
+      "grad_norm": 0.5934141278266907,
+      "learning_rate": 0.0004320129519697787,
+      "loss": 3.5357,
       "step": 26050
     },
     {
       "epoch": 2.8140161725067383,
-      "grad_norm": 0.5631428956985474,
-      "learning_rate": 0.0004316826767404209,
-      "loss": 3.503,
+      "grad_norm": 0.5788941383361816,
+      "learning_rate": 0.00043168915272531023,
+      "loss": 3.5101,
       "step": 26100
     },
     {
       "epoch": 2.8194070080862534,
-      "grad_norm": 0.5923662185668945,
-      "learning_rate": 0.0004313588774959525,
-      "loss": 3.5071,
+      "grad_norm": 0.5862647891044617,
+      "learning_rate": 0.00043136535348084183,
+      "loss": 3.5132,
       "step": 26150
     },
     {
       "epoch": 2.824797843665768,
-      "grad_norm": 0.5873034000396729,
-      "learning_rate": 0.00043103507825148404,
-      "loss": 3.5307,
+      "grad_norm": 0.6070460081100464,
+      "learning_rate": 0.0004310415542363734,
+      "loss": 3.5375,
       "step": 26200
     },
     {
       "epoch": 2.830188679245283,
-      "grad_norm": 0.5601552724838257,
-      "learning_rate": 0.0004307112790070156,
-      "loss": 3.5429,
+      "grad_norm": 0.6160635352134705,
+      "learning_rate": 0.000430717754991905,
+      "loss": 3.5495,
       "step": 26250
     },
     {
       "epoch": 2.835579514824798,
-      "grad_norm": 0.5437904596328735,
-      "learning_rate": 0.0004303874797625472,
-      "loss": 3.5588,
+      "grad_norm": 0.5690721869468689,
+      "learning_rate": 0.00043039395574743654,
+      "loss": 3.5639,
       "step": 26300
     },
     {
       "epoch": 2.8409703504043127,
-      "grad_norm": 0.7464920878410339,
-      "learning_rate": 0.00043006368051807874,
-      "loss": 3.529,
+      "grad_norm": 0.6696720719337463,
+      "learning_rate": 0.00043007015650296814,
+      "loss": 3.5328,
       "step": 26350
     },
     {
       "epoch": 2.8463611859838274,
-      "grad_norm": 0.5931203365325928,
-      "learning_rate": 0.00042973988127361035,
-      "loss": 3.5378,
+      "grad_norm": 0.6309919953346252,
+      "learning_rate": 0.0004297463572584997,
+      "loss": 3.5423,
       "step": 26400
     },
     {
       "epoch": 2.8517520215633425,
-      "grad_norm": 0.6171684265136719,
-      "learning_rate": 0.0004294160820291419,
-      "loss": 3.5438,
+      "grad_norm": 0.6136814951896667,
+      "learning_rate": 0.0004294225580140313,
+      "loss": 3.5511,
       "step": 26450
     },
     {
       "epoch": 2.857142857142857,
-      "grad_norm": 0.5586252808570862,
-      "learning_rate": 0.0004290922827846735,
-      "loss": 3.5151,
+      "grad_norm": 0.5879332423210144,
+      "learning_rate": 0.00042909875876956285,
+      "loss": 3.5206,
       "step": 26500
     },
     {
       "epoch": 2.862533692722372,
-      "grad_norm": 0.5578598976135254,
-      "learning_rate": 0.00042876848354020505,
-      "loss": 3.5306,
+      "grad_norm": 0.5740972757339478,
+      "learning_rate": 0.0004287749595250944,
+      "loss": 3.5365,
       "step": 26550
     },
     {
       "epoch": 2.867924528301887,
-      "grad_norm": 0.5919521450996399,
-      "learning_rate": 0.00042844468429573655,
-      "loss": 3.5387,
+      "grad_norm": 0.6117143034934998,
+      "learning_rate": 0.000428451160280626,
+      "loss": 3.5455,
       "step": 26600
     },
     {
       "epoch": 2.8733153638814017,
-      "grad_norm": 0.5993890762329102,
-      "learning_rate": 0.00042812088505126815,
-      "loss": 3.5445,
+      "grad_norm": 0.6194103956222534,
+      "learning_rate": 0.0004281273610361575,
+      "loss": 3.5499,
       "step": 26650
     },
     {
       "epoch": 2.8787061994609164,
-      "grad_norm": 0.5829864144325256,
-      "learning_rate": 0.0004277970858067997,
-      "loss": 3.5301,
+      "grad_norm": 0.5775049328804016,
+      "learning_rate": 0.0004278035617916891,
+      "loss": 3.5372,
       "step": 26700
     },
     {
       "epoch": 2.884097035040431,
-      "grad_norm": 0.6206557154655457,
-      "learning_rate": 0.0004274732865623313,
-      "loss": 3.5432,
+      "grad_norm": 0.6053679585456848,
+      "learning_rate": 0.00042747976254722066,
+      "loss": 3.548,
       "step": 26750
     },
     {
       "epoch": 2.889487870619946,
-      "grad_norm": 0.5877953171730042,
-      "learning_rate": 0.00042714948731786286,
-      "loss": 3.5212,
+      "grad_norm": 0.6154065728187561,
+      "learning_rate": 0.00042715596330275226,
+      "loss": 3.5265,
       "step": 26800
     },
     {
       "epoch": 2.894878706199461,
-      "grad_norm": 0.6274372935295105,
-      "learning_rate": 0.00042682568807339447,
-      "loss": 3.5414,
+      "grad_norm": 0.649990975856781,
+      "learning_rate": 0.0004268321640582838,
+      "loss": 3.5461,
       "step": 26850
     },
     {
       "epoch": 2.9002695417789757,
-      "grad_norm": 0.5919215679168701,
-      "learning_rate": 0.000426501888828926,
-      "loss": 3.5075,
+      "grad_norm": 0.6150764226913452,
+      "learning_rate": 0.0004265083648138154,
+      "loss": 3.5139,
       "step": 26900
     },
     {
       "epoch": 2.9056603773584904,
-      "grad_norm": 0.5824446678161621,
-      "learning_rate": 0.0004261780895844576,
-      "loss": 3.5275,
+      "grad_norm": 0.6371074914932251,
+      "learning_rate": 0.00042618456556934697,
+      "loss": 3.5355,
       "step": 26950
     },
     {
       "epoch": 2.9110512129380055,
-      "grad_norm": 0.5630289316177368,
-      "learning_rate": 0.00042585429033998917,
-      "loss": 3.5433,
+      "grad_norm": 0.5840454697608948,
+      "learning_rate": 0.0004258607663248785,
+      "loss": 3.5497,
       "step": 27000
     },
     {
       "epoch": 2.9110512129380055,
-      "eval_accuracy": 0.3697433868691464,
-      "eval_loss": 3.5023281574249268,
-      "eval_runtime": 183.854,
-      "eval_samples_per_second": 97.964,
-      "eval_steps_per_second": 6.124,
+      "eval_accuracy": 0.3689772753571665,
+      "eval_loss": 3.5088579654693604,
+      "eval_runtime": 181.3836,
+      "eval_samples_per_second": 99.298,
+      "eval_steps_per_second": 6.208,
       "step": 27000
     },
     {
       "epoch": 2.91644204851752,
-      "grad_norm": 0.5719215273857117,
-      "learning_rate": 0.0004255304910955207,
-      "loss": 3.5288,
+      "grad_norm": 0.6313679218292236,
+      "learning_rate": 0.0004255369670804101,
+      "loss": 3.5344,
       "step": 27050
     },
     {
       "epoch": 2.921832884097035,
-      "grad_norm": 0.5788906216621399,
-      "learning_rate": 0.0004252066918510523,
-      "loss": 3.5314,
+      "grad_norm": 0.6060076355934143,
+      "learning_rate": 0.0004252131678359417,
+      "loss": 3.5373,
       "step": 27100
     },
     {
       "epoch": 2.92722371967655,
-      "grad_norm": 0.6185005903244019,
-      "learning_rate": 0.0004248828926065839,
-      "loss": 3.5273,
+      "grad_norm": 0.6399513483047485,
+      "learning_rate": 0.0004248893685914733,
+      "loss": 3.5333,
       "step": 27150
     },
     {
       "epoch": 2.9326145552560647,
-      "grad_norm": 0.6660767793655396,
-      "learning_rate": 0.0004245590933621155,
-      "loss": 3.516,
+      "grad_norm": 0.6531930565834045,
+      "learning_rate": 0.00042456556934700483,
+      "loss": 3.5208,
       "step": 27200
     },
     {
       "epoch": 2.9380053908355794,
-      "grad_norm": 0.5727997422218323,
-      "learning_rate": 0.00042423529411764703,
-      "loss": 3.5295,
+      "grad_norm": 0.5938127636909485,
+      "learning_rate": 0.00042424177010253643,
+      "loss": 3.5356,
       "step": 27250
     },
     {
       "epoch": 2.9433962264150946,
-      "grad_norm": 0.5961682200431824,
-      "learning_rate": 0.00042391149487317864,
-      "loss": 3.5323,
+      "grad_norm": 0.585008442401886,
+      "learning_rate": 0.00042391797085806793,
+      "loss": 3.5371,
       "step": 27300
     },
     {
       "epoch": 2.9487870619946093,
-      "grad_norm": 0.6300620436668396,
-      "learning_rate": 0.00042358769562871013,
-      "loss": 3.5177,
+      "grad_norm": 0.6141608953475952,
+      "learning_rate": 0.0004235941716135995,
+      "loss": 3.5229,
       "step": 27350
     },
     {
       "epoch": 2.954177897574124,
-      "grad_norm": 0.6043277978897095,
-      "learning_rate": 0.00042326389638424174,
-      "loss": 3.5079,
+      "grad_norm": 0.6825471520423889,
+      "learning_rate": 0.0004232703723691311,
+      "loss": 3.5148,
       "step": 27400
     },
     {
       "epoch": 2.9595687331536387,
-      "grad_norm": 0.5596343874931335,
-      "learning_rate": 0.0004229400971397733,
-      "loss": 3.5159,
+      "grad_norm": 0.6025322675704956,
+      "learning_rate": 0.00042294657312466264,
+      "loss": 3.5217,
       "step": 27450
     },
     {
       "epoch": 2.964959568733154,
-      "grad_norm": 0.5775852203369141,
-      "learning_rate": 0.00042261629789530484,
-      "loss": 3.5263,
+      "grad_norm": 0.6103867888450623,
+      "learning_rate": 0.00042262277388019424,
+      "loss": 3.532,
       "step": 27500
     },
     {
       "epoch": 2.9703504043126685,
-      "grad_norm": 0.5495153069496155,
-      "learning_rate": 0.00042229249865083644,
-      "loss": 3.5161,
+      "grad_norm": 0.5813896656036377,
+      "learning_rate": 0.0004222989746357258,
+      "loss": 3.5235,
       "step": 27550
     },
     {
       "epoch": 2.975741239892183,
-      "grad_norm": 0.5802893042564392,
-      "learning_rate": 0.000421968699406368,
-      "loss": 3.5265,
+      "grad_norm": 0.5417213439941406,
+      "learning_rate": 0.0004219751753912574,
+      "loss": 3.5311,
       "step": 27600
     },
     {
       "epoch": 2.981132075471698,
-      "grad_norm": 0.6191282272338867,
-      "learning_rate": 0.0004216449001618996,
-      "loss": 3.5039,
+      "grad_norm": 0.6048957109451294,
+      "learning_rate": 0.00042165785213167835,
+      "loss": 3.5102,
       "step": 27650
     },
     {
       "epoch": 2.986522911051213,
-      "grad_norm": 0.5658198595046997,
-      "learning_rate": 0.00042132110091743115,
-      "loss": 3.5147,
+      "grad_norm": 0.575705885887146,
+      "learning_rate": 0.0004213340528872099,
+      "loss": 3.5226,
       "step": 27700
     },
     {
       "epoch": 2.9919137466307277,
-      "grad_norm": 0.6845032572746277,
-      "learning_rate": 0.00042099730167296275,
-      "loss": 3.521,
+      "grad_norm": 0.6198124289512634,
+      "learning_rate": 0.00042101025364274145,
+      "loss": 3.5271,
       "step": 27750
     },
     {
       "epoch": 2.9973045822102424,
-      "grad_norm": 0.5844449400901794,
-      "learning_rate": 0.0004206735024284943,
-      "loss": 3.5218,
+      "grad_norm": 0.5915241241455078,
+      "learning_rate": 0.00042068645439827305,
+      "loss": 3.5283,
       "step": 27800
     },
     {
       "epoch": 3.0026954177897576,
-      "grad_norm": 0.5795214772224426,
-      "learning_rate": 0.0004203497031840259,
-      "loss": 3.4854,
+      "grad_norm": 0.6294973492622375,
+      "learning_rate": 0.0004203626551538046,
+      "loss": 3.4931,
       "step": 27850
     },
     {
       "epoch": 3.0080862533692723,
-      "grad_norm": 0.7003495693206787,
-      "learning_rate": 0.00042002590393955746,
-      "loss": 3.4273,
+      "grad_norm": 0.7055127620697021,
+      "learning_rate": 0.0004200388559093362,
+      "loss": 3.4341,
       "step": 27900
     },
     {
       "epoch": 3.013477088948787,
-      "grad_norm": 0.549725353717804,
-      "learning_rate": 0.00041970210469508896,
-      "loss": 3.431,
+      "grad_norm": 0.5836459398269653,
+      "learning_rate": 0.0004197150566648677,
+      "loss": 3.4357,
       "step": 27950
     },
     {
       "epoch": 3.018867924528302,
-      "grad_norm": 0.5706000328063965,
-      "learning_rate": 0.00041937830545062056,
-      "loss": 3.4434,
+      "grad_norm": 0.6066904067993164,
+      "learning_rate": 0.00041939125742039936,
+      "loss": 3.4502,
       "step": 28000
     },
     {
       "epoch": 3.018867924528302,
-      "eval_accuracy": 0.37032381059585134,
-      "eval_loss": 3.499523639678955,
-      "eval_runtime": 183.7469,
-      "eval_samples_per_second": 98.021,
-      "eval_steps_per_second": 6.128,
+      "eval_accuracy": 0.3698292226505198,
+      "eval_loss": 3.504427671432495,
+      "eval_runtime": 180.9826,
+      "eval_samples_per_second": 99.518,
+      "eval_steps_per_second": 6.222,
       "step": 28000
     },
     {
       "epoch": 3.024258760107817,
-      "grad_norm": 0.6284692883491516,
-      "learning_rate": 0.0004190545062061521,
-      "loss": 3.4299,
+      "grad_norm": 0.6571897864341736,
+      "learning_rate": 0.00041906745817593086,
+      "loss": 3.4356,
       "step": 28050
     },
     {
       "epoch": 3.0296495956873315,
-      "grad_norm": 0.6581982374191284,
-      "learning_rate": 0.0004187307069616837,
-      "loss": 3.4238,
+      "grad_norm": 0.6707847118377686,
+      "learning_rate": 0.00041874365893146247,
+      "loss": 3.4299,
       "step": 28100
     },
     {
       "epoch": 3.035040431266846,
-      "grad_norm": 0.6106514930725098,
-      "learning_rate": 0.00041840690771721527,
-      "loss": 3.4642,
+      "grad_norm": 0.6154466867446899,
+      "learning_rate": 0.000418419859686994,
+      "loss": 3.4711,
       "step": 28150
     },
     {
       "epoch": 3.0404312668463613,
-      "grad_norm": 0.5879032611846924,
-      "learning_rate": 0.00041808310847274687,
-      "loss": 3.4451,
+      "grad_norm": 0.6235992908477783,
+      "learning_rate": 0.00041809606044252557,
+      "loss": 3.4527,
       "step": 28200
     },
     {
       "epoch": 3.045822102425876,
-      "grad_norm": 0.5639373660087585,
-      "learning_rate": 0.0004177593092282784,
-      "loss": 3.4552,
+      "grad_norm": 0.6089733242988586,
+      "learning_rate": 0.00041777226119805717,
+      "loss": 3.4613,
       "step": 28250
     },
     {
       "epoch": 3.0512129380053907,
-      "grad_norm": 0.593923807144165,
-      "learning_rate": 0.00041743550998381,
-      "loss": 3.4516,
+      "grad_norm": 0.6295673251152039,
+      "learning_rate": 0.0004174484619535887,
+      "loss": 3.4569,
       "step": 28300
     },
     {
       "epoch": 3.056603773584906,
-      "grad_norm": 0.6003124117851257,
-      "learning_rate": 0.0004171117107393416,
-      "loss": 3.4434,
+      "grad_norm": 0.5929367542266846,
+      "learning_rate": 0.00041712466270912033,
+      "loss": 3.4493,
       "step": 28350
     },
     {
       "epoch": 3.0619946091644206,
-      "grad_norm": 0.5810577869415283,
-      "learning_rate": 0.00041678791149487313,
-      "loss": 3.42,
+      "grad_norm": 0.613616406917572,
+      "learning_rate": 0.0004168008634646519,
+      "loss": 3.4263,
       "step": 28400
     },
     {
       "epoch": 3.0673854447439353,
-      "grad_norm": 0.606003999710083,
-      "learning_rate": 0.00041646411225040473,
-      "loss": 3.4639,
+      "grad_norm": 0.6133660674095154,
+      "learning_rate": 0.0004164770642201835,
+      "loss": 3.4705,
       "step": 28450
     },
     {
       "epoch": 3.07277628032345,
-      "grad_norm": 0.6627290844917297,
-      "learning_rate": 0.0004161403130059363,
-      "loss": 3.4576,
+      "grad_norm": 0.6664336323738098,
+      "learning_rate": 0.00041615326497571503,
+      "loss": 3.4629,
       "step": 28500
     },
     {
       "epoch": 3.078167115902965,
-      "grad_norm": 0.615278959274292,
-      "learning_rate": 0.0004158165137614679,
-      "loss": 3.4439,
+      "grad_norm": 0.6748630404472351,
+      "learning_rate": 0.0004158294657312466,
+      "loss": 3.4506,
       "step": 28550
     },
     {
       "epoch": 3.08355795148248,
-      "grad_norm": 0.5780900716781616,
-      "learning_rate": 0.00041549271451699944,
-      "loss": 3.4401,
+      "grad_norm": 0.6280421018600464,
+      "learning_rate": 0.0004155056664867782,
+      "loss": 3.4471,
       "step": 28600
     },
     {
       "epoch": 3.0889487870619945,
-      "grad_norm": 0.6600579023361206,
-      "learning_rate": 0.00041516891527253104,
-      "loss": 3.4401,
+      "grad_norm": 0.6666965484619141,
+      "learning_rate": 0.0004151818672423097,
+      "loss": 3.4467,
       "step": 28650
     },
     {
       "epoch": 3.0943396226415096,
-      "grad_norm": 0.5386844277381897,
-      "learning_rate": 0.00041484511602806254,
-      "loss": 3.45,
+      "grad_norm": 0.5865740776062012,
+      "learning_rate": 0.0004148580679978413,
+      "loss": 3.4561,
       "step": 28700
     },
     {
       "epoch": 3.0997304582210243,
-      "grad_norm": 0.5708300471305847,
-      "learning_rate": 0.0004145213167835941,
-      "loss": 3.4373,
+      "grad_norm": 0.6209290027618408,
+      "learning_rate": 0.00041453426875337284,
+      "loss": 3.4443,
       "step": 28750
     },
     {
       "epoch": 3.105121293800539,
-      "grad_norm": 0.5719059109687805,
-      "learning_rate": 0.0004141975175391257,
-      "loss": 3.4413,
+      "grad_norm": 0.5865198373794556,
+      "learning_rate": 0.00041421046950890445,
+      "loss": 3.4462,
       "step": 28800
     },
     {
       "epoch": 3.1105121293800537,
-      "grad_norm": 0.5947633385658264,
-      "learning_rate": 0.00041387371829465725,
-      "loss": 3.4252,
+      "grad_norm": 0.6128606200218201,
+      "learning_rate": 0.000413886670264436,
+      "loss": 3.4305,
       "step": 28850
     },
     {
       "epoch": 3.115902964959569,
-      "grad_norm": 0.6137687563896179,
-      "learning_rate": 0.00041354991905018885,
-      "loss": 3.4438,
+      "grad_norm": 0.5907280445098877,
+      "learning_rate": 0.0004135628710199676,
+      "loss": 3.4493,
       "step": 28900
     },
     {
       "epoch": 3.1212938005390836,
-      "grad_norm": 0.6112726330757141,
-      "learning_rate": 0.0004132261198057204,
-      "loss": 3.4593,
+      "grad_norm": 0.6502802968025208,
+      "learning_rate": 0.00041323907177549915,
+      "loss": 3.4671,
       "step": 28950
     },
     {
       "epoch": 3.1266846361185983,
-      "grad_norm": 0.6146498918533325,
-      "learning_rate": 0.000412902320561252,
-      "loss": 3.4286,
+      "grad_norm": 0.6189537644386292,
+      "learning_rate": 0.0004129152725310307,
+      "loss": 3.4347,
       "step": 29000
     },
     {
       "epoch": 3.1266846361185983,
-      "eval_accuracy": 0.371247034183611,
-      "eval_loss": 3.4940013885498047,
-      "eval_runtime": 183.9217,
-      "eval_samples_per_second": 97.928,
-      "eval_steps_per_second": 6.122,
+      "eval_accuracy": 0.3704489787226136,
+      "eval_loss": 3.5016098022460938,
+      "eval_runtime": 180.6679,
+      "eval_samples_per_second": 99.691,
+      "eval_steps_per_second": 6.232,
       "step": 29000
     },
     {
       "epoch": 3.1320754716981134,
-      "grad_norm": 0.6162520051002502,
-      "learning_rate": 0.00041258499730167296,
-      "loss": 3.4511,
+      "grad_norm": 0.677847146987915,
+      "learning_rate": 0.0004125914732865623,
+      "loss": 3.4562,
       "step": 29050
     },
     {
       "epoch": 3.137466307277628,
-      "grad_norm": 0.6008721590042114,
-      "learning_rate": 0.0004122611980572045,
-      "loss": 3.4405,
+      "grad_norm": 0.6556456089019775,
+      "learning_rate": 0.00041226767404209386,
+      "loss": 3.4467,
       "step": 29100
     },
     {
       "epoch": 3.142857142857143,
-      "grad_norm": 0.5806031227111816,
-      "learning_rate": 0.00041193739881273606,
-      "loss": 3.4575,
+      "grad_norm": 0.6047017574310303,
+      "learning_rate": 0.00041194387479762546,
+      "loss": 3.4633,
       "step": 29150
     },
     {
       "epoch": 3.1482479784366575,
-      "grad_norm": 0.5788300037384033,
-      "learning_rate": 0.00041161359956826766,
-      "loss": 3.4583,
+      "grad_norm": 0.5850408673286438,
+      "learning_rate": 0.000411620075553157,
+      "loss": 3.4652,
       "step": 29200
     },
     {
       "epoch": 3.1536388140161726,
-      "grad_norm": 0.6230771541595459,
-      "learning_rate": 0.0004112898003237992,
-      "loss": 3.4449,
+      "grad_norm": 0.5979166626930237,
+      "learning_rate": 0.0004112962763086886,
+      "loss": 3.4496,
       "step": 29250
     },
     {
       "epoch": 3.1590296495956873,
-      "grad_norm": 0.5821077227592468,
-      "learning_rate": 0.0004109660010793308,
-      "loss": 3.4518,
+      "grad_norm": 0.6348848938941956,
+      "learning_rate": 0.0004109724770642201,
+      "loss": 3.4571,
       "step": 29300
     },
     {
       "epoch": 3.164420485175202,
-      "grad_norm": 0.6511071920394897,
-      "learning_rate": 0.0004106422018348623,
-      "loss": 3.4639,
+      "grad_norm": 0.6686636209487915,
+      "learning_rate": 0.00041064867781975177,
+      "loss": 3.4705,
       "step": 29350
     },
     {
       "epoch": 3.169811320754717,
-      "grad_norm": 0.5678673982620239,
-      "learning_rate": 0.0004103184025903939,
-      "loss": 3.4581,
+      "grad_norm": 0.6021919250488281,
+      "learning_rate": 0.00041032487857528327,
+      "loss": 3.4636,
       "step": 29400
     },
     {
       "epoch": 3.175202156334232,
-      "grad_norm": 0.593774139881134,
-      "learning_rate": 0.00040999460334592547,
-      "loss": 3.4671,
+      "grad_norm": 0.5934717655181885,
+      "learning_rate": 0.0004100010793308148,
+      "loss": 3.4748,
       "step": 29450
     },
     {
       "epoch": 3.1805929919137466,
-      "grad_norm": 0.5744243860244751,
-      "learning_rate": 0.000409670804101457,
-      "loss": 3.4629,
+      "grad_norm": 0.6341081261634827,
+      "learning_rate": 0.0004096772800863464,
+      "loss": 3.4706,
       "step": 29500
     },
     {
       "epoch": 3.1859838274932613,
-      "grad_norm": 0.8164382576942444,
-      "learning_rate": 0.00040934700485698863,
-      "loss": 3.4392,
+      "grad_norm": 0.5976564884185791,
+      "learning_rate": 0.000409353480841878,
+      "loss": 3.4454,
       "step": 29550
     },
     {
       "epoch": 3.1913746630727764,
-      "grad_norm": 0.6126202940940857,
-      "learning_rate": 0.0004090232056125202,
-      "loss": 3.4606,
+      "grad_norm": 0.6214813590049744,
+      "learning_rate": 0.0004090296815974096,
+      "loss": 3.4676,
       "step": 29600
     },
     {
       "epoch": 3.196765498652291,
-      "grad_norm": 0.5741614103317261,
-      "learning_rate": 0.0004086994063680518,
-      "loss": 3.4472,
+      "grad_norm": 0.6009887456893921,
+      "learning_rate": 0.00040870588235294113,
+      "loss": 3.4544,
       "step": 29650
     },
     {
       "epoch": 3.202156334231806,
-      "grad_norm": 0.5594901442527771,
-      "learning_rate": 0.00040837560712358333,
-      "loss": 3.4545,
+      "grad_norm": 0.5742904543876648,
+      "learning_rate": 0.0004083885590933621,
+      "loss": 3.4598,
       "step": 29700
     },
     {
       "epoch": 3.207547169811321,
-      "grad_norm": 0.5645290613174438,
-      "learning_rate": 0.00040805180787911494,
-      "loss": 3.4618,
+      "grad_norm": 0.6009396910667419,
+      "learning_rate": 0.00040806475984889363,
+      "loss": 3.4684,
       "step": 29750
     },
     {
       "epoch": 3.2129380053908356,
-      "grad_norm": 0.5808700919151306,
-      "learning_rate": 0.0004077280086346465,
-      "loss": 3.4444,
+      "grad_norm": 0.6059123873710632,
+      "learning_rate": 0.00040774096060442524,
+      "loss": 3.4502,
       "step": 29800
     },
     {
       "epoch": 3.2183288409703503,
-      "grad_norm": 0.6301125288009644,
-      "learning_rate": 0.0004074042093901781,
-      "loss": 3.4651,
+      "grad_norm": 0.6594415903091431,
+      "learning_rate": 0.0004074171613599568,
+      "loss": 3.4727,
       "step": 29850
     },
     {
       "epoch": 3.223719676549865,
-      "grad_norm": 0.5941428542137146,
-      "learning_rate": 0.00040708041014570964,
-      "loss": 3.4735,
+      "grad_norm": 0.6165498495101929,
+      "learning_rate": 0.0004070933621154884,
+      "loss": 3.478,
       "step": 29900
     },
     {
       "epoch": 3.22911051212938,
-      "grad_norm": 0.5769610404968262,
-      "learning_rate": 0.00040675661090124114,
-      "loss": 3.4638,
+      "grad_norm": 0.6345921754837036,
+      "learning_rate": 0.00040676956287101994,
+      "loss": 3.4701,
       "step": 29950
     },
     {
       "epoch": 3.234501347708895,
-      "grad_norm": 0.6222258806228638,
-      "learning_rate": 0.0004064328116567728,
-      "loss": 3.4506,
+      "grad_norm": 0.6863733530044556,
+      "learning_rate": 0.00040644576362655155,
+      "loss": 3.4554,
       "step": 30000
     },
     {
       "epoch": 3.234501347708895,
-      "eval_accuracy": 0.3719082956588499,
-      "eval_loss": 3.4874448776245117,
-      "eval_runtime": 183.6565,
-      "eval_samples_per_second": 98.069,
-      "eval_steps_per_second": 6.131,
+      "eval_accuracy": 0.3711672829639552,
+      "eval_loss": 3.4930102825164795,
+      "eval_runtime": 181.1176,
+      "eval_samples_per_second": 99.444,
+      "eval_steps_per_second": 6.217,
       "step": 30000
     },
     {
       "epoch": 3.2398921832884096,
-      "grad_norm": 0.6240214109420776,
-      "learning_rate": 0.0004061090124123043,
-      "loss": 3.4467,
+      "grad_norm": 0.6440929174423218,
+      "learning_rate": 0.00040612196438208304,
+      "loss": 3.4528,
       "step": 30050
     },
     {
       "epoch": 3.2452830188679247,
-      "grad_norm": 0.608329713344574,
-      "learning_rate": 0.0004057852131678359,
-      "loss": 3.4485,
+      "grad_norm": 0.6343992948532104,
+      "learning_rate": 0.00040579816513761465,
+      "loss": 3.456,
       "step": 30100
     },
     {
       "epoch": 3.2506738544474394,
-      "grad_norm": 0.5454065799713135,
-      "learning_rate": 0.00040546141392336745,
-      "loss": 3.4499,
+      "grad_norm": 0.5912777185440063,
+      "learning_rate": 0.0004054743658931462,
+      "loss": 3.4573,
       "step": 30150
     },
     {
       "epoch": 3.256064690026954,
-      "grad_norm": 0.667070209980011,
-      "learning_rate": 0.00040513761467889906,
-      "loss": 3.4693,
+      "grad_norm": 0.6378411054611206,
+      "learning_rate": 0.00040515056664867775,
+      "loss": 3.4762,
       "step": 30200
     },
     {
       "epoch": 3.2614555256064692,
-      "grad_norm": 0.6167458295822144,
-      "learning_rate": 0.0004048138154344306,
-      "loss": 3.4631,
+      "grad_norm": 0.6168299317359924,
+      "learning_rate": 0.00040482676740420935,
+      "loss": 3.4697,
       "step": 30250
     },
     {
       "epoch": 3.266846361185984,
-      "grad_norm": 0.7352801561355591,
-      "learning_rate": 0.0004044900161899622,
-      "loss": 3.4654,
+      "grad_norm": 0.6656365990638733,
+      "learning_rate": 0.0004045029681597409,
+      "loss": 3.4709,
       "step": 30300
     },
     {
       "epoch": 3.2722371967654986,
-      "grad_norm": 0.6545478701591492,
-      "learning_rate": 0.00040416621694549376,
-      "loss": 3.4401,
+      "grad_norm": 0.6865881085395813,
+      "learning_rate": 0.0004041791689152725,
+      "loss": 3.4467,
       "step": 30350
     },
     {
       "epoch": 3.2776280323450133,
-      "grad_norm": 0.6410787105560303,
-      "learning_rate": 0.0004038424177010253,
-      "loss": 3.4697,
+      "grad_norm": 0.6933385729789734,
+      "learning_rate": 0.00040385536967080406,
+      "loss": 3.4779,
       "step": 30400
     },
     {
       "epoch": 3.2830188679245285,
-      "grad_norm": 0.5979618430137634,
-      "learning_rate": 0.0004035186184565569,
-      "loss": 3.4568,
+      "grad_norm": 0.5782449841499329,
+      "learning_rate": 0.00040353157042633567,
+      "loss": 3.4637,
       "step": 30450
     },
     {
       "epoch": 3.288409703504043,
-      "grad_norm": 0.6123748421669006,
-      "learning_rate": 0.00040319481921208847,
-      "loss": 3.4417,
+      "grad_norm": 0.607099711894989,
+      "learning_rate": 0.0004032077711818672,
+      "loss": 3.4467,
       "step": 30500
     },
     {
       "epoch": 3.293800539083558,
-      "grad_norm": 0.607407808303833,
-      "learning_rate": 0.00040287101996762007,
-      "loss": 3.4708,
+      "grad_norm": 0.6115757822990417,
+      "learning_rate": 0.0004028839719373988,
+      "loss": 3.4751,
       "step": 30550
     },
     {
       "epoch": 3.2991913746630726,
-      "grad_norm": 0.7021310329437256,
-      "learning_rate": 0.0004025472207231516,
-      "loss": 3.4618,
+      "grad_norm": 0.6959110498428345,
+      "learning_rate": 0.00040256017269293037,
+      "loss": 3.467,
       "step": 30600
     },
     {
       "epoch": 3.3045822102425877,
-      "grad_norm": 0.6325151920318604,
-      "learning_rate": 0.00040222342147868323,
-      "loss": 3.4408,
+      "grad_norm": 0.6443840265274048,
+      "learning_rate": 0.00040223637344846187,
+      "loss": 3.4477,
       "step": 30650
     },
     {
       "epoch": 3.3099730458221024,
-      "grad_norm": 0.6190793514251709,
-      "learning_rate": 0.0004018996222342147,
-      "loss": 3.4592,
+      "grad_norm": 0.6412681937217712,
+      "learning_rate": 0.00040191257420399347,
+      "loss": 3.4654,
       "step": 30700
     },
     {
       "epoch": 3.315363881401617,
-      "grad_norm": 0.5823326110839844,
-      "learning_rate": 0.00040157582298974633,
-      "loss": 3.4269,
+      "grad_norm": 0.6385542750358582,
+      "learning_rate": 0.000401588774959525,
+      "loss": 3.4351,
       "step": 30750
     },
     {
       "epoch": 3.3207547169811322,
-      "grad_norm": 0.5816695690155029,
-      "learning_rate": 0.0004012520237452779,
-      "loss": 3.4634,
+      "grad_norm": 0.5896449089050293,
+      "learning_rate": 0.00040126497571505663,
+      "loss": 3.4706,
       "step": 30800
     },
     {
       "epoch": 3.326145552560647,
-      "grad_norm": 0.6027852892875671,
-      "learning_rate": 0.00040092822450080943,
-      "loss": 3.4441,
+      "grad_norm": 0.6303786039352417,
+      "learning_rate": 0.0004009411764705882,
+      "loss": 3.45,
       "step": 30850
     },
     {
       "epoch": 3.3315363881401616,
-      "grad_norm": 0.6045222878456116,
-      "learning_rate": 0.00040060442525634103,
-      "loss": 3.4673,
+      "grad_norm": 0.6891083121299744,
+      "learning_rate": 0.0004006173772261198,
+      "loss": 3.4743,
       "step": 30900
     },
     {
       "epoch": 3.3369272237196768,
-      "grad_norm": 0.597585141658783,
-      "learning_rate": 0.0004002806260118726,
-      "loss": 3.4441,
+      "grad_norm": 0.6005822420120239,
+      "learning_rate": 0.00040029357798165133,
+      "loss": 3.4522,
       "step": 30950
     },
     {
       "epoch": 3.3423180592991915,
-      "grad_norm": 0.6446326375007629,
-      "learning_rate": 0.0003999568267674042,
-      "loss": 3.4563,
+      "grad_norm": 0.685102641582489,
+      "learning_rate": 0.00039996977873718294,
+      "loss": 3.4623,
       "step": 31000
     },
     {
       "epoch": 3.3423180592991915,
-      "eval_accuracy": 0.3723845212661408,
-      "eval_loss": 3.4811089038848877,
-      "eval_runtime": 183.6383,
-      "eval_samples_per_second": 98.079,
-      "eval_steps_per_second": 6.132,
+      "eval_accuracy": 0.3716806678588786,
+      "eval_loss": 3.487154483795166,
+      "eval_runtime": 181.2264,
+      "eval_samples_per_second": 99.384,
+      "eval_steps_per_second": 6.213,
       "step": 31000
     },
     {
       "epoch": 3.347708894878706,
-      "grad_norm": 0.6195496916770935,
-      "learning_rate": 0.00039963302752293574,
-      "loss": 3.4571,
+      "grad_norm": 0.6292726397514343,
+      "learning_rate": 0.0003996459794927145,
+      "loss": 3.4644,
       "step": 31050
     },
     {
       "epoch": 3.353099730458221,
-      "grad_norm": 0.6132037043571472,
-      "learning_rate": 0.00039930922827846735,
-      "loss": 3.4524,
+      "grad_norm": 0.6325132250785828,
+      "learning_rate": 0.00039932218024824604,
+      "loss": 3.4592,
       "step": 31100
     },
     {
       "epoch": 3.358490566037736,
-      "grad_norm": 0.5746621489524841,
-      "learning_rate": 0.0003989854290339989,
-      "loss": 3.463,
+      "grad_norm": 0.6206766963005066,
+      "learning_rate": 0.00039899838100377764,
+      "loss": 3.47,
       "step": 31150
     },
     {
       "epoch": 3.3638814016172507,
-      "grad_norm": 0.596961498260498,
-      "learning_rate": 0.00039866810577441985,
-      "loss": 3.4495,
+      "grad_norm": 0.6283134818077087,
+      "learning_rate": 0.0003986745817593092,
+      "loss": 3.4541,
       "step": 31200
     },
     {
       "epoch": 3.3692722371967654,
-      "grad_norm": 0.5609466433525085,
-      "learning_rate": 0.0003983443065299514,
-      "loss": 3.4515,
+      "grad_norm": 0.6028231382369995,
+      "learning_rate": 0.0003983507825148408,
+      "loss": 3.4582,
       "step": 31250
     },
     {
       "epoch": 3.37466307277628,
-      "grad_norm": 0.5683972835540771,
-      "learning_rate": 0.000398020507285483,
-      "loss": 3.4771,
+      "grad_norm": 0.6341281533241272,
+      "learning_rate": 0.00039802698327037235,
+      "loss": 3.4843,
       "step": 31300
     },
     {
       "epoch": 3.3800539083557952,
-      "grad_norm": 0.6360358595848083,
-      "learning_rate": 0.0003976967080410145,
-      "loss": 3.4486,
+      "grad_norm": 0.6757832765579224,
+      "learning_rate": 0.00039770318402590396,
+      "loss": 3.4552,
       "step": 31350
     },
     {
       "epoch": 3.38544474393531,
-      "grad_norm": 0.6113670468330383,
-      "learning_rate": 0.0003973729087965461,
-      "loss": 3.448,
+      "grad_norm": 0.6310288310050964,
+      "learning_rate": 0.00039737938478143545,
+      "loss": 3.4541,
       "step": 31400
     },
     {
       "epoch": 3.3908355795148246,
-      "grad_norm": 0.5796744227409363,
-      "learning_rate": 0.00039704910955207765,
-      "loss": 3.4534,
+      "grad_norm": 0.6138319969177246,
+      "learning_rate": 0.000397055585536967,
+      "loss": 3.4609,
       "step": 31450
     },
     {
       "epoch": 3.3962264150943398,
-      "grad_norm": 0.5753939151763916,
-      "learning_rate": 0.00039672531030760926,
-      "loss": 3.4462,
+      "grad_norm": 0.6245617270469666,
+      "learning_rate": 0.0003967317862924986,
+      "loss": 3.4549,
       "step": 31500
     },
     {
       "epoch": 3.4016172506738545,
-      "grad_norm": 0.6340164542198181,
-      "learning_rate": 0.0003964015110631408,
-      "loss": 3.4674,
+      "grad_norm": 0.6630147695541382,
+      "learning_rate": 0.00039640798704803016,
+      "loss": 3.4736,
       "step": 31550
     },
     {
       "epoch": 3.407008086253369,
-      "grad_norm": 0.5975732207298279,
-      "learning_rate": 0.00039607771181867236,
-      "loss": 3.445,
+      "grad_norm": 0.6623867154121399,
+      "learning_rate": 0.00039608418780356176,
+      "loss": 3.4511,
       "step": 31600
     },
     {
       "epoch": 3.4123989218328843,
-      "grad_norm": 0.5981366634368896,
-      "learning_rate": 0.00039575391257420397,
-      "loss": 3.4462,
+      "grad_norm": 0.621094822883606,
+      "learning_rate": 0.0003957603885590933,
+      "loss": 3.4536,
       "step": 31650
     },
     {
       "epoch": 3.417789757412399,
-      "grad_norm": 0.6517634987831116,
-      "learning_rate": 0.0003954301133297355,
-      "loss": 3.4725,
+      "grad_norm": 0.646723210811615,
+      "learning_rate": 0.00039544306529951426,
+      "loss": 3.4786,
       "step": 31700
     },
     {
       "epoch": 3.4231805929919137,
-      "grad_norm": 0.6406164169311523,
-      "learning_rate": 0.0003951063140852671,
-      "loss": 3.4464,
+      "grad_norm": 0.6829112768173218,
+      "learning_rate": 0.00039511926605504587,
+      "loss": 3.4525,
       "step": 31750
     },
     {
       "epoch": 3.4285714285714284,
-      "grad_norm": 0.6119227409362793,
-      "learning_rate": 0.00039478251484079867,
-      "loss": 3.4483,
+      "grad_norm": 0.6134275794029236,
+      "learning_rate": 0.0003947954668105774,
+      "loss": 3.4545,
       "step": 31800
     },
     {
       "epoch": 3.4339622641509435,
-      "grad_norm": 0.6088457107543945,
-      "learning_rate": 0.0003944587155963303,
-      "loss": 3.4796,
+      "grad_norm": 0.6556375026702881,
+      "learning_rate": 0.00039447166756610897,
+      "loss": 3.4852,
       "step": 31850
     },
     {
       "epoch": 3.439353099730458,
-      "grad_norm": 0.6567962169647217,
-      "learning_rate": 0.0003941349163518618,
-      "loss": 3.4615,
+      "grad_norm": 0.7237303853034973,
+      "learning_rate": 0.0003941478683216406,
+      "loss": 3.4677,
       "step": 31900
     },
     {
       "epoch": 3.444743935309973,
-      "grad_norm": 0.6715599894523621,
-      "learning_rate": 0.00039381111710739343,
-      "loss": 3.4414,
+      "grad_norm": 0.6524747014045715,
+      "learning_rate": 0.0003938240690771721,
+      "loss": 3.4474,
       "step": 31950
     },
     {
       "epoch": 3.450134770889488,
-      "grad_norm": 0.6421747803688049,
-      "learning_rate": 0.000393487317862925,
-      "loss": 3.4568,
+      "grad_norm": 0.6801609396934509,
+      "learning_rate": 0.00039350026983270373,
+      "loss": 3.4633,
       "step": 32000
     },
     {
       "epoch": 3.450134770889488,
-      "eval_accuracy": 0.3734006430729818,
-      "eval_loss": 3.473811388015747,
-      "eval_runtime": 183.9,
-      "eval_samples_per_second": 97.939,
-      "eval_steps_per_second": 6.123,
+      "eval_accuracy": 0.37275328916738315,
+      "eval_loss": 3.479112148284912,
+      "eval_runtime": 181.9617,
+      "eval_samples_per_second": 98.982,
+      "eval_steps_per_second": 6.188,
       "step": 32000
     },
     {
       "epoch": 3.4555256064690028,
-      "grad_norm": 0.593186616897583,
-      "learning_rate": 0.0003931635186184565,
-      "loss": 3.4593,
+      "grad_norm": 0.6095766425132751,
+      "learning_rate": 0.00039317647058823523,
+      "loss": 3.4656,
       "step": 32050
     },
     {
       "epoch": 3.4609164420485174,
-      "grad_norm": 0.697148859500885,
-      "learning_rate": 0.0003928397193739881,
-      "loss": 3.453,
+      "grad_norm": 0.72801673412323,
+      "learning_rate": 0.00039285267134376683,
+      "loss": 3.4583,
       "step": 32100
     },
     {
       "epoch": 3.466307277628032,
-      "grad_norm": 0.6392279863357544,
-      "learning_rate": 0.00039251592012951963,
-      "loss": 3.4662,
+      "grad_norm": 0.5965682864189148,
+      "learning_rate": 0.0003925288720992984,
+      "loss": 3.4721,
       "step": 32150
     },
     {
       "epoch": 3.4716981132075473,
-      "grad_norm": 0.5713638663291931,
-      "learning_rate": 0.00039219212088505124,
-      "loss": 3.4484,
+      "grad_norm": 0.6114888787269592,
+      "learning_rate": 0.00039220507285482993,
+      "loss": 3.4533,
       "step": 32200
     },
     {
       "epoch": 3.477088948787062,
-      "grad_norm": 0.6339240074157715,
-      "learning_rate": 0.0003918683216405828,
-      "loss": 3.4653,
+      "grad_norm": 0.6725350618362427,
+      "learning_rate": 0.00039188127361036154,
+      "loss": 3.4698,
       "step": 32250
     },
     {
       "epoch": 3.4824797843665767,
-      "grad_norm": 0.6048734784126282,
-      "learning_rate": 0.0003915445223961144,
-      "loss": 3.4757,
+      "grad_norm": 0.6084803342819214,
+      "learning_rate": 0.0003915574743658931,
+      "loss": 3.4831,
       "step": 32300
     },
     {
       "epoch": 3.487870619946092,
-      "grad_norm": 0.616313099861145,
-      "learning_rate": 0.00039122072315164594,
-      "loss": 3.4397,
+      "grad_norm": 0.6338387131690979,
+      "learning_rate": 0.0003912336751214247,
+      "loss": 3.4439,
       "step": 32350
     },
     {
       "epoch": 3.4932614555256065,
-      "grad_norm": 0.6406932473182678,
-      "learning_rate": 0.0003908969239071775,
-      "loss": 3.4706,
+      "grad_norm": 0.6302171349525452,
+      "learning_rate": 0.00039090987587695624,
+      "loss": 3.4765,
       "step": 32400
     },
     {
       "epoch": 3.498652291105121,
-      "grad_norm": 0.6056227087974548,
-      "learning_rate": 0.0003905731246627091,
-      "loss": 3.4449,
+      "grad_norm": 0.628710150718689,
+      "learning_rate": 0.00039058607663248785,
+      "loss": 3.4496,
       "step": 32450
     },
     {
       "epoch": 3.5040431266846364,
-      "grad_norm": 0.5830230712890625,
-      "learning_rate": 0.00039024932541824065,
-      "loss": 3.4704,
+      "grad_norm": 0.5954582095146179,
+      "learning_rate": 0.0003902622773880194,
+      "loss": 3.4751,
       "step": 32500
     },
     {
       "epoch": 3.509433962264151,
-      "grad_norm": 0.5504217147827148,
-      "learning_rate": 0.00038992552617377225,
-      "loss": 3.4591,
+      "grad_norm": 0.5757933259010315,
+      "learning_rate": 0.000389938478143551,
+      "loss": 3.4665,
       "step": 32550
     },
     {
       "epoch": 3.5148247978436657,
-      "grad_norm": 0.586064875125885,
-      "learning_rate": 0.0003896017269293038,
-      "loss": 3.4541,
+      "grad_norm": 0.6051267981529236,
+      "learning_rate": 0.00038961467889908255,
+      "loss": 3.4618,
       "step": 32600
     },
     {
       "epoch": 3.5202156334231804,
-      "grad_norm": 0.6093249917030334,
-      "learning_rate": 0.0003892779276848354,
-      "loss": 3.4655,
+      "grad_norm": 0.6285063028335571,
+      "learning_rate": 0.00038929087965461405,
+      "loss": 3.4725,
       "step": 32650
     },
     {
       "epoch": 3.525606469002695,
-      "grad_norm": 0.6133269667625427,
-      "learning_rate": 0.0003889541284403669,
-      "loss": 3.4392,
+      "grad_norm": 0.6470969915390015,
+      "learning_rate": 0.00038896708041014566,
+      "loss": 3.4467,
       "step": 32700
     },
     {
       "epoch": 3.5309973045822103,
-      "grad_norm": 0.6184484958648682,
-      "learning_rate": 0.0003886303291958985,
-      "loss": 3.4513,
+      "grad_norm": 0.8486336469650269,
+      "learning_rate": 0.0003886432811656772,
+      "loss": 3.4568,
       "step": 32750
     },
     {
       "epoch": 3.536388140161725,
-      "grad_norm": 0.6110262870788574,
-      "learning_rate": 0.00038830652995143006,
-      "loss": 3.4623,
+      "grad_norm": 0.564283549785614,
+      "learning_rate": 0.0003883194819212088,
+      "loss": 3.4685,
       "step": 32800
     },
     {
       "epoch": 3.5417789757412397,
-      "grad_norm": 0.6329076886177063,
-      "learning_rate": 0.0003879827307069616,
-      "loss": 3.4245,
+      "grad_norm": 0.6771153211593628,
+      "learning_rate": 0.00038799568267674036,
+      "loss": 3.4306,
       "step": 32850
     },
     {
       "epoch": 3.547169811320755,
-      "grad_norm": 0.6302708387374878,
-      "learning_rate": 0.0003876589314624932,
-      "loss": 3.4471,
+      "grad_norm": 0.6061602830886841,
+      "learning_rate": 0.00038767188343227197,
+      "loss": 3.455,
       "step": 32900
     },
     {
       "epoch": 3.5525606469002695,
-      "grad_norm": 0.6204167008399963,
-      "learning_rate": 0.00038733513221802477,
-      "loss": 3.4626,
+      "grad_norm": 0.6345692276954651,
+      "learning_rate": 0.0003873480841878035,
+      "loss": 3.4675,
       "step": 32950
     },
     {
       "epoch": 3.557951482479784,
-      "grad_norm": 0.635398805141449,
-      "learning_rate": 0.00038701133297355637,
-      "loss": 3.4606,
+      "grad_norm": 0.668306291103363,
+      "learning_rate": 0.0003870242849433351,
+      "loss": 3.4665,
       "step": 33000
     },
     {
       "epoch": 3.557951482479784,
-      "eval_accuracy": 0.37382395472390706,
-      "eval_loss": 3.468949556350708,
-      "eval_runtime": 183.5943,
-      "eval_samples_per_second": 98.102,
-      "eval_steps_per_second": 6.133,
+      "eval_accuracy": 0.3731642143890975,
+      "eval_loss": 3.475476026535034,
+      "eval_runtime": 181.4105,
+      "eval_samples_per_second": 99.283,
+      "eval_steps_per_second": 6.207,
       "step": 33000
     },
     {
       "epoch": 3.5633423180592994,
-      "grad_norm": 0.5870746374130249,
-      "learning_rate": 0.0003866875337290879,
-      "loss": 3.4628,
+      "grad_norm": 0.6180144548416138,
+      "learning_rate": 0.00038670048569886667,
+      "loss": 3.469,
       "step": 33050
     },
     {
       "epoch": 3.568733153638814,
-      "grad_norm": 0.6441285014152527,
-      "learning_rate": 0.00038636373448461953,
-      "loss": 3.4496,
+      "grad_norm": 0.6204782724380493,
+      "learning_rate": 0.0003863766864543982,
+      "loss": 3.4549,
       "step": 33100
     },
     {
       "epoch": 3.5741239892183287,
-      "grad_norm": 0.6544851660728455,
-      "learning_rate": 0.0003860399352401511,
-      "loss": 3.4598,
+      "grad_norm": 0.6758031249046326,
+      "learning_rate": 0.00038605288720992983,
+      "loss": 3.4642,
       "step": 33150
     },
     {
       "epoch": 3.579514824797844,
-      "grad_norm": 0.601290225982666,
-      "learning_rate": 0.0003857161359956827,
-      "loss": 3.4587,
+      "grad_norm": 0.5960129499435425,
+      "learning_rate": 0.0003857290879654614,
+      "loss": 3.4645,
       "step": 33200
     },
     {
       "epoch": 3.5849056603773586,
-      "grad_norm": 0.6271600127220154,
-      "learning_rate": 0.0003853988127361036,
-      "loss": 3.4506,
+      "grad_norm": 0.6631293296813965,
+      "learning_rate": 0.000385405288720993,
+      "loss": 3.4572,
       "step": 33250
     },
     {
       "epoch": 3.5902964959568733,
-      "grad_norm": 0.6283750534057617,
-      "learning_rate": 0.0003850750134916352,
-      "loss": 3.4413,
+      "grad_norm": 0.6704812049865723,
+      "learning_rate": 0.00038508148947652453,
+      "loss": 3.4463,
       "step": 33300
     },
     {
       "epoch": 3.595687331536388,
-      "grad_norm": 0.5692456364631653,
-      "learning_rate": 0.0003847512142471667,
-      "loss": 3.4524,
+      "grad_norm": 0.6017289161682129,
+      "learning_rate": 0.00038475769023205614,
+      "loss": 3.4586,
       "step": 33350
     },
     {
       "epoch": 3.601078167115903,
-      "grad_norm": 0.6084824204444885,
-      "learning_rate": 0.0003844274150026983,
-      "loss": 3.4519,
+      "grad_norm": 0.6118620038032532,
+      "learning_rate": 0.00038443389098758763,
+      "loss": 3.4575,
       "step": 33400
     },
     {
       "epoch": 3.606469002695418,
-      "grad_norm": 0.6203867197036743,
-      "learning_rate": 0.00038410361575822984,
-      "loss": 3.4636,
+      "grad_norm": 0.6353810429573059,
+      "learning_rate": 0.00038411009174311924,
+      "loss": 3.4685,
       "step": 33450
     },
     {
       "epoch": 3.6118598382749325,
-      "grad_norm": 0.5996057987213135,
-      "learning_rate": 0.00038377981651376144,
-      "loss": 3.4679,
+      "grad_norm": 0.6293622255325317,
+      "learning_rate": 0.0003837862924986508,
+      "loss": 3.4747,
       "step": 33500
     },
     {
       "epoch": 3.617250673854447,
-      "grad_norm": 0.5934234857559204,
-      "learning_rate": 0.000383456017269293,
-      "loss": 3.4491,
+      "grad_norm": 0.6343139410018921,
+      "learning_rate": 0.00038346249325418234,
+      "loss": 3.4543,
       "step": 33550
     },
     {
       "epoch": 3.6226415094339623,
-      "grad_norm": 0.6062605381011963,
-      "learning_rate": 0.00038313221802482454,
-      "loss": 3.451,
+      "grad_norm": 0.6264575123786926,
+      "learning_rate": 0.00038313869400971395,
+      "loss": 3.4553,
       "step": 33600
     },
     {
       "epoch": 3.628032345013477,
-      "grad_norm": 0.6698931455612183,
-      "learning_rate": 0.00038280841878035615,
-      "loss": 3.4472,
+      "grad_norm": 0.6940131187438965,
+      "learning_rate": 0.0003828148947652455,
+      "loss": 3.4546,
       "step": 33650
     },
     {
       "epoch": 3.6334231805929917,
-      "grad_norm": 0.6485372185707092,
-      "learning_rate": 0.0003824846195358877,
-      "loss": 3.4337,
+      "grad_norm": 0.6606136560440063,
+      "learning_rate": 0.00038249757150566645,
+      "loss": 3.4403,
       "step": 33700
     },
     {
       "epoch": 3.638814016172507,
-      "grad_norm": 0.6247066259384155,
-      "learning_rate": 0.0003821608202914193,
-      "loss": 3.4591,
+      "grad_norm": 0.6887759566307068,
+      "learning_rate": 0.00038217377226119805,
+      "loss": 3.4637,
       "step": 33750
     },
     {
       "epoch": 3.6442048517520216,
-      "grad_norm": 0.6120551228523254,
-      "learning_rate": 0.00038183702104695085,
-      "loss": 3.4388,
+      "grad_norm": 0.618350088596344,
+      "learning_rate": 0.0003818499730167296,
+      "loss": 3.4443,
       "step": 33800
     },
     {
       "epoch": 3.6495956873315363,
-      "grad_norm": 0.6113671660423279,
-      "learning_rate": 0.00038151322180248246,
-      "loss": 3.4713,
+      "grad_norm": 0.6543127298355103,
+      "learning_rate": 0.00038152617377226115,
+      "loss": 3.4763,
       "step": 33850
     },
     {
       "epoch": 3.6549865229110514,
-      "grad_norm": 0.5842575430870056,
-      "learning_rate": 0.000381189422558014,
-      "loss": 3.4447,
+      "grad_norm": 0.6391322612762451,
+      "learning_rate": 0.00038120237452779276,
+      "loss": 3.4501,
       "step": 33900
     },
     {
       "epoch": 3.660377358490566,
-      "grad_norm": 0.6196116209030151,
-      "learning_rate": 0.0003808656233135456,
-      "loss": 3.4459,
+      "grad_norm": 0.6833434104919434,
+      "learning_rate": 0.0003808785752833243,
+      "loss": 3.4538,
       "step": 33950
     },
     {
       "epoch": 3.665768194070081,
-      "grad_norm": 0.6134396195411682,
-      "learning_rate": 0.00038054182406907716,
-      "loss": 3.4583,
+      "grad_norm": 0.6446322202682495,
+      "learning_rate": 0.0003805547760388559,
+      "loss": 3.4633,
       "step": 34000
     },
     {
       "epoch": 3.665768194070081,
-      "eval_accuracy": 0.3740601661020157,
-      "eval_loss": 3.4624133110046387,
-      "eval_runtime": 184.0257,
-      "eval_samples_per_second": 97.872,
-      "eval_steps_per_second": 6.119,
+      "eval_accuracy": 0.37342280826209595,
+      "eval_loss": 3.4682791233062744,
+      "eval_runtime": 182.1113,
+      "eval_samples_per_second": 98.901,
+      "eval_steps_per_second": 6.183,
       "step": 34000
     },
     {
       "epoch": 3.671159029649596,
-      "grad_norm": 0.5997918248176575,
-      "learning_rate": 0.00038021802482460866,
-      "loss": 3.4519,
+      "grad_norm": 0.6396611332893372,
+      "learning_rate": 0.0003802309767943874,
+      "loss": 3.458,
       "step": 34050
     },
     {
       "epoch": 3.6765498652291106,
-      "grad_norm": 0.5810784101486206,
-      "learning_rate": 0.00037989422558014027,
-      "loss": 3.4694,
+      "grad_norm": 0.6037476658821106,
+      "learning_rate": 0.000379907177549919,
+      "loss": 3.4733,
       "step": 34100
     },
     {
       "epoch": 3.6819407008086253,
-      "grad_norm": 0.6275520324707031,
-      "learning_rate": 0.0003795704263356718,
-      "loss": 3.4657,
+      "grad_norm": 0.6792062520980835,
+      "learning_rate": 0.00037958337830545057,
+      "loss": 3.4704,
       "step": 34150
     },
     {
       "epoch": 3.68733153638814,
-      "grad_norm": 0.6059765815734863,
-      "learning_rate": 0.0003792466270912034,
-      "loss": 3.4458,
+      "grad_norm": 0.5976162552833557,
+      "learning_rate": 0.00037925957906098217,
+      "loss": 3.4503,
       "step": 34200
     },
     {
       "epoch": 3.6927223719676547,
-      "grad_norm": 0.6004732251167297,
-      "learning_rate": 0.00037892282784673497,
-      "loss": 3.4397,
+      "grad_norm": 0.6629889011383057,
+      "learning_rate": 0.0003789357798165137,
+      "loss": 3.443,
       "step": 34250
     },
     {
       "epoch": 3.69811320754717,
-      "grad_norm": 0.6046389937400818,
-      "learning_rate": 0.0003785990286022666,
-      "loss": 3.4392,
+      "grad_norm": 0.6140479445457458,
+      "learning_rate": 0.00037861198057204527,
+      "loss": 3.4448,
       "step": 34300
     },
     {
       "epoch": 3.7035040431266846,
-      "grad_norm": 0.6276936531066895,
-      "learning_rate": 0.00037827522935779813,
-      "loss": 3.4455,
+      "grad_norm": 0.6195974946022034,
+      "learning_rate": 0.0003782881813275769,
+      "loss": 3.4502,
       "step": 34350
     },
     {
       "epoch": 3.7088948787061993,
-      "grad_norm": 0.5531956553459167,
-      "learning_rate": 0.00037795143011332973,
-      "loss": 3.4406,
+      "grad_norm": 0.5922988653182983,
+      "learning_rate": 0.0003779643820831084,
+      "loss": 3.4468,
       "step": 34400
     },
     {
       "epoch": 3.7142857142857144,
-      "grad_norm": 0.576456606388092,
-      "learning_rate": 0.0003776276308688613,
-      "loss": 3.4619,
+      "grad_norm": 0.6114234924316406,
+      "learning_rate": 0.00037764058283864003,
+      "loss": 3.4674,
       "step": 34450
     },
     {
       "epoch": 3.719676549865229,
-      "grad_norm": 0.5860700011253357,
-      "learning_rate": 0.00037730383162439283,
-      "loss": 3.4433,
+      "grad_norm": 0.6167077422142029,
+      "learning_rate": 0.0003773167835941716,
+      "loss": 3.4503,
       "step": 34500
     },
     {
       "epoch": 3.725067385444744,
-      "grad_norm": 0.6148476004600525,
-      "learning_rate": 0.00037698003237992444,
-      "loss": 3.4385,
+      "grad_norm": 0.6252519488334656,
+      "learning_rate": 0.0003769929843497032,
+      "loss": 3.443,
       "step": 34550
     },
     {
       "epoch": 3.730458221024259,
-      "grad_norm": 0.6846517324447632,
-      "learning_rate": 0.000376656233135456,
-      "loss": 3.4564,
+      "grad_norm": 0.7280793786048889,
+      "learning_rate": 0.00037666918510523474,
+      "loss": 3.4633,
       "step": 34600
     },
     {
       "epoch": 3.7358490566037736,
-      "grad_norm": 0.6337952613830566,
-      "learning_rate": 0.0003763324338909876,
-      "loss": 3.4555,
+      "grad_norm": 0.5928522348403931,
+      "learning_rate": 0.00037634538586076634,
+      "loss": 3.462,
       "step": 34650
     },
     {
       "epoch": 3.7412398921832883,
-      "grad_norm": 0.6312965154647827,
-      "learning_rate": 0.0003760086346465191,
-      "loss": 3.4656,
+      "grad_norm": 0.6633599996566772,
+      "learning_rate": 0.00037602158661629784,
+      "loss": 3.4701,
       "step": 34700
     },
     {
       "epoch": 3.7466307277628035,
-      "grad_norm": 0.5984307527542114,
-      "learning_rate": 0.0003756848354020507,
-      "loss": 3.4617,
+      "grad_norm": 0.6255478262901306,
+      "learning_rate": 0.0003757042633567188,
+      "loss": 3.468,
       "step": 34750
     },
     {
       "epoch": 3.752021563342318,
-      "grad_norm": 0.5848837494850159,
-      "learning_rate": 0.00037536103615758224,
-      "loss": 3.4411,
+      "grad_norm": 0.6267266869544983,
+      "learning_rate": 0.00037538046411225034,
+      "loss": 3.4457,
       "step": 34800
     },
     {
       "epoch": 3.757412398921833,
-      "grad_norm": 0.6520708799362183,
-      "learning_rate": 0.0003750372369131138,
-      "loss": 3.4426,
+      "grad_norm": 0.6801672577857971,
+      "learning_rate": 0.00037505666486778195,
+      "loss": 3.4494,
       "step": 34850
     },
     {
       "epoch": 3.7628032345013476,
-      "grad_norm": 0.5814857482910156,
-      "learning_rate": 0.0003747134376686454,
-      "loss": 3.4595,
+      "grad_norm": 0.6275509595870972,
+      "learning_rate": 0.0003747328656233135,
+      "loss": 3.4646,
       "step": 34900
     },
     {
       "epoch": 3.7681940700808623,
-      "grad_norm": 0.6383914351463318,
-      "learning_rate": 0.00037438963842417695,
-      "loss": 3.4509,
+      "grad_norm": 0.6648680567741394,
+      "learning_rate": 0.0003744090663788451,
+      "loss": 3.457,
       "step": 34950
     },
     {
       "epoch": 3.7735849056603774,
-      "grad_norm": 0.6446661353111267,
-      "learning_rate": 0.00037406583917970856,
-      "loss": 3.459,
+      "grad_norm": 0.7071113586425781,
+      "learning_rate": 0.00037408526713437665,
+      "loss": 3.4656,
       "step": 35000
     },
     {
       "epoch": 3.7735849056603774,
-      "eval_accuracy": 0.3753658478548823,
-      "eval_loss": 3.4548258781433105,
-      "eval_runtime": 183.6058,
-      "eval_samples_per_second": 98.096,
-      "eval_steps_per_second": 6.133,
+      "eval_accuracy": 0.37444523193643026,
+      "eval_loss": 3.4612653255462646,
+      "eval_runtime": 181.5912,
+      "eval_samples_per_second": 99.184,
+      "eval_steps_per_second": 6.201,
       "step": 35000
     },
     {
       "epoch": 3.778975741239892,
-      "grad_norm": 0.6061877012252808,
-      "learning_rate": 0.0003737420399352401,
-      "loss": 3.4709,
+      "grad_norm": 0.5935749411582947,
+      "learning_rate": 0.0003737614678899082,
+      "loss": 3.4748,
       "step": 35050
     },
     {
       "epoch": 3.784366576819407,
-      "grad_norm": 0.6421599984169006,
-      "learning_rate": 0.0003734182406907717,
-      "loss": 3.4359,
+      "grad_norm": 0.7232018113136292,
+      "learning_rate": 0.0003734376686454398,
+      "loss": 3.4418,
       "step": 35100
     },
     {
       "epoch": 3.789757412398922,
-      "grad_norm": 0.6605613231658936,
-      "learning_rate": 0.00037309444144630326,
-      "loss": 3.457,
+      "grad_norm": 0.6240691542625427,
+      "learning_rate": 0.00037311386940097136,
+      "loss": 3.4651,
       "step": 35150
     },
     {
       "epoch": 3.7951482479784366,
-      "grad_norm": 0.6076908111572266,
-      "learning_rate": 0.00037277064220183487,
-      "loss": 3.4568,
+      "grad_norm": 0.6064972281455994,
+      "learning_rate": 0.00037279007015650296,
+      "loss": 3.4613,
       "step": 35200
     },
     {
       "epoch": 3.8005390835579513,
-      "grad_norm": 0.611754298210144,
-      "learning_rate": 0.0003724468429573664,
-      "loss": 3.4381,
+      "grad_norm": 0.609962522983551,
+      "learning_rate": 0.0003724662709120345,
+      "loss": 3.4435,
       "step": 35250
     },
     {
       "epoch": 3.8059299191374665,
-      "grad_norm": 0.6356371641159058,
-      "learning_rate": 0.00037212951969778737,
-      "loss": 3.4514,
+      "grad_norm": 0.6387305855751038,
+      "learning_rate": 0.0003721424716675661,
+      "loss": 3.4561,
       "step": 35300
     },
     {
       "epoch": 3.811320754716981,
-      "grad_norm": 0.671958863735199,
-      "learning_rate": 0.00037180572045331887,
-      "loss": 3.4529,
+      "grad_norm": 0.6870223879814148,
+      "learning_rate": 0.0003718186724230976,
+      "loss": 3.459,
       "step": 35350
     },
     {
       "epoch": 3.816711590296496,
-      "grad_norm": 0.6135779023170471,
-      "learning_rate": 0.00037148192120885047,
-      "loss": 3.4536,
+      "grad_norm": 0.6189145445823669,
+      "learning_rate": 0.0003714948731786293,
+      "loss": 3.4581,
       "step": 35400
     },
     {
       "epoch": 3.822102425876011,
-      "grad_norm": 0.683573305606842,
-      "learning_rate": 0.000371158121964382,
-      "loss": 3.4466,
+      "grad_norm": 0.6900521516799927,
+      "learning_rate": 0.00037117107393416077,
+      "loss": 3.4532,
       "step": 35450
     },
     {
       "epoch": 3.8274932614555257,
-      "grad_norm": 0.5822563171386719,
-      "learning_rate": 0.0003708343227199136,
-      "loss": 3.4516,
+      "grad_norm": 0.6049875020980835,
+      "learning_rate": 0.0003708472746896923,
+      "loss": 3.4568,
       "step": 35500
     },
     {
       "epoch": 3.8328840970350404,
-      "grad_norm": 0.6443066596984863,
-      "learning_rate": 0.0003705105234754452,
-      "loss": 3.4611,
+      "grad_norm": 0.6687650680541992,
+      "learning_rate": 0.0003705234754452239,
+      "loss": 3.4687,
       "step": 35550
     },
     {
       "epoch": 3.838274932614555,
-      "grad_norm": 0.642926037311554,
-      "learning_rate": 0.0003701867242309768,
-      "loss": 3.4675,
+      "grad_norm": 0.6466330289840698,
+      "learning_rate": 0.0003701996762007555,
+      "loss": 3.4726,
       "step": 35600
     },
     {
       "epoch": 3.8436657681940702,
-      "grad_norm": 0.6716592311859131,
-      "learning_rate": 0.00036986292498650833,
-      "loss": 3.4654,
+      "grad_norm": 0.6749017834663391,
+      "learning_rate": 0.0003698758769562871,
+      "loss": 3.4732,
       "step": 35650
     },
     {
       "epoch": 3.849056603773585,
-      "grad_norm": 0.5893973112106323,
-      "learning_rate": 0.0003695391257420399,
-      "loss": 3.4669,
+      "grad_norm": 0.6227940917015076,
+      "learning_rate": 0.00036955207771181863,
+      "loss": 3.475,
       "step": 35700
     },
     {
       "epoch": 3.8544474393530996,
-      "grad_norm": 0.6722517013549805,
-      "learning_rate": 0.0003692153264975715,
-      "loss": 3.4484,
+      "grad_norm": 0.6425681114196777,
+      "learning_rate": 0.00036922827846735024,
+      "loss": 3.4517,
       "step": 35750
     },
     {
       "epoch": 3.8598382749326143,
-      "grad_norm": 0.6753908395767212,
-      "learning_rate": 0.00036889152725310304,
-      "loss": 3.4549,
+      "grad_norm": 0.6994612216949463,
+      "learning_rate": 0.0003689044792228818,
+      "loss": 3.4622,
       "step": 35800
     },
     {
       "epoch": 3.8652291105121295,
-      "grad_norm": 0.6143805980682373,
-      "learning_rate": 0.00036856772800863464,
-      "loss": 3.4301,
+      "grad_norm": 0.6147516369819641,
+      "learning_rate": 0.0003685806799784134,
+      "loss": 3.4358,
       "step": 35850
     },
     {
       "epoch": 3.870619946091644,
-      "grad_norm": 0.6016655564308167,
-      "learning_rate": 0.0003682439287641662,
-      "loss": 3.4456,
+      "grad_norm": 0.6202683448791504,
+      "learning_rate": 0.00036825688073394494,
+      "loss": 3.4512,
       "step": 35900
     },
     {
       "epoch": 3.876010781671159,
-      "grad_norm": 0.6141828298568726,
-      "learning_rate": 0.0003679201295196978,
-      "loss": 3.4712,
+      "grad_norm": 0.6709093451499939,
+      "learning_rate": 0.0003679330814894765,
+      "loss": 3.4741,
       "step": 35950
     },
     {
       "epoch": 3.881401617250674,
-      "grad_norm": 0.5878786444664001,
-      "learning_rate": 0.00036759633027522935,
-      "loss": 3.4485,
+      "grad_norm": 0.6027911305427551,
+      "learning_rate": 0.0003676092822450081,
+      "loss": 3.4539,
       "step": 36000
     },
     {
       "epoch": 3.881401617250674,
-      "eval_accuracy": 0.3757107121208055,
-      "eval_loss": 3.449120044708252,
-      "eval_runtime": 184.1528,
-      "eval_samples_per_second": 97.805,
-      "eval_steps_per_second": 6.114,
+      "eval_accuracy": 0.3747746674922837,
+      "eval_loss": 3.455878734588623,
+      "eval_runtime": 181.8268,
+      "eval_samples_per_second": 99.056,
+      "eval_steps_per_second": 6.193,
       "step": 36000
     },
     {
       "epoch": 3.8867924528301887,
-      "grad_norm": 0.6707215309143066,
-      "learning_rate": 0.00036727253103076084,
-      "loss": 3.4586,
+      "grad_norm": 0.6970086097717285,
+      "learning_rate": 0.0003672854830005396,
+      "loss": 3.4633,
       "step": 36050
     },
     {
       "epoch": 3.8921832884097034,
-      "grad_norm": 0.6566523909568787,
-      "learning_rate": 0.00036694873178629245,
-      "loss": 3.4523,
+      "grad_norm": 0.6359192132949829,
+      "learning_rate": 0.0003669616837560712,
+      "loss": 3.4563,
       "step": 36100
     },
     {
       "epoch": 3.8975741239892185,
-      "grad_norm": 0.5984740257263184,
-      "learning_rate": 0.000366624932541824,
-      "loss": 3.4565,
+      "grad_norm": 0.6560056209564209,
+      "learning_rate": 0.00036663788451160275,
+      "loss": 3.4634,
       "step": 36150
     },
     {
       "epoch": 3.9029649595687332,
-      "grad_norm": 0.5919893383979797,
-      "learning_rate": 0.0003663011332973556,
-      "loss": 3.4475,
+      "grad_norm": 0.6394860148429871,
+      "learning_rate": 0.00036631408526713435,
+      "loss": 3.4526,
       "step": 36200
     },
     {
       "epoch": 3.908355795148248,
-      "grad_norm": 0.5963663458824158,
-      "learning_rate": 0.00036597733405288715,
-      "loss": 3.4628,
+      "grad_norm": 0.6794313788414001,
+      "learning_rate": 0.0003659902860226659,
+      "loss": 3.4686,
       "step": 36250
     },
     {
       "epoch": 3.913746630727763,
-      "grad_norm": 0.6338704824447632,
-      "learning_rate": 0.00036565353480841876,
-      "loss": 3.441,
+      "grad_norm": 0.6359961032867432,
+      "learning_rate": 0.00036566648677819745,
+      "loss": 3.4465,
       "step": 36300
     },
     {
       "epoch": 3.9191374663072778,
-      "grad_norm": 0.6194099187850952,
-      "learning_rate": 0.0003653297355639503,
-      "loss": 3.4647,
+      "grad_norm": 0.6093354821205139,
+      "learning_rate": 0.00036534268753372906,
+      "loss": 3.4699,
       "step": 36350
     },
     {
       "epoch": 3.9245283018867925,
-      "grad_norm": 0.6529048681259155,
-      "learning_rate": 0.0003650059363194819,
-      "loss": 3.465,
+      "grad_norm": 0.6388203501701355,
+      "learning_rate": 0.0003650188882892606,
+      "loss": 3.4683,
       "step": 36400
     },
     {
       "epoch": 3.929919137466307,
-      "grad_norm": 0.632722020149231,
-      "learning_rate": 0.00036468213707501347,
-      "loss": 3.4408,
+      "grad_norm": 0.6400043964385986,
+      "learning_rate": 0.0003646950890447922,
+      "loss": 3.4468,
       "step": 36450
     },
     {
       "epoch": 3.935309973045822,
-      "grad_norm": 0.6227572560310364,
-      "learning_rate": 0.000364358337830545,
-      "loss": 3.4502,
+      "grad_norm": 0.6427869200706482,
+      "learning_rate": 0.00036437128980032376,
+      "loss": 3.4568,
       "step": 36500
     },
     {
       "epoch": 3.940700808625337,
-      "grad_norm": 0.6491719484329224,
-      "learning_rate": 0.0003640345385860766,
-      "loss": 3.4509,
+      "grad_norm": 0.6727488040924072,
+      "learning_rate": 0.00036404749055585537,
+      "loss": 3.456,
       "step": 36550
     },
     {
       "epoch": 3.9460916442048517,
-      "grad_norm": 0.5961635112762451,
-      "learning_rate": 0.00036371073934160817,
-      "loss": 3.4478,
+      "grad_norm": 0.6401848196983337,
+      "learning_rate": 0.0003637236913113869,
+      "loss": 3.4561,
       "step": 36600
     },
     {
       "epoch": 3.9514824797843664,
-      "grad_norm": 0.6824967265129089,
-      "learning_rate": 0.0003633869400971398,
-      "loss": 3.4401,
+      "grad_norm": 0.7175589203834534,
+      "learning_rate": 0.0003633998920669185,
+      "loss": 3.4454,
       "step": 36650
     },
     {
       "epoch": 3.9568733153638815,
-      "grad_norm": 0.6491476893424988,
-      "learning_rate": 0.00036306314085267127,
-      "loss": 3.4488,
+      "grad_norm": 0.7020281553268433,
+      "learning_rate": 0.0003630760928224501,
+      "loss": 3.4544,
       "step": 36700
     },
     {
       "epoch": 3.9622641509433962,
-      "grad_norm": 0.6334487199783325,
-      "learning_rate": 0.00036273934160820293,
-      "loss": 3.4413,
+      "grad_norm": 0.6395590901374817,
+      "learning_rate": 0.00036275229357798157,
+      "loss": 3.4473,
       "step": 36750
     },
     {
       "epoch": 3.967654986522911,
-      "grad_norm": 0.6231988668441772,
-      "learning_rate": 0.00036241554236373443,
-      "loss": 3.4347,
+      "grad_norm": 0.6293502449989319,
+      "learning_rate": 0.0003624284943335132,
+      "loss": 3.4387,
       "step": 36800
     },
     {
       "epoch": 3.973045822102426,
-      "grad_norm": 0.6323678493499756,
-      "learning_rate": 0.00036209174311926603,
-      "loss": 3.4487,
+      "grad_norm": 0.6193983554840088,
+      "learning_rate": 0.00036210469508904473,
+      "loss": 3.4548,
       "step": 36850
     },
     {
       "epoch": 3.9784366576819408,
-      "grad_norm": 0.6366806030273438,
-      "learning_rate": 0.0003617679438747976,
-      "loss": 3.4424,
+      "grad_norm": 0.6645393371582031,
+      "learning_rate": 0.00036178089584457633,
+      "loss": 3.4466,
       "step": 36900
     },
     {
       "epoch": 3.9838274932614555,
-      "grad_norm": 0.6505522131919861,
-      "learning_rate": 0.00036144414463032913,
-      "loss": 3.4522,
+      "grad_norm": 0.6711632013320923,
+      "learning_rate": 0.0003614570966001079,
+      "loss": 3.4583,
       "step": 36950
     },
     {
       "epoch": 3.9892183288409706,
-      "grad_norm": 0.6528891921043396,
-      "learning_rate": 0.00036112034538586074,
-      "loss": 3.4463,
+      "grad_norm": 0.6945592761039734,
+      "learning_rate": 0.0003611332973556395,
+      "loss": 3.4528,
       "step": 37000
     },
     {
       "epoch": 3.9892183288409706,
-      "eval_accuracy": 0.3762563669234098,
-      "eval_loss": 3.4432127475738525,
-      "eval_runtime": 183.5509,
-      "eval_samples_per_second": 98.125,
-      "eval_steps_per_second": 6.135,
+      "eval_accuracy": 0.375789485464471,
+      "eval_loss": 3.447413206100464,
+      "eval_runtime": 186.1609,
+      "eval_samples_per_second": 96.75,
+      "eval_steps_per_second": 6.049,
       "step": 37000
     },
     {
       "epoch": 3.9946091644204853,
-      "grad_norm": 0.6302518248558044,
-      "learning_rate": 0.0003607965461413923,
-      "loss": 3.4597,
+      "grad_norm": 0.6951505541801453,
+      "learning_rate": 0.00036080949811117104,
+      "loss": 3.4635,
       "step": 37050
     },
     {
       "epoch": 4.0,
-      "grad_norm": 1.2225829362869263,
-      "learning_rate": 0.0003604727468969239,
-      "loss": 3.4339,
+      "grad_norm": 1.242596983909607,
+      "learning_rate": 0.00036048569886670264,
+      "loss": 3.4372,
       "step": 37100
     },
     {
       "epoch": 4.005390835579515,
-      "grad_norm": 0.6290598511695862,
-      "learning_rate": 0.00036014894765245544,
-      "loss": 3.3275,
+      "grad_norm": 0.6364685297012329,
+      "learning_rate": 0.0003601618996222342,
+      "loss": 3.3321,
       "step": 37150
     },
     {
       "epoch": 4.010781671159029,
-      "grad_norm": 0.6681774258613586,
-      "learning_rate": 0.00035982514840798705,
-      "loss": 3.3563,
+      "grad_norm": 0.6206088066101074,
+      "learning_rate": 0.00035983810037776574,
+      "loss": 3.3611,
       "step": 37200
     },
     {
       "epoch": 4.0161725067385445,
-      "grad_norm": 0.6569148898124695,
-      "learning_rate": 0.0003595013491635186,
-      "loss": 3.3489,
+      "grad_norm": 0.6396298408508301,
+      "learning_rate": 0.00035951430113329735,
+      "loss": 3.3547,
       "step": 37250
     },
     {
       "epoch": 4.02156334231806,
-      "grad_norm": 0.5988460183143616,
-      "learning_rate": 0.0003591775499190502,
-      "loss": 3.3536,
+      "grad_norm": 0.6321409344673157,
+      "learning_rate": 0.0003591905018888289,
+      "loss": 3.3586,
       "step": 37300
     },
     {
       "epoch": 4.026954177897574,
-      "grad_norm": 0.6734127998352051,
-      "learning_rate": 0.00035886022665947105,
-      "loss": 3.3418,
+      "grad_norm": 0.6794791221618652,
+      "learning_rate": 0.0003588667026443605,
+      "loss": 3.3464,
       "step": 37350
     },
     {
       "epoch": 4.032345013477089,
-      "grad_norm": 0.6245633363723755,
-      "learning_rate": 0.0003585364274150027,
-      "loss": 3.361,
+      "grad_norm": 0.6208573579788208,
+      "learning_rate": 0.000358542903399892,
+      "loss": 3.3657,
       "step": 37400
     },
     {
       "epoch": 4.037735849056604,
-      "grad_norm": 0.6742115616798401,
-      "learning_rate": 0.0003582126281705342,
-      "loss": 3.3584,
+      "grad_norm": 0.6329734921455383,
+      "learning_rate": 0.0003582191041554236,
+      "loss": 3.3646,
       "step": 37450
     },
     {
       "epoch": 4.0431266846361185,
-      "grad_norm": 0.5911452770233154,
-      "learning_rate": 0.0003578888289260658,
-      "loss": 3.3485,
+      "grad_norm": 0.6082785129547119,
+      "learning_rate": 0.00035789530491095516,
+      "loss": 3.3515,
       "step": 37500
     },
     {
       "epoch": 4.048517520215634,
-      "grad_norm": 0.6306824088096619,
-      "learning_rate": 0.00035756502968159736,
-      "loss": 3.3633,
+      "grad_norm": 0.6603065729141235,
+      "learning_rate": 0.00035757150566648676,
+      "loss": 3.3703,
       "step": 37550
     },
     {
       "epoch": 4.053908355795148,
-      "grad_norm": 0.6810359358787537,
-      "learning_rate": 0.00035724123043712896,
-      "loss": 3.3582,
+      "grad_norm": 0.6484035849571228,
+      "learning_rate": 0.0003572477064220183,
+      "loss": 3.3607,
       "step": 37600
     },
     {
       "epoch": 4.059299191374663,
-      "grad_norm": 0.6800394654273987,
-      "learning_rate": 0.0003569174311926605,
-      "loss": 3.3723,
+      "grad_norm": 0.6546576619148254,
+      "learning_rate": 0.00035692390717754986,
+      "loss": 3.3775,
       "step": 37650
     },
     {
       "epoch": 4.064690026954178,
-      "grad_norm": 0.6404582262039185,
-      "learning_rate": 0.00035659363194819206,
-      "loss": 3.3558,
+      "grad_norm": 0.6480041742324829,
+      "learning_rate": 0.00035660010793308147,
+      "loss": 3.3632,
       "step": 37700
     },
     {
       "epoch": 4.070080862533692,
-      "grad_norm": 0.7083536386489868,
-      "learning_rate": 0.00035626983270372367,
-      "loss": 3.3775,
+      "grad_norm": 0.6931122541427612,
+      "learning_rate": 0.000356276308688613,
+      "loss": 3.3812,
       "step": 37750
     },
     {
       "epoch": 4.0754716981132075,
-      "grad_norm": 0.6967944502830505,
-      "learning_rate": 0.0003559460334592552,
-      "loss": 3.3606,
+      "grad_norm": 0.695582389831543,
+      "learning_rate": 0.0003559525094441446,
+      "loss": 3.3666,
       "step": 37800
     },
     {
       "epoch": 4.080862533692723,
-      "grad_norm": 0.6590530872344971,
-      "learning_rate": 0.0003556222342147868,
-      "loss": 3.3595,
+      "grad_norm": 0.6651517152786255,
+      "learning_rate": 0.00035562871019967617,
+      "loss": 3.3643,
       "step": 37850
     },
     {
       "epoch": 4.086253369272237,
-      "grad_norm": 0.6163628101348877,
-      "learning_rate": 0.0003552984349703184,
-      "loss": 3.3587,
+      "grad_norm": 0.6387616991996765,
+      "learning_rate": 0.0003553049109552078,
+      "loss": 3.3655,
       "step": 37900
     },
     {
       "epoch": 4.091644204851752,
-      "grad_norm": 0.6308746337890625,
-      "learning_rate": 0.00035497463572585,
-      "loss": 3.3714,
+      "grad_norm": 0.6401187181472778,
+      "learning_rate": 0.00035498111171073933,
+      "loss": 3.3775,
       "step": 37950
     },
     {
       "epoch": 4.097035040431267,
-      "grad_norm": 0.6695423722267151,
-      "learning_rate": 0.00035465083648138153,
-      "loss": 3.3631,
+      "grad_norm": 0.7137041687965393,
+      "learning_rate": 0.0003546573124662708,
+      "loss": 3.3674,
       "step": 38000
     },
     {
       "epoch": 4.097035040431267,
-      "eval_accuracy": 0.37669934474702943,
-      "eval_loss": 3.446185827255249,
-      "eval_runtime": 183.7978,
-      "eval_samples_per_second": 97.994,
-      "eval_steps_per_second": 6.126,
+      "eval_accuracy": 0.37596365604363763,
+      "eval_loss": 3.452828884124756,
+      "eval_runtime": 186.0536,
+      "eval_samples_per_second": 96.805,
+      "eval_steps_per_second": 6.052,
       "step": 38000
     },
     {
       "epoch": 4.1024258760107815,
-      "grad_norm": 0.6689504384994507,
-      "learning_rate": 0.00035432703723691314,
-      "loss": 3.3665,
+      "grad_norm": 0.6992183923721313,
+      "learning_rate": 0.0003543335132218025,
+      "loss": 3.3712,
       "step": 38050
     },
     {
       "epoch": 4.107816711590297,
-      "grad_norm": 0.6203464865684509,
-      "learning_rate": 0.00035400323799244463,
-      "loss": 3.3684,
+      "grad_norm": 0.647977888584137,
+      "learning_rate": 0.000354009713977334,
+      "loss": 3.3769,
       "step": 38100
     },
     {
       "epoch": 4.113207547169812,
-      "grad_norm": 0.6366583704948425,
-      "learning_rate": 0.0003536794387479762,
-      "loss": 3.3796,
+      "grad_norm": 0.6405393481254578,
+      "learning_rate": 0.0003536859147328656,
+      "loss": 3.3852,
       "step": 38150
     },
     {
       "epoch": 4.118598382749326,
-      "grad_norm": 0.6430602669715881,
-      "learning_rate": 0.0003533556395035078,
-      "loss": 3.3595,
+      "grad_norm": 0.6509863138198853,
+      "learning_rate": 0.00035336211548839713,
+      "loss": 3.3633,
       "step": 38200
     },
     {
       "epoch": 4.123989218328841,
-      "grad_norm": 0.6412113904953003,
-      "learning_rate": 0.00035303184025903934,
-      "loss": 3.3632,
+      "grad_norm": 0.6734946966171265,
+      "learning_rate": 0.00035303831624392874,
+      "loss": 3.3685,
       "step": 38250
     },
     {
       "epoch": 4.129380053908355,
-      "grad_norm": 0.654319167137146,
-      "learning_rate": 0.00035270804101457094,
-      "loss": 3.3759,
+      "grad_norm": 0.678573489189148,
+      "learning_rate": 0.0003527145169994603,
+      "loss": 3.3807,
       "step": 38300
     },
     {
       "epoch": 4.1347708894878705,
-      "grad_norm": 0.6605011224746704,
-      "learning_rate": 0.0003523842417701025,
-      "loss": 3.3537,
+      "grad_norm": 0.6822803616523743,
+      "learning_rate": 0.0003523907177549919,
+      "loss": 3.359,
       "step": 38350
     },
     {
       "epoch": 4.140161725067386,
-      "grad_norm": 0.601024866104126,
-      "learning_rate": 0.0003520604425256341,
-      "loss": 3.3774,
+      "grad_norm": 0.6508968472480774,
+      "learning_rate": 0.00035206691851052345,
+      "loss": 3.3833,
       "step": 38400
     },
     {
       "epoch": 4.1455525606469,
-      "grad_norm": 0.6261166930198669,
-      "learning_rate": 0.00035173664328116565,
-      "loss": 3.3613,
+      "grad_norm": 0.6106385588645935,
+      "learning_rate": 0.000351743119266055,
+      "loss": 3.3665,
       "step": 38450
     },
     {
       "epoch": 4.150943396226415,
-      "grad_norm": 0.6527787446975708,
-      "learning_rate": 0.00035141284403669725,
-      "loss": 3.3736,
+      "grad_norm": 0.6760352253913879,
+      "learning_rate": 0.0003514193200215866,
+      "loss": 3.3777,
       "step": 38500
     },
     {
       "epoch": 4.15633423180593,
-      "grad_norm": 0.6282969117164612,
-      "learning_rate": 0.0003510890447922288,
-      "loss": 3.3998,
+      "grad_norm": 0.6681267023086548,
+      "learning_rate": 0.00035109552077711815,
+      "loss": 3.4062,
       "step": 38550
     },
     {
       "epoch": 4.1617250673854445,
-      "grad_norm": 0.6190927028656006,
+      "grad_norm": 0.6289014220237732,
       "learning_rate": 0.00035077172153264976,
-      "loss": 3.3699,
+      "loss": 3.3757,
       "step": 38600
     },
     {
       "epoch": 4.16711590296496,
-      "grad_norm": 0.6161026358604431,
+      "grad_norm": 0.6665849685668945,
       "learning_rate": 0.0003504479222881813,
-      "loss": 3.3852,
+      "loss": 3.3885,
       "step": 38650
     },
     {
       "epoch": 4.172506738544475,
-      "grad_norm": 0.7209351658821106,
+      "grad_norm": 0.7066333293914795,
       "learning_rate": 0.0003501241230437129,
-      "loss": 3.3512,
+      "loss": 3.3565,
       "step": 38700
     },
     {
       "epoch": 4.177897574123989,
-      "grad_norm": 0.625611424446106,
+      "grad_norm": 0.6751580238342285,
       "learning_rate": 0.0003498003237992444,
-      "loss": 3.3763,
+      "loss": 3.3814,
       "step": 38750
     },
     {
       "epoch": 4.183288409703504,
-      "grad_norm": 0.6233698129653931,
-      "learning_rate": 0.000349476524554776,
-      "loss": 3.3798,
+      "grad_norm": 0.6326186656951904,
+      "learning_rate": 0.00034948300053966536,
+      "loss": 3.3861,
       "step": 38800
     },
     {
       "epoch": 4.188679245283019,
-      "grad_norm": 0.6900334358215332,
-      "learning_rate": 0.00034915272531030756,
-      "loss": 3.3821,
+      "grad_norm": 0.6602861881256104,
+      "learning_rate": 0.0003491592012951969,
+      "loss": 3.3892,
       "step": 38850
     },
     {
       "epoch": 4.1940700808625335,
-      "grad_norm": 0.6348626613616943,
-      "learning_rate": 0.0003488289260658391,
-      "loss": 3.3889,
+      "grad_norm": 0.6448402404785156,
+      "learning_rate": 0.0003488354020507285,
+      "loss": 3.3944,
       "step": 38900
     },
     {
       "epoch": 4.199460916442049,
-      "grad_norm": 0.6387774348258972,
-      "learning_rate": 0.0003485051268213707,
-      "loss": 3.3821,
+      "grad_norm": 0.6615744829177856,
+      "learning_rate": 0.00034851160280626007,
+      "loss": 3.3882,
       "step": 38950
     },
     {
       "epoch": 4.204851752021563,
-      "grad_norm": 0.6410725712776184,
-      "learning_rate": 0.00034818132757690227,
-      "loss": 3.3635,
+      "grad_norm": 0.6682710647583008,
+      "learning_rate": 0.00034818780356179167,
+      "loss": 3.369,
       "step": 39000
     },
     {
       "epoch": 4.204851752021563,
-      "eval_accuracy": 0.3774232989385373,
-      "eval_loss": 3.443272113800049,
-      "eval_runtime": 183.5709,
-      "eval_samples_per_second": 98.115,
-      "eval_steps_per_second": 6.134,
+      "eval_accuracy": 0.3768517847486332,
+      "eval_loss": 3.448885440826416,
+      "eval_runtime": 185.6829,
+      "eval_samples_per_second": 96.999,
+      "eval_steps_per_second": 6.064,
       "step": 39000
     },
     {
       "epoch": 4.210242587601078,
-      "grad_norm": 0.6588302850723267,
-      "learning_rate": 0.0003478575283324339,
-      "loss": 3.3933,
+      "grad_norm": 0.6382125616073608,
+      "learning_rate": 0.0003478640043173232,
+      "loss": 3.398,
       "step": 39050
     },
     {
       "epoch": 4.215633423180593,
-      "grad_norm": 0.6347865462303162,
-      "learning_rate": 0.0003475337290879654,
-      "loss": 3.3851,
+      "grad_norm": 0.6248801946640015,
+      "learning_rate": 0.0003475402050728548,
+      "loss": 3.3883,
       "step": 39100
     },
     {
       "epoch": 4.2210242587601075,
-      "grad_norm": 0.6326488852500916,
-      "learning_rate": 0.00034720992984349703,
-      "loss": 3.3934,
+      "grad_norm": 0.635450541973114,
+      "learning_rate": 0.0003472164058283864,
+      "loss": 3.3998,
       "step": 39150
     },
     {
       "epoch": 4.226415094339623,
-      "grad_norm": 0.6076704263687134,
-      "learning_rate": 0.0003468861305990286,
-      "loss": 3.3747,
+      "grad_norm": 0.6210282444953918,
+      "learning_rate": 0.0003468926065839179,
+      "loss": 3.3785,
       "step": 39200
     },
     {
       "epoch": 4.231805929919138,
-      "grad_norm": 0.6273050308227539,
-      "learning_rate": 0.0003465623313545602,
-      "loss": 3.3727,
+      "grad_norm": 0.637699544429779,
+      "learning_rate": 0.00034656880733944953,
+      "loss": 3.3761,
       "step": 39250
     },
     {
       "epoch": 4.237196765498652,
-      "grad_norm": 0.6443808674812317,
-      "learning_rate": 0.00034623853211009173,
-      "loss": 3.384,
+      "grad_norm": 0.6527442336082458,
+      "learning_rate": 0.0003462450080949811,
+      "loss": 3.3886,
       "step": 39300
     },
     {
       "epoch": 4.242587601078167,
-      "grad_norm": 0.6146422624588013,
-      "learning_rate": 0.00034591473286562323,
-      "loss": 3.3702,
+      "grad_norm": 0.6105765700340271,
+      "learning_rate": 0.0003459212088505127,
+      "loss": 3.3776,
       "step": 39350
     },
     {
       "epoch": 4.247978436657682,
-      "grad_norm": 0.6165491938591003,
-      "learning_rate": 0.0003455909336211549,
-      "loss": 3.3923,
+      "grad_norm": 0.6594821810722351,
+      "learning_rate": 0.0003455974096060442,
+      "loss": 3.3958,
       "step": 39400
     },
     {
       "epoch": 4.2533692722371965,
-      "grad_norm": 0.6656982898712158,
-      "learning_rate": 0.0003452671343766864,
-      "loss": 3.3772,
+      "grad_norm": 0.6426079273223877,
+      "learning_rate": 0.0003452736103615758,
+      "loss": 3.3842,
       "step": 39450
     },
     {
       "epoch": 4.258760107816712,
-      "grad_norm": 0.6853116750717163,
-      "learning_rate": 0.000344943335132218,
-      "loss": 3.3865,
+      "grad_norm": 0.6598894596099854,
+      "learning_rate": 0.00034494981111710734,
+      "loss": 3.3918,
       "step": 39500
     },
     {
       "epoch": 4.264150943396227,
-      "grad_norm": 0.6323865056037903,
-      "learning_rate": 0.00034461953588774954,
-      "loss": 3.3894,
+      "grad_norm": 0.6500307321548462,
+      "learning_rate": 0.00034462601187263894,
+      "loss": 3.3938,
       "step": 39550
     },
     {
       "epoch": 4.269541778975741,
-      "grad_norm": 0.6815229654312134,
-      "learning_rate": 0.00034429573664328115,
-      "loss": 3.3976,
+      "grad_norm": 0.6460248827934265,
+      "learning_rate": 0.0003443022126281705,
+      "loss": 3.4021,
       "step": 39600
     },
     {
       "epoch": 4.274932614555256,
-      "grad_norm": 0.6554942727088928,
-      "learning_rate": 0.0003439719373988127,
-      "loss": 3.3929,
+      "grad_norm": 0.6311081647872925,
+      "learning_rate": 0.00034397841338370204,
+      "loss": 3.3994,
       "step": 39650
     },
     {
       "epoch": 4.280323450134771,
-      "grad_norm": 0.7081801891326904,
-      "learning_rate": 0.00034364813815434425,
-      "loss": 3.3724,
+      "grad_norm": 0.6745235919952393,
+      "learning_rate": 0.00034365461413923365,
+      "loss": 3.3775,
       "step": 39700
     },
     {
       "epoch": 4.285714285714286,
-      "grad_norm": 0.6656792163848877,
-      "learning_rate": 0.00034332433890987585,
-      "loss": 3.383,
+      "grad_norm": 0.6637439131736755,
+      "learning_rate": 0.0003433308148947652,
+      "loss": 3.3866,
       "step": 39750
     },
     {
       "epoch": 4.291105121293801,
-      "grad_norm": 0.6686912775039673,
-      "learning_rate": 0.0003430005396654074,
-      "loss": 3.3819,
+      "grad_norm": 0.6909286379814148,
+      "learning_rate": 0.0003430070156502968,
+      "loss": 3.3874,
       "step": 39800
     },
     {
       "epoch": 4.296495956873315,
-      "grad_norm": 0.6758848428726196,
-      "learning_rate": 0.000342676740420939,
-      "loss": 3.3823,
+      "grad_norm": 0.6249815225601196,
+      "learning_rate": 0.00034268321640582835,
+      "loss": 3.3868,
       "step": 39850
     },
     {
       "epoch": 4.30188679245283,
-      "grad_norm": 0.691242516040802,
-      "learning_rate": 0.00034235294117647056,
-      "loss": 3.3879,
+      "grad_norm": 0.7167820334434509,
+      "learning_rate": 0.00034235941716135996,
+      "loss": 3.3938,
       "step": 39900
     },
     {
       "epoch": 4.307277628032345,
-      "grad_norm": 0.6893483400344849,
-      "learning_rate": 0.00034202914193200216,
-      "loss": 3.3772,
+      "grad_norm": 0.7112882137298584,
+      "learning_rate": 0.0003420356179168915,
+      "loss": 3.383,
       "step": 39950
     },
     {
       "epoch": 4.3126684636118595,
-      "grad_norm": 0.5819874405860901,
-      "learning_rate": 0.0003417053426875337,
-      "loss": 3.3887,
+      "grad_norm": 0.617048978805542,
+      "learning_rate": 0.0003417118186724231,
+      "loss": 3.392,
       "step": 40000
     },
     {
       "epoch": 4.3126684636118595,
-      "eval_accuracy": 0.3775175009922725,
-      "eval_loss": 3.4382338523864746,
-      "eval_runtime": 184.2231,
-      "eval_samples_per_second": 97.767,
-      "eval_steps_per_second": 6.112,
+      "eval_accuracy": 0.3771873048662043,
+      "eval_loss": 3.441244125366211,
+      "eval_runtime": 186.4644,
+      "eval_samples_per_second": 96.592,
+      "eval_steps_per_second": 6.039,
       "step": 40000
     },
     {
       "epoch": 4.318059299191375,
-      "grad_norm": 0.6500782370567322,
-      "learning_rate": 0.0003413815434430653,
-      "loss": 3.3864,
+      "grad_norm": 0.66306471824646,
+      "learning_rate": 0.00034138801942795467,
+      "loss": 3.3909,
       "step": 40050
     },
     {
       "epoch": 4.32345013477089,
-      "grad_norm": 0.6664853692054749,
-      "learning_rate": 0.0003410577441985968,
-      "loss": 3.3832,
+      "grad_norm": 0.7126609683036804,
+      "learning_rate": 0.00034106422018348616,
+      "loss": 3.3868,
       "step": 40100
     },
     {
       "epoch": 4.328840970350404,
-      "grad_norm": 0.6425719857215881,
-      "learning_rate": 0.00034073394495412837,
-      "loss": 3.3954,
+      "grad_norm": 0.6458390951156616,
+      "learning_rate": 0.00034074042093901777,
+      "loss": 3.4017,
       "step": 40150
     },
     {
       "epoch": 4.334231805929919,
-      "grad_norm": 0.6592696905136108,
-      "learning_rate": 0.00034041014570965997,
-      "loss": 3.3912,
+      "grad_norm": 0.6810258030891418,
+      "learning_rate": 0.0003404166216945493,
+      "loss": 3.3971,
       "step": 40200
     },
     {
       "epoch": 4.339622641509434,
-      "grad_norm": 0.6312816739082336,
-      "learning_rate": 0.0003400863464651915,
-      "loss": 3.3784,
+      "grad_norm": 0.6616154909133911,
+      "learning_rate": 0.0003400928224500809,
+      "loss": 3.3827,
       "step": 40250
     },
     {
       "epoch": 4.345013477088949,
-      "grad_norm": 0.637238085269928,
-      "learning_rate": 0.0003397625472207231,
-      "loss": 3.3924,
+      "grad_norm": 0.6602355241775513,
+      "learning_rate": 0.00033976902320561247,
+      "loss": 3.3943,
       "step": 40300
     },
     {
       "epoch": 4.350404312668464,
-      "grad_norm": 0.5931002497673035,
-      "learning_rate": 0.0003394387479762547,
-      "loss": 3.3745,
+      "grad_norm": 0.6159692406654358,
+      "learning_rate": 0.0003394452239611441,
+      "loss": 3.3778,
       "step": 40350
     },
     {
       "epoch": 4.355795148247978,
-      "grad_norm": 0.6598531603813171,
-      "learning_rate": 0.0003391149487317863,
-      "loss": 3.3874,
+      "grad_norm": 0.6545746922492981,
+      "learning_rate": 0.00033912142471667563,
+      "loss": 3.3906,
       "step": 40400
     },
     {
       "epoch": 4.361185983827493,
-      "grad_norm": 0.6362660527229309,
-      "learning_rate": 0.00033879114948731783,
-      "loss": 3.3833,
+      "grad_norm": 0.6649766564369202,
+      "learning_rate": 0.00033879762547220723,
+      "loss": 3.3877,
       "step": 40450
     },
     {
       "epoch": 4.366576819407008,
-      "grad_norm": 0.623309850692749,
-      "learning_rate": 0.00033846735024284944,
-      "loss": 3.3982,
+      "grad_norm": 0.6569035053253174,
+      "learning_rate": 0.0003384738262277388,
+      "loss": 3.4027,
       "step": 40500
     },
     {
       "epoch": 4.3719676549865225,
-      "grad_norm": 0.6290128231048584,
-      "learning_rate": 0.000338143550998381,
-      "loss": 3.3897,
+      "grad_norm": 0.6234119534492493,
+      "learning_rate": 0.00033815002698327033,
+      "loss": 3.3961,
       "step": 40550
     },
     {
       "epoch": 4.377358490566038,
-      "grad_norm": 0.6916012167930603,
-      "learning_rate": 0.00033781975175391254,
-      "loss": 3.3787,
+      "grad_norm": 0.6875752806663513,
+      "learning_rate": 0.00033782622773880194,
+      "loss": 3.3831,
       "step": 40600
     },
     {
       "epoch": 4.382749326145553,
-      "grad_norm": 0.6463659405708313,
-      "learning_rate": 0.00033749595250944414,
-      "loss": 3.3871,
+      "grad_norm": 0.6854240298271179,
+      "learning_rate": 0.0003375024284943335,
+      "loss": 3.3916,
       "step": 40650
     },
     {
       "epoch": 4.388140161725067,
-      "grad_norm": 0.675667941570282,
-      "learning_rate": 0.00033717215326497564,
-      "loss": 3.3882,
+      "grad_norm": 0.6738277673721313,
+      "learning_rate": 0.0003371786292498651,
+      "loss": 3.3928,
       "step": 40700
     },
     {
       "epoch": 4.393530997304582,
-      "grad_norm": 0.6485020518302917,
-      "learning_rate": 0.0003368483540205073,
-      "loss": 3.3871,
+      "grad_norm": 0.6735571026802063,
+      "learning_rate": 0.0003368548300053966,
+      "loss": 3.39,
       "step": 40750
     },
     {
       "epoch": 4.398921832884097,
-      "grad_norm": 0.7107369303703308,
-      "learning_rate": 0.0003365245547760388,
-      "loss": 3.3755,
+      "grad_norm": 0.665442168712616,
+      "learning_rate": 0.0003365310307609282,
+      "loss": 3.3788,
       "step": 40800
     },
     {
       "epoch": 4.404312668463612,
-      "grad_norm": 0.6821770668029785,
-      "learning_rate": 0.0003362007555315704,
-      "loss": 3.3906,
+      "grad_norm": 0.6647260189056396,
+      "learning_rate": 0.0003362137075013491,
+      "loss": 3.3952,
       "step": 40850
     },
     {
       "epoch": 4.409703504043127,
-      "grad_norm": 0.6535558104515076,
-      "learning_rate": 0.00033587695628710195,
-      "loss": 3.3996,
+      "grad_norm": 0.6577963829040527,
+      "learning_rate": 0.0003358899082568807,
+      "loss": 3.4052,
       "step": 40900
     },
     {
       "epoch": 4.415094339622642,
-      "grad_norm": 0.6308371424674988,
-      "learning_rate": 0.00033555315704263355,
-      "loss": 3.3959,
+      "grad_norm": 0.645415186882019,
+      "learning_rate": 0.00033556610901241225,
+      "loss": 3.3999,
       "step": 40950
     },
     {
       "epoch": 4.420485175202156,
-      "grad_norm": 0.5909251570701599,
-      "learning_rate": 0.0003352293577981651,
-      "loss": 3.392,
+      "grad_norm": 0.6713573336601257,
+      "learning_rate": 0.00033524230976794385,
+      "loss": 3.3994,
       "step": 41000
     },
     {
       "epoch": 4.420485175202156,
-      "eval_accuracy": 0.37843866017516364,
-      "eval_loss": 3.4319839477539062,
-      "eval_runtime": 183.6983,
-      "eval_samples_per_second": 98.047,
-      "eval_steps_per_second": 6.13,
+      "eval_accuracy": 0.3774418785823536,
+      "eval_loss": 3.437533378601074,
+      "eval_runtime": 186.2561,
+      "eval_samples_per_second": 96.7,
+      "eval_steps_per_second": 6.045,
       "step": 41000
     },
     {
       "epoch": 4.425876010781671,
-      "grad_norm": 0.6647576689720154,
-      "learning_rate": 0.00033490555855369665,
-      "loss": 3.4006,
+      "grad_norm": 0.6576057076454163,
+      "learning_rate": 0.0003349185105234754,
+      "loss": 3.4064,
       "step": 41050
     },
     {
       "epoch": 4.431266846361186,
-      "grad_norm": 0.6007199883460999,
-      "learning_rate": 0.00033458175930922826,
-      "loss": 3.3697,
+      "grad_norm": 0.6560052633285522,
+      "learning_rate": 0.000334594711279007,
+      "loss": 3.3761,
       "step": 41100
     },
     {
       "epoch": 4.436657681940701,
-      "grad_norm": 0.6572251319885254,
-      "learning_rate": 0.0003342579600647598,
-      "loss": 3.3942,
+      "grad_norm": 0.6211264729499817,
+      "learning_rate": 0.00033427091203453856,
+      "loss": 3.3988,
       "step": 41150
     },
     {
       "epoch": 4.442048517520216,
-      "grad_norm": 0.6462733745574951,
-      "learning_rate": 0.0003339341608202914,
-      "loss": 3.406,
+      "grad_norm": 0.6420550346374512,
+      "learning_rate": 0.00033394711279007016,
+      "loss": 3.4094,
       "step": 41200
     },
     {
       "epoch": 4.44743935309973,
-      "grad_norm": 0.7077494859695435,
-      "learning_rate": 0.00033361036157582297,
-      "loss": 3.4034,
+      "grad_norm": 0.7038576006889343,
+      "learning_rate": 0.0003336233135456017,
+      "loss": 3.4068,
       "step": 41250
     },
     {
       "epoch": 4.452830188679245,
-      "grad_norm": 0.6289212703704834,
-      "learning_rate": 0.00033328656233135457,
-      "loss": 3.3715,
+      "grad_norm": 0.6897509098052979,
+      "learning_rate": 0.00033329951430113326,
+      "loss": 3.3771,
       "step": 41300
     },
     {
       "epoch": 4.45822102425876,
-      "grad_norm": 0.6613530516624451,
-      "learning_rate": 0.0003329627630868861,
-      "loss": 3.3796,
+      "grad_norm": 0.688103973865509,
+      "learning_rate": 0.00033297571505666487,
+      "loss": 3.3845,
       "step": 41350
     },
     {
       "epoch": 4.463611859838275,
-      "grad_norm": 0.653528094291687,
-      "learning_rate": 0.0003326389638424177,
-      "loss": 3.397,
+      "grad_norm": 0.6834924221038818,
+      "learning_rate": 0.00033265191581219637,
+      "loss": 3.4024,
       "step": 41400
     },
     {
       "epoch": 4.46900269541779,
-      "grad_norm": 0.6634072065353394,
-      "learning_rate": 0.0003323151645979492,
-      "loss": 3.4074,
+      "grad_norm": 0.6503341794013977,
+      "learning_rate": 0.00033232811656772797,
+      "loss": 3.4117,
       "step": 41450
     },
     {
       "epoch": 4.474393530997305,
-      "grad_norm": 0.643779456615448,
-      "learning_rate": 0.00033199136535348077,
-      "loss": 3.4053,
+      "grad_norm": 0.664341151714325,
+      "learning_rate": 0.0003320043173232595,
+      "loss": 3.4099,
       "step": 41500
     },
     {
       "epoch": 4.479784366576819,
-      "grad_norm": 0.6200483441352844,
-      "learning_rate": 0.0003316675661090124,
-      "loss": 3.3927,
+      "grad_norm": 0.6196551322937012,
+      "learning_rate": 0.0003316805180787911,
+      "loss": 3.3974,
       "step": 41550
     },
     {
       "epoch": 4.485175202156334,
-      "grad_norm": 0.7240083813667297,
-      "learning_rate": 0.00033134376686454393,
-      "loss": 3.3854,
+      "grad_norm": 0.6742070317268372,
+      "learning_rate": 0.0003313567188343227,
+      "loss": 3.3909,
       "step": 41600
     },
     {
       "epoch": 4.490566037735849,
-      "grad_norm": 0.6716325283050537,
-      "learning_rate": 0.00033101996762007553,
-      "loss": 3.398,
+      "grad_norm": 0.6695722341537476,
+      "learning_rate": 0.00033103291958985423,
+      "loss": 3.4033,
       "step": 41650
     },
     {
       "epoch": 4.495956873315364,
-      "grad_norm": 0.676361083984375,
-      "learning_rate": 0.0003306961683756071,
-      "loss": 3.3881,
+      "grad_norm": 0.6776498556137085,
+      "learning_rate": 0.00033070912034538583,
+      "loss": 3.3914,
       "step": 41700
     },
     {
       "epoch": 4.501347708894879,
-      "grad_norm": 0.6593371629714966,
-      "learning_rate": 0.0003303723691311387,
-      "loss": 3.3887,
+      "grad_norm": 0.6503829956054688,
+      "learning_rate": 0.0003303853211009174,
+      "loss": 3.3912,
       "step": 41750
     },
     {
       "epoch": 4.506738544474393,
-      "grad_norm": 0.7852563261985779,
-      "learning_rate": 0.00033004856988667024,
-      "loss": 3.4041,
+      "grad_norm": 0.7166768908500671,
+      "learning_rate": 0.000330061521856449,
+      "loss": 3.409,
       "step": 41800
     },
     {
       "epoch": 4.512129380053908,
-      "grad_norm": 0.651554524898529,
-      "learning_rate": 0.0003297247706422018,
-      "loss": 3.3813,
+      "grad_norm": 0.6912717223167419,
+      "learning_rate": 0.00032973772261198054,
+      "loss": 3.3862,
       "step": 41850
     },
     {
       "epoch": 4.517520215633423,
-      "grad_norm": 0.61871337890625,
-      "learning_rate": 0.0003294009713977334,
-      "loss": 3.3981,
+      "grad_norm": 0.6197432279586792,
+      "learning_rate": 0.00032941392336751214,
+      "loss": 3.4011,
       "step": 41900
     },
     {
       "epoch": 4.5229110512129385,
-      "grad_norm": 0.7037910223007202,
-      "learning_rate": 0.00032907717215326494,
-      "loss": 3.3993,
+      "grad_norm": 0.686069130897522,
+      "learning_rate": 0.0003290901241230437,
+      "loss": 3.4026,
       "step": 41950
     },
     {
       "epoch": 4.528301886792453,
-      "grad_norm": 0.6677039265632629,
-      "learning_rate": 0.00032875337290879655,
-      "loss": 3.3866,
+      "grad_norm": 0.6873455047607422,
+      "learning_rate": 0.0003287663248785753,
+      "loss": 3.3909,
       "step": 42000
     },
     {
       "epoch": 4.528301886792453,
-      "eval_accuracy": 0.37861880666316006,
-      "eval_loss": 3.4264838695526123,
-      "eval_runtime": 183.841,
-      "eval_samples_per_second": 97.971,
-      "eval_steps_per_second": 6.125,
+      "eval_accuracy": 0.37794363761828087,
+      "eval_loss": 3.4332618713378906,
+      "eval_runtime": 186.1089,
+      "eval_samples_per_second": 96.777,
+      "eval_steps_per_second": 6.05,
       "step": 42000
     },
     {
       "epoch": 4.533692722371968,
-      "grad_norm": 0.6433460116386414,
-      "learning_rate": 0.00032842957366432805,
-      "loss": 3.385,
+      "grad_norm": 0.680125892162323,
+      "learning_rate": 0.00032844252563410685,
+      "loss": 3.3878,
       "step": 42050
     },
     {
       "epoch": 4.539083557951482,
-      "grad_norm": 0.6643568277359009,
-      "learning_rate": 0.0003281057744198597,
-      "loss": 3.3937,
+      "grad_norm": 0.649057149887085,
+      "learning_rate": 0.00032811872638963834,
+      "loss": 3.3985,
       "step": 42100
     },
     {
       "epoch": 4.544474393530997,
-      "grad_norm": 0.6829423308372498,
-      "learning_rate": 0.0003277819751753912,
-      "loss": 3.393,
+      "grad_norm": 0.6688793897628784,
+      "learning_rate": 0.00032779492714516995,
+      "loss": 3.397,
       "step": 42150
     },
     {
       "epoch": 4.549865229110512,
-      "grad_norm": 0.6842221021652222,
-      "learning_rate": 0.0003274581759309228,
-      "loss": 3.3902,
+      "grad_norm": 0.6881921887397766,
+      "learning_rate": 0.0003274711279007015,
+      "loss": 3.3958,
       "step": 42200
     },
     {
       "epoch": 4.555256064690027,
-      "grad_norm": 0.6358245015144348,
-      "learning_rate": 0.00032713437668645436,
-      "loss": 3.4025,
+      "grad_norm": 0.6039194464683533,
+      "learning_rate": 0.0003271473286562331,
+      "loss": 3.4055,
       "step": 42250
     },
     {
       "epoch": 4.560646900269542,
-      "grad_norm": 0.7178597450256348,
-      "learning_rate": 0.0003268105774419859,
-      "loss": 3.38,
+      "grad_norm": 0.6900604367256165,
+      "learning_rate": 0.00032682352941176466,
+      "loss": 3.3825,
       "step": 42300
     },
     {
       "epoch": 4.566037735849057,
-      "grad_norm": 0.6716541051864624,
-      "learning_rate": 0.0003264867781975175,
-      "loss": 3.4133,
+      "grad_norm": 0.6447793245315552,
+      "learning_rate": 0.00032649973016729626,
+      "loss": 3.4166,
       "step": 42350
     },
     {
       "epoch": 4.571428571428571,
-      "grad_norm": 0.6570898294448853,
-      "learning_rate": 0.00032616297895304906,
-      "loss": 3.3732,
+      "grad_norm": 0.7399511933326721,
+      "learning_rate": 0.0003261759309228278,
+      "loss": 3.3784,
       "step": 42400
     },
     {
       "epoch": 4.576819407008086,
-      "grad_norm": 0.6482057571411133,
-      "learning_rate": 0.00032583917970858067,
-      "loss": 3.3954,
+      "grad_norm": 0.6312559843063354,
+      "learning_rate": 0.0003258521316783594,
+      "loss": 3.3998,
       "step": 42450
     },
     {
       "epoch": 4.5822102425876015,
-      "grad_norm": 0.6238645315170288,
-      "learning_rate": 0.0003255153804641122,
-      "loss": 3.4011,
+      "grad_norm": 0.6471340656280518,
+      "learning_rate": 0.00032552833243389097,
+      "loss": 3.4056,
       "step": 42500
     },
     {
       "epoch": 4.587601078167116,
-      "grad_norm": 0.6898269653320312,
-      "learning_rate": 0.0003251915812196438,
-      "loss": 3.3792,
+      "grad_norm": 0.6747329831123352,
+      "learning_rate": 0.0003252045331894225,
+      "loss": 3.3845,
       "step": 42550
     },
     {
       "epoch": 4.592991913746631,
-      "grad_norm": 0.6453298926353455,
-      "learning_rate": 0.00032486778197517537,
-      "loss": 3.4067,
+      "grad_norm": 0.6331080794334412,
+      "learning_rate": 0.0003248807339449541,
+      "loss": 3.4101,
       "step": 42600
     },
     {
       "epoch": 4.598382749326145,
-      "grad_norm": 0.6815885901451111,
-      "learning_rate": 0.000324543982730707,
-      "loss": 3.3896,
+      "grad_norm": 0.7317514419555664,
+      "learning_rate": 0.00032455693470048567,
+      "loss": 3.3953,
       "step": 42650
     },
     {
       "epoch": 4.60377358490566,
-      "grad_norm": 0.624555766582489,
-      "learning_rate": 0.0003242266594711278,
-      "loss": 3.3894,
+      "grad_norm": 0.628052294254303,
+      "learning_rate": 0.0003242331354560173,
+      "loss": 3.394,
       "step": 42700
     },
     {
       "epoch": 4.609164420485175,
-      "grad_norm": 0.6440547704696655,
-      "learning_rate": 0.0003239028602266595,
-      "loss": 3.3971,
+      "grad_norm": 0.6382932066917419,
+      "learning_rate": 0.0003239093362115488,
+      "loss": 3.4031,
       "step": 42750
     },
     {
       "epoch": 4.6145552560646905,
-      "grad_norm": 0.6493710875511169,
-      "learning_rate": 0.000323579060982191,
-      "loss": 3.3733,
+      "grad_norm": 0.6547942757606506,
+      "learning_rate": 0.00032358553696708043,
+      "loss": 3.3777,
       "step": 42800
     },
     {
       "epoch": 4.619946091644205,
-      "grad_norm": 0.6572422981262207,
-      "learning_rate": 0.0003232552617377226,
-      "loss": 3.3984,
+      "grad_norm": 0.7021990418434143,
+      "learning_rate": 0.00032326173772261193,
+      "loss": 3.4041,
       "step": 42850
     },
     {
       "epoch": 4.62533692722372,
-      "grad_norm": 0.6159889698028564,
-      "learning_rate": 0.00032293146249325413,
-      "loss": 3.3971,
+      "grad_norm": 0.6106559038162231,
+      "learning_rate": 0.0003229444144630329,
+      "loss": 3.4043,
       "step": 42900
     },
     {
       "epoch": 4.630727762803234,
-      "grad_norm": 0.7039486169815063,
-      "learning_rate": 0.00032260766324878574,
-      "loss": 3.3921,
+      "grad_norm": 0.7007409334182739,
+      "learning_rate": 0.00032262061521856443,
+      "loss": 3.3982,
       "step": 42950
     },
     {
       "epoch": 4.636118598382749,
-      "grad_norm": 0.681053876876831,
-      "learning_rate": 0.0003222838640043173,
-      "loss": 3.3958,
+      "grad_norm": 0.6991866827011108,
+      "learning_rate": 0.00032229681597409604,
+      "loss": 3.4005,
       "step": 43000
     },
     {
       "epoch": 4.636118598382749,
-      "eval_accuracy": 0.3791799988287219,
-      "eval_loss": 3.422306776046753,
-      "eval_runtime": 183.9806,
-      "eval_samples_per_second": 97.896,
-      "eval_steps_per_second": 6.12,
+      "eval_accuracy": 0.3783138180070648,
+      "eval_loss": 3.4264564514160156,
+      "eval_runtime": 186.4282,
+      "eval_samples_per_second": 96.611,
+      "eval_steps_per_second": 6.04,
       "step": 43000
     },
     {
       "epoch": 4.6415094339622645,
-      "grad_norm": 0.6886916160583496,
-      "learning_rate": 0.00032196006475984884,
-      "loss": 3.3878,
+      "grad_norm": 0.6871011853218079,
+      "learning_rate": 0.0003219730167296276,
+      "loss": 3.3931,
       "step": 43050
     },
     {
       "epoch": 4.646900269541779,
-      "grad_norm": 0.6496508121490479,
-      "learning_rate": 0.00032163626551538044,
-      "loss": 3.3883,
+      "grad_norm": 0.6913045048713684,
+      "learning_rate": 0.0003216492174851592,
+      "loss": 3.3939,
       "step": 43100
     },
     {
       "epoch": 4.652291105121294,
-      "grad_norm": 0.6063190698623657,
-      "learning_rate": 0.000321312466270912,
-      "loss": 3.406,
+      "grad_norm": 0.6591585874557495,
+      "learning_rate": 0.00032132541824069074,
+      "loss": 3.4101,
       "step": 43150
     },
     {
       "epoch": 4.657681940700809,
-      "grad_norm": 0.6703405380249023,
-      "learning_rate": 0.00032099514301133295,
-      "loss": 3.3759,
+      "grad_norm": 0.6364275217056274,
+      "learning_rate": 0.00032100161899622235,
+      "loss": 3.3783,
       "step": 43200
     },
     {
       "epoch": 4.663072776280323,
-      "grad_norm": 0.6515152454376221,
-      "learning_rate": 0.00032067134376686455,
-      "loss": 3.3939,
+      "grad_norm": 0.6496149897575378,
+      "learning_rate": 0.0003206778197517539,
+      "loss": 3.4001,
       "step": 43250
     },
     {
       "epoch": 4.668463611859838,
-      "grad_norm": 0.6611012816429138,
-      "learning_rate": 0.0003203475445223961,
-      "loss": 3.3818,
+      "grad_norm": 0.659415602684021,
+      "learning_rate": 0.00032035402050728545,
+      "loss": 3.3874,
       "step": 43300
     },
     {
       "epoch": 4.6738544474393535,
-      "grad_norm": 0.7121708393096924,
-      "learning_rate": 0.0003200237452779277,
-      "loss": 3.3905,
+      "grad_norm": 0.6965033411979675,
+      "learning_rate": 0.00032003022126281705,
+      "loss": 3.3946,
       "step": 43350
     },
     {
       "epoch": 4.679245283018868,
-      "grad_norm": 0.6578240394592285,
-      "learning_rate": 0.00031969994603345926,
-      "loss": 3.3888,
+      "grad_norm": 0.6237617135047913,
+      "learning_rate": 0.00031970642201834855,
+      "loss": 3.3927,
       "step": 43400
     },
     {
       "epoch": 4.684636118598383,
-      "grad_norm": 0.646023154258728,
-      "learning_rate": 0.00031937614678899075,
-      "loss": 3.3989,
+      "grad_norm": 0.6738497614860535,
+      "learning_rate": 0.0003193826227738802,
+      "loss": 3.4029,
       "step": 43450
     },
     {
       "epoch": 4.690026954177897,
-      "grad_norm": 0.6373788714408875,
-      "learning_rate": 0.00031905234754452236,
-      "loss": 3.3742,
+      "grad_norm": 0.6456235647201538,
+      "learning_rate": 0.0003190588235294117,
+      "loss": 3.3805,
       "step": 43500
     },
     {
       "epoch": 4.695417789757412,
-      "grad_norm": 0.6402143239974976,
-      "learning_rate": 0.0003187285483000539,
-      "loss": 3.3918,
+      "grad_norm": 0.6421175599098206,
+      "learning_rate": 0.0003187350242849433,
+      "loss": 3.397,
       "step": 43550
     },
     {
       "epoch": 4.7008086253369274,
-      "grad_norm": 0.7260469794273376,
-      "learning_rate": 0.0003184047490555855,
-      "loss": 3.4,
+      "grad_norm": 0.6876127123832703,
+      "learning_rate": 0.00031841122504047486,
+      "loss": 3.4036,
       "step": 43600
     },
     {
       "epoch": 4.706199460916442,
-      "grad_norm": 0.657831072807312,
-      "learning_rate": 0.00031808094981111706,
-      "loss": 3.3965,
+      "grad_norm": 0.6587170958518982,
+      "learning_rate": 0.00031808742579600646,
+      "loss": 3.3995,
       "step": 43650
     },
     {
       "epoch": 4.711590296495957,
-      "grad_norm": 0.6721468567848206,
-      "learning_rate": 0.00031775715056664867,
-      "loss": 3.3898,
+      "grad_norm": 0.6888542771339417,
+      "learning_rate": 0.000317763626551538,
+      "loss": 3.3954,
       "step": 43700
     },
     {
       "epoch": 4.716981132075472,
-      "grad_norm": 0.6404152512550354,
-      "learning_rate": 0.0003174333513221802,
-      "loss": 3.407,
+      "grad_norm": 0.6613134145736694,
+      "learning_rate": 0.00031743982730706957,
+      "loss": 3.4106,
       "step": 43750
     },
     {
       "epoch": 4.722371967654986,
-      "grad_norm": 0.6717345118522644,
-      "learning_rate": 0.00031710955207771177,
-      "loss": 3.3918,
+      "grad_norm": 0.7372632026672363,
+      "learning_rate": 0.00031711602806260117,
+      "loss": 3.3962,
       "step": 43800
     },
     {
       "epoch": 4.727762803234501,
-      "grad_norm": 0.6559435725212097,
-      "learning_rate": 0.0003167857528332434,
-      "loss": 3.3983,
+      "grad_norm": 0.6780330538749695,
+      "learning_rate": 0.0003167922288181327,
+      "loss": 3.4038,
       "step": 43850
     },
     {
       "epoch": 4.7331536388140165,
-      "grad_norm": 0.6558781862258911,
-      "learning_rate": 0.0003164619535887749,
-      "loss": 3.3887,
+      "grad_norm": 0.6436319947242737,
+      "learning_rate": 0.0003164684295736643,
+      "loss": 3.3938,
       "step": 43900
     },
     {
       "epoch": 4.738544474393531,
-      "grad_norm": 0.6264235377311707,
-      "learning_rate": 0.00031613815434430653,
-      "loss": 3.3874,
+      "grad_norm": 0.6325806975364685,
+      "learning_rate": 0.0003161446303291959,
+      "loss": 3.3918,
       "step": 43950
     },
     {
       "epoch": 4.743935309973046,
-      "grad_norm": 0.6540461778640747,
-      "learning_rate": 0.0003158143550998381,
-      "loss": 3.3875,
+      "grad_norm": 0.648617148399353,
+      "learning_rate": 0.0003158208310847275,
+      "loss": 3.3928,
       "step": 44000
     },
     {
       "epoch": 4.743935309973046,
-      "eval_accuracy": 0.3794974825669158,
-      "eval_loss": 3.418121099472046,
-      "eval_runtime": 183.5372,
-      "eval_samples_per_second": 98.133,
-      "eval_steps_per_second": 6.135,
+      "eval_accuracy": 0.3787626630866264,
+      "eval_loss": 3.4221608638763428,
+      "eval_runtime": 186.2275,
+      "eval_samples_per_second": 96.715,
+      "eval_steps_per_second": 6.046,
       "step": 44000
     },
     {
       "epoch": 4.74932614555256,
-      "grad_norm": 0.6550005078315735,
-      "learning_rate": 0.0003154905558553697,
-      "loss": 3.3855,
+      "grad_norm": 0.7008000612258911,
+      "learning_rate": 0.00031549703184025903,
+      "loss": 3.3916,
       "step": 44050
     },
     {
       "epoch": 4.754716981132075,
-      "grad_norm": 0.6489050984382629,
-      "learning_rate": 0.0003151667566109012,
-      "loss": 3.3913,
+      "grad_norm": 0.6235912442207336,
+      "learning_rate": 0.00031517323259579064,
+      "loss": 3.399,
       "step": 44100
     },
     {
       "epoch": 4.7601078167115904,
-      "grad_norm": 0.7171805500984192,
-      "learning_rate": 0.00031484295736643284,
-      "loss": 3.386,
+      "grad_norm": 0.7041214108467102,
+      "learning_rate": 0.00031484943335132213,
+      "loss": 3.3925,
       "step": 44150
     },
     {
       "epoch": 4.765498652291106,
-      "grad_norm": 0.7200286388397217,
-      "learning_rate": 0.00031451915812196434,
-      "loss": 3.3915,
+      "grad_norm": 0.7416334748268127,
+      "learning_rate": 0.0003145256341068537,
+      "loss": 3.3945,
       "step": 44200
     },
     {
       "epoch": 4.77088948787062,
-      "grad_norm": 0.6571013927459717,
-      "learning_rate": 0.0003141953588774959,
-      "loss": 3.3739,
+      "grad_norm": 0.6544662117958069,
+      "learning_rate": 0.0003142018348623853,
+      "loss": 3.3814,
       "step": 44250
     },
     {
       "epoch": 4.776280323450135,
-      "grad_norm": 0.7125272750854492,
-      "learning_rate": 0.0003138715596330275,
-      "loss": 3.3972,
+      "grad_norm": 0.7117622494697571,
+      "learning_rate": 0.00031387803561791684,
+      "loss": 3.403,
       "step": 44300
     },
     {
       "epoch": 4.781671159029649,
-      "grad_norm": 0.6634894013404846,
-      "learning_rate": 0.00031354776038855904,
-      "loss": 3.3768,
+      "grad_norm": 0.6284524202346802,
+      "learning_rate": 0.00031355423637344844,
+      "loss": 3.3816,
       "step": 44350
     },
     {
       "epoch": 4.787061994609164,
-      "grad_norm": 0.6476519703865051,
-      "learning_rate": 0.00031322396114409065,
-      "loss": 3.4184,
+      "grad_norm": 0.6240745186805725,
+      "learning_rate": 0.00031323043712898,
+      "loss": 3.4231,
       "step": 44400
     },
     {
       "epoch": 4.7924528301886795,
-      "grad_norm": 0.6488552689552307,
-      "learning_rate": 0.0003129001618996222,
-      "loss": 3.3751,
+      "grad_norm": 0.8229324817657471,
+      "learning_rate": 0.0003129066378845116,
+      "loss": 3.3803,
       "step": 44450
     },
     {
       "epoch": 4.797843665768194,
-      "grad_norm": 0.6651248335838318,
-      "learning_rate": 0.0003125763626551538,
-      "loss": 3.3902,
+      "grad_norm": 0.6355055570602417,
+      "learning_rate": 0.00031258283864004315,
+      "loss": 3.3949,
       "step": 44500
     },
     {
       "epoch": 4.803234501347709,
-      "grad_norm": 0.6538376212120056,
-      "learning_rate": 0.00031225256341068535,
-      "loss": 3.3826,
+      "grad_norm": 0.6870371103286743,
+      "learning_rate": 0.0003122590393955747,
+      "loss": 3.3877,
       "step": 44550
     },
     {
       "epoch": 4.808625336927224,
-      "grad_norm": 0.6490738391876221,
-      "learning_rate": 0.00031192876416621696,
-      "loss": 3.3966,
+      "grad_norm": 0.6802382469177246,
+      "learning_rate": 0.0003119352401511063,
+      "loss": 3.4016,
       "step": 44600
     },
     {
       "epoch": 4.814016172506738,
-      "grad_norm": 0.7015892267227173,
-      "learning_rate": 0.0003116049649217485,
-      "loss": 3.3805,
+      "grad_norm": 0.7622964978218079,
+      "learning_rate": 0.00031161144090663786,
+      "loss": 3.3853,
       "step": 44650
     },
     {
       "epoch": 4.819407008086253,
-      "grad_norm": 0.6848888993263245,
-      "learning_rate": 0.00031128116567728,
-      "loss": 3.3929,
+      "grad_norm": 0.6578140258789062,
+      "learning_rate": 0.00031128764166216946,
+      "loss": 3.4009,
       "step": 44700
     },
     {
       "epoch": 4.824797843665769,
-      "grad_norm": 0.722858726978302,
-      "learning_rate": 0.00031095736643281166,
-      "loss": 3.3867,
+      "grad_norm": 0.6855496764183044,
+      "learning_rate": 0.00031096384241770096,
+      "loss": 3.3922,
       "step": 44750
     },
     {
       "epoch": 4.830188679245283,
-      "grad_norm": 0.6747796535491943,
-      "learning_rate": 0.00031063356718834316,
-      "loss": 3.386,
+      "grad_norm": 0.7051732540130615,
+      "learning_rate": 0.0003106400431732326,
+      "loss": 3.3918,
       "step": 44800
     },
     {
       "epoch": 4.835579514824798,
-      "grad_norm": 0.6656998991966248,
-      "learning_rate": 0.00031030976794387476,
-      "loss": 3.3898,
+      "grad_norm": 0.6387345194816589,
+      "learning_rate": 0.0003103162439287641,
+      "loss": 3.3942,
       "step": 44850
     },
     {
       "epoch": 4.840970350404312,
-      "grad_norm": 0.6500997543334961,
-      "learning_rate": 0.0003099859686994063,
-      "loss": 3.3885,
+      "grad_norm": 0.6645002365112305,
+      "learning_rate": 0.00030999892066918506,
+      "loss": 3.3944,
       "step": 44900
     },
     {
       "epoch": 4.846361185983827,
-      "grad_norm": 0.6383869647979736,
-      "learning_rate": 0.0003096621694549379,
-      "loss": 3.3816,
+      "grad_norm": 0.6434149146080017,
+      "learning_rate": 0.0003096751214247166,
+      "loss": 3.3853,
       "step": 44950
     },
     {
       "epoch": 4.8517520215633425,
-      "grad_norm": 0.6764130592346191,
-      "learning_rate": 0.00030933837021046947,
-      "loss": 3.4055,
+      "grad_norm": 0.6710672378540039,
+      "learning_rate": 0.0003093513221802482,
+      "loss": 3.4106,
       "step": 45000
     },
     {
       "epoch": 4.8517520215633425,
-      "eval_accuracy": 0.3802766324254334,
-      "eval_loss": 3.4115288257598877,
-      "eval_runtime": 184.1198,
-      "eval_samples_per_second": 97.822,
-      "eval_steps_per_second": 6.116,
+      "eval_accuracy": 0.3795697367373124,
+      "eval_loss": 3.4175922870635986,
+      "eval_runtime": 186.3743,
+      "eval_samples_per_second": 96.639,
+      "eval_steps_per_second": 6.042,
       "step": 45000
     },
     {
       "epoch": 4.857142857142857,
-      "grad_norm": 0.7333670854568481,
-      "learning_rate": 0.0003090145709660011,
-      "loss": 3.3761,
+      "grad_norm": 0.7392635345458984,
+      "learning_rate": 0.00030902752293577977,
+      "loss": 3.3831,
       "step": 45050
     },
     {
       "epoch": 4.862533692722372,
-      "grad_norm": 0.666333794593811,
-      "learning_rate": 0.0003086907717215326,
-      "loss": 3.3802,
+      "grad_norm": 0.6971195936203003,
+      "learning_rate": 0.0003087037236913114,
+      "loss": 3.385,
       "step": 45100
     },
     {
       "epoch": 4.867924528301887,
-      "grad_norm": 0.6561823487281799,
-      "learning_rate": 0.0003083669724770642,
-      "loss": 3.3913,
+      "grad_norm": 0.6733442544937134,
+      "learning_rate": 0.0003083799244468429,
+      "loss": 3.397,
       "step": 45150
     },
     {
       "epoch": 4.873315363881401,
-      "grad_norm": 0.6639450192451477,
-      "learning_rate": 0.0003080431732325958,
-      "loss": 3.3775,
+      "grad_norm": 0.7128764390945435,
+      "learning_rate": 0.00030805612520237453,
+      "loss": 3.3829,
       "step": 45200
     },
     {
       "epoch": 4.878706199460916,
-      "grad_norm": 0.6635633707046509,
-      "learning_rate": 0.00030771937398812733,
-      "loss": 3.394,
+      "grad_norm": 0.7105869054794312,
+      "learning_rate": 0.0003077323259579061,
+      "loss": 3.3975,
       "step": 45250
     },
     {
       "epoch": 4.884097035040432,
-      "grad_norm": 0.6871430277824402,
-      "learning_rate": 0.00030739557474365894,
-      "loss": 3.3784,
+      "grad_norm": 0.6730533838272095,
+      "learning_rate": 0.00030740852671343763,
+      "loss": 3.3831,
       "step": 45300
     },
     {
       "epoch": 4.889487870619946,
-      "grad_norm": 0.6663231253623962,
-      "learning_rate": 0.0003070717754991905,
-      "loss": 3.3862,
+      "grad_norm": 0.6670714020729065,
+      "learning_rate": 0.00030708472746896924,
+      "loss": 3.3904,
       "step": 45350
     },
     {
       "epoch": 4.894878706199461,
-      "grad_norm": 0.6640026569366455,
-      "learning_rate": 0.0003067479762547221,
-      "loss": 3.3841,
+      "grad_norm": 0.7050979733467102,
+      "learning_rate": 0.00030676092822450073,
+      "loss": 3.3905,
       "step": 45400
     },
     {
       "epoch": 4.900269541778976,
-      "grad_norm": 0.7088777422904968,
-      "learning_rate": 0.0003064241770102536,
-      "loss": 3.3933,
+      "grad_norm": 0.7145324945449829,
+      "learning_rate": 0.0003064371289800324,
+      "loss": 3.3972,
       "step": 45450
     },
     {
       "epoch": 4.90566037735849,
-      "grad_norm": 0.6727973222732544,
-      "learning_rate": 0.00030610037776578514,
-      "loss": 3.3949,
+      "grad_norm": 0.6669186949729919,
+      "learning_rate": 0.0003061133297355639,
+      "loss": 3.3988,
       "step": 45500
     },
     {
       "epoch": 4.9110512129380055,
-      "grad_norm": 0.6286665201187134,
-      "learning_rate": 0.00030577657852131674,
-      "loss": 3.394,
+      "grad_norm": 0.667694091796875,
+      "learning_rate": 0.0003057895304910955,
+      "loss": 3.3995,
       "step": 45550
     },
     {
       "epoch": 4.916442048517521,
-      "grad_norm": 0.623975396156311,
-      "learning_rate": 0.0003054527792768483,
-      "loss": 3.3913,
+      "grad_norm": 0.729061484336853,
+      "learning_rate": 0.00030546573124662704,
+      "loss": 3.399,
       "step": 45600
     },
     {
       "epoch": 4.921832884097035,
-      "grad_norm": 0.6779341101646423,
-      "learning_rate": 0.0003051289800323799,
-      "loss": 3.406,
+      "grad_norm": 0.7021871209144592,
+      "learning_rate": 0.00030514193200215865,
+      "loss": 3.4118,
       "step": 45650
     },
     {
       "epoch": 4.92722371967655,
-      "grad_norm": 0.6502783894538879,
-      "learning_rate": 0.00030480518078791145,
-      "loss": 3.402,
+      "grad_norm": 0.6712191700935364,
+      "learning_rate": 0.0003048181327576902,
+      "loss": 3.4067,
       "step": 45700
     },
     {
       "epoch": 4.932614555256064,
-      "grad_norm": 0.656953752040863,
-      "learning_rate": 0.00030448138154344305,
-      "loss": 3.3801,
+      "grad_norm": 0.6800783276557922,
+      "learning_rate": 0.00030449433351322175,
+      "loss": 3.3845,
       "step": 45750
     },
     {
       "epoch": 4.938005390835579,
-      "grad_norm": 0.6836448311805725,
-      "learning_rate": 0.0003041575822989746,
-      "loss": 3.4076,
+      "grad_norm": 0.7037836313247681,
+      "learning_rate": 0.00030417053426875335,
+      "loss": 3.4125,
       "step": 45800
     },
     {
       "epoch": 4.943396226415095,
-      "grad_norm": 0.669844925403595,
-      "learning_rate": 0.0003038337830545062,
-      "loss": 3.3775,
+      "grad_norm": 0.6804696321487427,
+      "learning_rate": 0.0003038467350242849,
+      "loss": 3.3838,
       "step": 45850
     },
     {
       "epoch": 4.948787061994609,
-      "grad_norm": 0.6585819721221924,
-      "learning_rate": 0.00030350998381003776,
-      "loss": 3.3735,
+      "grad_norm": 0.6646173000335693,
+      "learning_rate": 0.0003035229357798165,
+      "loss": 3.3785,
       "step": 45900
     },
     {
       "epoch": 4.954177897574124,
-      "grad_norm": 0.6468183398246765,
-      "learning_rate": 0.0003031861845655693,
-      "loss": 3.3727,
+      "grad_norm": 0.6985288262367249,
+      "learning_rate": 0.00030319913653534806,
+      "loss": 3.3781,
       "step": 45950
     },
     {
       "epoch": 4.959568733153639,
-      "grad_norm": 0.6753085255622864,
-      "learning_rate": 0.0003028623853211009,
-      "loss": 3.3721,
+      "grad_norm": 0.6994397640228271,
+      "learning_rate": 0.00030287533729087966,
+      "loss": 3.3782,
       "step": 46000
     },
     {
       "epoch": 4.959568733153639,
-      "eval_accuracy": 0.3807414494795038,
-      "eval_loss": 3.407444953918457,
-      "eval_runtime": 183.8672,
-      "eval_samples_per_second": 97.957,
-      "eval_steps_per_second": 6.124,
+      "eval_accuracy": 0.38011039350707726,
+      "eval_loss": 3.413032293319702,
+      "eval_runtime": 186.0054,
+      "eval_samples_per_second": 96.831,
+      "eval_steps_per_second": 6.054,
       "step": 46000
     },
     {
       "epoch": 4.964959568733153,
-      "grad_norm": 0.6388835906982422,
-      "learning_rate": 0.00030253858607663247,
-      "loss": 3.3795,
+      "grad_norm": 0.6314168572425842,
+      "learning_rate": 0.0003025515380464112,
+      "loss": 3.3852,
       "step": 46050
     },
     {
       "epoch": 4.9703504043126685,
-      "grad_norm": 0.6600316166877747,
-      "learning_rate": 0.00030221478683216407,
-      "loss": 3.3892,
+      "grad_norm": 0.6810317039489746,
+      "learning_rate": 0.0003022277388019428,
+      "loss": 3.3939,
       "step": 46100
     },
     {
       "epoch": 4.975741239892184,
-      "grad_norm": 0.6317592263221741,
-      "learning_rate": 0.00030189098758769557,
-      "loss": 3.388,
+      "grad_norm": 0.6387427449226379,
+      "learning_rate": 0.0003019039395574743,
+      "loss": 3.3923,
       "step": 46150
     },
     {
       "epoch": 4.981132075471698,
-      "grad_norm": 0.670013964176178,
-      "learning_rate": 0.00030156718834322717,
-      "loss": 3.3928,
+      "grad_norm": 0.6579132080078125,
+      "learning_rate": 0.00030158014031300587,
+      "loss": 3.3979,
       "step": 46200
     },
     {
       "epoch": 4.986522911051213,
-      "grad_norm": 0.65596604347229,
-      "learning_rate": 0.0003012433890987587,
-      "loss": 3.3911,
+      "grad_norm": 0.6629548072814941,
+      "learning_rate": 0.00030125634106853747,
+      "loss": 3.3975,
       "step": 46250
     },
     {
       "epoch": 4.991913746630727,
-      "grad_norm": 0.6848031878471375,
-      "learning_rate": 0.0003009195898542903,
-      "loss": 3.3845,
+      "grad_norm": 0.7346848845481873,
+      "learning_rate": 0.000300932541824069,
+      "loss": 3.3909,
       "step": 46300
     },
     {
       "epoch": 4.997304582210242,
-      "grad_norm": 0.6780948638916016,
-      "learning_rate": 0.0003005957906098219,
-      "loss": 3.3775,
+      "grad_norm": 0.6879866719245911,
+      "learning_rate": 0.0003006087425796006,
+      "loss": 3.383,
       "step": 46350
     },
     {
       "epoch": 5.002695417789758,
-      "grad_norm": 0.7113416790962219,
-      "learning_rate": 0.00030027199136535343,
-      "loss": 3.3608,
+      "grad_norm": 0.6756317019462585,
+      "learning_rate": 0.0003002849433351322,
+      "loss": 3.3666,
       "step": 46400
     },
     {
       "epoch": 5.008086253369272,
-      "grad_norm": 0.641831636428833,
-      "learning_rate": 0.00029994819212088503,
-      "loss": 3.2879,
+      "grad_norm": 0.6570309400558472,
+      "learning_rate": 0.00029996114409066373,
+      "loss": 3.2941,
       "step": 46450
     },
     {
       "epoch": 5.013477088948787,
-      "grad_norm": 0.6682975888252258,
-      "learning_rate": 0.0002996243928764166,
-      "loss": 3.295,
+      "grad_norm": 0.6881490349769592,
+      "learning_rate": 0.00029963734484619533,
+      "loss": 3.2995,
       "step": 46500
     },
     {
       "epoch": 5.018867924528302,
-      "grad_norm": 0.6478630900382996,
-      "learning_rate": 0.0002993005936319482,
-      "loss": 3.3073,
+      "grad_norm": 0.6954234838485718,
+      "learning_rate": 0.0002993135456017269,
+      "loss": 3.3126,
       "step": 46550
     },
     {
       "epoch": 5.024258760107816,
-      "grad_norm": 0.6761943697929382,
-      "learning_rate": 0.00029897679438747974,
-      "loss": 3.2986,
+      "grad_norm": 0.6882918477058411,
+      "learning_rate": 0.0002989897463572585,
+      "loss": 3.3068,
       "step": 46600
     },
     {
       "epoch": 5.0296495956873315,
-      "grad_norm": 0.7165064215660095,
-      "learning_rate": 0.0002986529951430113,
-      "loss": 3.286,
+      "grad_norm": 0.7332518100738525,
+      "learning_rate": 0.00029866594711279004,
+      "loss": 3.2919,
       "step": 46650
     },
     {
       "epoch": 5.035040431266847,
-      "grad_norm": 0.6782529354095459,
-      "learning_rate": 0.0002983291958985429,
-      "loss": 3.297,
+      "grad_norm": 0.6864347457885742,
+      "learning_rate": 0.00029834214786832164,
+      "loss": 3.3014,
       "step": 46700
     },
     {
       "epoch": 5.040431266846361,
-      "grad_norm": 0.6282607316970825,
-      "learning_rate": 0.00029800539665407444,
-      "loss": 3.2874,
+      "grad_norm": 0.6287637948989868,
+      "learning_rate": 0.0002980183486238532,
+      "loss": 3.2935,
       "step": 46750
     },
     {
       "epoch": 5.045822102425876,
-      "grad_norm": 0.7574204802513123,
-      "learning_rate": 0.000297681597409606,
-      "loss": 3.296,
+      "grad_norm": 0.7623891234397888,
+      "learning_rate": 0.0002977010253642741,
+      "loss": 3.3022,
       "step": 46800
     },
     {
       "epoch": 5.051212938005391,
-      "grad_norm": 0.6350822448730469,
-      "learning_rate": 0.0002973577981651376,
-      "loss": 3.3101,
+      "grad_norm": 0.6717135906219482,
+      "learning_rate": 0.0002973772261198057,
+      "loss": 3.3148,
       "step": 46850
     },
     {
       "epoch": 5.056603773584905,
-      "grad_norm": 0.648804783821106,
-      "learning_rate": 0.00029703399892066915,
-      "loss": 3.3177,
+      "grad_norm": 0.6661005020141602,
+      "learning_rate": 0.00029705342687533725,
+      "loss": 3.3229,
       "step": 46900
     },
     {
       "epoch": 5.061994609164421,
-      "grad_norm": 0.6287740468978882,
-      "learning_rate": 0.00029671019967620076,
-      "loss": 3.3077,
+      "grad_norm": 0.6788833141326904,
+      "learning_rate": 0.00029672962763086885,
+      "loss": 3.3148,
       "step": 46950
     },
     {
       "epoch": 5.067385444743936,
-      "grad_norm": 0.671330988407135,
-      "learning_rate": 0.0002963864004317323,
-      "loss": 3.3139,
+      "grad_norm": 0.7043373584747314,
+      "learning_rate": 0.0002964058283864004,
+      "loss": 3.3204,
       "step": 47000
     },
     {
       "epoch": 5.067385444743936,
-      "eval_accuracy": 0.3809589725609084,
-      "eval_loss": 3.4086132049560547,
-      "eval_runtime": 183.6167,
-      "eval_samples_per_second": 98.09,
-      "eval_steps_per_second": 6.132,
+      "eval_accuracy": 0.38039267370961927,
+      "eval_loss": 3.415480136871338,
+      "eval_runtime": 186.1322,
+      "eval_samples_per_second": 96.765,
+      "eval_steps_per_second": 6.049,
       "step": 47000
     },
     {
       "epoch": 5.07277628032345,
-      "grad_norm": 0.6910858750343323,
-      "learning_rate": 0.00029606260118726386,
-      "loss": 3.2859,
+      "grad_norm": 0.6791506409645081,
+      "learning_rate": 0.000296082029141932,
+      "loss": 3.2917,
       "step": 47050
     },
     {
       "epoch": 5.078167115902965,
-      "grad_norm": 0.643031120300293,
-      "learning_rate": 0.0002957388019427954,
-      "loss": 3.2986,
+      "grad_norm": 0.6270334124565125,
+      "learning_rate": 0.00029575822989746356,
+      "loss": 3.3065,
       "step": 47100
     },
     {
       "epoch": 5.083557951482479,
-      "grad_norm": 0.7260094881057739,
-      "learning_rate": 0.000295415002698327,
-      "loss": 3.3228,
+      "grad_norm": 0.72202068567276,
+      "learning_rate": 0.0002954344306529951,
+      "loss": 3.3291,
       "step": 47150
     },
     {
       "epoch": 5.0889487870619945,
-      "grad_norm": 0.6842207908630371,
-      "learning_rate": 0.00029509120345385856,
-      "loss": 3.3173,
+      "grad_norm": 0.713234007358551,
+      "learning_rate": 0.00029511063140852666,
+      "loss": 3.323,
       "step": 47200
     },
     {
       "epoch": 5.09433962264151,
-      "grad_norm": 0.6923130750656128,
-      "learning_rate": 0.0002947738801942795,
-      "loss": 3.3,
+      "grad_norm": 0.6992060542106628,
+      "learning_rate": 0.00029478683216405826,
+      "loss": 3.3067,
       "step": 47250
     },
     {
       "epoch": 5.099730458221024,
-      "grad_norm": 0.652860701084137,
-      "learning_rate": 0.0002944500809498111,
-      "loss": 3.3124,
+      "grad_norm": 0.6967647671699524,
+      "learning_rate": 0.0002944630329195898,
+      "loss": 3.3178,
       "step": 47300
     },
     {
       "epoch": 5.105121293800539,
-      "grad_norm": 0.7349758744239807,
-      "learning_rate": 0.00029412628170534267,
-      "loss": 3.314,
+      "grad_norm": 0.7404730916023254,
+      "learning_rate": 0.0002941392336751214,
+      "loss": 3.3213,
       "step": 47350
     },
     {
       "epoch": 5.110512129380054,
-      "grad_norm": 0.6797593235969543,
-      "learning_rate": 0.0002938024824608742,
-      "loss": 3.3092,
+      "grad_norm": 0.6786648631095886,
+      "learning_rate": 0.00029381543443065297,
+      "loss": 3.3157,
       "step": 47400
     },
     {
       "epoch": 5.115902964959568,
-      "grad_norm": 0.7050797939300537,
-      "learning_rate": 0.00029347868321640577,
-      "loss": 3.316,
+      "grad_norm": 0.6873419880867004,
+      "learning_rate": 0.0002934916351861846,
+      "loss": 3.3232,
       "step": 47450
     },
     {
       "epoch": 5.121293800539084,
-      "grad_norm": 0.7074447870254517,
-      "learning_rate": 0.0002931548839719374,
-      "loss": 3.3116,
+      "grad_norm": 0.6754744648933411,
+      "learning_rate": 0.0002931678359417161,
+      "loss": 3.3163,
       "step": 47500
     },
     {
       "epoch": 5.126684636118599,
-      "grad_norm": 0.7114849090576172,
-      "learning_rate": 0.0002928310847274689,
-      "loss": 3.3062,
+      "grad_norm": 0.695446789264679,
+      "learning_rate": 0.0002928440366972477,
+      "loss": 3.3111,
       "step": 47550
     },
     {
       "epoch": 5.132075471698113,
-      "grad_norm": 0.7198069095611572,
-      "learning_rate": 0.00029250728548300053,
-      "loss": 3.3031,
+      "grad_norm": 0.70115065574646,
+      "learning_rate": 0.0002925202374527792,
+      "loss": 3.3089,
       "step": 47600
     },
     {
       "epoch": 5.137466307277628,
-      "grad_norm": 0.6557970643043518,
-      "learning_rate": 0.0002921834862385321,
-      "loss": 3.3167,
+      "grad_norm": 0.6705079078674316,
+      "learning_rate": 0.00029219643820831083,
+      "loss": 3.3255,
       "step": 47650
     },
     {
       "epoch": 5.142857142857143,
-      "grad_norm": 0.6805382370948792,
-      "learning_rate": 0.0002918596869940637,
-      "loss": 3.3198,
+      "grad_norm": 0.6612884402275085,
+      "learning_rate": 0.0002918726389638424,
+      "loss": 3.3242,
       "step": 47700
     },
     {
       "epoch": 5.1482479784366575,
-      "grad_norm": 0.6633416414260864,
-      "learning_rate": 0.0002915358877495952,
-      "loss": 3.3215,
+      "grad_norm": 0.6707556247711182,
+      "learning_rate": 0.000291548839719374,
+      "loss": 3.3269,
       "step": 47750
     },
     {
       "epoch": 5.153638814016173,
-      "grad_norm": 0.6749739646911621,
-      "learning_rate": 0.0002912120885051268,
-      "loss": 3.3283,
+      "grad_norm": 0.7169354557991028,
+      "learning_rate": 0.00029122504047490554,
+      "loss": 3.3362,
       "step": 47800
     },
     {
       "epoch": 5.159029649595688,
-      "grad_norm": 0.7563609480857849,
-      "learning_rate": 0.00029088828926065834,
-      "loss": 3.3213,
+      "grad_norm": 0.6918734312057495,
+      "learning_rate": 0.0002909012412304371,
+      "loss": 3.3266,
       "step": 47850
     },
     {
       "epoch": 5.164420485175202,
-      "grad_norm": 0.7092325687408447,
-      "learning_rate": 0.00029056449001618994,
-      "loss": 3.3062,
+      "grad_norm": 0.6942093968391418,
+      "learning_rate": 0.0002905774419859687,
+      "loss": 3.3104,
       "step": 47900
     },
     {
       "epoch": 5.169811320754717,
-      "grad_norm": 0.7057313919067383,
-      "learning_rate": 0.0002902406907717215,
-      "loss": 3.3222,
+      "grad_norm": 0.6852614283561707,
+      "learning_rate": 0.00029025364274150024,
+      "loss": 3.3282,
       "step": 47950
     },
     {
       "epoch": 5.175202156334231,
-      "grad_norm": 0.6930835247039795,
-      "learning_rate": 0.0002899168915272531,
-      "loss": 3.3123,
+      "grad_norm": 0.7430343627929688,
+      "learning_rate": 0.0002899298434970318,
+      "loss": 3.3185,
       "step": 48000
     },
     {
       "epoch": 5.175202156334231,
-      "eval_accuracy": 0.3812406008461235,
-      "eval_loss": 3.4109530448913574,
-      "eval_runtime": 183.8217,
-      "eval_samples_per_second": 97.981,
-      "eval_steps_per_second": 6.126,
+      "eval_accuracy": 0.38073080149649796,
+      "eval_loss": 3.417128324508667,
+      "eval_runtime": 185.6412,
+      "eval_samples_per_second": 97.021,
+      "eval_steps_per_second": 6.065,
       "step": 48000
     },
     {
       "epoch": 5.180592991913747,
-      "grad_norm": 0.6858886480331421,
-      "learning_rate": 0.00028959309228278465,
-      "loss": 3.3438,
+      "grad_norm": 0.7521916627883911,
+      "learning_rate": 0.0002896060442525634,
+      "loss": 3.3496,
       "step": 48050
     },
     {
       "epoch": 5.185983827493262,
-      "grad_norm": 0.7007202506065369,
-      "learning_rate": 0.00028926929303831625,
-      "loss": 3.3228,
+      "grad_norm": 0.7258300185203552,
+      "learning_rate": 0.00028928224500809495,
+      "loss": 3.3288,
       "step": 48100
     },
     {
       "epoch": 5.191374663072776,
-      "grad_norm": 0.6837854981422424,
-      "learning_rate": 0.0002889454937938478,
-      "loss": 3.3293,
+      "grad_norm": 0.6726123690605164,
+      "learning_rate": 0.0002889584457636265,
+      "loss": 3.335,
       "step": 48150
     },
     {
       "epoch": 5.196765498652291,
-      "grad_norm": 0.6684266924858093,
-      "learning_rate": 0.00028862169454937935,
-      "loss": 3.3229,
+      "grad_norm": 0.7019742727279663,
+      "learning_rate": 0.0002886346465191581,
+      "loss": 3.3274,
       "step": 48200
     },
     {
       "epoch": 5.202156334231806,
-      "grad_norm": 0.6720783710479736,
-      "learning_rate": 0.0002882978953049109,
-      "loss": 3.3037,
+      "grad_norm": 0.7249247431755066,
+      "learning_rate": 0.00028831084727468965,
+      "loss": 3.309,
       "step": 48250
     },
     {
       "epoch": 5.2075471698113205,
-      "grad_norm": 0.652512788772583,
-      "learning_rate": 0.0002879740960604425,
-      "loss": 3.3299,
+      "grad_norm": 0.6396780014038086,
+      "learning_rate": 0.00028798704803022126,
+      "loss": 3.3375,
       "step": 48300
     },
     {
       "epoch": 5.212938005390836,
-      "grad_norm": 0.6807644367218018,
-      "learning_rate": 0.00028765029681597406,
-      "loss": 3.317,
+      "grad_norm": 0.6878575086593628,
+      "learning_rate": 0.0002876632487857528,
+      "loss": 3.3209,
       "step": 48350
     },
     {
       "epoch": 5.218328840970351,
-      "grad_norm": 0.6457434892654419,
-      "learning_rate": 0.00028732649757150566,
-      "loss": 3.3284,
+      "grad_norm": 0.6487157344818115,
+      "learning_rate": 0.0002873394495412844,
+      "loss": 3.3338,
       "step": 48400
     },
     {
       "epoch": 5.223719676549865,
-      "grad_norm": 0.7819386720657349,
-      "learning_rate": 0.0002870026983270372,
-      "loss": 3.3175,
+      "grad_norm": 0.7928135991096497,
+      "learning_rate": 0.0002870156502968159,
+      "loss": 3.3244,
       "step": 48450
     },
     {
       "epoch": 5.22911051212938,
-      "grad_norm": 0.6470807790756226,
-      "learning_rate": 0.00028667889908256877,
-      "loss": 3.3181,
+      "grad_norm": 0.6876332759857178,
+      "learning_rate": 0.0002866918510523475,
+      "loss": 3.3252,
       "step": 48500
     },
     {
       "epoch": 5.234501347708895,
-      "grad_norm": 0.6543563604354858,
-      "learning_rate": 0.00028635509983810037,
-      "loss": 3.3024,
+      "grad_norm": 0.6853935718536377,
+      "learning_rate": 0.00028636805180787907,
+      "loss": 3.3085,
       "step": 48550
     },
     {
       "epoch": 5.2398921832884096,
-      "grad_norm": 0.6741393208503723,
-      "learning_rate": 0.0002860313005936319,
-      "loss": 3.3287,
+      "grad_norm": 0.7154141664505005,
+      "learning_rate": 0.00028604425256341067,
+      "loss": 3.3337,
       "step": 48600
     },
     {
       "epoch": 5.245283018867925,
-      "grad_norm": 0.6699310541152954,
-      "learning_rate": 0.00028570750134916347,
-      "loss": 3.3222,
+      "grad_norm": 0.6808615922927856,
+      "learning_rate": 0.0002857204533189422,
+      "loss": 3.3278,
       "step": 48650
     },
     {
       "epoch": 5.250673854447439,
-      "grad_norm": 0.6818548440933228,
-      "learning_rate": 0.0002853837021046951,
-      "loss": 3.3207,
+      "grad_norm": 0.6630486249923706,
+      "learning_rate": 0.0002853966540744738,
+      "loss": 3.3267,
       "step": 48700
     },
     {
       "epoch": 5.256064690026954,
-      "grad_norm": 0.6772786974906921,
-      "learning_rate": 0.00028505990286022663,
-      "loss": 3.3115,
+      "grad_norm": 0.660525918006897,
+      "learning_rate": 0.0002850728548300054,
+      "loss": 3.3157,
       "step": 48750
     },
     {
       "epoch": 5.261455525606469,
-      "grad_norm": 0.6387109756469727,
-      "learning_rate": 0.0002847361036157582,
-      "loss": 3.3251,
+      "grad_norm": 0.6710373163223267,
+      "learning_rate": 0.000284749055585537,
+      "loss": 3.3319,
       "step": 48800
     },
     {
       "epoch": 5.2668463611859835,
-      "grad_norm": 0.6841785907745361,
-      "learning_rate": 0.0002844123043712898,
-      "loss": 3.3394,
+      "grad_norm": 0.7242515683174133,
+      "learning_rate": 0.0002844252563410685,
+      "loss": 3.344,
       "step": 48850
     },
     {
       "epoch": 5.272237196765499,
-      "grad_norm": 0.6936202049255371,
-      "learning_rate": 0.00028408850512682133,
-      "loss": 3.3173,
+      "grad_norm": 0.665414035320282,
+      "learning_rate": 0.0002841014570966001,
+      "loss": 3.3239,
       "step": 48900
     },
     {
       "epoch": 5.277628032345014,
-      "grad_norm": 0.6657775044441223,
-      "learning_rate": 0.00028376470588235294,
-      "loss": 3.3317,
+      "grad_norm": 0.676421582698822,
+      "learning_rate": 0.00028377765785213163,
+      "loss": 3.3396,
       "step": 48950
     },
     {
       "epoch": 5.283018867924528,
-      "grad_norm": 0.6841873526573181,
-      "learning_rate": 0.0002834409066378845,
-      "loss": 3.3424,
+      "grad_norm": 0.678449809551239,
+      "learning_rate": 0.00028345385860766324,
+      "loss": 3.3484,
       "step": 49000
     },
     {
       "epoch": 5.283018867924528,
-      "eval_accuracy": 0.38162577533342584,
-      "eval_loss": 3.40484619140625,
-      "eval_runtime": 183.6772,
-      "eval_samples_per_second": 98.058,
-      "eval_steps_per_second": 6.13,
+      "eval_accuracy": 0.38096386194086,
+      "eval_loss": 3.4122438430786133,
+      "eval_runtime": 186.0239,
+      "eval_samples_per_second": 96.821,
+      "eval_steps_per_second": 6.053,
       "step": 49000
     },
     {
       "epoch": 5.288409703504043,
-      "grad_norm": 0.6730915307998657,
-      "learning_rate": 0.00028311710739341604,
-      "loss": 3.336,
+      "grad_norm": 0.708109438419342,
+      "learning_rate": 0.0002831300593631948,
+      "loss": 3.3418,
       "step": 49050
     },
     {
       "epoch": 5.293800539083558,
-      "grad_norm": 0.7052405476570129,
-      "learning_rate": 0.00028279330814894764,
-      "loss": 3.3385,
+      "grad_norm": 0.7314214110374451,
+      "learning_rate": 0.0002828062601187264,
+      "loss": 3.3452,
       "step": 49100
     },
     {
       "epoch": 5.2991913746630726,
-      "grad_norm": 0.6705338954925537,
-      "learning_rate": 0.0002824695089044792,
-      "loss": 3.3141,
+      "grad_norm": 0.670226514339447,
+      "learning_rate": 0.00028248246087425794,
+      "loss": 3.32,
       "step": 49150
     },
     {
       "epoch": 5.304582210242588,
-      "grad_norm": 0.6564133763313293,
-      "learning_rate": 0.00028214570966001075,
-      "loss": 3.3388,
+      "grad_norm": 0.6688907742500305,
+      "learning_rate": 0.0002821586616297895,
+      "loss": 3.3438,
       "step": 49200
     },
     {
       "epoch": 5.309973045822103,
-      "grad_norm": 0.6649319529533386,
-      "learning_rate": 0.0002818283864004317,
-      "loss": 3.3264,
+      "grad_norm": 0.7018823623657227,
+      "learning_rate": 0.0002818348623853211,
+      "loss": 3.3322,
       "step": 49250
     },
     {
       "epoch": 5.315363881401617,
-      "grad_norm": 0.7245952486991882,
-      "learning_rate": 0.0002815045871559633,
-      "loss": 3.3236,
+      "grad_norm": 0.6848539113998413,
+      "learning_rate": 0.00028151106314085265,
+      "loss": 3.3303,
       "step": 49300
     },
     {
       "epoch": 5.320754716981132,
-      "grad_norm": 0.6721333861351013,
-      "learning_rate": 0.00028118078791149485,
-      "loss": 3.3403,
+      "grad_norm": 0.6737995147705078,
+      "learning_rate": 0.0002811872638963842,
+      "loss": 3.3441,
       "step": 49350
     },
     {
       "epoch": 5.3261455525606465,
-      "grad_norm": 0.6484825015068054,
-      "learning_rate": 0.0002808569886670264,
-      "loss": 3.3165,
+      "grad_norm": 0.6838564276695251,
+      "learning_rate": 0.0002808634646519158,
+      "loss": 3.321,
       "step": 49400
     },
     {
       "epoch": 5.331536388140162,
-      "grad_norm": 0.6736767292022705,
-      "learning_rate": 0.00028053318942255795,
-      "loss": 3.316,
+      "grad_norm": 0.6920115947723389,
+      "learning_rate": 0.00028053966540744736,
+      "loss": 3.3198,
       "step": 49450
     },
     {
       "epoch": 5.336927223719677,
-      "grad_norm": 0.6363009810447693,
-      "learning_rate": 0.00028020939017808956,
-      "loss": 3.343,
+      "grad_norm": 0.6798560619354248,
+      "learning_rate": 0.0002802158661629789,
+      "loss": 3.3476,
       "step": 49500
     },
     {
       "epoch": 5.342318059299191,
-      "grad_norm": 0.698854923248291,
-      "learning_rate": 0.0002798855909336211,
-      "loss": 3.3392,
+      "grad_norm": 0.7122732400894165,
+      "learning_rate": 0.0002798920669185105,
+      "loss": 3.3455,
       "step": 49550
     },
     {
       "epoch": 5.347708894878706,
-      "grad_norm": 0.6946191787719727,
-      "learning_rate": 0.0002795617916891527,
-      "loss": 3.3336,
+      "grad_norm": 0.7223125696182251,
+      "learning_rate": 0.00027956826767404206,
+      "loss": 3.339,
       "step": 49600
     },
     {
       "epoch": 5.353099730458221,
-      "grad_norm": 0.6818442940711975,
-      "learning_rate": 0.00027923799244468426,
-      "loss": 3.321,
+      "grad_norm": 0.6990898251533508,
+      "learning_rate": 0.00027924446842957367,
+      "loss": 3.3258,
       "step": 49650
     },
     {
       "epoch": 5.3584905660377355,
-      "grad_norm": 0.6966042518615723,
-      "learning_rate": 0.00027891419320021587,
-      "loss": 3.3464,
+      "grad_norm": 0.7318446040153503,
+      "learning_rate": 0.0002789206691851052,
+      "loss": 3.3524,
       "step": 49700
     },
     {
       "epoch": 5.363881401617251,
-      "grad_norm": 0.6865735054016113,
-      "learning_rate": 0.0002785903939557474,
-      "loss": 3.3428,
+      "grad_norm": 0.6932084560394287,
+      "learning_rate": 0.00027859686994063677,
+      "loss": 3.3505,
       "step": 49750
     },
     {
       "epoch": 5.369272237196766,
-      "grad_norm": 0.6621612310409546,
-      "learning_rate": 0.00027826659471127897,
-      "loss": 3.3162,
+      "grad_norm": 0.6720565557479858,
+      "learning_rate": 0.0002782730706961683,
+      "loss": 3.3227,
       "step": 49800
     },
     {
       "epoch": 5.37466307277628,
-      "grad_norm": 0.6775622963905334,
-      "learning_rate": 0.0002779427954668105,
-      "loss": 3.348,
+      "grad_norm": 0.6870189309120178,
+      "learning_rate": 0.0002779492714516999,
+      "loss": 3.3545,
       "step": 49850
     },
     {
       "epoch": 5.380053908355795,
-      "grad_norm": 0.6511054039001465,
-      "learning_rate": 0.0002776189962223421,
-      "loss": 3.3274,
+      "grad_norm": 0.6661623120307922,
+      "learning_rate": 0.00027762547220723147,
+      "loss": 3.3326,
       "step": 49900
     },
     {
       "epoch": 5.38544474393531,
-      "grad_norm": 0.7020261883735657,
-      "learning_rate": 0.0002772951969778737,
-      "loss": 3.3409,
+      "grad_norm": 0.688973069190979,
+      "learning_rate": 0.0002773016729627631,
+      "loss": 3.3452,
       "step": 49950
     },
     {
       "epoch": 5.390835579514825,
-      "grad_norm": 0.6863080263137817,
-      "learning_rate": 0.0002769713977334053,
-      "loss": 3.3402,
+      "grad_norm": 0.6956173181533813,
+      "learning_rate": 0.00027697787371829463,
+      "loss": 3.3482,
       "step": 50000
     },
     {
       "epoch": 5.390835579514825,
-      "eval_accuracy": 0.3820112757793916,
-      "eval_loss": 3.4008445739746094,
-      "eval_runtime": 183.6696,
-      "eval_samples_per_second": 98.062,
-      "eval_steps_per_second": 6.131,
+      "eval_accuracy": 0.3812600497130423,
+      "eval_loss": 3.406048059463501,
+      "eval_runtime": 186.1428,
+      "eval_samples_per_second": 96.759,
+      "eval_steps_per_second": 6.049,
       "step": 50000
     },
     {
       "epoch": 5.39622641509434,
-      "grad_norm": 0.6434275507926941,
-      "learning_rate": 0.00027664759848893683,
-      "loss": 3.3234,
+      "grad_norm": 0.7032738327980042,
+      "learning_rate": 0.00027665407447382623,
+      "loss": 3.3279,
       "step": 50050
     },
     {
       "epoch": 5.401617250673855,
-      "grad_norm": 0.7132904529571533,
-      "learning_rate": 0.00027632379924446844,
-      "loss": 3.3187,
+      "grad_norm": 0.7030700445175171,
+      "learning_rate": 0.0002763302752293578,
+      "loss": 3.3238,
       "step": 50100
     },
     {
       "epoch": 5.407008086253369,
-      "grad_norm": 0.6604653000831604,
-      "learning_rate": 0.000276,
-      "loss": 3.3325,
+      "grad_norm": 0.6637178659439087,
+      "learning_rate": 0.0002760129519697787,
+      "loss": 3.3398,
       "step": 50150
     },
     {
       "epoch": 5.412398921832884,
-      "grad_norm": 0.702043890953064,
-      "learning_rate": 0.00027567620075553154,
-      "loss": 3.33,
+      "grad_norm": 0.7277248501777649,
+      "learning_rate": 0.0002756891527253103,
+      "loss": 3.3353,
       "step": 50200
     },
     {
       "epoch": 5.4177897574123985,
-      "grad_norm": 0.6745196580886841,
-      "learning_rate": 0.0002753524015110631,
-      "loss": 3.3392,
+      "grad_norm": 0.6890369057655334,
+      "learning_rate": 0.00027536535348084184,
+      "loss": 3.3436,
       "step": 50250
     },
     {
       "epoch": 5.423180592991914,
-      "grad_norm": 0.6715883612632751,
-      "learning_rate": 0.0002750286022665947,
-      "loss": 3.3213,
+      "grad_norm": 0.7098992466926575,
+      "learning_rate": 0.00027504155423637344,
+      "loss": 3.3257,
       "step": 50300
     },
     {
       "epoch": 5.428571428571429,
-      "grad_norm": 0.6874634623527527,
-      "learning_rate": 0.00027470480302212624,
-      "loss": 3.3519,
+      "grad_norm": 0.6998778581619263,
+      "learning_rate": 0.000274717754991905,
+      "loss": 3.3558,
       "step": 50350
     },
     {
       "epoch": 5.433962264150943,
-      "grad_norm": 0.6819542050361633,
-      "learning_rate": 0.00027438100377765785,
-      "loss": 3.3322,
+      "grad_norm": 0.7060012817382812,
+      "learning_rate": 0.0002743939557474366,
+      "loss": 3.3394,
       "step": 50400
     },
     {
       "epoch": 5.439353099730458,
-      "grad_norm": 0.7074691653251648,
-      "learning_rate": 0.0002740572045331894,
-      "loss": 3.3365,
+      "grad_norm": 0.7231426239013672,
+      "learning_rate": 0.00027407015650296815,
+      "loss": 3.3418,
       "step": 50450
     },
     {
       "epoch": 5.444743935309973,
-      "grad_norm": 0.7091408967971802,
-      "learning_rate": 0.00027373340528872095,
-      "loss": 3.3268,
+      "grad_norm": 0.7161968350410461,
+      "learning_rate": 0.0002737463572584997,
+      "loss": 3.3312,
       "step": 50500
     },
     {
       "epoch": 5.450134770889488,
-      "grad_norm": 0.6710211634635925,
-      "learning_rate": 0.00027340960604425255,
-      "loss": 3.3326,
+      "grad_norm": 0.6985761523246765,
+      "learning_rate": 0.00027342255801403125,
+      "loss": 3.3374,
       "step": 50550
     },
     {
       "epoch": 5.455525606469003,
-      "grad_norm": 0.6949871182441711,
-      "learning_rate": 0.0002730858067997841,
-      "loss": 3.3433,
+      "grad_norm": 0.7191507816314697,
+      "learning_rate": 0.00027309875876956285,
+      "loss": 3.3479,
       "step": 50600
     },
     {
       "epoch": 5.460916442048518,
-      "grad_norm": 0.7077759504318237,
-      "learning_rate": 0.00027276200755531565,
-      "loss": 3.3538,
+      "grad_norm": 0.713187038898468,
+      "learning_rate": 0.0002727749595250944,
+      "loss": 3.359,
       "step": 50650
     },
     {
       "epoch": 5.466307277628032,
-      "grad_norm": 0.7271976470947266,
-      "learning_rate": 0.00027243820831084726,
-      "loss": 3.3146,
+      "grad_norm": 0.7666817307472229,
+      "learning_rate": 0.000272451160280626,
+      "loss": 3.3219,
       "step": 50700
     },
     {
       "epoch": 5.471698113207547,
-      "grad_norm": 0.6720749735832214,
-      "learning_rate": 0.0002721144090663788,
-      "loss": 3.3346,
+      "grad_norm": 0.7090429067611694,
+      "learning_rate": 0.00027212736103615756,
+      "loss": 3.3392,
       "step": 50750
     },
     {
       "epoch": 5.4770889487870615,
-      "grad_norm": 0.6647154092788696,
-      "learning_rate": 0.00027179060982191036,
-      "loss": 3.3343,
+      "grad_norm": 0.691491425037384,
+      "learning_rate": 0.00027180356179168916,
+      "loss": 3.3398,
       "step": 50800
     },
     {
       "epoch": 5.482479784366577,
-      "grad_norm": 0.633385419845581,
-      "learning_rate": 0.00027146681057744197,
-      "loss": 3.3266,
+      "grad_norm": 0.635062575340271,
+      "learning_rate": 0.0002714797625472207,
+      "loss": 3.3324,
       "step": 50850
     },
     {
       "epoch": 5.487870619946092,
-      "grad_norm": 0.7008586525917053,
-      "learning_rate": 0.0002711430113329735,
-      "loss": 3.3252,
+      "grad_norm": 0.7192666530609131,
+      "learning_rate": 0.00027115596330275226,
+      "loss": 3.3308,
       "step": 50900
     },
     {
       "epoch": 5.493261455525606,
-      "grad_norm": 0.6773853898048401,
-      "learning_rate": 0.0002708192120885051,
-      "loss": 3.3414,
+      "grad_norm": 0.7424925565719604,
+      "learning_rate": 0.0002708321640582838,
+      "loss": 3.3472,
       "step": 50950
     },
     {
       "epoch": 5.498652291105121,
-      "grad_norm": 0.6934987306594849,
-      "learning_rate": 0.00027049541284403667,
-      "loss": 3.3439,
+      "grad_norm": 0.6961691379547119,
+      "learning_rate": 0.0002705083648138154,
+      "loss": 3.3511,
       "step": 51000
     },
     {
       "epoch": 5.498652291105121,
-      "eval_accuracy": 0.38229887997343653,
-      "eval_loss": 3.3964803218841553,
-      "eval_runtime": 184.2643,
-      "eval_samples_per_second": 97.745,
-      "eval_steps_per_second": 6.111,
+      "eval_accuracy": 0.3816840132812944,
+      "eval_loss": 3.4023196697235107,
+      "eval_runtime": 185.977,
+      "eval_samples_per_second": 96.845,
+      "eval_steps_per_second": 6.055,
       "step": 51000
     },
     {
       "epoch": 5.504043126684636,
-      "grad_norm": 0.6530951857566833,
-      "learning_rate": 0.0002701716135995683,
-      "loss": 3.3507,
+      "grad_norm": 0.7173027992248535,
+      "learning_rate": 0.00027018456556934697,
+      "loss": 3.3559,
       "step": 51050
     },
     {
       "epoch": 5.509433962264151,
-      "grad_norm": 0.6792340874671936,
-      "learning_rate": 0.0002698478143550998,
-      "loss": 3.3366,
+      "grad_norm": 0.6625124216079712,
+      "learning_rate": 0.0002698607663248786,
+      "loss": 3.3429,
       "step": 51100
     },
     {
       "epoch": 5.514824797843666,
-      "grad_norm": 0.6732293367385864,
-      "learning_rate": 0.0002695240151106314,
-      "loss": 3.3349,
+      "grad_norm": 0.6566953659057617,
+      "learning_rate": 0.0002695369670804101,
+      "loss": 3.3397,
       "step": 51150
     },
     {
       "epoch": 5.520215633423181,
-      "grad_norm": 0.6773748993873596,
-      "learning_rate": 0.00026920021586616293,
-      "loss": 3.3177,
+      "grad_norm": 0.694071888923645,
+      "learning_rate": 0.0002692131678359417,
+      "loss": 3.3227,
       "step": 51200
     },
     {
       "epoch": 5.525606469002695,
-      "grad_norm": 0.6929314136505127,
-      "learning_rate": 0.00026887641662169453,
-      "loss": 3.338,
+      "grad_norm": 0.7035288214683533,
+      "learning_rate": 0.0002688893685914733,
+      "loss": 3.3436,
       "step": 51250
     },
     {
       "epoch": 5.53099730458221,
-      "grad_norm": 0.6713197231292725,
-      "learning_rate": 0.0002685590933621155,
-      "loss": 3.33,
+      "grad_norm": 0.7029026746749878,
+      "learning_rate": 0.00026856556934700483,
+      "loss": 3.3368,
       "step": 51300
     },
     {
       "epoch": 5.536388140161725,
-      "grad_norm": 0.6788355708122253,
-      "learning_rate": 0.00026823529411764704,
-      "loss": 3.3213,
+      "grad_norm": 0.7325018048286438,
+      "learning_rate": 0.0002682417701025364,
+      "loss": 3.3267,
       "step": 51350
     },
     {
       "epoch": 5.54177897574124,
-      "grad_norm": 0.7311290502548218,
-      "learning_rate": 0.0002679114948731786,
-      "loss": 3.3131,
+      "grad_norm": 0.7371904850006104,
+      "learning_rate": 0.000267917970858068,
+      "loss": 3.3168,
       "step": 51400
     },
     {
       "epoch": 5.547169811320755,
-      "grad_norm": 0.6731926202774048,
-      "learning_rate": 0.0002675876956287102,
-      "loss": 3.3122,
+      "grad_norm": 0.7338621616363525,
+      "learning_rate": 0.00026759417161359954,
+      "loss": 3.3169,
       "step": 51450
     },
     {
       "epoch": 5.55256064690027,
-      "grad_norm": 0.7098152041435242,
-      "learning_rate": 0.00026726389638424174,
-      "loss": 3.3427,
+      "grad_norm": 0.7411488890647888,
+      "learning_rate": 0.0002672703723691311,
+      "loss": 3.3481,
       "step": 51500
     },
     {
       "epoch": 5.557951482479784,
-      "grad_norm": 0.7369858026504517,
-      "learning_rate": 0.0002669400971397733,
-      "loss": 3.3459,
+      "grad_norm": 0.7501685619354248,
+      "learning_rate": 0.0002669465731246627,
+      "loss": 3.3501,
       "step": 51550
     },
     {
       "epoch": 5.563342318059299,
-      "grad_norm": 0.6638474464416504,
-      "learning_rate": 0.0002666162978953049,
-      "loss": 3.3208,
+      "grad_norm": 0.6812242865562439,
+      "learning_rate": 0.00026662277388019424,
+      "loss": 3.3249,
       "step": 51600
     },
     {
       "epoch": 5.568733153638814,
-      "grad_norm": 0.6835918426513672,
-      "learning_rate": 0.00026629249865083645,
-      "loss": 3.3362,
+      "grad_norm": 0.6850436925888062,
+      "learning_rate": 0.00026629897463572585,
+      "loss": 3.3417,
       "step": 51650
     },
     {
       "epoch": 5.574123989218329,
-      "grad_norm": 0.7660952806472778,
-      "learning_rate": 0.00026596869940636805,
-      "loss": 3.3392,
+      "grad_norm": 0.7204802632331848,
+      "learning_rate": 0.0002659751753912574,
+      "loss": 3.3435,
       "step": 51700
     },
     {
       "epoch": 5.579514824797844,
-      "grad_norm": 0.6250906586647034,
-      "learning_rate": 0.0002656449001618996,
-      "loss": 3.3312,
+      "grad_norm": 0.6956599354743958,
+      "learning_rate": 0.00026565137614678895,
+      "loss": 3.3369,
       "step": 51750
     },
     {
       "epoch": 5.584905660377358,
-      "grad_norm": 0.6872050762176514,
-      "learning_rate": 0.0002653211009174312,
-      "loss": 3.3224,
+      "grad_norm": 0.6965302228927612,
+      "learning_rate": 0.0002653275769023205,
+      "loss": 3.3288,
       "step": 51800
     },
     {
       "epoch": 5.590296495956873,
-      "grad_norm": 0.7721320986747742,
-      "learning_rate": 0.0002649973016729627,
-      "loss": 3.333,
+      "grad_norm": 0.7415390014648438,
+      "learning_rate": 0.0002650037776578521,
+      "loss": 3.3369,
       "step": 51850
     },
     {
       "epoch": 5.595687331536388,
-      "grad_norm": 0.6676695346832275,
-      "learning_rate": 0.0002646735024284943,
-      "loss": 3.3208,
+      "grad_norm": 0.7478113174438477,
+      "learning_rate": 0.00026467997841338366,
+      "loss": 3.3257,
       "step": 51900
     },
     {
       "epoch": 5.601078167115903,
-      "grad_norm": 0.7069310545921326,
-      "learning_rate": 0.00026434970318402586,
-      "loss": 3.3394,
+      "grad_norm": 0.7053236365318298,
+      "learning_rate": 0.00026435617916891526,
+      "loss": 3.3436,
       "step": 51950
     },
     {
       "epoch": 5.606469002695418,
-      "grad_norm": 0.6760640144348145,
-      "learning_rate": 0.00026402590393955746,
-      "loss": 3.3189,
+      "grad_norm": 0.6879009008407593,
+      "learning_rate": 0.0002640323799244468,
+      "loss": 3.3221,
       "step": 52000
     },
     {
       "epoch": 5.606469002695418,
-      "eval_accuracy": 0.3827234954590155,
-      "eval_loss": 3.3922245502471924,
-      "eval_runtime": 183.5334,
-      "eval_samples_per_second": 98.135,
-      "eval_steps_per_second": 6.135,
+      "eval_accuracy": 0.3822317324887672,
+      "eval_loss": 3.396566390991211,
+      "eval_runtime": 186.417,
+      "eval_samples_per_second": 96.617,
+      "eval_steps_per_second": 6.04,
       "step": 52000
     },
     {
       "epoch": 5.611859838274933,
-      "grad_norm": 0.7278321981430054,
-      "learning_rate": 0.000263702104695089,
-      "loss": 3.335,
+      "grad_norm": 0.7406420707702637,
+      "learning_rate": 0.0002637085806799784,
+      "loss": 3.3409,
       "step": 52050
     },
     {
       "epoch": 5.617250673854447,
-      "grad_norm": 0.7073381543159485,
-      "learning_rate": 0.0002633783054506206,
-      "loss": 3.3279,
+      "grad_norm": 0.701263427734375,
+      "learning_rate": 0.00026338478143550997,
+      "loss": 3.3317,
       "step": 52100
     },
     {
       "epoch": 5.622641509433962,
-      "grad_norm": 0.7076802253723145,
-      "learning_rate": 0.00026305450620615217,
-      "loss": 3.3472,
+      "grad_norm": 0.736813485622406,
+      "learning_rate": 0.00026306098219104157,
+      "loss": 3.3537,
       "step": 52150
     },
     {
       "epoch": 5.628032345013477,
-      "grad_norm": 0.7682281732559204,
-      "learning_rate": 0.0002627307069616837,
-      "loss": 3.3518,
+      "grad_norm": 0.7299323081970215,
+      "learning_rate": 0.00026273718294657307,
+      "loss": 3.3561,
       "step": 52200
     },
     {
       "epoch": 5.633423180592992,
-      "grad_norm": 0.6639510989189148,
-      "learning_rate": 0.0002624069077172153,
-      "loss": 3.3299,
+      "grad_norm": 0.6701496839523315,
+      "learning_rate": 0.00026241338370210467,
+      "loss": 3.335,
       "step": 52250
     },
     {
       "epoch": 5.638814016172507,
-      "grad_norm": 0.6877766847610474,
-      "learning_rate": 0.0002620831084727469,
-      "loss": 3.3161,
+      "grad_norm": 0.6939319372177124,
+      "learning_rate": 0.0002620895844576362,
+      "loss": 3.3223,
       "step": 52300
     },
     {
       "epoch": 5.644204851752022,
-      "grad_norm": 0.6921620965003967,
-      "learning_rate": 0.0002617593092282784,
-      "loss": 3.3437,
+      "grad_norm": 0.7171927690505981,
+      "learning_rate": 0.00026176578521316783,
+      "loss": 3.3477,
       "step": 52350
     },
     {
       "epoch": 5.649595687331536,
-      "grad_norm": 0.6771333813667297,
-      "learning_rate": 0.00026143550998381003,
-      "loss": 3.3196,
+      "grad_norm": 0.6944140791893005,
+      "learning_rate": 0.0002614419859686994,
+      "loss": 3.3225,
       "step": 52400
     },
     {
       "epoch": 5.654986522911051,
-      "grad_norm": 0.7312127351760864,
-      "learning_rate": 0.0002611117107393416,
-      "loss": 3.3373,
+      "grad_norm": 0.7691469788551331,
+      "learning_rate": 0.000261118186724231,
+      "loss": 3.3444,
       "step": 52450
     },
     {
       "epoch": 5.660377358490566,
-      "grad_norm": 0.7787958383560181,
-      "learning_rate": 0.00026078791149487313,
-      "loss": 3.3444,
+      "grad_norm": 0.7626478672027588,
+      "learning_rate": 0.00026079438747976253,
+      "loss": 3.3496,
       "step": 52500
     },
     {
       "epoch": 5.665768194070081,
-      "grad_norm": 0.6783169507980347,
-      "learning_rate": 0.00026046411225040474,
-      "loss": 3.3308,
+      "grad_norm": 0.6976443529129028,
+      "learning_rate": 0.0002604705882352941,
+      "loss": 3.3376,
       "step": 52550
     },
     {
       "epoch": 5.671159029649596,
-      "grad_norm": 0.6774412393569946,
-      "learning_rate": 0.0002601403130059363,
-      "loss": 3.3452,
+      "grad_norm": 0.6761083006858826,
+      "learning_rate": 0.00026014678899082563,
+      "loss": 3.3484,
       "step": 52600
     },
     {
       "epoch": 5.67654986522911,
-      "grad_norm": 0.7152056097984314,
-      "learning_rate": 0.0002598165137614679,
-      "loss": 3.3428,
+      "grad_norm": 0.716553807258606,
+      "learning_rate": 0.00025982298974635724,
+      "loss": 3.351,
       "step": 52650
     },
     {
       "epoch": 5.681940700808625,
-      "grad_norm": 0.7162232398986816,
-      "learning_rate": 0.00025949271451699944,
-      "loss": 3.3342,
+      "grad_norm": 0.6996124982833862,
+      "learning_rate": 0.0002594991905018888,
+      "loss": 3.3394,
       "step": 52700
     },
     {
       "epoch": 5.6873315363881405,
-      "grad_norm": 0.7322828769683838,
-      "learning_rate": 0.000259168915272531,
-      "loss": 3.3471,
+      "grad_norm": 0.7624792456626892,
+      "learning_rate": 0.0002591753912574204,
+      "loss": 3.3528,
       "step": 52750
     },
     {
       "epoch": 5.692722371967655,
-      "grad_norm": 0.7306454181671143,
-      "learning_rate": 0.0002588451160280626,
-      "loss": 3.3263,
+      "grad_norm": 0.7466551065444946,
+      "learning_rate": 0.00025885159201295195,
+      "loss": 3.3324,
       "step": 52800
     },
     {
       "epoch": 5.69811320754717,
-      "grad_norm": 0.7096463441848755,
-      "learning_rate": 0.00025852131678359415,
-      "loss": 3.326,
+      "grad_norm": 0.6931333541870117,
+      "learning_rate": 0.0002585277927684835,
+      "loss": 3.3283,
       "step": 52850
     },
     {
       "epoch": 5.703504043126685,
-      "grad_norm": 0.7072291374206543,
-      "learning_rate": 0.0002581975175391257,
-      "loss": 3.3271,
+      "grad_norm": 0.729158878326416,
+      "learning_rate": 0.0002582039935240151,
+      "loss": 3.3323,
       "step": 52900
     },
     {
       "epoch": 5.708894878706199,
-      "grad_norm": 0.7425927519798279,
-      "learning_rate": 0.0002578737182946573,
-      "loss": 3.343,
+      "grad_norm": 0.730543315410614,
+      "learning_rate": 0.00025788019427954665,
+      "loss": 3.35,
       "step": 52950
     },
     {
       "epoch": 5.714285714285714,
-      "grad_norm": 0.7667298316955566,
-      "learning_rate": 0.00025754991905018885,
-      "loss": 3.3341,
+      "grad_norm": 0.81474769115448,
+      "learning_rate": 0.00025755639503507826,
+      "loss": 3.339,
       "step": 53000
     },
     {
       "epoch": 5.714285714285714,
-      "eval_accuracy": 0.38343777954350794,
-      "eval_loss": 3.385892391204834,
-      "eval_runtime": 183.7702,
-      "eval_samples_per_second": 98.008,
-      "eval_steps_per_second": 6.127,
+      "eval_accuracy": 0.3826263597773094,
+      "eval_loss": 3.3922922611236572,
+      "eval_runtime": 186.0334,
+      "eval_samples_per_second": 96.816,
+      "eval_steps_per_second": 6.053,
       "step": 53000
     },
     {
       "epoch": 5.719676549865229,
-      "grad_norm": 0.666398286819458,
-      "learning_rate": 0.00025722611980572046,
-      "loss": 3.3137,
+      "grad_norm": 0.697928786277771,
+      "learning_rate": 0.00025723907177549915,
+      "loss": 3.3179,
       "step": 53050
     },
     {
       "epoch": 5.725067385444744,
-      "grad_norm": 0.7412814497947693,
-      "learning_rate": 0.000256902320561252,
-      "loss": 3.327,
+      "grad_norm": 0.7605249881744385,
+      "learning_rate": 0.00025691527253103076,
+      "loss": 3.3357,
       "step": 53100
     },
     {
       "epoch": 5.730458221024259,
-      "grad_norm": 0.7019506692886353,
-      "learning_rate": 0.00025657852131678356,
-      "loss": 3.3345,
+      "grad_norm": 0.7238236665725708,
+      "learning_rate": 0.0002565914732865623,
+      "loss": 3.3368,
       "step": 53150
     },
     {
       "epoch": 5.735849056603773,
-      "grad_norm": 0.7068491578102112,
-      "learning_rate": 0.0002562547220723151,
-      "loss": 3.3335,
+      "grad_norm": 0.7202304601669312,
+      "learning_rate": 0.00025626767404209386,
+      "loss": 3.3395,
       "step": 53200
     },
     {
       "epoch": 5.741239892183288,
-      "grad_norm": 0.6603403091430664,
-      "learning_rate": 0.0002559309228278467,
-      "loss": 3.3237,
+      "grad_norm": 0.7189924120903015,
+      "learning_rate": 0.00025594387479762546,
+      "loss": 3.3297,
       "step": 53250
     },
     {
       "epoch": 5.7466307277628035,
-      "grad_norm": 0.6823389530181885,
-      "learning_rate": 0.00025561359956826767,
-      "loss": 3.3499,
+      "grad_norm": 0.7561414837837219,
+      "learning_rate": 0.000255620075553157,
+      "loss": 3.3566,
       "step": 53300
     },
     {
       "epoch": 5.752021563342318,
-      "grad_norm": 0.6933403611183167,
-      "learning_rate": 0.0002552898003237992,
-      "loss": 3.327,
+      "grad_norm": 0.7732231020927429,
+      "learning_rate": 0.00025529627630868857,
+      "loss": 3.332,
       "step": 53350
     },
     {
       "epoch": 5.757412398921833,
-      "grad_norm": 0.6974020004272461,
-      "learning_rate": 0.0002549660010793308,
-      "loss": 3.3137,
+      "grad_norm": 0.6987825036048889,
+      "learning_rate": 0.00025497247706422017,
+      "loss": 3.3203,
       "step": 53400
     },
     {
       "epoch": 5.762803234501348,
-      "grad_norm": 0.6720824241638184,
-      "learning_rate": 0.0002546422018348624,
-      "loss": 3.3461,
+      "grad_norm": 0.6938470602035522,
+      "learning_rate": 0.0002546486778197517,
+      "loss": 3.3521,
       "step": 53450
     },
     {
       "epoch": 5.768194070080862,
-      "grad_norm": 0.7010537385940552,
-      "learning_rate": 0.0002543184025903939,
-      "loss": 3.3161,
+      "grad_norm": 0.7128288149833679,
+      "learning_rate": 0.00025432487857528327,
+      "loss": 3.3194,
       "step": 53500
     },
     {
       "epoch": 5.773584905660377,
-      "grad_norm": 0.6914909482002258,
-      "learning_rate": 0.0002539946033459255,
-      "loss": 3.3437,
+      "grad_norm": 0.7418397068977356,
+      "learning_rate": 0.0002540010793308149,
+      "loss": 3.3489,
       "step": 53550
     },
     {
       "epoch": 5.7789757412398925,
-      "grad_norm": 0.6911107301712036,
-      "learning_rate": 0.0002536708041014571,
-      "loss": 3.3169,
+      "grad_norm": 0.730099618434906,
+      "learning_rate": 0.0002536772800863464,
+      "loss": 3.321,
       "step": 53600
     },
     {
       "epoch": 5.784366576819407,
-      "grad_norm": 0.6994060277938843,
-      "learning_rate": 0.00025334700485698863,
-      "loss": 3.3496,
+      "grad_norm": 0.6976033449172974,
+      "learning_rate": 0.00025335348084187803,
+      "loss": 3.3539,
       "step": 53650
     },
     {
       "epoch": 5.789757412398922,
-      "grad_norm": 0.7192985415458679,
-      "learning_rate": 0.00025302320561252023,
-      "loss": 3.3223,
+      "grad_norm": 0.743530809879303,
+      "learning_rate": 0.0002530296815974096,
+      "loss": 3.3296,
       "step": 53700
     },
     {
       "epoch": 5.795148247978437,
-      "grad_norm": 0.7265368103981018,
-      "learning_rate": 0.0002526994063680518,
-      "loss": 3.3361,
+      "grad_norm": 0.7820248603820801,
+      "learning_rate": 0.0002527058823529412,
+      "loss": 3.3406,
       "step": 53750
     },
     {
       "epoch": 5.800539083557951,
-      "grad_norm": 0.7237759232521057,
-      "learning_rate": 0.0002523756071235834,
-      "loss": 3.329,
+      "grad_norm": 0.7171027064323425,
+      "learning_rate": 0.0002523820831084727,
+      "loss": 3.3329,
       "step": 53800
     },
     {
       "epoch": 5.8059299191374665,
-      "grad_norm": 0.698350191116333,
-      "learning_rate": 0.00025205180787911494,
-      "loss": 3.3318,
+      "grad_norm": 0.7321844696998596,
+      "learning_rate": 0.0002520582838640043,
+      "loss": 3.3382,
       "step": 53850
     },
     {
       "epoch": 5.811320754716981,
-      "grad_norm": 0.6901752948760986,
-      "learning_rate": 0.0002517280086346465,
-      "loss": 3.3252,
+      "grad_norm": 0.7202467322349548,
+      "learning_rate": 0.00025173448461953584,
+      "loss": 3.3285,
       "step": 53900
     },
     {
       "epoch": 5.816711590296496,
-      "grad_norm": 0.6650895476341248,
-      "learning_rate": 0.00025140420939017804,
-      "loss": 3.3167,
+      "grad_norm": 0.6996687650680542,
+      "learning_rate": 0.00025141068537506744,
+      "loss": 3.324,
       "step": 53950
     },
     {
       "epoch": 5.822102425876011,
-      "grad_norm": 0.6690970063209534,
-      "learning_rate": 0.00025108041014570965,
-      "loss": 3.3311,
+      "grad_norm": 0.6719265580177307,
+      "learning_rate": 0.000251086886130599,
+      "loss": 3.3373,
       "step": 54000
     },
     {
       "epoch": 5.822102425876011,
-      "eval_accuracy": 0.384017551352886,
-      "eval_loss": 3.3810160160064697,
-      "eval_runtime": 183.9693,
-      "eval_samples_per_second": 97.902,
-      "eval_steps_per_second": 6.121,
+      "eval_accuracy": 0.38304413013095606,
+      "eval_loss": 3.3882386684417725,
+      "eval_runtime": 186.9287,
+      "eval_samples_per_second": 96.352,
+      "eval_steps_per_second": 6.024,
       "step": 54000
     },
     {
       "epoch": 5.827493261455525,
-      "grad_norm": 0.6725430488586426,
-      "learning_rate": 0.0002507566109012412,
-      "loss": 3.3441,
+      "grad_norm": 0.7260991930961609,
+      "learning_rate": 0.0002507630868861306,
+      "loss": 3.3503,
       "step": 54050
     },
     {
       "epoch": 5.83288409703504,
-      "grad_norm": 0.6652386784553528,
-      "learning_rate": 0.0002504328116567728,
-      "loss": 3.3496,
+      "grad_norm": 0.7028260827064514,
+      "learning_rate": 0.00025043928764166215,
+      "loss": 3.3547,
       "step": 54100
     },
     {
       "epoch": 5.8382749326145555,
-      "grad_norm": 0.7626335620880127,
-      "learning_rate": 0.00025010901241230435,
-      "loss": 3.3252,
+      "grad_norm": 0.7298499345779419,
+      "learning_rate": 0.00025011548839719375,
+      "loss": 3.3293,
       "step": 54150
     },
     {
       "epoch": 5.84366576819407,
-      "grad_norm": 0.6746981143951416,
-      "learning_rate": 0.0002497852131678359,
-      "loss": 3.3347,
+      "grad_norm": 0.7049230933189392,
+      "learning_rate": 0.0002497916891527253,
+      "loss": 3.3382,
       "step": 54200
     },
     {
       "epoch": 5.849056603773585,
-      "grad_norm": 0.7410542368888855,
-      "learning_rate": 0.0002494614139233675,
-      "loss": 3.3627,
+      "grad_norm": 0.7368071675300598,
+      "learning_rate": 0.00024946788990825686,
+      "loss": 3.3683,
       "step": 54250
     },
     {
       "epoch": 5.8544474393531,
-      "grad_norm": 0.6656700372695923,
-      "learning_rate": 0.00024913761467889906,
-      "loss": 3.3415,
+      "grad_norm": 0.6987283825874329,
+      "learning_rate": 0.0002491440906637884,
+      "loss": 3.3453,
       "step": 54300
     },
     {
       "epoch": 5.859838274932614,
-      "grad_norm": 0.6493524312973022,
-      "learning_rate": 0.0002488138154344306,
-      "loss": 3.3432,
+      "grad_norm": 0.6668088436126709,
+      "learning_rate": 0.00024882029141932,
+      "loss": 3.3486,
       "step": 54350
     },
     {
       "epoch": 5.8652291105121295,
-      "grad_norm": 0.7260926961898804,
-      "learning_rate": 0.0002484900161899622,
-      "loss": 3.3449,
+      "grad_norm": 0.7445617318153381,
+      "learning_rate": 0.00024849649217485156,
+      "loss": 3.3484,
       "step": 54400
     },
     {
       "epoch": 5.870619946091644,
-      "grad_norm": 0.7246960401535034,
-      "learning_rate": 0.00024816621694549376,
-      "loss": 3.3164,
+      "grad_norm": 1.6818408966064453,
+      "learning_rate": 0.00024817269293038317,
+      "loss": 3.3226,
       "step": 54450
     },
     {
       "epoch": 5.876010781671159,
-      "grad_norm": 0.7035079002380371,
-      "learning_rate": 0.00024784241770102537,
-      "loss": 3.3455,
+      "grad_norm": 0.7447675466537476,
+      "learning_rate": 0.0002478488936859147,
+      "loss": 3.3501,
       "step": 54500
     },
     {
       "epoch": 5.881401617250674,
-      "grad_norm": 0.6615075469017029,
-      "learning_rate": 0.0002475186184565569,
-      "loss": 3.332,
+      "grad_norm": 0.6854404211044312,
+      "learning_rate": 0.00024752509444144627,
+      "loss": 3.3368,
       "step": 54550
     },
     {
       "epoch": 5.886792452830189,
-      "grad_norm": 0.6515076160430908,
-      "learning_rate": 0.00024719481921208847,
-      "loss": 3.3244,
+      "grad_norm": 0.681835412979126,
+      "learning_rate": 0.00024720129519697787,
+      "loss": 3.3296,
       "step": 54600
     },
     {
       "epoch": 5.892183288409703,
-      "grad_norm": 0.6946157813072205,
-      "learning_rate": 0.0002468710199676201,
-      "loss": 3.3463,
+      "grad_norm": 0.7360410690307617,
+      "learning_rate": 0.0002468774959525094,
+      "loss": 3.3525,
       "step": 54650
     },
     {
       "epoch": 5.8975741239892185,
-      "grad_norm": 0.6916877627372742,
-      "learning_rate": 0.0002465472207231516,
-      "loss": 3.3441,
+      "grad_norm": 0.7209084033966064,
+      "learning_rate": 0.00024655369670804097,
+      "loss": 3.3474,
       "step": 54700
     },
     {
       "epoch": 5.902964959568733,
-      "grad_norm": 0.7089888453483582,
-      "learning_rate": 0.0002462234214786832,
-      "loss": 3.3255,
+      "grad_norm": 0.7356888651847839,
+      "learning_rate": 0.0002462298974635726,
+      "loss": 3.3308,
       "step": 54750
     },
     {
       "epoch": 5.908355795148248,
-      "grad_norm": 0.6872110366821289,
-      "learning_rate": 0.0002458996222342148,
-      "loss": 3.3591,
+      "grad_norm": 0.7450288534164429,
+      "learning_rate": 0.00024590609821910413,
+      "loss": 3.3639,
       "step": 54800
     },
     {
       "epoch": 5.913746630727763,
-      "grad_norm": 0.692903459072113,
-      "learning_rate": 0.00024557582298974633,
-      "loss": 3.3261,
+      "grad_norm": 0.7401671409606934,
+      "learning_rate": 0.0002455822989746357,
+      "loss": 3.3298,
       "step": 54850
     },
     {
       "epoch": 5.919137466307277,
-      "grad_norm": 0.696246862411499,
-      "learning_rate": 0.0002452520237452779,
-      "loss": 3.3555,
+      "grad_norm": 0.7494671940803528,
+      "learning_rate": 0.0002452584997301673,
+      "loss": 3.3588,
       "step": 54900
     },
     {
       "epoch": 5.9245283018867925,
-      "grad_norm": 0.6855026483535767,
-      "learning_rate": 0.0002449282245008095,
-      "loss": 3.3346,
+      "grad_norm": 0.6836948990821838,
+      "learning_rate": 0.00024493470048569883,
+      "loss": 3.339,
       "step": 54950
     },
     {
       "epoch": 5.929919137466308,
-      "grad_norm": 0.736362099647522,
-      "learning_rate": 0.00024460442525634104,
-      "loss": 3.3347,
+      "grad_norm": 0.7340713143348694,
+      "learning_rate": 0.00024461090124123044,
+      "loss": 3.3391,
       "step": 55000
     },
     {
       "epoch": 5.929919137466308,
-      "eval_accuracy": 0.38405981732624583,
-      "eval_loss": 3.3812179565429688,
-      "eval_runtime": 183.7282,
-      "eval_samples_per_second": 98.031,
-      "eval_steps_per_second": 6.129,
+      "eval_accuracy": 0.383468202352096,
+      "eval_loss": 3.3851757049560547,
+      "eval_runtime": 187.0023,
+      "eval_samples_per_second": 96.314,
+      "eval_steps_per_second": 6.021,
       "step": 55000
     },
     {
       "epoch": 5.935309973045822,
-      "grad_norm": 0.702866792678833,
-      "learning_rate": 0.00024428062601187264,
-      "loss": 3.3561,
+      "grad_norm": 0.7362047433853149,
+      "learning_rate": 0.000244287101996762,
+      "loss": 3.3615,
       "step": 55050
     },
     {
       "epoch": 5.940700808625337,
-      "grad_norm": 0.7051444053649902,
-      "learning_rate": 0.0002439568267674042,
-      "loss": 3.3295,
+      "grad_norm": 0.7279862761497498,
+      "learning_rate": 0.00024396330275229354,
+      "loss": 3.3346,
       "step": 55100
     },
     {
       "epoch": 5.946091644204852,
-      "grad_norm": 0.7245047092437744,
-      "learning_rate": 0.00024363302752293574,
-      "loss": 3.3433,
+      "grad_norm": 0.7830991148948669,
+      "learning_rate": 0.00024363950350782512,
+      "loss": 3.3483,
       "step": 55150
     },
     {
       "epoch": 5.951482479784366,
-      "grad_norm": 0.7037590742111206,
-      "learning_rate": 0.00024330922827846732,
-      "loss": 3.3271,
+      "grad_norm": 0.7136879563331604,
+      "learning_rate": 0.0002433157042633567,
+      "loss": 3.3333,
       "step": 55200
     },
     {
       "epoch": 5.9568733153638815,
-      "grad_norm": 0.6878694295883179,
-      "learning_rate": 0.0002429854290339989,
-      "loss": 3.3147,
+      "grad_norm": 0.7100470066070557,
+      "learning_rate": 0.00024299190501888827,
+      "loss": 3.3194,
       "step": 55250
     },
     {
       "epoch": 5.962264150943396,
-      "grad_norm": 0.6708479523658752,
+      "grad_norm": 0.6911134719848633,
       "learning_rate": 0.00024266810577441985,
-      "loss": 3.3289,
+      "loss": 3.3342,
       "step": 55300
     },
     {
       "epoch": 5.967654986522911,
-      "grad_norm": 0.701370358467102,
+      "grad_norm": 0.7091890573501587,
       "learning_rate": 0.00024234430652995143,
-      "loss": 3.3266,
+      "loss": 3.333,
       "step": 55350
     },
     {
       "epoch": 5.973045822102426,
-      "grad_norm": 0.6717433929443359,
+      "grad_norm": 0.6774744391441345,
       "learning_rate": 0.00024202050728548298,
-      "loss": 3.3393,
+      "loss": 3.3436,
       "step": 55400
     },
     {
       "epoch": 5.97843665768194,
-      "grad_norm": 0.6901810169219971,
+      "grad_norm": 0.7220950722694397,
       "learning_rate": 0.00024169670804101456,
-      "loss": 3.3462,
+      "loss": 3.3533,
       "step": 55450
     },
     {
       "epoch": 5.9838274932614555,
-      "grad_norm": 0.6869121789932251,
+      "grad_norm": 0.716070294380188,
       "learning_rate": 0.0002413729087965461,
-      "loss": 3.3139,
+      "loss": 3.3188,
       "step": 55500
     },
     {
       "epoch": 5.989218328840971,
-      "grad_norm": 0.7031119465827942,
+      "grad_norm": 0.7205971479415894,
       "learning_rate": 0.00024104910955207768,
-      "loss": 3.336,
+      "loss": 3.3414,
       "step": 55550
     },
     {
       "epoch": 5.994609164420485,
-      "grad_norm": 0.6958100199699402,
+      "grad_norm": 0.7215341329574585,
       "learning_rate": 0.00024072531030760926,
-      "loss": 3.3504,
+      "loss": 3.3542,
       "step": 55600
     },
     {
       "epoch": 6.0,
-      "grad_norm": 1.451558232307434,
+      "grad_norm": 1.6747649908065796,
       "learning_rate": 0.00024040151106314084,
-      "loss": 3.3206,
+      "loss": 3.3237,
       "step": 55650
     },
     {
       "epoch": 6.005390835579515,
-      "grad_norm": 0.6931526064872742,
+      "grad_norm": 0.7102381587028503,
       "learning_rate": 0.0002400777118186724,
-      "loss": 3.2332,
+      "loss": 3.2383,
       "step": 55700
     },
     {
       "epoch": 6.010781671159029,
-      "grad_norm": 0.7126358151435852,
-      "learning_rate": 0.00023975391257420397,
-      "loss": 3.2352,
+      "grad_norm": 0.7376989126205444,
+      "learning_rate": 0.00023976038855909334,
+      "loss": 3.2401,
       "step": 55750
     },
     {
       "epoch": 6.0161725067385445,
-      "grad_norm": 0.7339010238647461,
-      "learning_rate": 0.00023943011332973555,
-      "loss": 3.2684,
+      "grad_norm": 0.7144143581390381,
+      "learning_rate": 0.00023943658931462492,
+      "loss": 3.2727,
       "step": 55800
     },
     {
       "epoch": 6.02156334231806,
-      "grad_norm": 0.7038446068763733,
-      "learning_rate": 0.00023910631408526712,
-      "loss": 3.2338,
+      "grad_norm": 0.6902917623519897,
+      "learning_rate": 0.00023911279007015647,
+      "loss": 3.2385,
       "step": 55850
     },
     {
       "epoch": 6.026954177897574,
-      "grad_norm": 0.7177115082740784,
-      "learning_rate": 0.0002387825148407987,
-      "loss": 3.2345,
+      "grad_norm": 0.7157079577445984,
+      "learning_rate": 0.00023878899082568805,
+      "loss": 3.2399,
       "step": 55900
     },
     {
       "epoch": 6.032345013477089,
-      "grad_norm": 0.7070705890655518,
-      "learning_rate": 0.00023845871559633025,
-      "loss": 3.2623,
+      "grad_norm": 0.7330350279808044,
+      "learning_rate": 0.00023846519158121963,
+      "loss": 3.2678,
       "step": 55950
     },
     {
       "epoch": 6.037735849056604,
-      "grad_norm": 0.7214942574501038,
-      "learning_rate": 0.0002381349163518618,
-      "loss": 3.2357,
+      "grad_norm": 0.7658732533454895,
+      "learning_rate": 0.0002381413923367512,
+      "loss": 3.2419,
       "step": 56000
     },
     {
       "epoch": 6.037735849056604,
-      "eval_accuracy": 0.38422996774856333,
-      "eval_loss": 3.3826565742492676,
-      "eval_runtime": 183.9778,
-      "eval_samples_per_second": 97.898,
-      "eval_steps_per_second": 6.12,
+      "eval_accuracy": 0.3834734176907111,
+      "eval_loss": 3.3884003162384033,
+      "eval_runtime": 186.2326,
+      "eval_samples_per_second": 96.712,
+      "eval_steps_per_second": 6.046,
       "step": 56000
     },
     {
       "epoch": 6.0431266846361185,
-      "grad_norm": 0.6994975209236145,
-      "learning_rate": 0.00023781111710739338,
-      "loss": 3.2527,
+      "grad_norm": 0.714589536190033,
+      "learning_rate": 0.00023781759309228275,
+      "loss": 3.255,
       "step": 56050
     },
     {
       "epoch": 6.048517520215634,
-      "grad_norm": 0.6794683933258057,
-      "learning_rate": 0.00023748731786292496,
-      "loss": 3.2424,
+      "grad_norm": 0.7134836912155151,
+      "learning_rate": 0.00023749379384781433,
+      "loss": 3.2471,
       "step": 56100
     },
     {
       "epoch": 6.053908355795148,
-      "grad_norm": 0.6963556408882141,
-      "learning_rate": 0.00023716351861845654,
-      "loss": 3.2607,
+      "grad_norm": 0.7150063514709473,
+      "learning_rate": 0.0002371699946033459,
+      "loss": 3.2665,
       "step": 56150
     },
     {
       "epoch": 6.059299191374663,
-      "grad_norm": 0.673784613609314,
-      "learning_rate": 0.0002368397193739881,
-      "loss": 3.2584,
+      "grad_norm": 0.7194061875343323,
+      "learning_rate": 0.0002368461953588775,
+      "loss": 3.2651,
       "step": 56200
     },
     {
       "epoch": 6.064690026954178,
-      "grad_norm": 0.7150440812110901,
-      "learning_rate": 0.0002365159201295197,
-      "loss": 3.2551,
+      "grad_norm": 0.7277020812034607,
+      "learning_rate": 0.00023652239611440904,
+      "loss": 3.2589,
       "step": 56250
     },
     {
       "epoch": 6.070080862533692,
-      "grad_norm": 0.7197666764259338,
-      "learning_rate": 0.00023619212088505127,
-      "loss": 3.2584,
+      "grad_norm": 0.7941170334815979,
+      "learning_rate": 0.00023619859686994062,
+      "loss": 3.2641,
       "step": 56300
     },
     {
       "epoch": 6.0754716981132075,
-      "grad_norm": 0.6947667002677917,
-      "learning_rate": 0.0002358683216405828,
-      "loss": 3.2609,
+      "grad_norm": 0.7519837021827698,
+      "learning_rate": 0.00023587479762547217,
+      "loss": 3.2658,
       "step": 56350
     },
     {
       "epoch": 6.080862533692723,
-      "grad_norm": 0.7144685983657837,
-      "learning_rate": 0.00023554452239611437,
-      "loss": 3.2739,
+      "grad_norm": 0.7735520005226135,
+      "learning_rate": 0.00023555099838100374,
+      "loss": 3.2783,
       "step": 56400
     },
     {
       "epoch": 6.086253369272237,
-      "grad_norm": 0.7323464155197144,
-      "learning_rate": 0.00023522072315164595,
-      "loss": 3.2539,
+      "grad_norm": 0.7712603807449341,
+      "learning_rate": 0.00023522719913653532,
+      "loss": 3.2605,
       "step": 56450
     },
     {
       "epoch": 6.091644204851752,
-      "grad_norm": 0.7219662070274353,
-      "learning_rate": 0.00023489692390717752,
-      "loss": 3.2484,
+      "grad_norm": 0.758263349533081,
+      "learning_rate": 0.0002349033998920669,
+      "loss": 3.2532,
       "step": 56500
     },
     {
       "epoch": 6.097035040431267,
-      "grad_norm": 0.7247258424758911,
-      "learning_rate": 0.0002345731246627091,
-      "loss": 3.2491,
+      "grad_norm": 0.8331606388092041,
+      "learning_rate": 0.00023457960064759848,
+      "loss": 3.2544,
       "step": 56550
     },
     {
       "epoch": 6.1024258760107815,
-      "grad_norm": 0.761326014995575,
-      "learning_rate": 0.00023424932541824068,
-      "loss": 3.2653,
+      "grad_norm": 0.7820115089416504,
+      "learning_rate": 0.00023425580140313005,
+      "loss": 3.2708,
       "step": 56600
     },
     {
       "epoch": 6.107816711590297,
-      "grad_norm": 0.706102728843689,
-      "learning_rate": 0.00023392552617377226,
-      "loss": 3.2608,
+      "grad_norm": 0.7105745673179626,
+      "learning_rate": 0.00023393200215866163,
+      "loss": 3.266,
       "step": 56650
     },
     {
       "epoch": 6.113207547169812,
-      "grad_norm": 0.6789826154708862,
-      "learning_rate": 0.00023360172692930384,
-      "loss": 3.2632,
+      "grad_norm": 0.6970210075378418,
+      "learning_rate": 0.00023360820291419316,
+      "loss": 3.2667,
       "step": 56700
     },
     {
       "epoch": 6.118598382749326,
-      "grad_norm": 0.7313393354415894,
-      "learning_rate": 0.00023327792768483539,
-      "loss": 3.2674,
+      "grad_norm": 0.7292172908782959,
+      "learning_rate": 0.00023328440366972473,
+      "loss": 3.2725,
       "step": 56750
     },
     {
       "epoch": 6.123989218328841,
-      "grad_norm": 0.7755120992660522,
+      "grad_norm": 0.7377158403396606,
       "learning_rate": 0.0002329606044252563,
-      "loss": 3.2678,
+      "loss": 3.2739,
       "step": 56800
     },
     {
       "epoch": 6.129380053908355,
-      "grad_norm": 0.6921339631080627,
+      "grad_norm": 0.7170320749282837,
       "learning_rate": 0.0002326368051807879,
-      "loss": 3.2611,
+      "loss": 3.2636,
       "step": 56850
     },
     {
       "epoch": 6.1347708894878705,
-      "grad_norm": 0.6921306252479553,
+      "grad_norm": 0.7189595103263855,
       "learning_rate": 0.00023231300593631947,
-      "loss": 3.2698,
+      "loss": 3.2756,
       "step": 56900
     },
     {
       "epoch": 6.140161725067386,
-      "grad_norm": 0.7285143733024597,
-      "learning_rate": 0.00023198920669185104,
-      "loss": 3.2675,
+      "grad_norm": 0.719249427318573,
+      "learning_rate": 0.00023199568267674042,
+      "loss": 3.2726,
       "step": 56950
     },
     {
       "epoch": 6.1455525606469,
-      "grad_norm": 0.7773783802986145,
-      "learning_rate": 0.00023166540744738262,
-      "loss": 3.2576,
+      "grad_norm": 0.7608470320701599,
+      "learning_rate": 0.000231671883432272,
+      "loss": 3.2621,
       "step": 57000
     },
     {
       "epoch": 6.1455525606469,
-      "eval_accuracy": 0.3845002961334457,
-      "eval_loss": 3.383111000061035,
-      "eval_runtime": 185.0654,
-      "eval_samples_per_second": 97.322,
-      "eval_steps_per_second": 6.084,
+      "eval_accuracy": 0.38364454598901887,
+      "eval_loss": 3.387451171875,
+      "eval_runtime": 186.9336,
+      "eval_samples_per_second": 96.35,
+      "eval_steps_per_second": 6.024,
       "step": 57000
     },
     {
       "epoch": 6.150943396226415,
-      "grad_norm": 0.724994957447052,
-      "learning_rate": 0.0002313416082029142,
-      "loss": 3.2746,
+      "grad_norm": 0.7869667410850525,
+      "learning_rate": 0.00023134808418780352,
+      "loss": 3.2809,
       "step": 57050
     },
     {
       "epoch": 6.15633423180593,
-      "grad_norm": 0.69971764087677,
-      "learning_rate": 0.00023101780895844572,
-      "loss": 3.278,
+      "grad_norm": 0.7261604070663452,
+      "learning_rate": 0.0002310242849433351,
+      "loss": 3.2829,
       "step": 57100
     },
     {
       "epoch": 6.1617250673854445,
-      "grad_norm": 0.7013365626335144,
-      "learning_rate": 0.0002306940097139773,
-      "loss": 3.2754,
+      "grad_norm": 0.7406666278839111,
+      "learning_rate": 0.00023070048569886667,
+      "loss": 3.2817,
       "step": 57150
     },
     {
       "epoch": 6.16711590296496,
-      "grad_norm": 0.7439168095588684,
-      "learning_rate": 0.00023037021046950888,
-      "loss": 3.2606,
+      "grad_norm": 0.7367774844169617,
+      "learning_rate": 0.00023037668645439825,
+      "loss": 3.2659,
       "step": 57200
     },
     {
       "epoch": 6.172506738544475,
-      "grad_norm": 0.6989354491233826,
-      "learning_rate": 0.00023004641122504046,
-      "loss": 3.2701,
+      "grad_norm": 0.7706501483917236,
+      "learning_rate": 0.00023005288720992983,
+      "loss": 3.2735,
       "step": 57250
     },
     {
       "epoch": 6.177897574123989,
-      "grad_norm": 0.7361764311790466,
-      "learning_rate": 0.00022972261198057203,
-      "loss": 3.2753,
+      "grad_norm": 0.7444599270820618,
+      "learning_rate": 0.0002297290879654614,
+      "loss": 3.2795,
       "step": 57300
     },
     {
       "epoch": 6.183288409703504,
-      "grad_norm": 0.7835276126861572,
-      "learning_rate": 0.0002293988127361036,
-      "loss": 3.281,
+      "grad_norm": 0.7758175134658813,
+      "learning_rate": 0.00022940528872099299,
+      "loss": 3.2834,
       "step": 57350
     },
     {
       "epoch": 6.188679245283019,
-      "grad_norm": 0.7380451560020447,
-      "learning_rate": 0.00022907501349163516,
-      "loss": 3.2676,
+      "grad_norm": 0.7569961547851562,
+      "learning_rate": 0.00022908148947652454,
+      "loss": 3.2711,
       "step": 57400
     },
     {
       "epoch": 6.1940700808625335,
-      "grad_norm": 0.7116109132766724,
-      "learning_rate": 0.00022875121424716674,
-      "loss": 3.2601,
+      "grad_norm": 0.7522600889205933,
+      "learning_rate": 0.0002287576902320561,
+      "loss": 3.2661,
       "step": 57450
     },
     {
       "epoch": 6.199460916442049,
-      "grad_norm": 0.7472301125526428,
-      "learning_rate": 0.00022842741500269832,
-      "loss": 3.2762,
+      "grad_norm": 0.7756637930870056,
+      "learning_rate": 0.00022843389098758766,
+      "loss": 3.2804,
       "step": 57500
     },
     {
       "epoch": 6.204851752021563,
-      "grad_norm": 0.7332467436790466,
-      "learning_rate": 0.00022810361575822987,
-      "loss": 3.2991,
+      "grad_norm": 0.8059285879135132,
+      "learning_rate": 0.00022811009174311924,
+      "loss": 3.3043,
       "step": 57550
     },
     {
       "epoch": 6.210242587601078,
-      "grad_norm": 0.6753149032592773,
-      "learning_rate": 0.00022777981651376145,
-      "loss": 3.2702,
+      "grad_norm": 0.7188214659690857,
+      "learning_rate": 0.00022778629249865082,
+      "loss": 3.2737,
       "step": 57600
     },
     {
       "epoch": 6.215633423180593,
-      "grad_norm": 0.7721616625785828,
-      "learning_rate": 0.00022745601726929302,
-      "loss": 3.2752,
+      "grad_norm": 0.8081936836242676,
+      "learning_rate": 0.0002274624932541824,
+      "loss": 3.2793,
       "step": 57650
     },
     {
       "epoch": 6.2210242587601075,
-      "grad_norm": 0.7246688604354858,
-      "learning_rate": 0.00022713221802482457,
-      "loss": 3.2696,
+      "grad_norm": 0.7585772275924683,
+      "learning_rate": 0.00022713869400971397,
+      "loss": 3.2733,
       "step": 57700
     },
     {
       "epoch": 6.226415094339623,
-      "grad_norm": 0.6795578002929688,
-      "learning_rate": 0.00022680841878035615,
-      "loss": 3.2596,
+      "grad_norm": 0.7177000045776367,
+      "learning_rate": 0.00022681489476524553,
+      "loss": 3.2653,
       "step": 57750
     },
     {
       "epoch": 6.231805929919138,
-      "grad_norm": 0.7370179891586304,
-      "learning_rate": 0.00022648461953588773,
-      "loss": 3.2688,
+      "grad_norm": 0.781024694442749,
+      "learning_rate": 0.0002264910955207771,
+      "loss": 3.2724,
       "step": 57800
     },
     {
       "epoch": 6.237196765498652,
-      "grad_norm": 0.6983503699302673,
-      "learning_rate": 0.0002261608202914193,
-      "loss": 3.2753,
+      "grad_norm": 0.7386572957038879,
+      "learning_rate": 0.00022616729627630868,
+      "loss": 3.2803,
       "step": 57850
     },
     {
       "epoch": 6.242587601078167,
-      "grad_norm": 0.7345255613327026,
-      "learning_rate": 0.00022583702104695088,
-      "loss": 3.2752,
+      "grad_norm": 0.7951345443725586,
+      "learning_rate": 0.00022584349703184023,
+      "loss": 3.2791,
       "step": 57900
     },
     {
       "epoch": 6.247978436657682,
-      "grad_norm": 0.7167491316795349,
-      "learning_rate": 0.00022551322180248246,
-      "loss": 3.2838,
+      "grad_norm": 0.7476143836975098,
+      "learning_rate": 0.0002255196977873718,
+      "loss": 3.2882,
       "step": 57950
     },
     {
       "epoch": 6.2533692722371965,
-      "grad_norm": 0.7173773646354675,
-      "learning_rate": 0.00022518942255801399,
-      "loss": 3.2761,
+      "grad_norm": 0.7600144743919373,
+      "learning_rate": 0.0002251958985429034,
+      "loss": 3.2821,
       "step": 58000
     },
     {
       "epoch": 6.2533692722371965,
-      "eval_accuracy": 0.38490709254542316,
-      "eval_loss": 3.379185914993286,
-      "eval_runtime": 184.7659,
-      "eval_samples_per_second": 97.48,
-      "eval_steps_per_second": 6.094,
+      "eval_accuracy": 0.3843726289902637,
+      "eval_loss": 3.3831236362457275,
+      "eval_runtime": 186.414,
+      "eval_samples_per_second": 96.618,
+      "eval_steps_per_second": 6.04,
       "step": 58000
     },
     {
       "epoch": 6.258760107816712,
-      "grad_norm": 0.7806297540664673,
-      "learning_rate": 0.00022486562331354556,
-      "loss": 3.268,
+      "grad_norm": 0.7771358489990234,
+      "learning_rate": 0.00022487209929843494,
+      "loss": 3.2721,
       "step": 58050
     },
     {
       "epoch": 6.264150943396227,
-      "grad_norm": 0.7050454616546631,
-      "learning_rate": 0.00022454182406907714,
-      "loss": 3.2624,
+      "grad_norm": 0.8165920376777649,
+      "learning_rate": 0.00022454830005396652,
+      "loss": 3.2678,
       "step": 58100
     },
     {
       "epoch": 6.269541778975741,
-      "grad_norm": 0.6826767325401306,
-      "learning_rate": 0.00022421802482460872,
-      "loss": 3.2622,
+      "grad_norm": 0.7891263365745544,
+      "learning_rate": 0.0002242245008094981,
+      "loss": 3.2671,
       "step": 58150
     },
     {
       "epoch": 6.274932614555256,
-      "grad_norm": 0.7257795333862305,
-      "learning_rate": 0.0002238942255801403,
-      "loss": 3.2999,
+      "grad_norm": 0.7737285494804382,
+      "learning_rate": 0.00022390070156502967,
+      "loss": 3.3039,
       "step": 58200
     },
     {
       "epoch": 6.280323450134771,
-      "grad_norm": 0.7288961410522461,
-      "learning_rate": 0.00022357042633567187,
-      "loss": 3.2753,
+      "grad_norm": 0.7451758980751038,
+      "learning_rate": 0.00022357690232056125,
+      "loss": 3.2814,
       "step": 58250
     },
     {
       "epoch": 6.285714285714286,
-      "grad_norm": 0.7374154925346375,
-      "learning_rate": 0.00022324662709120345,
-      "loss": 3.262,
+      "grad_norm": 0.7483622431755066,
+      "learning_rate": 0.0002232531030760928,
+      "loss": 3.2665,
       "step": 58300
     },
     {
       "epoch": 6.291105121293801,
-      "grad_norm": 0.7150074243545532,
-      "learning_rate": 0.00022292282784673503,
-      "loss": 3.2877,
+      "grad_norm": 0.7253767848014832,
+      "learning_rate": 0.00022292930383162435,
+      "loss": 3.293,
       "step": 58350
     },
     {
       "epoch": 6.296495956873315,
-      "grad_norm": 0.7542361617088318,
-      "learning_rate": 0.00022259902860226655,
-      "loss": 3.2719,
+      "grad_norm": 0.7356411814689636,
+      "learning_rate": 0.00022260550458715593,
+      "loss": 3.276,
       "step": 58400
     },
     {
       "epoch": 6.30188679245283,
-      "grad_norm": 0.7025394439697266,
-      "learning_rate": 0.00022227522935779813,
-      "loss": 3.2783,
+      "grad_norm": 0.7129693627357483,
+      "learning_rate": 0.0002222817053426875,
+      "loss": 3.2837,
       "step": 58450
     },
     {
       "epoch": 6.307277628032345,
-      "grad_norm": 0.7773345708847046,
-      "learning_rate": 0.0002219514301133297,
-      "loss": 3.2935,
+      "grad_norm": 0.782540500164032,
+      "learning_rate": 0.00022195790609821908,
+      "loss": 3.2998,
       "step": 58500
     },
     {
       "epoch": 6.3126684636118595,
-      "grad_norm": 0.7307718396186829,
-      "learning_rate": 0.00022162763086886129,
-      "loss": 3.2527,
+      "grad_norm": 0.7661550045013428,
+      "learning_rate": 0.00022163410685375066,
+      "loss": 3.2574,
       "step": 58550
     },
     {
       "epoch": 6.318059299191375,
-      "grad_norm": 0.7027918696403503,
-      "learning_rate": 0.00022130383162439286,
-      "loss": 3.2868,
+      "grad_norm": 0.8685694932937622,
+      "learning_rate": 0.00022131030760928224,
+      "loss": 3.2924,
       "step": 58600
     },
     {
       "epoch": 6.32345013477089,
-      "grad_norm": 0.7043139934539795,
-      "learning_rate": 0.00022098003237992444,
-      "loss": 3.2659,
+      "grad_norm": 0.7420215606689453,
+      "learning_rate": 0.00022098650836481382,
+      "loss": 3.2702,
       "step": 58650
     },
     {
       "epoch": 6.328840970350404,
-      "grad_norm": 0.6800243854522705,
-      "learning_rate": 0.00022065623313545602,
-      "loss": 3.2702,
+      "grad_norm": 0.7648389935493469,
+      "learning_rate": 0.0002206627091203454,
+      "loss": 3.2739,
       "step": 58700
     },
     {
       "epoch": 6.334231805929919,
-      "grad_norm": 0.7141492962837219,
-      "learning_rate": 0.00022033243389098757,
-      "loss": 3.2907,
+      "grad_norm": 0.738979697227478,
+      "learning_rate": 0.00022033890987587692,
+      "loss": 3.2933,
       "step": 58750
     },
     {
       "epoch": 6.339622641509434,
-      "grad_norm": 0.7111701369285583,
-      "learning_rate": 0.00022000863464651915,
-      "loss": 3.2788,
+      "grad_norm": 0.7525094151496887,
+      "learning_rate": 0.0002200151106314085,
+      "loss": 3.2846,
       "step": 58800
     },
     {
       "epoch": 6.345013477088949,
-      "grad_norm": 0.7029032111167908,
-      "learning_rate": 0.0002196848354020507,
-      "loss": 3.2843,
+      "grad_norm": 0.7786487936973572,
+      "learning_rate": 0.00021969131138694007,
+      "loss": 3.2893,
       "step": 58850
     },
     {
       "epoch": 6.350404312668464,
-      "grad_norm": 0.7716480493545532,
-      "learning_rate": 0.00021936103615758227,
-      "loss": 3.29,
+      "grad_norm": 0.8033542633056641,
+      "learning_rate": 0.00021936751214247165,
+      "loss": 3.2947,
       "step": 58900
     },
     {
       "epoch": 6.355795148247978,
-      "grad_norm": 0.6854252815246582,
-      "learning_rate": 0.00021903723691311385,
-      "loss": 3.3005,
+      "grad_norm": 0.7372086048126221,
+      "learning_rate": 0.00021904371289800323,
+      "loss": 3.307,
       "step": 58950
     },
     {
       "epoch": 6.361185983827493,
-      "grad_norm": 0.6938695311546326,
-      "learning_rate": 0.00021871343766864543,
-      "loss": 3.289,
+      "grad_norm": 0.754686713218689,
+      "learning_rate": 0.0002187199136535348,
+      "loss": 3.2945,
       "step": 59000
     },
     {
       "epoch": 6.361185983827493,
-      "eval_accuracy": 0.3852709710667139,
-      "eval_loss": 3.37485408782959,
-      "eval_runtime": 185.1638,
-      "eval_samples_per_second": 97.271,
-      "eval_steps_per_second": 6.081,
+      "eval_accuracy": 0.3843762145355616,
+      "eval_loss": 3.380387544631958,
+      "eval_runtime": 186.6504,
+      "eval_samples_per_second": 96.496,
+      "eval_steps_per_second": 6.033,
       "step": 59000
     },
     {
       "epoch": 6.366576819407008,
-      "grad_norm": 0.7637062668800354,
-      "learning_rate": 0.00021838963842417698,
-      "loss": 3.2879,
+      "grad_norm": 0.7646997570991516,
+      "learning_rate": 0.00021839611440906638,
+      "loss": 3.2918,
       "step": 59050
     },
     {
       "epoch": 6.3719676549865225,
-      "grad_norm": 0.71692955493927,
-      "learning_rate": 0.00021806583917970856,
-      "loss": 3.2819,
+      "grad_norm": 0.7432668209075928,
+      "learning_rate": 0.00021807231516459793,
+      "loss": 3.2872,
       "step": 59100
     },
     {
       "epoch": 6.377358490566038,
-      "grad_norm": 0.7055894732475281,
-      "learning_rate": 0.00021774203993524014,
-      "loss": 3.2831,
+      "grad_norm": 0.7253616452217102,
+      "learning_rate": 0.00021774851592012948,
+      "loss": 3.2899,
       "step": 59150
     },
     {
       "epoch": 6.382749326145553,
-      "grad_norm": 0.6954150199890137,
-      "learning_rate": 0.00021741824069077171,
-      "loss": 3.2762,
+      "grad_norm": 0.6919981837272644,
+      "learning_rate": 0.00021742471667566106,
+      "loss": 3.2819,
       "step": 59200
     },
     {
       "epoch": 6.388140161725067,
-      "grad_norm": 0.7507545948028564,
-      "learning_rate": 0.00021709444144630326,
-      "loss": 3.2795,
+      "grad_norm": 0.8177418112754822,
+      "learning_rate": 0.00021710091743119264,
+      "loss": 3.2829,
       "step": 59250
     },
     {
       "epoch": 6.393530997304582,
-      "grad_norm": 0.7440924644470215,
-      "learning_rate": 0.00021677064220183484,
-      "loss": 3.2736,
+      "grad_norm": 0.7494319677352905,
+      "learning_rate": 0.00021677711818672422,
+      "loss": 3.2766,
       "step": 59300
     },
     {
       "epoch": 6.398921832884097,
-      "grad_norm": 0.7490218281745911,
-      "learning_rate": 0.00021644684295736642,
-      "loss": 3.2633,
+      "grad_norm": 0.7261015772819519,
+      "learning_rate": 0.0002164533189422558,
+      "loss": 3.2709,
       "step": 59350
     },
     {
       "epoch": 6.404312668463612,
-      "grad_norm": 0.7519710659980774,
-      "learning_rate": 0.00021612304371289797,
-      "loss": 3.2532,
+      "grad_norm": 0.7578741908073425,
+      "learning_rate": 0.00021612951969778734,
+      "loss": 3.2586,
       "step": 59400
     },
     {
       "epoch": 6.409703504043127,
-      "grad_norm": 0.7268811464309692,
-      "learning_rate": 0.00021579924446842955,
-      "loss": 3.2831,
+      "grad_norm": 0.7677342891693115,
+      "learning_rate": 0.00021580572045331892,
+      "loss": 3.2886,
       "step": 59450
     },
     {
       "epoch": 6.415094339622642,
-      "grad_norm": 0.7365671396255493,
-      "learning_rate": 0.00021547544522396113,
-      "loss": 3.2734,
+      "grad_norm": 0.7580781579017639,
+      "learning_rate": 0.0002154819212088505,
+      "loss": 3.2793,
       "step": 59500
     },
     {
       "epoch": 6.420485175202156,
-      "grad_norm": 0.784694492816925,
-      "learning_rate": 0.0002151516459794927,
-      "loss": 3.2898,
+      "grad_norm": 0.7613184452056885,
+      "learning_rate": 0.00021515812196438208,
+      "loss": 3.2956,
       "step": 59550
     },
     {
       "epoch": 6.425876010781671,
-      "grad_norm": 0.7644874453544617,
-      "learning_rate": 0.00021482784673502428,
-      "loss": 3.2836,
+      "grad_norm": 0.7674439549446106,
+      "learning_rate": 0.00021483432271991363,
+      "loss": 3.2894,
       "step": 59600
     },
     {
       "epoch": 6.431266846361186,
-      "grad_norm": 0.7010469436645508,
-      "learning_rate": 0.00021450404749055586,
-      "loss": 3.2646,
+      "grad_norm": 0.756726086139679,
+      "learning_rate": 0.0002145105234754452,
+      "loss": 3.2695,
       "step": 59650
     },
     {
       "epoch": 6.436657681940701,
-      "grad_norm": 0.7214942574501038,
-      "learning_rate": 0.00021418024824608738,
-      "loss": 3.2619,
+      "grad_norm": 0.7872341871261597,
+      "learning_rate": 0.00021418672423097676,
+      "loss": 3.2683,
       "step": 59700
     },
     {
       "epoch": 6.442048517520216,
-      "grad_norm": 0.7231842875480652,
-      "learning_rate": 0.00021385644900161896,
-      "loss": 3.278,
+      "grad_norm": 0.7990399599075317,
+      "learning_rate": 0.00021386292498650833,
+      "loss": 3.284,
       "step": 59750
     },
     {
       "epoch": 6.44743935309973,
-      "grad_norm": 0.728193998336792,
-      "learning_rate": 0.00021353264975715054,
-      "loss": 3.2837,
+      "grad_norm": 0.7387508749961853,
+      "learning_rate": 0.0002135391257420399,
+      "loss": 3.2896,
       "step": 59800
     },
     {
       "epoch": 6.452830188679245,
-      "grad_norm": 0.768265426158905,
-      "learning_rate": 0.00021320885051268211,
-      "loss": 3.2929,
+      "grad_norm": 0.7749733328819275,
+      "learning_rate": 0.0002132153264975715,
+      "loss": 3.2969,
       "step": 59850
     },
     {
       "epoch": 6.45822102425876,
-      "grad_norm": 0.7742714285850525,
-      "learning_rate": 0.0002128850512682137,
-      "loss": 3.2794,
+      "grad_norm": 0.7537945508956909,
+      "learning_rate": 0.00021289152725310307,
+      "loss": 3.2855,
       "step": 59900
     },
     {
       "epoch": 6.463611859838275,
-      "grad_norm": 0.6836204528808594,
-      "learning_rate": 0.00021256125202374527,
-      "loss": 3.2765,
+      "grad_norm": 0.7070581912994385,
+      "learning_rate": 0.00021256772800863464,
+      "loss": 3.2825,
       "step": 59950
     },
     {
       "epoch": 6.46900269541779,
-      "grad_norm": 0.9617937207221985,
-      "learning_rate": 0.00021223745277927685,
-      "loss": 3.2893,
+      "grad_norm": 0.7940769195556641,
+      "learning_rate": 0.0002122439287641662,
+      "loss": 3.2938,
       "step": 60000
     },
     {
       "epoch": 6.46900269541779,
-      "eval_accuracy": 0.38564234663726377,
-      "eval_loss": 3.3716087341308594,
-      "eval_runtime": 184.7027,
-      "eval_samples_per_second": 97.513,
-      "eval_steps_per_second": 6.096,
+      "eval_accuracy": 0.3849972744423092,
+      "eval_loss": 3.375958204269409,
+      "eval_runtime": 186.8862,
+      "eval_samples_per_second": 96.374,
+      "eval_steps_per_second": 6.025,
       "step": 60000
     },
     {
       "epoch": 6.474393530997305,
-      "grad_norm": 0.7370874881744385,
-      "learning_rate": 0.00021191365353480843,
-      "loss": 3.2907,
+      "grad_norm": 0.7654575109481812,
+      "learning_rate": 0.00021192012951969775,
+      "loss": 3.2953,
       "step": 60050
     },
     {
       "epoch": 6.479784366576819,
-      "grad_norm": 0.7189035415649414,
-      "learning_rate": 0.00021158985429033995,
-      "loss": 3.2801,
+      "grad_norm": 0.711226761341095,
+      "learning_rate": 0.00021159633027522932,
+      "loss": 3.285,
       "step": 60100
     },
     {
       "epoch": 6.485175202156334,
-      "grad_norm": 0.7072881460189819,
-      "learning_rate": 0.00021126605504587153,
-      "loss": 3.2779,
+      "grad_norm": 0.7320746183395386,
+      "learning_rate": 0.0002112725310307609,
+      "loss": 3.2819,
       "step": 60150
     },
     {
       "epoch": 6.490566037735849,
-      "grad_norm": 0.711500883102417,
-      "learning_rate": 0.0002109422558014031,
-      "loss": 3.2626,
+      "grad_norm": 0.7347570061683655,
+      "learning_rate": 0.00021094873178629248,
+      "loss": 3.2687,
       "step": 60200
     },
     {
       "epoch": 6.495956873315364,
-      "grad_norm": 0.7228755950927734,
-      "learning_rate": 0.00021061845655693468,
-      "loss": 3.2895,
+      "grad_norm": 0.744023323059082,
+      "learning_rate": 0.00021062493254182406,
+      "loss": 3.2949,
       "step": 60250
     },
     {
       "epoch": 6.501347708894879,
-      "grad_norm": 0.7080203890800476,
-      "learning_rate": 0.00021029465731246626,
-      "loss": 3.2776,
+      "grad_norm": 0.743115246295929,
+      "learning_rate": 0.00021030113329735563,
+      "loss": 3.2837,
       "step": 60300
     },
     {
       "epoch": 6.506738544474393,
-      "grad_norm": 0.7516891360282898,
-      "learning_rate": 0.00020997085806799784,
-      "loss": 3.2753,
+      "grad_norm": 0.8300098776817322,
+      "learning_rate": 0.0002099773340528872,
+      "loss": 3.2811,
       "step": 60350
     },
     {
       "epoch": 6.512129380053908,
-      "grad_norm": 0.6993798017501831,
-      "learning_rate": 0.0002096470588235294,
-      "loss": 3.2688,
+      "grad_norm": 0.7603579163551331,
+      "learning_rate": 0.0002096535348084188,
+      "loss": 3.273,
       "step": 60400
     },
     {
       "epoch": 6.517520215633423,
-      "grad_norm": 0.7494493126869202,
-      "learning_rate": 0.00020932325957906097,
-      "loss": 3.2716,
+      "grad_norm": 0.7887403964996338,
+      "learning_rate": 0.0002093297355639503,
+      "loss": 3.2769,
       "step": 60450
     },
     {
       "epoch": 6.5229110512129385,
-      "grad_norm": 0.7326953411102295,
-      "learning_rate": 0.00020899946033459254,
-      "loss": 3.2797,
+      "grad_norm": 0.7563978433609009,
+      "learning_rate": 0.0002090059363194819,
+      "loss": 3.2838,
       "step": 60500
     },
     {
       "epoch": 6.528301886792453,
-      "grad_norm": 0.7331379652023315,
-      "learning_rate": 0.0002086756610901241,
-      "loss": 3.272,
+      "grad_norm": 0.7663004994392395,
+      "learning_rate": 0.00020868213707501347,
+      "loss": 3.2772,
       "step": 60550
     },
     {
       "epoch": 6.533692722371968,
-      "grad_norm": 0.7465304136276245,
-      "learning_rate": 0.00020835186184565567,
-      "loss": 3.2809,
+      "grad_norm": 0.7374081015586853,
+      "learning_rate": 0.00020835833783054505,
+      "loss": 3.2846,
       "step": 60600
     },
     {
       "epoch": 6.539083557951482,
-      "grad_norm": 0.6983660459518433,
-      "learning_rate": 0.00020802806260118725,
-      "loss": 3.283,
+      "grad_norm": 0.7471432089805603,
+      "learning_rate": 0.00020803453858607662,
+      "loss": 3.289,
       "step": 60650
     },
     {
       "epoch": 6.544474393530997,
-      "grad_norm": 0.7236700654029846,
-      "learning_rate": 0.00020770426335671883,
-      "loss": 3.2601,
+      "grad_norm": 0.8024073839187622,
+      "learning_rate": 0.0002077107393416082,
+      "loss": 3.2654,
       "step": 60700
     },
     {
       "epoch": 6.549865229110512,
-      "grad_norm": 0.794366717338562,
+      "grad_norm": 0.8139125108718872,
       "learning_rate": 0.00020738694009713975,
-      "loss": 3.2861,
+      "loss": 3.2911,
       "step": 60750
     },
     {
       "epoch": 6.555256064690027,
-      "grad_norm": 0.7264977097511292,
+      "grad_norm": 0.7971754670143127,
       "learning_rate": 0.00020706314085267133,
-      "loss": 3.3007,
+      "loss": 3.3059,
       "step": 60800
     },
     {
       "epoch": 6.560646900269542,
-      "grad_norm": 0.7221018075942993,
+      "grad_norm": 0.8010126352310181,
       "learning_rate": 0.00020673934160820288,
-      "loss": 3.2797,
+      "loss": 3.2843,
       "step": 60850
     },
     {
       "epoch": 6.566037735849057,
-      "grad_norm": 0.7276595830917358,
+      "grad_norm": 0.7944552302360535,
       "learning_rate": 0.00020641554236373446,
-      "loss": 3.2759,
+      "loss": 3.2799,
       "step": 60900
     },
     {
       "epoch": 6.571428571428571,
-      "grad_norm": 0.7512069940567017,
+      "grad_norm": 0.7295520901679993,
       "learning_rate": 0.00020609174311926604,
-      "loss": 3.2959,
+      "loss": 3.3003,
       "step": 60950
     },
     {
       "epoch": 6.576819407008086,
-      "grad_norm": 0.7264222502708435,
+      "grad_norm": 0.7202658653259277,
       "learning_rate": 0.0002057679438747976,
-      "loss": 3.2819,
+      "loss": 3.2862,
       "step": 61000
     },
     {
       "epoch": 6.576819407008086,
-      "eval_accuracy": 0.38633196651622226,
-      "eval_loss": 3.366112470626831,
-      "eval_runtime": 184.763,
-      "eval_samples_per_second": 97.482,
-      "eval_steps_per_second": 6.094,
+      "eval_accuracy": 0.3855968210772695,
+      "eval_loss": 3.371309518814087,
+      "eval_runtime": 185.7636,
+      "eval_samples_per_second": 96.957,
+      "eval_steps_per_second": 6.061,
       "step": 61000
     },
     {
       "epoch": 6.5822102425876015,
-      "grad_norm": 0.7526605725288391,
+      "grad_norm": 0.8448360562324524,
       "learning_rate": 0.00020544414463032916,
-      "loss": 3.2952,
+      "loss": 3.3007,
       "step": 61050
     },
     {
       "epoch": 6.587601078167116,
-      "grad_norm": 0.7065187692642212,
+      "grad_norm": 0.7245503664016724,
       "learning_rate": 0.00020512034538586074,
-      "loss": 3.263,
+      "loss": 3.2678,
       "step": 61100
     },
     {
       "epoch": 6.592991913746631,
-      "grad_norm": 0.767744779586792,
+      "grad_norm": 0.8263252973556519,
       "learning_rate": 0.00020479654614139232,
-      "loss": 3.3016,
+      "loss": 3.3039,
       "step": 61150
     },
     {
       "epoch": 6.598382749326145,
-      "grad_norm": 0.7380595803260803,
+      "grad_norm": 0.7636867165565491,
       "learning_rate": 0.0002044727468969239,
-      "loss": 3.2851,
+      "loss": 3.2895,
       "step": 61200
     },
     {
       "epoch": 6.60377358490566,
-      "grad_norm": 0.6881290674209595,
+      "grad_norm": 0.7171377539634705,
       "learning_rate": 0.00020414894765245547,
-      "loss": 3.2706,
+      "loss": 3.2764,
       "step": 61250
     },
     {
       "epoch": 6.609164420485175,
-      "grad_norm": 0.6972199082374573,
+      "grad_norm": 0.752569854259491,
       "learning_rate": 0.00020382514840798702,
-      "loss": 3.2987,
+      "loss": 3.3033,
       "step": 61300
     },
     {
       "epoch": 6.6145552560646905,
-      "grad_norm": 0.7250159382820129,
+      "grad_norm": 0.7721526622772217,
       "learning_rate": 0.0002035013491635186,
-      "loss": 3.2942,
+      "loss": 3.299,
       "step": 61350
     },
     {
       "epoch": 6.619946091644205,
-      "grad_norm": 0.7611603736877441,
+      "grad_norm": 0.7868698835372925,
       "learning_rate": 0.00020317754991905015,
-      "loss": 3.2798,
+      "loss": 3.2842,
       "step": 61400
     },
     {
       "epoch": 6.62533692722372,
-      "grad_norm": 0.7625829577445984,
+      "grad_norm": 0.7732236385345459,
       "learning_rate": 0.00020285375067458173,
-      "loss": 3.2658,
+      "loss": 3.2712,
       "step": 61450
     },
     {
       "epoch": 6.630727762803234,
-      "grad_norm": 0.7896705865859985,
+      "grad_norm": 0.7886024713516235,
       "learning_rate": 0.0002025299514301133,
-      "loss": 3.2919,
+      "loss": 3.2966,
       "step": 61500
     },
     {
       "epoch": 6.636118598382749,
-      "grad_norm": 0.75810706615448,
+      "grad_norm": 0.8080036640167236,
       "learning_rate": 0.00020220615218564489,
-      "loss": 3.278,
+      "loss": 3.2826,
       "step": 61550
     },
     {
       "epoch": 6.6415094339622645,
-      "grad_norm": 0.7625492215156555,
+      "grad_norm": 0.8146077394485474,
       "learning_rate": 0.00020188235294117646,
-      "loss": 3.2772,
+      "loss": 3.2837,
       "step": 61600
     },
     {
       "epoch": 6.646900269541779,
-      "grad_norm": 0.7140247821807861,
+      "grad_norm": 0.7391796112060547,
       "learning_rate": 0.00020155855369670804,
-      "loss": 3.2875,
+      "loss": 3.293,
       "step": 61650
     },
     {
       "epoch": 6.652291105121294,
-      "grad_norm": 0.7512028217315674,
+      "grad_norm": 0.7737180590629578,
       "learning_rate": 0.00020123475445223956,
-      "loss": 3.2969,
+      "loss": 3.303,
       "step": 61700
     },
     {
       "epoch": 6.657681940700809,
-      "grad_norm": 0.7267550826072693,
+      "grad_norm": 0.7490894198417664,
       "learning_rate": 0.00020091095520777114,
-      "loss": 3.2861,
+      "loss": 3.2906,
       "step": 61750
     },
     {
       "epoch": 6.663072776280323,
-      "grad_norm": 0.7511851787567139,
+      "grad_norm": 0.7744179964065552,
       "learning_rate": 0.00020058715596330272,
-      "loss": 3.2806,
+      "loss": 3.2851,
       "step": 61800
     },
     {
       "epoch": 6.668463611859838,
-      "grad_norm": 0.777556836605072,
+      "grad_norm": 0.8192709684371948,
       "learning_rate": 0.0002002633567188343,
-      "loss": 3.2767,
+      "loss": 3.2823,
       "step": 61850
     },
     {
       "epoch": 6.6738544474393535,
-      "grad_norm": 0.7368999123573303,
+      "grad_norm": 0.7854858040809631,
       "learning_rate": 0.00019993955747436588,
-      "loss": 3.274,
+      "loss": 3.2798,
       "step": 61900
     },
     {
       "epoch": 6.679245283018868,
-      "grad_norm": 0.7757512331008911,
+      "grad_norm": 0.8101821541786194,
       "learning_rate": 0.00019961575822989745,
-      "loss": 3.2747,
+      "loss": 3.2795,
       "step": 61950
     },
     {
       "epoch": 6.684636118598383,
-      "grad_norm": 0.7147890329360962,
+      "grad_norm": 0.7239598035812378,
       "learning_rate": 0.00019929195898542903,
-      "loss": 3.2998,
+      "loss": 3.3055,
       "step": 62000
     },
     {
       "epoch": 6.684636118598383,
-      "eval_accuracy": 0.38667759135236013,
-      "eval_loss": 3.3609557151794434,
-      "eval_runtime": 185.0014,
-      "eval_samples_per_second": 97.356,
-      "eval_steps_per_second": 6.086,
+      "eval_accuracy": 0.3858110845720396,
+      "eval_loss": 3.365727424621582,
+      "eval_runtime": 185.8358,
+      "eval_samples_per_second": 96.919,
+      "eval_steps_per_second": 6.059,
       "step": 62000
     },
     {
       "epoch": 6.690026954177897,
-      "grad_norm": 0.7304677367210388,
+      "grad_norm": 0.7538943290710449,
       "learning_rate": 0.0001989681597409606,
-      "loss": 3.2825,
+      "loss": 3.2865,
       "step": 62050
     },
     {
       "epoch": 6.695417789757412,
-      "grad_norm": 0.7841406464576721,
+      "grad_norm": 0.7734880447387695,
       "learning_rate": 0.00019864436049649216,
-      "loss": 3.2635,
+      "loss": 3.2683,
       "step": 62100
     },
     {
       "epoch": 6.7008086253369274,
-      "grad_norm": 0.7494974732398987,
+      "grad_norm": 0.7974568605422974,
       "learning_rate": 0.0001983205612520237,
-      "loss": 3.3044,
+      "loss": 3.3102,
       "step": 62150
     },
     {
       "epoch": 6.706199460916442,
-      "grad_norm": 0.7478502988815308,
+      "grad_norm": 0.7609190940856934,
       "learning_rate": 0.0001979967620075553,
-      "loss": 3.2954,
+      "loss": 3.3002,
       "step": 62200
     },
     {
       "epoch": 6.711590296495957,
-      "grad_norm": 0.7201170325279236,
+      "grad_norm": 0.7522209286689758,
       "learning_rate": 0.00019767296276308686,
-      "loss": 3.2798,
+      "loss": 3.2854,
       "step": 62250
     },
     {
       "epoch": 6.716981132075472,
-      "grad_norm": 0.7911655902862549,
+      "grad_norm": 0.7567285895347595,
       "learning_rate": 0.00019734916351861844,
-      "loss": 3.2812,
+      "loss": 3.2866,
       "step": 62300
     },
     {
       "epoch": 6.722371967654986,
-      "grad_norm": 0.8429829478263855,
+      "grad_norm": 0.819766104221344,
       "learning_rate": 0.00019702536427415002,
-      "loss": 3.3016,
+      "loss": 3.3088,
       "step": 62350
     },
     {
       "epoch": 6.727762803234501,
-      "grad_norm": 0.7752325534820557,
+      "grad_norm": 0.7971532940864563,
       "learning_rate": 0.0001967015650296816,
-      "loss": 3.2843,
+      "loss": 3.2892,
       "step": 62400
     },
     {
       "epoch": 6.7331536388140165,
-      "grad_norm": 0.7346392273902893,
+      "grad_norm": 0.779016375541687,
       "learning_rate": 0.00019637776578521315,
-      "loss": 3.2862,
+      "loss": 3.2908,
       "step": 62450
     },
     {
       "epoch": 6.738544474393531,
-      "grad_norm": 0.7346991300582886,
+      "grad_norm": 0.762880265712738,
       "learning_rate": 0.00019605396654074473,
-      "loss": 3.2908,
+      "loss": 3.2959,
       "step": 62500
     },
     {
       "epoch": 6.743935309973046,
-      "grad_norm": 0.7638818025588989,
+      "grad_norm": 0.783698320388794,
       "learning_rate": 0.0001957301672962763,
-      "loss": 3.2882,
+      "loss": 3.2927,
       "step": 62550
     },
     {
       "epoch": 6.74932614555256,
-      "grad_norm": 0.7539122104644775,
+      "grad_norm": 0.7661332488059998,
       "learning_rate": 0.00019540636805180785,
-      "loss": 3.2801,
+      "loss": 3.2859,
       "step": 62600
     },
     {
       "epoch": 6.754716981132075,
-      "grad_norm": 0.7578554153442383,
+      "grad_norm": 0.8325770497322083,
       "learning_rate": 0.00019508256880733943,
-      "loss": 3.2811,
+      "loss": 3.2861,
       "step": 62650
     },
     {
       "epoch": 6.7601078167115904,
-      "grad_norm": 0.7557489275932312,
+      "grad_norm": 0.8068282008171082,
       "learning_rate": 0.000194758769562871,
-      "loss": 3.2902,
+      "loss": 3.2958,
       "step": 62700
     },
     {
       "epoch": 6.765498652291106,
-      "grad_norm": 0.7585029006004333,
+      "grad_norm": 0.7944379448890686,
       "learning_rate": 0.00019443497031840256,
-      "loss": 3.2708,
+      "loss": 3.2744,
       "step": 62750
     },
     {
       "epoch": 6.77088948787062,
-      "grad_norm": 0.7600674629211426,
+      "grad_norm": 0.7637803554534912,
       "learning_rate": 0.00019411117107393414,
-      "loss": 3.2731,
+      "loss": 3.277,
       "step": 62800
     },
     {
       "epoch": 6.776280323450135,
-      "grad_norm": 0.7798412442207336,
+      "grad_norm": 0.8579869866371155,
       "learning_rate": 0.00019378737182946572,
-      "loss": 3.2889,
+      "loss": 3.2961,
       "step": 62850
     },
     {
       "epoch": 6.781671159029649,
-      "grad_norm": 0.8409514427185059,
+      "grad_norm": 0.8637164235115051,
       "learning_rate": 0.0001934635725849973,
-      "loss": 3.2864,
+      "loss": 3.2927,
       "step": 62900
     },
     {
       "epoch": 6.787061994609164,
-      "grad_norm": 0.7428311705589294,
-      "learning_rate": 0.00019313977334052887,
-      "loss": 3.3036,
+      "grad_norm": 0.7535831332206726,
+      "learning_rate": 0.00019314624932541822,
+      "loss": 3.3076,
       "step": 62950
     },
     {
       "epoch": 6.7924528301886795,
-      "grad_norm": 0.7958593964576721,
-      "learning_rate": 0.00019281597409606042,
-      "loss": 3.2773,
+      "grad_norm": 0.7850025296211243,
+      "learning_rate": 0.0001928224500809498,
+      "loss": 3.2823,
       "step": 63000
     },
     {
       "epoch": 6.7924528301886795,
-      "eval_accuracy": 0.38693759771290015,
-      "eval_loss": 3.3576364517211914,
-      "eval_runtime": 185.2969,
-      "eval_samples_per_second": 97.201,
-      "eval_steps_per_second": 6.077,
+      "eval_accuracy": 0.3859229970464885,
+      "eval_loss": 3.363136053085327,
+      "eval_runtime": 186.2459,
+      "eval_samples_per_second": 96.705,
+      "eval_steps_per_second": 6.046,
       "step": 63000
     },
     {
       "epoch": 6.797843665768194,
-      "grad_norm": 0.757053017616272,
-      "learning_rate": 0.00019249217485159197,
-      "loss": 3.291,
+      "grad_norm": 0.8071801662445068,
+      "learning_rate": 0.00019249865083648137,
+      "loss": 3.2953,
       "step": 63050
     },
     {
       "epoch": 6.803234501347709,
-      "grad_norm": 0.703620433807373,
-      "learning_rate": 0.00019216837560712355,
-      "loss": 3.2964,
+      "grad_norm": 0.7797868847846985,
+      "learning_rate": 0.00019217485159201292,
+      "loss": 3.3027,
       "step": 63100
     },
     {
       "epoch": 6.808625336927224,
-      "grad_norm": 0.7836263179779053,
-      "learning_rate": 0.00019184457636265513,
-      "loss": 3.2749,
+      "grad_norm": 0.8164698481559753,
+      "learning_rate": 0.0001918510523475445,
+      "loss": 3.2777,
       "step": 63150
     },
     {
       "epoch": 6.814016172506738,
-      "grad_norm": 0.7800487279891968,
-      "learning_rate": 0.0001915207771181867,
-      "loss": 3.2889,
+      "grad_norm": 0.7647949457168579,
+      "learning_rate": 0.00019152725310307608,
+      "loss": 3.2946,
       "step": 63200
     },
     {
       "epoch": 6.819407008086253,
-      "grad_norm": 0.7118497490882874,
-      "learning_rate": 0.00019119697787371828,
-      "loss": 3.2898,
+      "grad_norm": 0.7452571988105774,
+      "learning_rate": 0.00019120345385860766,
+      "loss": 3.2955,
       "step": 63250
     },
     {
       "epoch": 6.824797843665769,
-      "grad_norm": 0.7383244037628174,
-      "learning_rate": 0.00019087317862924986,
-      "loss": 3.2806,
+      "grad_norm": 0.7728260159492493,
+      "learning_rate": 0.00019087965461413923,
+      "loss": 3.2871,
       "step": 63300
     },
     {
       "epoch": 6.830188679245283,
-      "grad_norm": 0.8963047862052917,
-      "learning_rate": 0.00019054937938478144,
-      "loss": 3.2644,
+      "grad_norm": 0.7668321132659912,
+      "learning_rate": 0.00019055585536967079,
+      "loss": 3.2701,
       "step": 63350
     },
     {
       "epoch": 6.835579514824798,
-      "grad_norm": 0.7486133575439453,
-      "learning_rate": 0.00019022558014031302,
-      "loss": 3.295,
+      "grad_norm": 0.7463417053222656,
+      "learning_rate": 0.00019023205612520234,
+      "loss": 3.3004,
       "step": 63400
     },
     {
       "epoch": 6.840970350404312,
-      "grad_norm": 0.7660999894142151,
-      "learning_rate": 0.00018990178089584454,
-      "loss": 3.2999,
+      "grad_norm": 0.7865289449691772,
+      "learning_rate": 0.00018990825688073391,
+      "loss": 3.3048,
       "step": 63450
     },
     {
       "epoch": 6.846361185983827,
-      "grad_norm": 0.7520319223403931,
-      "learning_rate": 0.00018957798165137612,
-      "loss": 3.2928,
+      "grad_norm": 0.7706369757652283,
+      "learning_rate": 0.0001895844576362655,
+      "loss": 3.297,
       "step": 63500
     },
     {
       "epoch": 6.8517520215633425,
-      "grad_norm": 0.7498881220817566,
-      "learning_rate": 0.0001892541824069077,
-      "loss": 3.2875,
+      "grad_norm": 0.7762321829795837,
+      "learning_rate": 0.00018926065839179707,
+      "loss": 3.2919,
       "step": 63550
     },
     {
       "epoch": 6.857142857142857,
-      "grad_norm": 0.7376906275749207,
-      "learning_rate": 0.00018893038316243927,
-      "loss": 3.2675,
+      "grad_norm": 0.7470084428787231,
+      "learning_rate": 0.00018893685914732865,
+      "loss": 3.2721,
       "step": 63600
     },
     {
       "epoch": 6.862533692722372,
-      "grad_norm": 0.7770941853523254,
-      "learning_rate": 0.00018860658391797085,
-      "loss": 3.2695,
+      "grad_norm": 0.7886165380477905,
+      "learning_rate": 0.00018861305990286022,
+      "loss": 3.2759,
       "step": 63650
     },
     {
       "epoch": 6.867924528301887,
-      "grad_norm": 0.7603805661201477,
-      "learning_rate": 0.00018828278467350243,
-      "loss": 3.278,
+      "grad_norm": 0.7840895056724548,
+      "learning_rate": 0.0001882892606583918,
+      "loss": 3.282,
       "step": 63700
     },
     {
       "epoch": 6.873315363881401,
-      "grad_norm": 0.7244340181350708,
-      "learning_rate": 0.000187958985429034,
-      "loss": 3.2774,
+      "grad_norm": 0.7776980996131897,
+      "learning_rate": 0.00018796546141392333,
+      "loss": 3.2817,
       "step": 63750
     },
     {
       "epoch": 6.878706199460916,
-      "grad_norm": 0.8087451457977295,
-      "learning_rate": 0.00018763518618456556,
-      "loss": 3.2684,
+      "grad_norm": 0.8188679814338684,
+      "learning_rate": 0.0001876416621694549,
+      "loss": 3.2735,
       "step": 63800
     },
     {
       "epoch": 6.884097035040432,
-      "grad_norm": 0.7538607120513916,
-      "learning_rate": 0.0001873113869400971,
-      "loss": 3.2879,
+      "grad_norm": 0.7789130210876465,
+      "learning_rate": 0.00018731786292498648,
+      "loss": 3.2928,
       "step": 63850
     },
     {
       "epoch": 6.889487870619946,
-      "grad_norm": 0.7293424606323242,
-      "learning_rate": 0.00018698758769562868,
-      "loss": 3.2841,
+      "grad_norm": 0.7601881623268127,
+      "learning_rate": 0.00018699406368051806,
+      "loss": 3.2882,
       "step": 63900
     },
     {
       "epoch": 6.894878706199461,
-      "grad_norm": 0.8110401034355164,
-      "learning_rate": 0.00018666378845116026,
-      "loss": 3.302,
+      "grad_norm": 0.8311821818351746,
+      "learning_rate": 0.00018667026443604964,
+      "loss": 3.3062,
       "step": 63950
     },
     {
       "epoch": 6.900269541778976,
-      "grad_norm": 0.7437310814857483,
-      "learning_rate": 0.00018633998920669184,
-      "loss": 3.2882,
+      "grad_norm": 0.7588828802108765,
+      "learning_rate": 0.00018634646519158121,
+      "loss": 3.2918,
       "step": 64000
     },
     {
       "epoch": 6.900269541778976,
-      "eval_accuracy": 0.38752508387731305,
-      "eval_loss": 3.353229522705078,
-      "eval_runtime": 184.6958,
-      "eval_samples_per_second": 97.517,
-      "eval_steps_per_second": 6.097,
+      "eval_accuracy": 0.3866121823138958,
+      "eval_loss": 3.3591983318328857,
+      "eval_runtime": 185.487,
+      "eval_samples_per_second": 97.101,
+      "eval_steps_per_second": 6.071,
       "step": 64000
     },
     {
       "epoch": 6.90566037735849,
-      "grad_norm": 0.7894251346588135,
-      "learning_rate": 0.00018601618996222342,
-      "loss": 3.2894,
+      "grad_norm": 0.7953882217407227,
+      "learning_rate": 0.0001860226659471128,
+      "loss": 3.293,
       "step": 64050
     },
     {
       "epoch": 6.9110512129380055,
-      "grad_norm": 0.7877970933914185,
-      "learning_rate": 0.00018569239071775497,
-      "loss": 3.2938,
+      "grad_norm": 0.8116249442100525,
+      "learning_rate": 0.00018569886670264434,
+      "loss": 3.2978,
       "step": 64100
     },
     {
       "epoch": 6.916442048517521,
-      "grad_norm": 0.7726397514343262,
-      "learning_rate": 0.00018536859147328655,
-      "loss": 3.2889,
+      "grad_norm": 0.794157087802887,
+      "learning_rate": 0.00018537506745817592,
+      "loss": 3.2924,
       "step": 64150
     },
     {
       "epoch": 6.921832884097035,
-      "grad_norm": 0.7730385065078735,
-      "learning_rate": 0.00018504479222881812,
-      "loss": 3.2796,
+      "grad_norm": 0.7764679789543152,
+      "learning_rate": 0.00018505126821370747,
+      "loss": 3.285,
       "step": 64200
     },
     {
       "epoch": 6.92722371967655,
-      "grad_norm": 0.7642452716827393,
-      "learning_rate": 0.0001847209929843497,
-      "loss": 3.264,
+      "grad_norm": 0.7788585424423218,
+      "learning_rate": 0.00018472746896923905,
+      "loss": 3.2677,
       "step": 64250
     },
     {
       "epoch": 6.932614555256064,
-      "grad_norm": 0.7660939693450928,
-      "learning_rate": 0.00018439719373988125,
-      "loss": 3.306,
+      "grad_norm": 0.7901521325111389,
+      "learning_rate": 0.00018440366972477063,
+      "loss": 3.3108,
       "step": 64300
     },
     {
       "epoch": 6.938005390835579,
-      "grad_norm": 0.7493078112602234,
-      "learning_rate": 0.00018407339449541283,
-      "loss": 3.2839,
+      "grad_norm": 0.8074705004692078,
+      "learning_rate": 0.0001840798704803022,
+      "loss": 3.2881,
       "step": 64350
     },
     {
       "epoch": 6.943396226415095,
-      "grad_norm": 0.7475608587265015,
-      "learning_rate": 0.00018374959525094438,
-      "loss": 3.2844,
+      "grad_norm": 0.7702144384384155,
+      "learning_rate": 0.00018375607123583378,
+      "loss": 3.2889,
       "step": 64400
     },
     {
       "epoch": 6.948787061994609,
-      "grad_norm": 0.7330486178398132,
-      "learning_rate": 0.00018342579600647596,
-      "loss": 3.2808,
+      "grad_norm": 0.7345027923583984,
+      "learning_rate": 0.00018343227199136533,
+      "loss": 3.2853,
       "step": 64450
     },
     {
       "epoch": 6.954177897574124,
-      "grad_norm": 1.2176107168197632,
-      "learning_rate": 0.00018310199676200753,
-      "loss": 3.2776,
+      "grad_norm": 0.7992977499961853,
+      "learning_rate": 0.0001831084727468969,
+      "loss": 3.2799,
       "step": 64500
     },
     {
       "epoch": 6.959568733153639,
-      "grad_norm": 0.710422933101654,
-      "learning_rate": 0.0001827781975175391,
-      "loss": 3.2817,
+      "grad_norm": 0.7189365029335022,
+      "learning_rate": 0.0001827846735024285,
+      "loss": 3.2869,
       "step": 64550
     },
     {
       "epoch": 6.964959568733153,
-      "grad_norm": 0.7552564144134521,
-      "learning_rate": 0.0001824543982730707,
-      "loss": 3.2731,
+      "grad_norm": 0.788296103477478,
+      "learning_rate": 0.00018246087425796004,
+      "loss": 3.2776,
       "step": 64600
     },
     {
       "epoch": 6.9703504043126685,
-      "grad_norm": 0.7585906386375427,
-      "learning_rate": 0.00018213059902860227,
-      "loss": 3.2843,
+      "grad_norm": 0.7620663642883301,
+      "learning_rate": 0.00018213707501349162,
+      "loss": 3.2895,
       "step": 64650
     },
     {
       "epoch": 6.975741239892184,
-      "grad_norm": 0.7237086892127991,
-      "learning_rate": 0.00018180679978413382,
-      "loss": 3.2802,
+      "grad_norm": 0.7858600616455078,
+      "learning_rate": 0.0001818132757690232,
+      "loss": 3.2845,
       "step": 64700
     },
     {
       "epoch": 6.981132075471698,
-      "grad_norm": 0.8057036995887756,
-      "learning_rate": 0.00018148300053966537,
-      "loss": 3.2895,
+      "grad_norm": 0.7973349094390869,
+      "learning_rate": 0.00018148947652455474,
+      "loss": 3.296,
       "step": 64750
     },
     {
       "epoch": 6.986522911051213,
-      "grad_norm": 0.7535995244979858,
+      "grad_norm": 0.7475130558013916,
       "learning_rate": 0.00018116567728008632,
-      "loss": 3.277,
+      "loss": 3.2835,
       "step": 64800
     },
     {
       "epoch": 6.991913746630727,
-      "grad_norm": 0.7500599026679993,
+      "grad_norm": 0.7695804834365845,
       "learning_rate": 0.0001808418780356179,
-      "loss": 3.2713,
+      "loss": 3.2782,
       "step": 64850
     },
     {
       "epoch": 6.997304582210242,
-      "grad_norm": 0.8092964291572571,
+      "grad_norm": 0.8086309432983398,
       "learning_rate": 0.00018051807879114948,
-      "loss": 3.2817,
+      "loss": 3.2881,
       "step": 64900
     },
     {
       "epoch": 7.002695417789758,
-      "grad_norm": 0.7631925940513611,
+      "grad_norm": 0.8140087723731995,
       "learning_rate": 0.0001802007555315704,
-      "loss": 3.2317,
+      "loss": 3.2371,
       "step": 64950
     },
     {
       "epoch": 7.008086253369272,
-      "grad_norm": 0.7095732092857361,
+      "grad_norm": 0.763765811920166,
       "learning_rate": 0.00017987695628710198,
-      "loss": 3.1839,
+      "loss": 3.187,
       "step": 65000
     },
     {
       "epoch": 7.008086253369272,
-      "eval_accuracy": 0.3875124801423266,
-      "eval_loss": 3.354387044906616,
-      "eval_runtime": 184.7817,
-      "eval_samples_per_second": 97.472,
-      "eval_steps_per_second": 6.094,
+      "eval_accuracy": 0.3865791518360002,
+      "eval_loss": 3.360050678253174,
+      "eval_runtime": 186.6484,
+      "eval_samples_per_second": 96.497,
+      "eval_steps_per_second": 6.033,
       "step": 65000
     },
     {
       "epoch": 7.013477088948787,
-      "grad_norm": 0.9158305525779724,
+      "grad_norm": 0.8818866610527039,
       "learning_rate": 0.00017955315704263356,
-      "loss": 3.2062,
+      "loss": 3.2108,
       "step": 65050
     },
     {
       "epoch": 7.018867924528302,
-      "grad_norm": 0.7549683451652527,
+      "grad_norm": 0.819786548614502,
       "learning_rate": 0.0001792293577981651,
-      "loss": 3.2018,
+      "loss": 3.2072,
       "step": 65100
     },
     {
       "epoch": 7.024258760107816,
-      "grad_norm": 0.7751328349113464,
+      "grad_norm": 0.7957449555397034,
       "learning_rate": 0.00017890555855369668,
-      "loss": 3.2119,
+      "loss": 3.2149,
       "step": 65150
     },
     {
       "epoch": 7.0296495956873315,
-      "grad_norm": 0.7429438829421997,
+      "grad_norm": 0.7758747339248657,
       "learning_rate": 0.00017858175930922826,
-      "loss": 3.2053,
+      "loss": 3.2107,
       "step": 65200
     },
     {
       "epoch": 7.035040431266847,
-      "grad_norm": 0.7653865218162537,
+      "grad_norm": 0.8405669331550598,
       "learning_rate": 0.00017825796006475984,
-      "loss": 3.2041,
+      "loss": 3.2103,
       "step": 65250
     },
     {
       "epoch": 7.040431266846361,
-      "grad_norm": 0.7375260591506958,
+      "grad_norm": 0.8363875150680542,
       "learning_rate": 0.00017793416082029142,
-      "loss": 3.1978,
+      "loss": 3.2034,
       "step": 65300
     },
     {
       "epoch": 7.045822102425876,
-      "grad_norm": 0.7559249997138977,
+      "grad_norm": 0.7539160251617432,
       "learning_rate": 0.000177610361575823,
-      "loss": 3.2147,
+      "loss": 3.2192,
       "step": 65350
     },
     {
       "epoch": 7.051212938005391,
-      "grad_norm": 0.7759532332420349,
+      "grad_norm": 0.7640693187713623,
       "learning_rate": 0.00017728656233135452,
-      "loss": 3.1921,
+      "loss": 3.199,
       "step": 65400
     },
     {
       "epoch": 7.056603773584905,
-      "grad_norm": 0.7644091844558716,
+      "grad_norm": 0.775389552116394,
       "learning_rate": 0.0001769627630868861,
-      "loss": 3.1814,
+      "loss": 3.1876,
       "step": 65450
     },
     {
       "epoch": 7.061994609164421,
-      "grad_norm": 0.7937024235725403,
+      "grad_norm": 0.7880697250366211,
       "learning_rate": 0.00017663896384241767,
-      "loss": 3.1995,
+      "loss": 3.205,
       "step": 65500
     },
     {
       "epoch": 7.067385444743936,
-      "grad_norm": 0.7552359104156494,
+      "grad_norm": 0.7930703163146973,
       "learning_rate": 0.00017631516459794925,
-      "loss": 3.1924,
+      "loss": 3.1946,
       "step": 65550
     },
     {
       "epoch": 7.07277628032345,
-      "grad_norm": 0.7388563752174377,
+      "grad_norm": 0.8107736110687256,
       "learning_rate": 0.00017599136535348083,
-      "loss": 3.2205,
+      "loss": 3.2241,
       "step": 65600
     },
     {
       "epoch": 7.078167115902965,
-      "grad_norm": 0.7597045302391052,
+      "grad_norm": 0.7506124377250671,
       "learning_rate": 0.0001756675661090124,
-      "loss": 3.2121,
+      "loss": 3.2173,
       "step": 65650
     },
     {
       "epoch": 7.083557951482479,
-      "grad_norm": 0.8127196431159973,
+      "grad_norm": 0.8445967435836792,
       "learning_rate": 0.00017534376686454398,
-      "loss": 3.2061,
+      "loss": 3.2112,
       "step": 65700
     },
     {
       "epoch": 7.0889487870619945,
-      "grad_norm": 0.7847875356674194,
+      "grad_norm": 0.8482174873352051,
       "learning_rate": 0.00017501996762007556,
-      "loss": 3.1891,
+      "loss": 3.1935,
       "step": 65750
     },
     {
       "epoch": 7.09433962264151,
-      "grad_norm": 0.7454573512077332,
+      "grad_norm": 0.7698912620544434,
       "learning_rate": 0.00017469616837560709,
-      "loss": 3.2224,
+      "loss": 3.2249,
       "step": 65800
     },
     {
       "epoch": 7.099730458221024,
-      "grad_norm": 0.739266574382782,
+      "grad_norm": 0.8019884824752808,
       "learning_rate": 0.00017437236913113866,
-      "loss": 3.2168,
+      "loss": 3.2234,
       "step": 65850
     },
     {
       "epoch": 7.105121293800539,
-      "grad_norm": 0.7397430539131165,
+      "grad_norm": 0.7337784767150879,
       "learning_rate": 0.00017404856988667024,
-      "loss": 3.2299,
+      "loss": 3.2342,
       "step": 65900
     },
     {
       "epoch": 7.110512129380054,
-      "grad_norm": 0.7771823406219482,
+      "grad_norm": 0.7504554986953735,
       "learning_rate": 0.00017372477064220182,
-      "loss": 3.2127,
+      "loss": 3.2179,
       "step": 65950
     },
     {
       "epoch": 7.115902964959568,
-      "grad_norm": 0.8010558485984802,
+      "grad_norm": 0.8045883774757385,
       "learning_rate": 0.0001734009713977334,
-      "loss": 3.2252,
+      "loss": 3.2306,
       "step": 66000
     },
     {
       "epoch": 7.115902964959568,
-      "eval_accuracy": 0.3877516251484063,
-      "eval_loss": 3.3572838306427,
-      "eval_runtime": 184.947,
-      "eval_samples_per_second": 97.385,
-      "eval_steps_per_second": 6.088,
+      "eval_accuracy": 0.38693835828311485,
+      "eval_loss": 3.3626465797424316,
+      "eval_runtime": 186.9971,
+      "eval_samples_per_second": 96.317,
+      "eval_steps_per_second": 6.021,
       "step": 66000
     },
     {
       "epoch": 7.121293800539084,
-      "grad_norm": 0.7524279952049255,
+      "grad_norm": 0.8456098437309265,
       "learning_rate": 0.00017307717215326497,
-      "loss": 3.2191,
+      "loss": 3.2233,
       "step": 66050
     },
     {
       "epoch": 7.126684636118599,
-      "grad_norm": 0.7754007577896118,
+      "grad_norm": 0.8226268291473389,
       "learning_rate": 0.00017275337290879655,
-      "loss": 3.2188,
+      "loss": 3.2241,
       "step": 66100
     },
     {
       "epoch": 7.132075471698113,
-      "grad_norm": 0.8016265630722046,
+      "grad_norm": 0.8094230890274048,
       "learning_rate": 0.0001724295736643281,
-      "loss": 3.2193,
+      "loss": 3.2227,
       "step": 66150
     },
     {
       "epoch": 7.137466307277628,
-      "grad_norm": 0.7195385098457336,
+      "grad_norm": 0.7555733323097229,
       "learning_rate": 0.00017210577441985968,
-      "loss": 3.2144,
+      "loss": 3.2196,
       "step": 66200
     },
     {
       "epoch": 7.142857142857143,
-      "grad_norm": 0.7985889911651611,
+      "grad_norm": 0.7971102595329285,
       "learning_rate": 0.00017178197517539123,
-      "loss": 3.2224,
+      "loss": 3.2266,
       "step": 66250
     },
     {
       "epoch": 7.1482479784366575,
-      "grad_norm": 0.7573408484458923,
+      "grad_norm": 0.7829456329345703,
       "learning_rate": 0.0001714581759309228,
-      "loss": 3.2186,
+      "loss": 3.2225,
       "step": 66300
     },
     {
       "epoch": 7.153638814016173,
-      "grad_norm": 0.8037809729576111,
+      "grad_norm": 0.7997987866401672,
       "learning_rate": 0.00017113437668645439,
-      "loss": 3.2227,
+      "loss": 3.2304,
       "step": 66350
     },
     {
       "epoch": 7.159029649595688,
-      "grad_norm": 0.7795762419700623,
+      "grad_norm": 0.7970221638679504,
       "learning_rate": 0.00017081057744198596,
-      "loss": 3.2077,
+      "loss": 3.2125,
       "step": 66400
     },
     {
       "epoch": 7.164420485175202,
-      "grad_norm": 0.7938854098320007,
+      "grad_norm": 0.8575058579444885,
       "learning_rate": 0.00017048677819751751,
-      "loss": 3.2172,
+      "loss": 3.2224,
       "step": 66450
     },
     {
       "epoch": 7.169811320754717,
-      "grad_norm": 0.7674042582511902,
+      "grad_norm": 0.8168599009513855,
       "learning_rate": 0.0001701629789530491,
-      "loss": 3.2177,
+      "loss": 3.2232,
       "step": 66500
     },
     {
       "epoch": 7.175202156334231,
-      "grad_norm": 0.782508134841919,
+      "grad_norm": 0.7927587032318115,
       "learning_rate": 0.00016983917970858067,
-      "loss": 3.2215,
+      "loss": 3.226,
       "step": 66550
     },
     {
       "epoch": 7.180592991913747,
-      "grad_norm": 0.7263006567955017,
+      "grad_norm": 0.7701125741004944,
       "learning_rate": 0.00016951538046411225,
-      "loss": 3.2079,
+      "loss": 3.2134,
       "step": 66600
     },
     {
       "epoch": 7.185983827493262,
-      "grad_norm": 0.7729055881500244,
+      "grad_norm": 0.8022675514221191,
       "learning_rate": 0.0001691915812196438,
-      "loss": 3.215,
+      "loss": 3.221,
       "step": 66650
     },
     {
       "epoch": 7.191374663072776,
-      "grad_norm": 0.7278570532798767,
+      "grad_norm": 0.7721037864685059,
       "learning_rate": 0.00016886778197517538,
-      "loss": 3.2179,
+      "loss": 3.222,
       "step": 66700
     },
     {
       "epoch": 7.196765498652291,
-      "grad_norm": 0.7499024271965027,
+      "grad_norm": 0.8144177794456482,
       "learning_rate": 0.00016854398273070693,
-      "loss": 3.2216,
+      "loss": 3.2264,
       "step": 66750
     },
     {
       "epoch": 7.202156334231806,
-      "grad_norm": 0.8059422373771667,
+      "grad_norm": 0.771833062171936,
       "learning_rate": 0.0001682201834862385,
-      "loss": 3.211,
+      "loss": 3.2167,
       "step": 66800
     },
     {
       "epoch": 7.2075471698113205,
-      "grad_norm": 0.7714557647705078,
+      "grad_norm": 0.7803823351860046,
       "learning_rate": 0.00016789638424177008,
-      "loss": 3.2301,
+      "loss": 3.2348,
       "step": 66850
     },
     {
       "epoch": 7.212938005390836,
-      "grad_norm": 0.7622218728065491,
+      "grad_norm": 0.8095530867576599,
       "learning_rate": 0.00016757258499730166,
-      "loss": 3.2388,
+      "loss": 3.2439,
       "step": 66900
     },
     {
       "epoch": 7.218328840970351,
-      "grad_norm": 0.7789103984832764,
+      "grad_norm": 0.8291968107223511,
       "learning_rate": 0.00016724878575283324,
-      "loss": 3.227,
+      "loss": 3.2322,
       "step": 66950
     },
     {
       "epoch": 7.223719676549865,
-      "grad_norm": 0.7892781496047974,
-      "learning_rate": 0.00016692498650836481,
-      "loss": 3.2308,
+      "grad_norm": 0.7933746576309204,
+      "learning_rate": 0.00016693146249325416,
+      "loss": 3.2341,
       "step": 67000
     },
     {
       "epoch": 7.223719676549865,
-      "eval_accuracy": 0.388054006135194,
-      "eval_loss": 3.3542799949645996,
-      "eval_runtime": 184.5856,
-      "eval_samples_per_second": 97.575,
-      "eval_steps_per_second": 6.1,
+      "eval_accuracy": 0.3871245893328289,
+      "eval_loss": 3.360367774963379,
+      "eval_runtime": 186.5559,
+      "eval_samples_per_second": 96.545,
+      "eval_steps_per_second": 6.036,
       "step": 67000
     },
     {
       "epoch": 7.22911051212938,
-      "grad_norm": 0.772681474685669,
-      "learning_rate": 0.0001666011872638964,
-      "loss": 3.1994,
+      "grad_norm": 0.7940529584884644,
+      "learning_rate": 0.00016660766324878574,
+      "loss": 3.2052,
       "step": 67050
     },
     {
       "epoch": 7.234501347708895,
-      "grad_norm": 0.7405024766921997,
-      "learning_rate": 0.00016627738801942792,
-      "loss": 3.2434,
+      "grad_norm": 0.7628929615020752,
+      "learning_rate": 0.0001662838640043173,
+      "loss": 3.2478,
       "step": 67100
     },
     {
       "epoch": 7.2398921832884096,
-      "grad_norm": 0.7997522354125977,
+      "grad_norm": 0.7867326140403748,
       "learning_rate": 0.00016596006475984887,
-      "loss": 3.2065,
+      "loss": 3.211,
       "step": 67150
     },
     {
       "epoch": 7.245283018867925,
-      "grad_norm": 0.8065983057022095,
+      "grad_norm": 0.8237667679786682,
       "learning_rate": 0.00016563626551538045,
-      "loss": 3.2268,
+      "loss": 3.2299,
       "step": 67200
     },
     {
       "epoch": 7.250673854447439,
-      "grad_norm": 0.7845258116722107,
+      "grad_norm": 0.7866894602775574,
       "learning_rate": 0.00016531246627091202,
-      "loss": 3.2351,
+      "loss": 3.2405,
       "step": 67250
     },
     {
       "epoch": 7.256064690026954,
-      "grad_norm": 0.7305331826210022,
+      "grad_norm": 0.7794045805931091,
       "learning_rate": 0.0001649886670264436,
-      "loss": 3.237,
+      "loss": 3.2427,
       "step": 67300
     },
     {
       "epoch": 7.261455525606469,
-      "grad_norm": 0.7165262699127197,
+      "grad_norm": 0.7607665657997131,
       "learning_rate": 0.00016466486778197518,
-      "loss": 3.2162,
+      "loss": 3.2209,
       "step": 67350
     },
     {
       "epoch": 7.2668463611859835,
-      "grad_norm": 0.7781831622123718,
+      "grad_norm": 0.7799270153045654,
       "learning_rate": 0.0001643410685375067,
-      "loss": 3.2421,
+      "loss": 3.2466,
       "step": 67400
     },
     {
       "epoch": 7.272237196765499,
-      "grad_norm": 0.8223204016685486,
+      "grad_norm": 0.8593279123306274,
       "learning_rate": 0.00016401726929303828,
-      "loss": 3.2393,
+      "loss": 3.2451,
       "step": 67450
     },
     {
       "epoch": 7.277628032345014,
-      "grad_norm": 0.7840092778205872,
+      "grad_norm": 0.7918068170547485,
       "learning_rate": 0.00016369347004856986,
-      "loss": 3.229,
+      "loss": 3.2333,
       "step": 67500
     },
     {
       "epoch": 7.283018867924528,
-      "grad_norm": 0.7795689702033997,
+      "grad_norm": 0.8057292103767395,
       "learning_rate": 0.00016336967080410143,
-      "loss": 3.2406,
+      "loss": 3.2436,
       "step": 67550
     },
     {
       "epoch": 7.288409703504043,
-      "grad_norm": 0.8028538823127747,
+      "grad_norm": 0.8262519240379333,
       "learning_rate": 0.000163045871559633,
-      "loss": 3.2256,
+      "loss": 3.2315,
       "step": 67600
     },
     {
       "epoch": 7.293800539083558,
-      "grad_norm": 0.7559919953346252,
+      "grad_norm": 0.836014986038208,
       "learning_rate": 0.0001627220723151646,
-      "loss": 3.228,
+      "loss": 3.2335,
       "step": 67650
     },
     {
       "epoch": 7.2991913746630726,
-      "grad_norm": 0.7705023884773254,
+      "grad_norm": 0.8312990665435791,
       "learning_rate": 0.00016239827307069617,
-      "loss": 3.2099,
+      "loss": 3.2138,
       "step": 67700
     },
     {
       "epoch": 7.304582210242588,
-      "grad_norm": 0.7943682670593262,
+      "grad_norm": 0.8154436349868774,
       "learning_rate": 0.00016207447382622775,
-      "loss": 3.2247,
+      "loss": 3.2299,
       "step": 67750
     },
     {
       "epoch": 7.309973045822103,
-      "grad_norm": 0.7587892413139343,
+      "grad_norm": 0.7973268628120422,
       "learning_rate": 0.00016175067458175932,
-      "loss": 3.2219,
+      "loss": 3.2276,
       "step": 67800
     },
     {
       "epoch": 7.315363881401617,
-      "grad_norm": 0.8514966368675232,
+      "grad_norm": 0.8404545187950134,
       "learning_rate": 0.00016142687533729085,
-      "loss": 3.2224,
+      "loss": 3.2248,
       "step": 67850
     },
     {
       "epoch": 7.320754716981132,
-      "grad_norm": 0.7471888065338135,
+      "grad_norm": 0.8025550246238708,
       "learning_rate": 0.00016110307609282242,
-      "loss": 3.237,
+      "loss": 3.2427,
       "step": 67900
     },
     {
       "epoch": 7.3261455525606465,
-      "grad_norm": 0.7487116456031799,
+      "grad_norm": 0.7733309268951416,
       "learning_rate": 0.000160779276848354,
-      "loss": 3.2122,
+      "loss": 3.2186,
       "step": 67950
     },
     {
       "epoch": 7.331536388140162,
-      "grad_norm": 0.8473713397979736,
+      "grad_norm": 0.848850667476654,
       "learning_rate": 0.00016045547760388558,
-      "loss": 3.2282,
+      "loss": 3.2319,
       "step": 68000
     },
     {
       "epoch": 7.331536388140162,
-      "eval_accuracy": 0.3882555572420898,
-      "eval_loss": 3.352534532546997,
-      "eval_runtime": 184.9997,
-      "eval_samples_per_second": 97.357,
-      "eval_steps_per_second": 6.086,
+      "eval_accuracy": 0.38735754112430315,
+      "eval_loss": 3.356987714767456,
+      "eval_runtime": 186.7755,
+      "eval_samples_per_second": 96.431,
+      "eval_steps_per_second": 6.029,
       "step": 68000
     },
     {
       "epoch": 7.336927223719677,
-      "grad_norm": 0.7282635569572449,
+      "grad_norm": 0.7472612261772156,
       "learning_rate": 0.00016013167835941716,
-      "loss": 3.2332,
+      "loss": 3.2364,
       "step": 68050
     },
     {
       "epoch": 7.342318059299191,
-      "grad_norm": 0.7940888404846191,
+      "grad_norm": 0.8511797189712524,
       "learning_rate": 0.00015980787911494873,
-      "loss": 3.2392,
+      "loss": 3.2425,
       "step": 68100
     },
     {
       "epoch": 7.347708894878706,
-      "grad_norm": 0.8099002242088318,
+      "grad_norm": 0.8311546444892883,
       "learning_rate": 0.00015948407987048029,
-      "loss": 3.2271,
+      "loss": 3.2313,
       "step": 68150
     },
     {
       "epoch": 7.353099730458221,
-      "grad_norm": 0.7751094102859497,
+      "grad_norm": 0.8016986846923828,
       "learning_rate": 0.00015916028062601186,
-      "loss": 3.2459,
+      "loss": 3.2514,
       "step": 68200
     },
     {
       "epoch": 7.3584905660377355,
-      "grad_norm": 0.7973043918609619,
+      "grad_norm": 0.7910669445991516,
       "learning_rate": 0.00015883648138154344,
-      "loss": 3.2324,
+      "loss": 3.238,
       "step": 68250
     },
     {
       "epoch": 7.363881401617251,
-      "grad_norm": 0.787239670753479,
+      "grad_norm": 0.7893770337104797,
       "learning_rate": 0.000158512682137075,
-      "loss": 3.2199,
+      "loss": 3.2246,
       "step": 68300
     },
     {
       "epoch": 7.369272237196766,
-      "grad_norm": 0.7251448631286621,
+      "grad_norm": 0.7780228853225708,
       "learning_rate": 0.00015818888289260657,
-      "loss": 3.2329,
+      "loss": 3.2366,
       "step": 68350
     },
     {
       "epoch": 7.37466307277628,
-      "grad_norm": 0.7860444188117981,
+      "grad_norm": 0.8126387596130371,
       "learning_rate": 0.00015786508364813815,
-      "loss": 3.225,
+      "loss": 3.2298,
       "step": 68400
     },
     {
       "epoch": 7.380053908355795,
-      "grad_norm": 0.8149095773696899,
+      "grad_norm": 0.854413628578186,
       "learning_rate": 0.0001575412844036697,
-      "loss": 3.2095,
+      "loss": 3.2146,
       "step": 68450
     },
     {
       "epoch": 7.38544474393531,
-      "grad_norm": 0.8455824255943298,
+      "grad_norm": 0.8384074568748474,
       "learning_rate": 0.00015721748515920127,
-      "loss": 3.2296,
+      "loss": 3.2344,
       "step": 68500
     },
     {
       "epoch": 7.390835579514825,
-      "grad_norm": 0.7530245780944824,
+      "grad_norm": 0.7914549112319946,
       "learning_rate": 0.00015689368591473285,
-      "loss": 3.2297,
+      "loss": 3.2363,
       "step": 68550
     },
     {
       "epoch": 7.39622641509434,
-      "grad_norm": 0.7616867423057556,
+      "grad_norm": 0.7769351601600647,
       "learning_rate": 0.00015656988667026443,
-      "loss": 3.2266,
+      "loss": 3.2326,
       "step": 68600
     },
     {
       "epoch": 7.401617250673855,
-      "grad_norm": 0.7562954425811768,
+      "grad_norm": 0.7804869413375854,
       "learning_rate": 0.000156246087425796,
-      "loss": 3.2275,
+      "loss": 3.2307,
       "step": 68650
     },
     {
       "epoch": 7.407008086253369,
-      "grad_norm": 0.7328892350196838,
+      "grad_norm": 0.747808575630188,
       "learning_rate": 0.00015592228818132756,
-      "loss": 3.2381,
+      "loss": 3.2423,
       "step": 68700
     },
     {
       "epoch": 7.412398921832884,
-      "grad_norm": 0.8182516694068909,
+      "grad_norm": 0.8189519047737122,
       "learning_rate": 0.0001555984889368591,
-      "loss": 3.2225,
+      "loss": 3.2273,
       "step": 68750
     },
     {
       "epoch": 7.4177897574123985,
-      "grad_norm": 0.8202837109565735,
+      "grad_norm": 0.790206253528595,
       "learning_rate": 0.0001552746896923907,
-      "loss": 3.2338,
+      "loss": 3.2398,
       "step": 68800
     },
     {
       "epoch": 7.423180592991914,
-      "grad_norm": 0.775603711605072,
+      "grad_norm": 0.7983890175819397,
       "learning_rate": 0.00015495089044792226,
-      "loss": 3.2351,
+      "loss": 3.2414,
       "step": 68850
     },
     {
       "epoch": 7.428571428571429,
-      "grad_norm": 0.776694118976593,
+      "grad_norm": 0.782149076461792,
       "learning_rate": 0.00015462709120345384,
-      "loss": 3.23,
+      "loss": 3.2356,
       "step": 68900
     },
     {
       "epoch": 7.433962264150943,
-      "grad_norm": 0.786277711391449,
+      "grad_norm": 0.7900369763374329,
       "learning_rate": 0.00015430329195898542,
-      "loss": 3.2334,
+      "loss": 3.2367,
       "step": 68950
     },
     {
       "epoch": 7.439353099730458,
-      "grad_norm": 0.7692110538482666,
+      "grad_norm": 0.782538652420044,
       "learning_rate": 0.000153979492714517,
-      "loss": 3.2182,
+      "loss": 3.224,
       "step": 69000
     },
     {
       "epoch": 7.439353099730458,
-      "eval_accuracy": 0.388615632912307,
-      "eval_loss": 3.346757173538208,
-      "eval_runtime": 184.8049,
-      "eval_samples_per_second": 97.46,
-      "eval_steps_per_second": 6.093,
+      "eval_accuracy": 0.387778136453033,
+      "eval_loss": 3.3513643741607666,
+      "eval_runtime": 186.9452,
+      "eval_samples_per_second": 96.344,
+      "eval_steps_per_second": 6.023,
       "step": 69000
     },
     {
       "epoch": 7.444743935309973,
-      "grad_norm": 0.8124752640724182,
-      "learning_rate": 0.00015365569347004858,
-      "loss": 3.2203,
+      "grad_norm": 0.8281074166297913,
+      "learning_rate": 0.00015366216945493792,
+      "loss": 3.2243,
       "step": 69050
     },
     {
       "epoch": 7.450134770889488,
-      "grad_norm": 0.7867438793182373,
-      "learning_rate": 0.00015333189422558015,
-      "loss": 3.2298,
+      "grad_norm": 0.8465034365653992,
+      "learning_rate": 0.00015333837021046947,
+      "loss": 3.2335,
       "step": 69100
     },
     {
       "epoch": 7.455525606469003,
-      "grad_norm": 0.7936617136001587,
-      "learning_rate": 0.00015300809498111168,
-      "loss": 3.2292,
+      "grad_norm": 0.8020473122596741,
+      "learning_rate": 0.00015301457096600105,
+      "loss": 3.2337,
       "step": 69150
     },
     {
       "epoch": 7.460916442048518,
-      "grad_norm": 0.776745617389679,
-      "learning_rate": 0.00015268429573664325,
-      "loss": 3.2207,
+      "grad_norm": 0.7596130967140198,
+      "learning_rate": 0.00015269077172153263,
+      "loss": 3.2251,
       "step": 69200
     },
     {
       "epoch": 7.466307277628032,
-      "grad_norm": 0.798880934715271,
-      "learning_rate": 0.00015236049649217483,
-      "loss": 3.2378,
+      "grad_norm": 0.8672736287117004,
+      "learning_rate": 0.0001523669724770642,
+      "loss": 3.2425,
       "step": 69250
     },
     {
       "epoch": 7.471698113207547,
-      "grad_norm": 0.7821260690689087,
-      "learning_rate": 0.0001520366972477064,
-      "loss": 3.2253,
+      "grad_norm": 0.8052238821983337,
+      "learning_rate": 0.00015204317323259578,
+      "loss": 3.231,
       "step": 69300
     },
     {
       "epoch": 7.4770889487870615,
-      "grad_norm": 0.767823338508606,
-      "learning_rate": 0.000151712898003238,
-      "loss": 3.2376,
+      "grad_norm": 0.8050634860992432,
+      "learning_rate": 0.00015171937398812736,
+      "loss": 3.2423,
       "step": 69350
     },
     {
       "epoch": 7.482479784366577,
-      "grad_norm": 0.7894257307052612,
-      "learning_rate": 0.00015138909875876956,
-      "loss": 3.2231,
+      "grad_norm": 0.8076505064964294,
+      "learning_rate": 0.00015139557474365894,
+      "loss": 3.2262,
       "step": 69400
     },
     {
       "epoch": 7.487870619946092,
-      "grad_norm": 0.8050634860992432,
-      "learning_rate": 0.00015106529951430114,
-      "loss": 3.2228,
+      "grad_norm": 0.8427257537841797,
+      "learning_rate": 0.00015107177549919046,
+      "loss": 3.2278,
       "step": 69450
     },
     {
       "epoch": 7.493261455525606,
-      "grad_norm": 0.7465404868125916,
-      "learning_rate": 0.0001507415002698327,
-      "loss": 3.2237,
+      "grad_norm": 0.8119729161262512,
+      "learning_rate": 0.00015074797625472204,
+      "loss": 3.2298,
       "step": 69500
     },
     {
       "epoch": 7.498652291105121,
-      "grad_norm": 0.8405658006668091,
-      "learning_rate": 0.00015041770102536424,
-      "loss": 3.2376,
+      "grad_norm": 0.8911359310150146,
+      "learning_rate": 0.00015042417701025362,
+      "loss": 3.2424,
       "step": 69550
     },
     {
       "epoch": 7.504043126684636,
-      "grad_norm": 0.7720454335212708,
-      "learning_rate": 0.00015009390178089582,
-      "loss": 3.2108,
+      "grad_norm": 0.8080430030822754,
+      "learning_rate": 0.0001501003777657852,
+      "loss": 3.2146,
       "step": 69600
     },
     {
       "epoch": 7.509433962264151,
-      "grad_norm": 0.7862969636917114,
-      "learning_rate": 0.0001497701025364274,
-      "loss": 3.2289,
+      "grad_norm": 0.8010596632957458,
+      "learning_rate": 0.00014977657852131677,
+      "loss": 3.2353,
       "step": 69650
     },
     {
       "epoch": 7.514824797843666,
-      "grad_norm": 0.761938214302063,
-      "learning_rate": 0.00014944630329195898,
-      "loss": 3.238,
+      "grad_norm": 0.7891853451728821,
+      "learning_rate": 0.00014945277927684835,
+      "loss": 3.2431,
       "step": 69700
     },
     {
       "epoch": 7.520215633423181,
-      "grad_norm": 0.8123347163200378,
-      "learning_rate": 0.00014912250404749055,
-      "loss": 3.244,
+      "grad_norm": 0.8413270711898804,
+      "learning_rate": 0.00014912898003237993,
+      "loss": 3.2482,
       "step": 69750
     },
     {
       "epoch": 7.525606469002695,
-      "grad_norm": 0.8718035817146301,
-      "learning_rate": 0.0001487987048030221,
-      "loss": 3.2249,
+      "grad_norm": 0.8815584182739258,
+      "learning_rate": 0.00014880518078791148,
+      "loss": 3.2316,
       "step": 69800
     },
     {
       "epoch": 7.53099730458221,
-      "grad_norm": 0.755567193031311,
-      "learning_rate": 0.00014847490555855368,
-      "loss": 3.227,
+      "grad_norm": 0.8128268718719482,
+      "learning_rate": 0.00014848138154344306,
+      "loss": 3.2334,
       "step": 69850
     },
     {
       "epoch": 7.536388140161725,
-      "grad_norm": 0.7703545093536377,
-      "learning_rate": 0.00014815110631408526,
-      "loss": 3.2319,
+      "grad_norm": 0.8221144080162048,
+      "learning_rate": 0.00014815758229897463,
+      "loss": 3.2361,
       "step": 69900
     },
     {
       "epoch": 7.54177897574124,
-      "grad_norm": 0.7938928604125977,
-      "learning_rate": 0.0001478273070696168,
-      "loss": 3.2262,
+      "grad_norm": 0.81682950258255,
+      "learning_rate": 0.00014784025903939556,
+      "loss": 3.2317,
       "step": 69950
     },
     {
       "epoch": 7.547169811320755,
-      "grad_norm": 0.7993608713150024,
-      "learning_rate": 0.0001475035078251484,
-      "loss": 3.2238,
+      "grad_norm": 0.8862870931625366,
+      "learning_rate": 0.00014751645979492714,
+      "loss": 3.2291,
       "step": 70000
     },
     {
       "epoch": 7.547169811320755,
-      "eval_accuracy": 0.3888174013249785,
-      "eval_loss": 3.3437108993530273,
-      "eval_runtime": 184.8174,
-      "eval_samples_per_second": 97.453,
-      "eval_steps_per_second": 6.093,
+      "eval_accuracy": 0.38790612955487846,
+      "eval_loss": 3.348696708679199,
+      "eval_runtime": 186.5493,
+      "eval_samples_per_second": 96.548,
+      "eval_steps_per_second": 6.036,
       "step": 70000
     },
     {
       "epoch": 7.55256064690027,
-      "grad_norm": 0.78631591796875,
-      "learning_rate": 0.00014717970858067997,
-      "loss": 3.2313,
+      "grad_norm": 0.8094576001167297,
+      "learning_rate": 0.00014719266055045871,
+      "loss": 3.2372,
       "step": 70050
     },
     {
       "epoch": 7.557951482479784,
-      "grad_norm": 0.7872461080551147,
-      "learning_rate": 0.00014685590933621154,
-      "loss": 3.2356,
+      "grad_norm": 0.815657377243042,
+      "learning_rate": 0.00014686886130599027,
+      "loss": 3.2401,
       "step": 70100
     },
     {
       "epoch": 7.563342318059299,
-      "grad_norm": 0.7788323760032654,
-      "learning_rate": 0.0001465321100917431,
-      "loss": 3.2341,
+      "grad_norm": 0.783902108669281,
+      "learning_rate": 0.00014654506206152184,
+      "loss": 3.2398,
       "step": 70150
     },
     {
       "epoch": 7.568733153638814,
-      "grad_norm": 0.8200633525848389,
-      "learning_rate": 0.00014620831084727467,
-      "loss": 3.2179,
+      "grad_norm": 0.8154528737068176,
+      "learning_rate": 0.00014622126281705342,
+      "loss": 3.2235,
       "step": 70200
     },
     {
       "epoch": 7.574123989218329,
-      "grad_norm": 0.806707501411438,
-      "learning_rate": 0.00014588451160280625,
-      "loss": 3.2329,
+      "grad_norm": 0.8692945837974548,
+      "learning_rate": 0.000145897463572585,
+      "loss": 3.2384,
       "step": 70250
     },
     {
       "epoch": 7.579514824797844,
-      "grad_norm": 0.7861495614051819,
-      "learning_rate": 0.0001455607123583378,
-      "loss": 3.2287,
+      "grad_norm": 0.7604818940162659,
+      "learning_rate": 0.00014557366432811658,
+      "loss": 3.2334,
       "step": 70300
     },
     {
       "epoch": 7.584905660377358,
-      "grad_norm": 0.8309231996536255,
-      "learning_rate": 0.00014523691311386938,
-      "loss": 3.2446,
+      "grad_norm": 0.8642157316207886,
+      "learning_rate": 0.00014524986508364813,
+      "loss": 3.2482,
       "step": 70350
     },
     {
       "epoch": 7.590296495956873,
-      "grad_norm": 0.7764883637428284,
-      "learning_rate": 0.00014491311386940096,
-      "loss": 3.2258,
+      "grad_norm": 0.8210476636886597,
+      "learning_rate": 0.0001449260658391797,
+      "loss": 3.231,
       "step": 70400
     },
     {
       "epoch": 7.595687331536388,
-      "grad_norm": 0.7940296530723572,
-      "learning_rate": 0.00014458931462493253,
-      "loss": 3.2383,
+      "grad_norm": 0.7866239547729492,
+      "learning_rate": 0.00014460226659471128,
+      "loss": 3.2424,
       "step": 70450
     },
     {
       "epoch": 7.601078167115903,
-      "grad_norm": 0.8480539321899414,
-      "learning_rate": 0.00014426551538046408,
-      "loss": 3.2404,
+      "grad_norm": 0.8430171608924866,
+      "learning_rate": 0.00014427846735024283,
+      "loss": 3.2461,
       "step": 70500
     },
     {
       "epoch": 7.606469002695418,
-      "grad_norm": 0.7697724103927612,
-      "learning_rate": 0.00014394171613599566,
-      "loss": 3.2288,
+      "grad_norm": 0.7713740468025208,
+      "learning_rate": 0.0001439546681057744,
+      "loss": 3.2347,
       "step": 70550
     },
     {
       "epoch": 7.611859838274933,
-      "grad_norm": 0.7845086455345154,
-      "learning_rate": 0.00014361791689152724,
-      "loss": 3.2292,
+      "grad_norm": 0.7943990230560303,
+      "learning_rate": 0.000143630868861306,
+      "loss": 3.2344,
       "step": 70600
     },
     {
       "epoch": 7.617250673854447,
-      "grad_norm": 0.7915938496589661,
-      "learning_rate": 0.00014329411764705882,
-      "loss": 3.2291,
+      "grad_norm": 0.8576189875602722,
+      "learning_rate": 0.00014330706961683754,
+      "loss": 3.2354,
       "step": 70650
     },
     {
       "epoch": 7.622641509433962,
-      "grad_norm": 0.7984021306037903,
-      "learning_rate": 0.0001429703184025904,
-      "loss": 3.2314,
+      "grad_norm": 0.791982889175415,
+      "learning_rate": 0.00014298327037236912,
+      "loss": 3.2369,
       "step": 70700
     },
     {
       "epoch": 7.628032345013477,
-      "grad_norm": 0.8050603866577148,
-      "learning_rate": 0.00014264651915812194,
-      "loss": 3.2322,
+      "grad_norm": 0.8384832143783569,
+      "learning_rate": 0.0001426594711279007,
+      "loss": 3.2379,
       "step": 70750
     },
     {
       "epoch": 7.633423180592992,
-      "grad_norm": 0.8283957839012146,
-      "learning_rate": 0.00014232271991365352,
-      "loss": 3.2444,
+      "grad_norm": 0.8308936953544617,
+      "learning_rate": 0.00014233567188343224,
+      "loss": 3.2495,
       "step": 70800
     },
     {
       "epoch": 7.638814016172507,
-      "grad_norm": 0.7875947952270508,
-      "learning_rate": 0.0001419989206691851,
-      "loss": 3.2222,
+      "grad_norm": 0.7926676273345947,
+      "learning_rate": 0.00014201187263896382,
+      "loss": 3.2267,
       "step": 70850
     },
     {
       "epoch": 7.644204851752022,
-      "grad_norm": 0.8083146810531616,
-      "learning_rate": 0.00014168159740960602,
-      "loss": 3.2393,
+      "grad_norm": 0.8410943746566772,
+      "learning_rate": 0.0001416880733944954,
+      "loss": 3.2433,
       "step": 70900
     },
     {
       "epoch": 7.649595687331536,
-      "grad_norm": 0.7859264612197876,
-      "learning_rate": 0.0001413577981651376,
-      "loss": 3.2201,
+      "grad_norm": 0.8225048780441284,
+      "learning_rate": 0.00014136427415002695,
+      "loss": 3.2261,
       "step": 70950
     },
     {
       "epoch": 7.654986522911051,
-      "grad_norm": 0.7985530495643616,
-      "learning_rate": 0.00014103399892066918,
-      "loss": 3.2513,
+      "grad_norm": 0.8333801627159119,
+      "learning_rate": 0.00014104047490555853,
+      "loss": 3.2551,
       "step": 71000
     },
     {
       "epoch": 7.654986522911051,
-      "eval_accuracy": 0.38947138305673384,
-      "eval_loss": 3.3401072025299072,
-      "eval_runtime": 184.8335,
-      "eval_samples_per_second": 97.444,
-      "eval_steps_per_second": 6.092,
+      "eval_accuracy": 0.38870070812346574,
+      "eval_loss": 3.345031261444092,
+      "eval_runtime": 186.8863,
+      "eval_samples_per_second": 96.374,
+      "eval_steps_per_second": 6.025,
       "step": 71000
     },
     {
       "epoch": 7.660377358490566,
-      "grad_norm": 0.7884781956672668,
-      "learning_rate": 0.00014071019967620073,
-      "loss": 3.2415,
+      "grad_norm": 0.8508845567703247,
+      "learning_rate": 0.0001407166756610901,
+      "loss": 3.2491,
       "step": 71050
     },
     {
       "epoch": 7.665768194070081,
-      "grad_norm": 0.7903490662574768,
-      "learning_rate": 0.0001403864004317323,
-      "loss": 3.2499,
+      "grad_norm": 0.8326047658920288,
+      "learning_rate": 0.00014039287641662168,
+      "loss": 3.2552,
       "step": 71100
     },
     {
       "epoch": 7.671159029649596,
-      "grad_norm": 0.7995983958244324,
-      "learning_rate": 0.00014006260118726389,
-      "loss": 3.233,
+      "grad_norm": 0.819631814956665,
+      "learning_rate": 0.00014006907717215326,
+      "loss": 3.2368,
       "step": 71150
     },
     {
       "epoch": 7.67654986522911,
-      "grad_norm": 0.7693688273429871,
-      "learning_rate": 0.00013973880194279546,
-      "loss": 3.2488,
+      "grad_norm": 0.8281809091567993,
+      "learning_rate": 0.0001397452779276848,
+      "loss": 3.252,
       "step": 71200
     },
     {
       "epoch": 7.681940700808625,
-      "grad_norm": 0.7897132039070129,
-      "learning_rate": 0.00013941500269832704,
-      "loss": 3.2375,
+      "grad_norm": 0.7906057834625244,
+      "learning_rate": 0.0001394214786832164,
+      "loss": 3.2415,
       "step": 71250
     },
     {
       "epoch": 7.6873315363881405,
-      "grad_norm": 0.8603083491325378,
-      "learning_rate": 0.0001390912034538586,
-      "loss": 3.2326,
+      "grad_norm": 0.8822153210639954,
+      "learning_rate": 0.00013909767943874797,
+      "loss": 3.2378,
       "step": 71300
     },
     {
       "epoch": 7.692722371967655,
-      "grad_norm": 0.7997660636901855,
-      "learning_rate": 0.00013876740420939017,
-      "loss": 3.2474,
+      "grad_norm": 0.8512835502624512,
+      "learning_rate": 0.00013877388019427954,
+      "loss": 3.2507,
       "step": 71350
     },
     {
       "epoch": 7.69811320754717,
-      "grad_norm": 0.8124864101409912,
-      "learning_rate": 0.00013844360496492175,
-      "loss": 3.2522,
+      "grad_norm": 0.8629444241523743,
+      "learning_rate": 0.0001384500809498111,
+      "loss": 3.2571,
       "step": 71400
     },
     {
       "epoch": 7.703504043126685,
-      "grad_norm": 0.7789794206619263,
-      "learning_rate": 0.00013811980572045333,
-      "loss": 3.2292,
+      "grad_norm": 0.7869336605072021,
+      "learning_rate": 0.00013812628170534267,
+      "loss": 3.2327,
       "step": 71450
     },
     {
       "epoch": 7.708894878706199,
-      "grad_norm": 0.8284264206886292,
-      "learning_rate": 0.00013779600647598488,
-      "loss": 3.2384,
+      "grad_norm": 0.8382977247238159,
+      "learning_rate": 0.00013780248246087425,
+      "loss": 3.2431,
       "step": 71500
     },
     {
       "epoch": 7.714285714285714,
-      "grad_norm": 0.7541694045066833,
-      "learning_rate": 0.00013747220723151645,
-      "loss": 3.2273,
+      "grad_norm": 0.7827866673469543,
+      "learning_rate": 0.00013747868321640583,
+      "loss": 3.2334,
       "step": 71550
     },
     {
       "epoch": 7.719676549865229,
-      "grad_norm": 0.8304938673973083,
-      "learning_rate": 0.00013714840798704803,
-      "loss": 3.2218,
+      "grad_norm": 0.8907617926597595,
+      "learning_rate": 0.00013715488397193738,
+      "loss": 3.2262,
       "step": 71600
     },
     {
       "epoch": 7.725067385444744,
-      "grad_norm": 0.8099672198295593,
-      "learning_rate": 0.00013682460874257958,
-      "loss": 3.244,
+      "grad_norm": 0.8571818470954895,
+      "learning_rate": 0.00013683108472746896,
+      "loss": 3.2491,
       "step": 71650
     },
     {
       "epoch": 7.730458221024259,
-      "grad_norm": 0.8018233180046082,
-      "learning_rate": 0.00013650080949811116,
-      "loss": 3.2449,
+      "grad_norm": 0.8379048109054565,
+      "learning_rate": 0.00013650728548300053,
+      "loss": 3.2492,
       "step": 71700
     },
     {
       "epoch": 7.735849056603773,
-      "grad_norm": 0.8082437515258789,
-      "learning_rate": 0.00013617701025364274,
-      "loss": 3.2408,
+      "grad_norm": 0.862058699131012,
+      "learning_rate": 0.0001361834862385321,
+      "loss": 3.2455,
       "step": 71750
     },
     {
       "epoch": 7.741239892183288,
-      "grad_norm": 0.7693942785263062,
-      "learning_rate": 0.0001358532110091743,
-      "loss": 3.2298,
+      "grad_norm": 0.8231572508811951,
+      "learning_rate": 0.0001358596869940637,
+      "loss": 3.2324,
       "step": 71800
     },
     {
       "epoch": 7.7466307277628035,
-      "grad_norm": 0.7721502780914307,
-      "learning_rate": 0.00013552941176470587,
-      "loss": 3.2468,
+      "grad_norm": 0.8185727000236511,
+      "learning_rate": 0.00013553588774959524,
+      "loss": 3.252,
       "step": 71850
     },
     {
       "epoch": 7.752021563342318,
-      "grad_norm": 0.76530522108078,
-      "learning_rate": 0.00013520561252023744,
-      "loss": 3.2439,
+      "grad_norm": 0.801917314529419,
+      "learning_rate": 0.00013521208850512682,
+      "loss": 3.249,
       "step": 71900
     },
     {
       "epoch": 7.757412398921833,
-      "grad_norm": 0.786440908908844,
-      "learning_rate": 0.00013488181327576902,
-      "loss": 3.244,
+      "grad_norm": 0.8169463872909546,
+      "learning_rate": 0.0001348882892606584,
+      "loss": 3.2506,
       "step": 71950
     },
     {
       "epoch": 7.762803234501348,
-      "grad_norm": 0.7962386012077332,
-      "learning_rate": 0.00013455801403130057,
-      "loss": 3.2518,
+      "grad_norm": 0.8260762691497803,
+      "learning_rate": 0.00013456449001618995,
+      "loss": 3.2557,
       "step": 72000
     },
     {
       "epoch": 7.762803234501348,
-      "eval_accuracy": 0.39001149656205963,
-      "eval_loss": 3.3362839221954346,
-      "eval_runtime": 184.9207,
-      "eval_samples_per_second": 97.399,
-      "eval_steps_per_second": 6.089,
+      "eval_accuracy": 0.38898505273087625,
+      "eval_loss": 3.341965675354004,
+      "eval_runtime": 186.5684,
+      "eval_samples_per_second": 96.538,
+      "eval_steps_per_second": 6.035,
       "step": 72000
     },
     {
       "epoch": 7.768194070080862,
-      "grad_norm": 0.8243362903594971,
-      "learning_rate": 0.00013423421478683215,
-      "loss": 3.2376,
+      "grad_norm": 0.8457881212234497,
+      "learning_rate": 0.00013424069077172152,
+      "loss": 3.2403,
       "step": 72050
     },
     {
       "epoch": 7.773584905660377,
-      "grad_norm": 0.7997215390205383,
-      "learning_rate": 0.00013391041554236373,
-      "loss": 3.2442,
+      "grad_norm": 0.8046347498893738,
+      "learning_rate": 0.00013392336751214248,
+      "loss": 3.249,
       "step": 72100
     },
     {
       "epoch": 7.7789757412398925,
-      "grad_norm": 0.8042483329772949,
-      "learning_rate": 0.00013358661629789528,
-      "loss": 3.2313,
+      "grad_norm": 0.8065981864929199,
+      "learning_rate": 0.00013359956826767403,
+      "loss": 3.2348,
       "step": 72150
     },
     {
       "epoch": 7.784366576819407,
-      "grad_norm": 0.7405474781990051,
-      "learning_rate": 0.00013326281705342685,
-      "loss": 3.2229,
+      "grad_norm": 0.8088060021400452,
+      "learning_rate": 0.0001332757690232056,
+      "loss": 3.2262,
       "step": 72200
     },
     {
       "epoch": 7.789757412398922,
-      "grad_norm": 0.784062385559082,
-      "learning_rate": 0.00013293901780895843,
-      "loss": 3.2333,
+      "grad_norm": 0.8394667506217957,
+      "learning_rate": 0.00013295196977873718,
+      "loss": 3.2385,
       "step": 72250
     },
     {
       "epoch": 7.795148247978437,
-      "grad_norm": 0.8437650203704834,
-      "learning_rate": 0.00013261521856449,
-      "loss": 3.2217,
+      "grad_norm": 0.8281553387641907,
+      "learning_rate": 0.00013262817053426876,
+      "loss": 3.2268,
       "step": 72300
     },
     {
       "epoch": 7.800539083557951,
-      "grad_norm": 0.7900862097740173,
-      "learning_rate": 0.00013229141932002156,
-      "loss": 3.2094,
+      "grad_norm": 0.8217343091964722,
+      "learning_rate": 0.0001323043712898003,
+      "loss": 3.2141,
       "step": 72350
     },
     {
       "epoch": 7.8059299191374665,
-      "grad_norm": 0.80324786901474,
-      "learning_rate": 0.00013196762007555314,
-      "loss": 3.2356,
+      "grad_norm": 0.8292942643165588,
+      "learning_rate": 0.0001319805720453319,
+      "loss": 3.2382,
       "step": 72400
     },
     {
       "epoch": 7.811320754716981,
-      "grad_norm": 0.8037505149841309,
-      "learning_rate": 0.00013164382083108472,
-      "loss": 3.2382,
+      "grad_norm": 0.7839635014533997,
+      "learning_rate": 0.00013165677280086346,
+      "loss": 3.2435,
       "step": 72450
     },
     {
       "epoch": 7.816711590296496,
-      "grad_norm": 0.8118758797645569,
-      "learning_rate": 0.0001313200215866163,
-      "loss": 3.2327,
+      "grad_norm": 0.8212515711784363,
+      "learning_rate": 0.00013133297355639502,
+      "loss": 3.2369,
       "step": 72500
     },
     {
       "epoch": 7.822102425876011,
-      "grad_norm": 0.8497474789619446,
-      "learning_rate": 0.00013099622234214784,
-      "loss": 3.2207,
+      "grad_norm": 0.8480151295661926,
+      "learning_rate": 0.0001310091743119266,
+      "loss": 3.2246,
       "step": 72550
     },
     {
       "epoch": 7.827493261455525,
-      "grad_norm": 0.7857134938240051,
-      "learning_rate": 0.00013067242309767942,
-      "loss": 3.2506,
+      "grad_norm": 0.8105837106704712,
+      "learning_rate": 0.00013068537506745817,
+      "loss": 3.2547,
       "step": 72600
     },
     {
       "epoch": 7.83288409703504,
-      "grad_norm": 0.8757586479187012,
-      "learning_rate": 0.000130348623853211,
-      "loss": 3.2352,
+      "grad_norm": 0.8260267972946167,
+      "learning_rate": 0.00013036157582298972,
+      "loss": 3.2396,
       "step": 72650
     },
     {
       "epoch": 7.8382749326145555,
-      "grad_norm": 0.7962746024131775,
-      "learning_rate": 0.00013002482460874258,
-      "loss": 3.2401,
+      "grad_norm": 0.8459762334823608,
+      "learning_rate": 0.0001300377765785213,
+      "loss": 3.2442,
       "step": 72700
     },
     {
       "epoch": 7.84366576819407,
-      "grad_norm": 0.8165475726127625,
-      "learning_rate": 0.00012970102536427413,
-      "loss": 3.2086,
+      "grad_norm": 0.8333255648612976,
+      "learning_rate": 0.00012971397733405288,
+      "loss": 3.2129,
       "step": 72750
     },
     {
       "epoch": 7.849056603773585,
-      "grad_norm": 0.7961111068725586,
-      "learning_rate": 0.0001293772261198057,
-      "loss": 3.2404,
+      "grad_norm": 0.8110676407814026,
+      "learning_rate": 0.00012939017808958443,
+      "loss": 3.2455,
       "step": 72800
     },
     {
       "epoch": 7.8544474393531,
-      "grad_norm": 0.8070231676101685,
-      "learning_rate": 0.00012905342687533728,
-      "loss": 3.2371,
+      "grad_norm": 0.804604172706604,
+      "learning_rate": 0.000129066378845116,
+      "loss": 3.244,
       "step": 72850
     },
     {
       "epoch": 7.859838274932614,
-      "grad_norm": 0.7886551022529602,
-      "learning_rate": 0.00012872962763086886,
-      "loss": 3.2402,
+      "grad_norm": 0.8135696053504944,
+      "learning_rate": 0.00012874257960064758,
+      "loss": 3.2445,
       "step": 72900
     },
     {
       "epoch": 7.8652291105121295,
-      "grad_norm": 0.8020292520523071,
-      "learning_rate": 0.00012840582838640044,
-      "loss": 3.2398,
+      "grad_norm": 0.8256625533103943,
+      "learning_rate": 0.00012841878035617916,
+      "loss": 3.2455,
       "step": 72950
     },
     {
       "epoch": 7.870619946091644,
-      "grad_norm": 0.8239029049873352,
-      "learning_rate": 0.000128082029141932,
-      "loss": 3.2309,
+      "grad_norm": 0.8487560749053955,
+      "learning_rate": 0.0001280949811117107,
+      "loss": 3.2361,
       "step": 73000
     },
     {
       "epoch": 7.870619946091644,
-      "eval_accuracy": 0.39013916370524165,
-      "eval_loss": 3.332185983657837,
-      "eval_runtime": 184.799,
-      "eval_samples_per_second": 97.463,
-      "eval_steps_per_second": 6.093,
+      "eval_accuracy": 0.3893036229979483,
+      "eval_loss": 3.337681770324707,
+      "eval_runtime": 189.3296,
+      "eval_samples_per_second": 95.13,
+      "eval_steps_per_second": 5.947,
       "step": 73000
     },
     {
       "epoch": 7.876010781671159,
-      "grad_norm": 0.850242018699646,
-      "learning_rate": 0.00012775822989746357,
-      "loss": 3.2415,
+      "grad_norm": 0.8776509761810303,
+      "learning_rate": 0.0001277711818672423,
+      "loss": 3.2452,
       "step": 73050
     },
     {
       "epoch": 7.881401617250674,
-      "grad_norm": 0.7743697166442871,
-      "learning_rate": 0.0001274409066378845,
-      "loss": 3.2358,
+      "grad_norm": 0.7829513549804688,
+      "learning_rate": 0.00012744738262277387,
+      "loss": 3.2412,
       "step": 73100
     },
     {
       "epoch": 7.886792452830189,
-      "grad_norm": 0.7658908367156982,
-      "learning_rate": 0.00012711710739341607,
-      "loss": 3.2371,
+      "grad_norm": 0.8267134428024292,
+      "learning_rate": 0.00012712358337830544,
+      "loss": 3.2404,
       "step": 73150
     },
     {
       "epoch": 7.892183288409703,
-      "grad_norm": 0.8552940487861633,
-      "learning_rate": 0.00012679330814894765,
-      "loss": 3.2308,
+      "grad_norm": 0.866329550743103,
+      "learning_rate": 0.00012679978413383702,
+      "loss": 3.2352,
       "step": 73200
     },
     {
       "epoch": 7.8975741239892185,
-      "grad_norm": 0.791025698184967,
-      "learning_rate": 0.00012646950890447922,
-      "loss": 3.2368,
+      "grad_norm": 0.8073294162750244,
+      "learning_rate": 0.00012647598488936857,
+      "loss": 3.2424,
       "step": 73250
     },
     {
       "epoch": 7.902964959568733,
-      "grad_norm": 0.8062611222267151,
-      "learning_rate": 0.00012614570966001077,
-      "loss": 3.2394,
+      "grad_norm": 0.8263236284255981,
+      "learning_rate": 0.00012615218564490015,
+      "loss": 3.2442,
       "step": 73300
     },
     {
       "epoch": 7.908355795148248,
-      "grad_norm": 0.8666378259658813,
-      "learning_rate": 0.00012582191041554235,
-      "loss": 3.2175,
+      "grad_norm": 0.853076696395874,
+      "learning_rate": 0.00012582838640043173,
+      "loss": 3.221,
       "step": 73350
     },
     {
       "epoch": 7.913746630727763,
-      "grad_norm": 0.8134450316429138,
-      "learning_rate": 0.00012549811117107393,
-      "loss": 3.2308,
+      "grad_norm": 0.8882546424865723,
+      "learning_rate": 0.0001255045871559633,
+      "loss": 3.2356,
       "step": 73400
     },
     {
       "epoch": 7.919137466307277,
-      "grad_norm": 0.822311282157898,
-      "learning_rate": 0.0001251743119266055,
-      "loss": 3.2146,
+      "grad_norm": 0.8196534514427185,
+      "learning_rate": 0.00012518078791149486,
+      "loss": 3.2181,
       "step": 73450
     },
     {
       "epoch": 7.9245283018867925,
-      "grad_norm": 0.8420822620391846,
-      "learning_rate": 0.00012485051268213706,
-      "loss": 3.2317,
+      "grad_norm": 0.8723163604736328,
+      "learning_rate": 0.00012485698866702643,
+      "loss": 3.2369,
       "step": 73500
     },
     {
       "epoch": 7.929919137466308,
-      "grad_norm": 0.7920177578926086,
-      "learning_rate": 0.00012452671343766864,
-      "loss": 3.2235,
+      "grad_norm": 0.8038402795791626,
+      "learning_rate": 0.000124533189422558,
+      "loss": 3.2301,
       "step": 73550
     },
     {
       "epoch": 7.935309973045822,
-      "grad_norm": 0.8194290995597839,
-      "learning_rate": 0.00012420291419320021,
-      "loss": 3.2235,
+      "grad_norm": 0.8190316557884216,
+      "learning_rate": 0.0001242093901780896,
+      "loss": 3.2278,
       "step": 73600
     },
     {
       "epoch": 7.940700808625337,
-      "grad_norm": 0.8023219704627991,
-      "learning_rate": 0.0001238791149487318,
-      "loss": 3.2397,
+      "grad_norm": 0.8468826413154602,
+      "learning_rate": 0.00012388559093362114,
+      "loss": 3.2444,
       "step": 73650
     },
     {
       "epoch": 7.946091644204852,
-      "grad_norm": 1.2114663124084473,
-      "learning_rate": 0.00012355531570426334,
-      "loss": 3.2168,
+      "grad_norm": 0.8245653510093689,
+      "learning_rate": 0.00012356179168915272,
+      "loss": 3.2201,
       "step": 73700
     },
     {
       "epoch": 7.951482479784366,
-      "grad_norm": 0.8042489290237427,
-      "learning_rate": 0.00012323151645979492,
-      "loss": 3.2347,
+      "grad_norm": 0.8021673560142517,
+      "learning_rate": 0.0001232379924446843,
+      "loss": 3.2398,
       "step": 73750
     },
     {
       "epoch": 7.9568733153638815,
-      "grad_norm": 0.8316463232040405,
-      "learning_rate": 0.0001229077172153265,
-      "loss": 3.2324,
+      "grad_norm": 0.8257600665092468,
+      "learning_rate": 0.00012291419320021587,
+      "loss": 3.2373,
       "step": 73800
     },
     {
       "epoch": 7.962264150943396,
-      "grad_norm": 0.8015411496162415,
-      "learning_rate": 0.00012258391797085805,
-      "loss": 3.221,
+      "grad_norm": 0.8508664965629578,
+      "learning_rate": 0.00012259039395574742,
+      "loss": 3.2246,
       "step": 73850
     },
     {
       "epoch": 7.967654986522911,
-      "grad_norm": 0.8200330138206482,
-      "learning_rate": 0.00012226011872638963,
-      "loss": 3.2256,
+      "grad_norm": 0.8783078789710999,
+      "learning_rate": 0.000122266594711279,
+      "loss": 3.2311,
       "step": 73900
     },
     {
       "epoch": 7.973045822102426,
-      "grad_norm": 0.7811427712440491,
-      "learning_rate": 0.00012193631948192119,
-      "loss": 3.228,
+      "grad_norm": 0.7954983115196228,
+      "learning_rate": 0.00012194279546681056,
+      "loss": 3.2315,
       "step": 73950
     },
     {
       "epoch": 7.97843665768194,
-      "grad_norm": 0.8133488297462463,
-      "learning_rate": 0.00012161252023745277,
-      "loss": 3.2519,
+      "grad_norm": 0.855838418006897,
+      "learning_rate": 0.00012161899622234214,
+      "loss": 3.2561,
       "step": 74000
     },
     {
       "epoch": 7.97843665768194,
-      "eval_accuracy": 0.39057507909115335,
-      "eval_loss": 3.328582525253296,
-      "eval_runtime": 185.1358,
-      "eval_samples_per_second": 97.285,
-      "eval_steps_per_second": 6.082,
+      "eval_accuracy": 0.3899309847721891,
+      "eval_loss": 3.332972526550293,
+      "eval_runtime": 189.8927,
+      "eval_samples_per_second": 94.848,
+      "eval_steps_per_second": 5.93,
       "step": 74000
     },
     {
       "epoch": 7.9838274932614555,
-      "grad_norm": 0.8701046109199524,
-      "learning_rate": 0.00012128872099298435,
-      "loss": 3.2368,
+      "grad_norm": 0.9047753810882568,
+      "learning_rate": 0.00012129519697787372,
+      "loss": 3.242,
       "step": 74050
     },
     {
       "epoch": 7.989218328840971,
-      "grad_norm": 0.8377940058708191,
-      "learning_rate": 0.00012096492174851591,
-      "loss": 3.2394,
+      "grad_norm": 0.8472272753715515,
+      "learning_rate": 0.00012097139773340527,
+      "loss": 3.2458,
       "step": 74100
     },
     {
       "epoch": 7.994609164420485,
-      "grad_norm": 0.8556270599365234,
-      "learning_rate": 0.00012064112250404749,
-      "loss": 3.2404,
+      "grad_norm": 0.8983229398727417,
+      "learning_rate": 0.00012064759848893685,
+      "loss": 3.2466,
       "step": 74150
     },
     {
       "epoch": 8.0,
-      "grad_norm": 1.6344754695892334,
-      "learning_rate": 0.00012031732325957905,
-      "loss": 3.2383,
+      "grad_norm": 1.6775580644607544,
+      "learning_rate": 0.00012032379924446843,
+      "loss": 3.2418,
       "step": 74200
     },
     {
       "epoch": 8.005390835579515,
-      "grad_norm": 0.801658034324646,
-      "learning_rate": 0.00011999352401511062,
-      "loss": 3.1543,
+      "grad_norm": 0.8066301941871643,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 3.1583,
       "step": 74250
     },
     {
       "epoch": 8.01078167115903,
-      "grad_norm": 0.7476630210876465,
-      "learning_rate": 0.00011966972477064219,
-      "loss": 3.1637,
+      "grad_norm": 0.8377830386161804,
+      "learning_rate": 0.00011967620075553155,
+      "loss": 3.168,
       "step": 74300
     },
     {
       "epoch": 8.016172506738544,
-      "grad_norm": 0.8232449293136597,
-      "learning_rate": 0.00011934592552617377,
-      "loss": 3.1622,
+      "grad_norm": 0.8650294542312622,
+      "learning_rate": 0.00011935240151106313,
+      "loss": 3.1651,
       "step": 74350
     },
     {
       "epoch": 8.021563342318059,
-      "grad_norm": 0.8696622252464294,
-      "learning_rate": 0.00011902212628170532,
-      "loss": 3.1583,
+      "grad_norm": 0.8577993512153625,
+      "learning_rate": 0.00011902860226659471,
+      "loss": 3.1631,
       "step": 74400
     },
     {
       "epoch": 8.026954177897574,
-      "grad_norm": 0.8294352293014526,
-      "learning_rate": 0.0001186983270372369,
-      "loss": 3.1607,
+      "grad_norm": 0.8454034924507141,
+      "learning_rate": 0.00011870480302212627,
+      "loss": 3.165,
       "step": 74450
     },
     {
       "epoch": 8.032345013477089,
-      "grad_norm": 0.7915950417518616,
-      "learning_rate": 0.00011837452779276848,
-      "loss": 3.1799,
+      "grad_norm": 0.8800636529922485,
+      "learning_rate": 0.00011838100377765784,
+      "loss": 3.1856,
       "step": 74500
     },
     {
       "epoch": 8.037735849056604,
-      "grad_norm": 0.8357707858085632,
-      "learning_rate": 0.00011805072854830005,
-      "loss": 3.158,
+      "grad_norm": 0.8559412956237793,
+      "learning_rate": 0.00011805720453318941,
+      "loss": 3.1619,
       "step": 74550
     },
     {
       "epoch": 8.04312668463612,
-      "grad_norm": 0.8107123970985413,
-      "learning_rate": 0.0001177269293038316,
-      "loss": 3.1785,
+      "grad_norm": 0.8575382828712463,
+      "learning_rate": 0.00011773340528872098,
+      "loss": 3.1835,
       "step": 74600
     },
     {
       "epoch": 8.048517520215633,
-      "grad_norm": 0.8061670064926147,
-      "learning_rate": 0.00011740313005936318,
-      "loss": 3.175,
+      "grad_norm": 0.8358049392700195,
+      "learning_rate": 0.00011740960604425256,
+      "loss": 3.1801,
       "step": 74650
     },
     {
       "epoch": 8.053908355795148,
-      "grad_norm": 0.7983211874961853,
-      "learning_rate": 0.00011707933081489476,
-      "loss": 3.1642,
+      "grad_norm": 0.791158139705658,
+      "learning_rate": 0.00011708580679978412,
+      "loss": 3.1668,
       "step": 74700
     },
     {
       "epoch": 8.059299191374663,
-      "grad_norm": 0.8365211486816406,
-      "learning_rate": 0.00011675553157042632,
-      "loss": 3.1511,
+      "grad_norm": 0.870159387588501,
+      "learning_rate": 0.00011676200755531568,
+      "loss": 3.1553,
       "step": 74750
     },
     {
       "epoch": 8.064690026954178,
-      "grad_norm": 0.8603684306144714,
-      "learning_rate": 0.00011643173232595789,
-      "loss": 3.1714,
+      "grad_norm": 0.8641995787620544,
+      "learning_rate": 0.00011643820831084726,
+      "loss": 3.1753,
       "step": 74800
     },
     {
       "epoch": 8.070080862533693,
-      "grad_norm": 0.7773607969284058,
-      "learning_rate": 0.00011610793308148947,
-      "loss": 3.1512,
+      "grad_norm": 0.8149543404579163,
+      "learning_rate": 0.00011611440906637884,
+      "loss": 3.1567,
       "step": 74850
     },
     {
       "epoch": 8.075471698113208,
-      "grad_norm": 0.7770172953605652,
-      "learning_rate": 0.00011578413383702104,
-      "loss": 3.152,
+      "grad_norm": 0.7908215522766113,
+      "learning_rate": 0.00011579060982191042,
+      "loss": 3.1551,
       "step": 74900
     },
     {
       "epoch": 8.080862533692722,
-      "grad_norm": 0.78849858045578,
-      "learning_rate": 0.00011546033459255261,
-      "loss": 3.1676,
+      "grad_norm": 0.7902390360832214,
+      "learning_rate": 0.00011546681057744197,
+      "loss": 3.1726,
       "step": 74950
     },
     {
       "epoch": 8.086253369272237,
-      "grad_norm": 0.814720630645752,
-      "learning_rate": 0.00011513653534808419,
-      "loss": 3.1583,
+      "grad_norm": 0.8608523607254028,
+      "learning_rate": 0.00011514301133297355,
+      "loss": 3.1611,
       "step": 75000
     },
     {
       "epoch": 8.086253369272237,
-      "eval_accuracy": 0.3906881867473682,
-      "eval_loss": 3.3347113132476807,
-      "eval_runtime": 184.6754,
-      "eval_samples_per_second": 97.528,
-      "eval_steps_per_second": 6.097,
+      "eval_accuracy": 0.38982102804972085,
+      "eval_loss": 3.3394863605499268,
+      "eval_runtime": 186.8464,
+      "eval_samples_per_second": 96.395,
+      "eval_steps_per_second": 6.026,
       "step": 75000
     },
     {
       "epoch": 8.091644204851752,
-      "grad_norm": 0.8267271518707275,
-      "learning_rate": 0.00011481273610361575,
-      "loss": 3.165,
+      "grad_norm": 0.9693127870559692,
+      "learning_rate": 0.00011481921208850512,
+      "loss": 3.1673,
       "step": 75050
     },
     {
       "epoch": 8.097035040431267,
-      "grad_norm": 0.8477714657783508,
-      "learning_rate": 0.00011448893685914731,
-      "loss": 3.1511,
+      "grad_norm": 0.8476482033729553,
+      "learning_rate": 0.00011449541284403669,
+      "loss": 3.1551,
       "step": 75100
     },
     {
       "epoch": 8.102425876010782,
-      "grad_norm": 0.7959654331207275,
-      "learning_rate": 0.00011416513761467889,
-      "loss": 3.1745,
+      "grad_norm": 0.8478171825408936,
+      "learning_rate": 0.00011417161359956825,
+      "loss": 3.1792,
       "step": 75150
     },
     {
       "epoch": 8.107816711590296,
-      "grad_norm": 0.8167853951454163,
-      "learning_rate": 0.00011384133837021047,
-      "loss": 3.1673,
+      "grad_norm": 0.8267400860786438,
+      "learning_rate": 0.00011384781435509983,
+      "loss": 3.1717,
       "step": 75200
     },
     {
       "epoch": 8.11320754716981,
-      "grad_norm": 0.7871206402778625,
-      "learning_rate": 0.00011351753912574202,
-      "loss": 3.1734,
+      "grad_norm": 0.8523077964782715,
+      "learning_rate": 0.0001135240151106314,
+      "loss": 3.1777,
       "step": 75250
     },
     {
       "epoch": 8.118598382749326,
-      "grad_norm": 0.7872896194458008,
-      "learning_rate": 0.0001131937398812736,
-      "loss": 3.1801,
+      "grad_norm": 0.8074910640716553,
+      "learning_rate": 0.00011320021586616297,
+      "loss": 3.1837,
       "step": 75300
     },
     {
       "epoch": 8.123989218328841,
-      "grad_norm": 0.7880840301513672,
-      "learning_rate": 0.00011286994063680517,
-      "loss": 3.1839,
+      "grad_norm": 0.8084198832511902,
+      "learning_rate": 0.00011287641662169454,
+      "loss": 3.1891,
       "step": 75350
     },
     {
       "epoch": 8.129380053908356,
-      "grad_norm": 0.883547306060791,
-      "learning_rate": 0.00011254614139233675,
-      "loss": 3.1856,
+      "grad_norm": 0.8906427621841431,
+      "learning_rate": 0.0001125526173772261,
+      "loss": 3.1887,
       "step": 75400
     },
     {
       "epoch": 8.134770889487871,
-      "grad_norm": 0.8173453211784363,
-      "learning_rate": 0.0001122223421478683,
-      "loss": 3.1721,
+      "grad_norm": 0.8376168012619019,
+      "learning_rate": 0.00011222881813275768,
+      "loss": 3.1756,
       "step": 75450
     },
     {
       "epoch": 8.140161725067385,
-      "grad_norm": 0.7894411087036133,
-      "learning_rate": 0.00011189854290339988,
-      "loss": 3.1769,
+      "grad_norm": 0.8163133859634399,
+      "learning_rate": 0.00011190501888828925,
+      "loss": 3.1814,
       "step": 75500
     },
     {
       "epoch": 8.1455525606469,
-      "grad_norm": 0.7918045520782471,
-      "learning_rate": 0.00011157474365893146,
-      "loss": 3.1614,
+      "grad_norm": 0.7996161580085754,
+      "learning_rate": 0.00011158121964382083,
+      "loss": 3.1666,
       "step": 75550
     },
     {
       "epoch": 8.150943396226415,
-      "grad_norm": 0.825491726398468,
-      "learning_rate": 0.00011125094441446302,
-      "loss": 3.1587,
+      "grad_norm": 0.8337327837944031,
+      "learning_rate": 0.00011125742039935238,
+      "loss": 3.1657,
       "step": 75600
     },
     {
       "epoch": 8.15633423180593,
-      "grad_norm": 0.8395944833755493,
-      "learning_rate": 0.00011092714516999459,
-      "loss": 3.1649,
+      "grad_norm": 0.8075405955314636,
+      "learning_rate": 0.00011093362115488396,
+      "loss": 3.1692,
       "step": 75650
     },
     {
       "epoch": 8.161725067385445,
-      "grad_norm": 0.8326493501663208,
-      "learning_rate": 0.00011060334592552616,
-      "loss": 3.1851,
+      "grad_norm": 0.8634636402130127,
+      "learning_rate": 0.00011060982191041554,
+      "loss": 3.1888,
       "step": 75700
     },
     {
       "epoch": 8.167115902964959,
-      "grad_norm": 0.795487642288208,
-      "learning_rate": 0.00011027954668105773,
-      "loss": 3.1673,
+      "grad_norm": 0.780362069606781,
+      "learning_rate": 0.00011028602266594712,
+      "loss": 3.1705,
       "step": 75750
     },
     {
       "epoch": 8.172506738544474,
-      "grad_norm": 0.8097192049026489,
-      "learning_rate": 0.0001099557474365893,
-      "loss": 3.1614,
+      "grad_norm": 0.8302311897277832,
+      "learning_rate": 0.00010996222342147867,
+      "loss": 3.1654,
       "step": 75800
     },
     {
       "epoch": 8.177897574123989,
-      "grad_norm": 0.8157842755317688,
-      "learning_rate": 0.00010963194819212088,
-      "loss": 3.1657,
+      "grad_norm": 0.8145652413368225,
+      "learning_rate": 0.00010963842417701024,
+      "loss": 3.1686,
       "step": 75850
     },
     {
       "epoch": 8.183288409703504,
-      "grad_norm": 0.7984054088592529,
-      "learning_rate": 0.00010930814894765243,
-      "loss": 3.1681,
+      "grad_norm": 0.8234553933143616,
+      "learning_rate": 0.00010931462493254182,
+      "loss": 3.1723,
       "step": 75900
     },
     {
       "epoch": 8.18867924528302,
-      "grad_norm": 0.8114879727363586,
-      "learning_rate": 0.00010898434970318401,
-      "loss": 3.1797,
+      "grad_norm": 0.8523650765419006,
+      "learning_rate": 0.00010899082568807339,
+      "loss": 3.184,
       "step": 75950
     },
     {
       "epoch": 8.194070080862534,
-      "grad_norm": 0.7779016494750977,
-      "learning_rate": 0.00010866055045871559,
-      "loss": 3.1571,
+      "grad_norm": 0.7952580451965332,
+      "learning_rate": 0.00010866702644360495,
+      "loss": 3.1624,
       "step": 76000
     },
     {
       "epoch": 8.194070080862534,
-      "eval_accuracy": 0.3906270151715287,
-      "eval_loss": 3.3333170413970947,
-      "eval_runtime": 185.1654,
-      "eval_samples_per_second": 97.27,
-      "eval_steps_per_second": 6.081,
+      "eval_accuracy": 0.3898963245009763,
+      "eval_loss": 3.3378188610076904,
+      "eval_runtime": 187.3408,
+      "eval_samples_per_second": 96.14,
+      "eval_steps_per_second": 6.01,
       "step": 76000
     },
     {
       "epoch": 8.199460916442048,
-      "grad_norm": 0.8344588279724121,
-      "learning_rate": 0.00010833675121424717,
-      "loss": 3.1813,
+      "grad_norm": 0.86629718542099,
+      "learning_rate": 0.00010834322719913653,
+      "loss": 3.1843,
       "step": 76050
     },
     {
       "epoch": 8.204851752021563,
-      "grad_norm": 0.8400987386703491,
-      "learning_rate": 0.00010801295196977872,
-      "loss": 3.1983,
+      "grad_norm": 0.8768171668052673,
+      "learning_rate": 0.00010802590393955745,
+      "loss": 3.2064,
       "step": 76100
     },
     {
       "epoch": 8.210242587601078,
-      "grad_norm": 0.7751808762550354,
-      "learning_rate": 0.0001076891527253103,
-      "loss": 3.1688,
+      "grad_norm": 0.7895368933677673,
+      "learning_rate": 0.00010770210469508903,
+      "loss": 3.1727,
       "step": 76150
     },
     {
       "epoch": 8.215633423180593,
-      "grad_norm": 0.8149611353874207,
-      "learning_rate": 0.00010736535348084187,
-      "loss": 3.1826,
+      "grad_norm": 0.8563736081123352,
+      "learning_rate": 0.00010737830545062061,
+      "loss": 3.1873,
       "step": 76200
     },
     {
       "epoch": 8.221024258760108,
-      "grad_norm": 0.8001996874809265,
-      "learning_rate": 0.00010704155423637345,
-      "loss": 3.1785,
+      "grad_norm": 0.8298239707946777,
+      "learning_rate": 0.00010705450620615219,
+      "loss": 3.1816,
       "step": 76250
     },
     {
       "epoch": 8.226415094339623,
-      "grad_norm": 0.8465657830238342,
-      "learning_rate": 0.000106717754991905,
-      "loss": 3.1606,
+      "grad_norm": 0.9139582514762878,
+      "learning_rate": 0.00010673070696168375,
+      "loss": 3.1634,
       "step": 76300
     },
     {
       "epoch": 8.231805929919137,
-      "grad_norm": 0.8427553772926331,
-      "learning_rate": 0.00010639395574743658,
-      "loss": 3.1798,
+      "grad_norm": 0.8484526872634888,
+      "learning_rate": 0.00010640690771721531,
+      "loss": 3.1833,
       "step": 76350
     },
     {
       "epoch": 8.237196765498652,
-      "grad_norm": 0.7900227308273315,
-      "learning_rate": 0.00010607015650296816,
-      "loss": 3.1745,
+      "grad_norm": 0.8191407918930054,
+      "learning_rate": 0.00010608310847274689,
+      "loss": 3.1796,
       "step": 76400
     },
     {
       "epoch": 8.242587601078167,
-      "grad_norm": 0.7699751257896423,
-      "learning_rate": 0.00010574635725849972,
-      "loss": 3.175,
+      "grad_norm": 0.8060504198074341,
+      "learning_rate": 0.00010575930922827846,
+      "loss": 3.1798,
       "step": 76450
     },
     {
       "epoch": 8.247978436657682,
-      "grad_norm": 0.8397601246833801,
-      "learning_rate": 0.00010542255801403128,
-      "loss": 3.164,
+      "grad_norm": 0.882337212562561,
+      "learning_rate": 0.00010543550998381003,
+      "loss": 3.1668,
       "step": 76500
     },
     {
       "epoch": 8.253369272237197,
-      "grad_norm": 0.874344527721405,
-      "learning_rate": 0.00010509875876956286,
-      "loss": 3.1902,
+      "grad_norm": 0.8530251979827881,
+      "learning_rate": 0.0001051117107393416,
+      "loss": 3.1929,
       "step": 76550
     },
     {
       "epoch": 8.25876010781671,
-      "grad_norm": 0.7956122159957886,
-      "learning_rate": 0.00010477495952509443,
-      "loss": 3.1647,
+      "grad_norm": 0.868147075176239,
+      "learning_rate": 0.00010478791149487316,
+      "loss": 3.1715,
       "step": 76600
     },
     {
       "epoch": 8.264150943396226,
-      "grad_norm": 0.7708241939544678,
-      "learning_rate": 0.000104451160280626,
-      "loss": 3.1618,
+      "grad_norm": 0.8064590692520142,
+      "learning_rate": 0.00010446411225040474,
+      "loss": 3.1666,
       "step": 76650
     },
     {
       "epoch": 8.269541778975741,
-      "grad_norm": 0.8957728743553162,
-      "learning_rate": 0.00010412736103615758,
-      "loss": 3.1832,
+      "grad_norm": 0.8385329842567444,
+      "learning_rate": 0.00010414031300593632,
+      "loss": 3.1886,
       "step": 76700
     },
     {
       "epoch": 8.274932614555256,
-      "grad_norm": 0.8150855898857117,
-      "learning_rate": 0.00010380356179168913,
-      "loss": 3.1755,
+      "grad_norm": 0.807418167591095,
+      "learning_rate": 0.00010381651376146787,
+      "loss": 3.1801,
       "step": 76750
     },
     {
       "epoch": 8.280323450134771,
-      "grad_norm": 0.8756967186927795,
-      "learning_rate": 0.00010347976254722071,
-      "loss": 3.1797,
+      "grad_norm": 0.8563040494918823,
+      "learning_rate": 0.00010349271451699945,
+      "loss": 3.184,
       "step": 76800
     },
     {
       "epoch": 8.285714285714286,
-      "grad_norm": 0.7948899865150452,
-      "learning_rate": 0.00010315596330275229,
-      "loss": 3.1907,
+      "grad_norm": 0.8556200265884399,
+      "learning_rate": 0.00010316891527253102,
+      "loss": 3.194,
       "step": 76850
     },
     {
       "epoch": 8.2911051212938,
-      "grad_norm": 0.8002045154571533,
-      "learning_rate": 0.00010283216405828387,
-      "loss": 3.2053,
+      "grad_norm": 0.8641937375068665,
+      "learning_rate": 0.0001028451160280626,
+      "loss": 3.2111,
       "step": 76900
     },
     {
       "epoch": 8.296495956873315,
-      "grad_norm": 0.8174263834953308,
-      "learning_rate": 0.00010250836481381542,
-      "loss": 3.1878,
+      "grad_norm": 0.8860243558883667,
+      "learning_rate": 0.00010252131678359416,
+      "loss": 3.1899,
       "step": 76950
     },
     {
       "epoch": 8.30188679245283,
-      "grad_norm": 0.7678391933441162,
-      "learning_rate": 0.000102184565569347,
-      "loss": 3.1771,
+      "grad_norm": 0.8518773913383484,
+      "learning_rate": 0.00010219751753912573,
+      "loss": 3.1808,
       "step": 77000
     },
     {
       "epoch": 8.30188679245283,
-      "eval_accuracy": 0.3910068656673281,
-      "eval_loss": 3.3296449184417725,
-      "eval_runtime": 184.8006,
-      "eval_samples_per_second": 97.462,
-      "eval_steps_per_second": 6.093,
+      "eval_accuracy": 0.39007462388987985,
+      "eval_loss": 3.334059238433838,
+      "eval_runtime": 187.5237,
+      "eval_samples_per_second": 96.047,
+      "eval_steps_per_second": 6.005,
       "step": 77000
     },
     {
       "epoch": 8.307277628032345,
-      "grad_norm": 0.8229707479476929,
-      "learning_rate": 0.00010186724230976793,
-      "loss": 3.1903,
+      "grad_norm": 0.8479878306388855,
+      "learning_rate": 0.0001018737182946573,
+      "loss": 3.1943,
       "step": 77050
     },
     {
       "epoch": 8.31266846361186,
-      "grad_norm": 0.8444199562072754,
-      "learning_rate": 0.0001015434430652995,
-      "loss": 3.1739,
+      "grad_norm": 0.8413503170013428,
+      "learning_rate": 0.00010154991905018887,
+      "loss": 3.1781,
       "step": 77100
     },
     {
       "epoch": 8.318059299191376,
-      "grad_norm": 0.8374699354171753,
-      "learning_rate": 0.00010121964382083107,
-      "loss": 3.1627,
+      "grad_norm": 0.8530333042144775,
+      "learning_rate": 0.00010122611980572045,
+      "loss": 3.1658,
       "step": 77150
     },
     {
       "epoch": 8.323450134770889,
-      "grad_norm": 0.8763797283172607,
-      "learning_rate": 0.00010089584457636265,
-      "loss": 3.1634,
+      "grad_norm": 0.8598548769950867,
+      "learning_rate": 0.00010090232056125201,
+      "loss": 3.1676,
       "step": 77200
     },
     {
       "epoch": 8.328840970350404,
-      "grad_norm": 0.844142496585846,
-      "learning_rate": 0.00010057204533189423,
-      "loss": 3.1628,
+      "grad_norm": 0.8489903211593628,
+      "learning_rate": 0.00010057852131678359,
+      "loss": 3.1663,
       "step": 77250
     },
     {
       "epoch": 8.33423180592992,
-      "grad_norm": 0.8463607430458069,
-      "learning_rate": 0.00010024824608742578,
-      "loss": 3.1713,
+      "grad_norm": 0.8233048915863037,
+      "learning_rate": 0.00010025472207231515,
+      "loss": 3.1748,
       "step": 77300
     },
     {
       "epoch": 8.339622641509434,
-      "grad_norm": 0.816167414188385,
-      "learning_rate": 9.992444684295736e-05,
-      "loss": 3.1854,
+      "grad_norm": 0.8634786009788513,
+      "learning_rate": 9.993092282784673e-05,
+      "loss": 3.1883,
       "step": 77350
     },
     {
       "epoch": 8.34501347708895,
-      "grad_norm": 0.868462860584259,
-      "learning_rate": 9.960064759848894e-05,
-      "loss": 3.1387,
+      "grad_norm": 0.8366324305534363,
+      "learning_rate": 9.96071235833783e-05,
+      "loss": 3.144,
       "step": 77400
     },
     {
       "epoch": 8.350404312668463,
-      "grad_norm": 0.8351553082466125,
-      "learning_rate": 9.92768483540205e-05,
-      "loss": 3.1847,
+      "grad_norm": 0.8811054229736328,
+      "learning_rate": 9.928332433890986e-05,
+      "loss": 3.1893,
       "step": 77450
     },
     {
       "epoch": 8.355795148247978,
-      "grad_norm": 0.8340010046958923,
-      "learning_rate": 9.895304910955206e-05,
-      "loss": 3.1897,
+      "grad_norm": 0.8160362243652344,
+      "learning_rate": 9.895952509444144e-05,
+      "loss": 3.1928,
       "step": 77500
     },
     {
       "epoch": 8.361185983827493,
-      "grad_norm": 0.8664101362228394,
-      "learning_rate": 9.862924986508364e-05,
-      "loss": 3.1756,
+      "grad_norm": 0.8781470060348511,
+      "learning_rate": 9.863572584997302e-05,
+      "loss": 3.1806,
       "step": 77550
     },
     {
       "epoch": 8.366576819407008,
-      "grad_norm": 0.9245196580886841,
-      "learning_rate": 9.83054506206152e-05,
-      "loss": 3.1933,
+      "grad_norm": 0.9126551747322083,
+      "learning_rate": 9.831192660550457e-05,
+      "loss": 3.1956,
       "step": 77600
     },
     {
       "epoch": 8.371967654986523,
-      "grad_norm": 0.8184807300567627,
-      "learning_rate": 9.798165137614678e-05,
-      "loss": 3.1595,
+      "grad_norm": 0.9115016460418701,
+      "learning_rate": 9.798812736103614e-05,
+      "loss": 3.1629,
       "step": 77650
     },
     {
       "epoch": 8.377358490566039,
-      "grad_norm": 0.8540685176849365,
-      "learning_rate": 9.765785213167835e-05,
-      "loss": 3.1975,
+      "grad_norm": 0.8540595173835754,
+      "learning_rate": 9.766432811656772e-05,
+      "loss": 3.2022,
       "step": 77700
     },
     {
       "epoch": 8.382749326145552,
-      "grad_norm": 0.8300887942314148,
-      "learning_rate": 9.733405288720992e-05,
-      "loss": 3.1769,
+      "grad_norm": 0.8928021788597107,
+      "learning_rate": 9.73405288720993e-05,
+      "loss": 3.1811,
       "step": 77750
     },
     {
       "epoch": 8.388140161725067,
-      "grad_norm": 0.8534173369407654,
-      "learning_rate": 9.701025364274149e-05,
-      "loss": 3.1714,
+      "grad_norm": 0.912022054195404,
+      "learning_rate": 9.701672962763086e-05,
+      "loss": 3.1762,
       "step": 77800
     },
     {
       "epoch": 8.393530997304582,
-      "grad_norm": 0.8436583280563354,
-      "learning_rate": 9.668645439827307e-05,
-      "loss": 3.1783,
+      "grad_norm": 0.9288665652275085,
+      "learning_rate": 9.669293038316243e-05,
+      "loss": 3.1843,
       "step": 77850
     },
     {
       "epoch": 8.398921832884097,
-      "grad_norm": 0.8904436230659485,
-      "learning_rate": 9.636265515380464e-05,
-      "loss": 3.176,
+      "grad_norm": 0.9502650499343872,
+      "learning_rate": 9.6369131138694e-05,
+      "loss": 3.1813,
       "step": 77900
     },
     {
       "epoch": 8.404312668463612,
-      "grad_norm": 0.9166279435157776,
-      "learning_rate": 9.60388559093362e-05,
-      "loss": 3.1848,
+      "grad_norm": 0.9512535929679871,
+      "learning_rate": 9.604533189422557e-05,
+      "loss": 3.1882,
       "step": 77950
     },
     {
       "epoch": 8.409703504043126,
-      "grad_norm": 0.8319751024246216,
-      "learning_rate": 9.571505666486777e-05,
-      "loss": 3.1673,
+      "grad_norm": 0.886163055896759,
+      "learning_rate": 9.572153264975715e-05,
+      "loss": 3.1735,
       "step": 78000
     },
     {
       "epoch": 8.409703504043126,
-      "eval_accuracy": 0.3911980947498816,
-      "eval_loss": 3.3280577659606934,
-      "eval_runtime": 184.9642,
-      "eval_samples_per_second": 97.376,
-      "eval_steps_per_second": 6.088,
+      "eval_accuracy": 0.39053053140714944,
+      "eval_loss": 3.330672025680542,
+      "eval_runtime": 187.6606,
+      "eval_samples_per_second": 95.976,
+      "eval_steps_per_second": 6.0,
       "step": 78000
     },
     {
       "epoch": 8.415094339622641,
-      "grad_norm": 0.8120921850204468,
-      "learning_rate": 9.539125742039935e-05,
-      "loss": 3.1822,
+      "grad_norm": 0.8564638495445251,
+      "learning_rate": 9.539773340528871e-05,
+      "loss": 3.1861,
       "step": 78050
     },
     {
       "epoch": 8.420485175202156,
-      "grad_norm": 0.8287458419799805,
-      "learning_rate": 9.506745817593093e-05,
-      "loss": 3.1704,
+      "grad_norm": 0.7967727184295654,
+      "learning_rate": 9.508041014570966e-05,
+      "loss": 3.1744,
       "step": 78100
     },
     {
       "epoch": 8.425876010781671,
-      "grad_norm": 0.8519405722618103,
-      "learning_rate": 9.474365893146248e-05,
-      "loss": 3.1791,
+      "grad_norm": 0.8836035132408142,
+      "learning_rate": 9.475661090124121e-05,
+      "loss": 3.1818,
       "step": 78150
     },
     {
       "epoch": 8.431266846361186,
-      "grad_norm": 0.9082578420639038,
-      "learning_rate": 9.441985968699406e-05,
-      "loss": 3.1903,
+      "grad_norm": 0.9195553064346313,
+      "learning_rate": 9.443928764166217e-05,
+      "loss": 3.1958,
       "step": 78200
     },
     {
       "epoch": 8.436657681940702,
-      "grad_norm": 0.8822596669197083,
-      "learning_rate": 9.409606044252563e-05,
-      "loss": 3.2137,
+      "grad_norm": 0.8804324865341187,
+      "learning_rate": 9.411548839719374e-05,
+      "loss": 3.2166,
       "step": 78250
     },
     {
       "epoch": 8.442048517520215,
-      "grad_norm": 0.8615216016769409,
-      "learning_rate": 9.37722611980572e-05,
-      "loss": 3.1818,
+      "grad_norm": 0.8681713938713074,
+      "learning_rate": 9.37916891527253e-05,
+      "loss": 3.1861,
       "step": 78300
     },
     {
       "epoch": 8.44743935309973,
-      "grad_norm": 0.8587093353271484,
-      "learning_rate": 9.344846195358876e-05,
-      "loss": 3.1776,
+      "grad_norm": 0.849361002445221,
+      "learning_rate": 9.346788990825687e-05,
+      "loss": 3.1819,
       "step": 78350
     },
     {
       "epoch": 8.452830188679245,
-      "grad_norm": 0.8063041567802429,
-      "learning_rate": 9.312466270912034e-05,
-      "loss": 3.1846,
+      "grad_norm": 0.9182830452919006,
+      "learning_rate": 9.314409066378845e-05,
+      "loss": 3.1871,
       "step": 78400
     },
     {
       "epoch": 8.45822102425876,
-      "grad_norm": 0.8895376324653625,
-      "learning_rate": 9.28008634646519e-05,
-      "loss": 3.1933,
+      "grad_norm": 0.9341931939125061,
+      "learning_rate": 9.282029141932001e-05,
+      "loss": 3.1961,
       "step": 78450
     },
     {
       "epoch": 8.463611859838275,
-      "grad_norm": 0.8046665787696838,
-      "learning_rate": 9.247706422018348e-05,
-      "loss": 3.1872,
+      "grad_norm": 0.8424123525619507,
+      "learning_rate": 9.249649217485158e-05,
+      "loss": 3.1912,
       "step": 78500
     },
     {
       "epoch": 8.46900269541779,
-      "grad_norm": 0.8373346924781799,
-      "learning_rate": 9.215326497571505e-05,
-      "loss": 3.1793,
+      "grad_norm": 0.8723587989807129,
+      "learning_rate": 9.217269293038316e-05,
+      "loss": 3.1831,
       "step": 78550
     },
     {
       "epoch": 8.474393530997304,
-      "grad_norm": 0.8134214878082275,
-      "learning_rate": 9.182946573124661e-05,
-      "loss": 3.1889,
+      "grad_norm": 0.8485751748085022,
+      "learning_rate": 9.184889368591473e-05,
+      "loss": 3.1916,
       "step": 78600
     },
     {
       "epoch": 8.479784366576819,
-      "grad_norm": 0.8032200932502747,
-      "learning_rate": 9.150566648677819e-05,
-      "loss": 3.1765,
+      "grad_norm": 0.8241350054740906,
+      "learning_rate": 9.15250944414463e-05,
+      "loss": 3.1807,
       "step": 78650
     },
     {
       "epoch": 8.485175202156334,
-      "grad_norm": 0.8005744218826294,
-      "learning_rate": 9.118186724230976e-05,
-      "loss": 3.1795,
+      "grad_norm": 0.8575042486190796,
+      "learning_rate": 9.120129519697786e-05,
+      "loss": 3.1848,
       "step": 78700
     },
     {
       "epoch": 8.49056603773585,
-      "grad_norm": 0.8798218965530396,
-      "learning_rate": 9.085806799784134e-05,
-      "loss": 3.189,
+      "grad_norm": 0.8988580107688904,
+      "learning_rate": 9.087749595250944e-05,
+      "loss": 3.1934,
       "step": 78750
     },
     {
       "epoch": 8.495956873315365,
-      "grad_norm": 0.8683854937553406,
-      "learning_rate": 9.053426875337289e-05,
-      "loss": 3.1934,
+      "grad_norm": 0.8846838474273682,
+      "learning_rate": 9.0553696708041e-05,
+      "loss": 3.1993,
       "step": 78800
     },
     {
       "epoch": 8.501347708894878,
-      "grad_norm": 0.813085675239563,
-      "learning_rate": 9.021046950890447e-05,
-      "loss": 3.1887,
+      "grad_norm": 0.8227244019508362,
+      "learning_rate": 9.022989746357258e-05,
+      "loss": 3.1927,
       "step": 78850
     },
     {
       "epoch": 8.506738544474393,
-      "grad_norm": 0.8143117427825928,
-      "learning_rate": 8.988667026443605e-05,
-      "loss": 3.1808,
+      "grad_norm": 0.8900447487831116,
+      "learning_rate": 8.990609821910416e-05,
+      "loss": 3.1854,
       "step": 78900
     },
     {
       "epoch": 8.512129380053908,
-      "grad_norm": 0.8620724678039551,
-      "learning_rate": 8.956287101996761e-05,
-      "loss": 3.1792,
+      "grad_norm": 0.8625020384788513,
+      "learning_rate": 8.958229897463571e-05,
+      "loss": 3.1838,
       "step": 78950
     },
     {
       "epoch": 8.517520215633423,
-      "grad_norm": 0.8598374724388123,
-      "learning_rate": 8.923907177549918e-05,
-      "loss": 3.1806,
+      "grad_norm": 0.8397684097290039,
+      "learning_rate": 8.925849973016729e-05,
+      "loss": 3.185,
       "step": 79000
     },
     {
       "epoch": 8.517520215633423,
-      "eval_accuracy": 0.3915852249891646,
-      "eval_loss": 3.322683095932007,
-      "eval_runtime": 184.9558,
-      "eval_samples_per_second": 97.38,
-      "eval_steps_per_second": 6.088,
+      "eval_accuracy": 0.39076543895060434,
+      "eval_loss": 3.3277270793914795,
+      "eval_runtime": 185.0461,
+      "eval_samples_per_second": 97.332,
+      "eval_steps_per_second": 6.085,
       "step": 79000
     },
     {
       "epoch": 8.522911051212938,
-      "grad_norm": 0.8073253035545349,
-      "learning_rate": 8.891527253103075e-05,
-      "loss": 3.1988,
+      "grad_norm": 0.8842145800590515,
+      "learning_rate": 8.893470048569886e-05,
+      "loss": 3.2027,
       "step": 79050
     },
     {
       "epoch": 8.528301886792454,
-      "grad_norm": 0.8984984755516052,
-      "learning_rate": 8.859147328656233e-05,
-      "loss": 3.1804,
+      "grad_norm": 0.8830186128616333,
+      "learning_rate": 8.861090124123044e-05,
+      "loss": 3.1838,
       "step": 79100
     },
     {
       "epoch": 8.533692722371967,
-      "grad_norm": 0.843517541885376,
-      "learning_rate": 8.82676740420939e-05,
-      "loss": 3.1896,
+      "grad_norm": 0.8847579956054688,
+      "learning_rate": 8.828710199676199e-05,
+      "loss": 3.1935,
       "step": 79150
     },
     {
       "epoch": 8.539083557951482,
-      "grad_norm": 0.8529115319252014,
-      "learning_rate": 8.794387479762546e-05,
-      "loss": 3.2095,
+      "grad_norm": 0.889342725276947,
+      "learning_rate": 8.796330275229357e-05,
+      "loss": 3.2124,
       "step": 79200
     },
     {
       "epoch": 8.544474393530997,
-      "grad_norm": 0.8642679452896118,
-      "learning_rate": 8.762007555315704e-05,
-      "loss": 3.1868,
+      "grad_norm": 0.8822177648544312,
+      "learning_rate": 8.763950350782515e-05,
+      "loss": 3.1901,
       "step": 79250
     },
     {
       "epoch": 8.549865229110512,
-      "grad_norm": 0.8819208145141602,
-      "learning_rate": 8.72962763086886e-05,
-      "loss": 3.1621,
+      "grad_norm": 0.8914117813110352,
+      "learning_rate": 8.731570426335671e-05,
+      "loss": 3.1681,
       "step": 79300
     },
     {
       "epoch": 8.555256064690028,
-      "grad_norm": 0.8184266686439514,
-      "learning_rate": 8.697247706422018e-05,
-      "loss": 3.1841,
+      "grad_norm": 0.8982561826705933,
+      "learning_rate": 8.699190501888828e-05,
+      "loss": 3.1881,
       "step": 79350
     },
     {
       "epoch": 8.560646900269543,
-      "grad_norm": 0.9021525979042053,
-      "learning_rate": 8.664867781975174e-05,
-      "loss": 3.1821,
+      "grad_norm": 0.87257981300354,
+      "learning_rate": 8.666810577441985e-05,
+      "loss": 3.1854,
       "step": 79400
     },
     {
       "epoch": 8.566037735849056,
-      "grad_norm": 0.823969841003418,
-      "learning_rate": 8.632487857528331e-05,
-      "loss": 3.1666,
+      "grad_norm": 0.8853148818016052,
+      "learning_rate": 8.634430652995142e-05,
+      "loss": 3.1713,
       "step": 79450
     },
     {
       "epoch": 8.571428571428571,
-      "grad_norm": 0.8354356288909912,
-      "learning_rate": 8.600107933081489e-05,
-      "loss": 3.1763,
+      "grad_norm": 0.8575986623764038,
+      "learning_rate": 8.6020507285483e-05,
+      "loss": 3.179,
       "step": 79500
     },
     {
       "epoch": 8.576819407008086,
-      "grad_norm": 0.9359567165374756,
-      "learning_rate": 8.567728008634646e-05,
-      "loss": 3.176,
+      "grad_norm": 0.9323446750640869,
+      "learning_rate": 8.569670804101456e-05,
+      "loss": 3.1808,
       "step": 79550
     },
     {
       "epoch": 8.582210242587601,
-      "grad_norm": 0.8543683290481567,
-      "learning_rate": 8.535348084187804e-05,
-      "loss": 3.1616,
+      "grad_norm": 0.8653596639633179,
+      "learning_rate": 8.537290879654612e-05,
+      "loss": 3.1643,
       "step": 79600
     },
     {
       "epoch": 8.587601078167117,
-      "grad_norm": 0.8099145293235779,
-      "learning_rate": 8.502968159740959e-05,
-      "loss": 3.1743,
+      "grad_norm": 0.8415989279747009,
+      "learning_rate": 8.50491095520777e-05,
+      "loss": 3.1776,
       "step": 79650
     },
     {
       "epoch": 8.59299191374663,
-      "grad_norm": 0.8237737417221069,
-      "learning_rate": 8.470588235294117e-05,
-      "loss": 3.173,
+      "grad_norm": 0.8409903049468994,
+      "learning_rate": 8.472531030760928e-05,
+      "loss": 3.1777,
       "step": 79700
     },
     {
       "epoch": 8.598382749326145,
-      "grad_norm": 0.8273354768753052,
-      "learning_rate": 8.438208310847275e-05,
-      "loss": 3.1819,
+      "grad_norm": 0.878783643245697,
+      "learning_rate": 8.440151106314086e-05,
+      "loss": 3.1863,
       "step": 79750
     },
     {
       "epoch": 8.60377358490566,
-      "grad_norm": 0.8201135396957397,
-      "learning_rate": 8.405828386400431e-05,
-      "loss": 3.1865,
+      "grad_norm": 0.868650496006012,
+      "learning_rate": 8.407771181867241e-05,
+      "loss": 3.1884,
       "step": 79800
     },
     {
       "epoch": 8.609164420485175,
-      "grad_norm": 0.835292637348175,
-      "learning_rate": 8.373448461953587e-05,
-      "loss": 3.165,
+      "grad_norm": 0.8431050181388855,
+      "learning_rate": 8.375391257420398e-05,
+      "loss": 3.1687,
       "step": 79850
     },
     {
       "epoch": 8.61455525606469,
-      "grad_norm": 0.8473345041275024,
-      "learning_rate": 8.341068537506745e-05,
-      "loss": 3.2078,
+      "grad_norm": 0.8444762825965881,
+      "learning_rate": 8.343011332973556e-05,
+      "loss": 3.2114,
       "step": 79900
     },
     {
       "epoch": 8.619946091644206,
-      "grad_norm": 0.8289116024971008,
-      "learning_rate": 8.308688613059902e-05,
-      "loss": 3.1844,
+      "grad_norm": 0.8640735745429993,
+      "learning_rate": 8.310631408526714e-05,
+      "loss": 3.1881,
       "step": 79950
     },
     {
       "epoch": 8.625336927223719,
-      "grad_norm": 0.8131082653999329,
-      "learning_rate": 8.27630868861306e-05,
-      "loss": 3.1973,
+      "grad_norm": 0.8508297204971313,
+      "learning_rate": 8.278251484079869e-05,
+      "loss": 3.201,
       "step": 80000
     },
     {
       "epoch": 8.625336927223719,
-      "eval_accuracy": 0.39217618804598753,
-      "eval_loss": 3.3202123641967773,
-      "eval_runtime": 184.6219,
-      "eval_samples_per_second": 97.556,
-      "eval_steps_per_second": 6.099,
+      "eval_accuracy": 0.39141974664102314,
+      "eval_loss": 3.3241944313049316,
+      "eval_runtime": 185.1703,
+      "eval_samples_per_second": 97.267,
+      "eval_steps_per_second": 6.081,
       "step": 80000
     },
     {
       "epoch": 8.630727762803234,
-      "grad_norm": 0.8290890455245972,
-      "learning_rate": 8.243928764166216e-05,
-      "loss": 3.1902,
+      "grad_norm": 0.8537095189094543,
+      "learning_rate": 8.245871559633027e-05,
+      "loss": 3.1927,
       "step": 80050
     },
     {
       "epoch": 8.63611859838275,
-      "grad_norm": 0.8842519521713257,
-      "learning_rate": 8.211548839719374e-05,
-      "loss": 3.1727,
+      "grad_norm": 0.8756033182144165,
+      "learning_rate": 8.213491635186185e-05,
+      "loss": 3.1767,
       "step": 80100
     },
     {
       "epoch": 8.641509433962264,
-      "grad_norm": 0.8551476001739502,
-      "learning_rate": 8.179816513761467e-05,
-      "loss": 3.1815,
+      "grad_norm": 0.8737725615501404,
+      "learning_rate": 8.181111710739341e-05,
+      "loss": 3.1864,
       "step": 80150
     },
     {
       "epoch": 8.64690026954178,
-      "grad_norm": 0.8610032796859741,
-      "learning_rate": 8.147436589314624e-05,
-      "loss": 3.1982,
+      "grad_norm": 0.8691494464874268,
+      "learning_rate": 8.148731786292497e-05,
+      "loss": 3.2019,
       "step": 80200
     },
     {
       "epoch": 8.652291105121293,
-      "grad_norm": 0.8614827394485474,
-      "learning_rate": 8.115056664867782e-05,
-      "loss": 3.1849,
+      "grad_norm": 0.8661626577377319,
+      "learning_rate": 8.116351861845655e-05,
+      "loss": 3.1894,
       "step": 80250
     },
     {
       "epoch": 8.657681940700808,
-      "grad_norm": 0.8478200435638428,
-      "learning_rate": 8.082676740420938e-05,
-      "loss": 3.1728,
+      "grad_norm": 0.8543850183486938,
+      "learning_rate": 8.083971937398812e-05,
+      "loss": 3.1777,
       "step": 80300
     },
     {
       "epoch": 8.663072776280323,
-      "grad_norm": 0.8627998232841492,
-      "learning_rate": 8.050296815974096e-05,
-      "loss": 3.1857,
+      "grad_norm": 0.8591643571853638,
+      "learning_rate": 8.05159201295197e-05,
+      "loss": 3.1899,
       "step": 80350
     },
     {
       "epoch": 8.668463611859838,
-      "grad_norm": 0.8678106069564819,
-      "learning_rate": 8.017916891527252e-05,
-      "loss": 3.1834,
+      "grad_norm": 0.9583737850189209,
+      "learning_rate": 8.019212088505126e-05,
+      "loss": 3.1867,
       "step": 80400
     },
     {
       "epoch": 8.673854447439354,
-      "grad_norm": 0.8390142321586609,
-      "learning_rate": 7.985536967080409e-05,
-      "loss": 3.1827,
+      "grad_norm": 0.8576064705848694,
+      "learning_rate": 7.986832164058282e-05,
+      "loss": 3.1855,
       "step": 80450
     },
     {
       "epoch": 8.679245283018869,
-      "grad_norm": 0.8010692000389099,
-      "learning_rate": 7.953157042633566e-05,
-      "loss": 3.1803,
+      "grad_norm": 0.8681039214134216,
+      "learning_rate": 7.95445223961144e-05,
+      "loss": 3.1856,
       "step": 80500
     },
     {
       "epoch": 8.684636118598382,
-      "grad_norm": 0.9025766253471375,
-      "learning_rate": 7.920777118186724e-05,
-      "loss": 3.1869,
+      "grad_norm": 0.9212780594825745,
+      "learning_rate": 7.922072315164598e-05,
+      "loss": 3.1902,
       "step": 80550
     },
     {
       "epoch": 8.690026954177897,
-      "grad_norm": 0.83486008644104,
-      "learning_rate": 7.888397193739879e-05,
-      "loss": 3.1919,
+      "grad_norm": 0.8667877316474915,
+      "learning_rate": 7.889692390717755e-05,
+      "loss": 3.1975,
       "step": 80600
     },
     {
       "epoch": 8.695417789757412,
-      "grad_norm": 0.8118559122085571,
-      "learning_rate": 7.856017269293037e-05,
-      "loss": 3.1851,
+      "grad_norm": 0.8530960083007812,
+      "learning_rate": 7.85731246627091e-05,
+      "loss": 3.1916,
       "step": 80650
     },
     {
       "epoch": 8.700808625336927,
-      "grad_norm": 0.8715183734893799,
-      "learning_rate": 7.823637344846195e-05,
-      "loss": 3.1723,
+      "grad_norm": 0.8906033635139465,
+      "learning_rate": 7.824932541824068e-05,
+      "loss": 3.1747,
       "step": 80700
     },
     {
       "epoch": 8.706199460916443,
-      "grad_norm": 0.867709219455719,
-      "learning_rate": 7.791257420399353e-05,
-      "loss": 3.1792,
+      "grad_norm": 0.8984014391899109,
+      "learning_rate": 7.792552617377226e-05,
+      "loss": 3.1822,
       "step": 80750
     },
     {
       "epoch": 8.711590296495956,
-      "grad_norm": 0.8244636654853821,
-      "learning_rate": 7.758877495952508e-05,
-      "loss": 3.1918,
+      "grad_norm": 0.8454893231391907,
+      "learning_rate": 7.760172692930382e-05,
+      "loss": 3.1962,
       "step": 80800
     },
     {
       "epoch": 8.716981132075471,
-      "grad_norm": 0.8116188049316406,
-      "learning_rate": 7.726497571505665e-05,
-      "loss": 3.1863,
+      "grad_norm": 0.8464348912239075,
+      "learning_rate": 7.727792768483539e-05,
+      "loss": 3.1895,
       "step": 80850
     },
     {
       "epoch": 8.722371967654986,
-      "grad_norm": 0.8385442495346069,
-      "learning_rate": 7.694117647058823e-05,
-      "loss": 3.1733,
+      "grad_norm": 0.8532600402832031,
+      "learning_rate": 7.695412844036697e-05,
+      "loss": 3.1819,
       "step": 80900
     },
     {
       "epoch": 8.727762803234501,
-      "grad_norm": 0.8450207710266113,
-      "learning_rate": 7.661737722611981e-05,
-      "loss": 3.1752,
+      "grad_norm": 0.9035687446594238,
+      "learning_rate": 7.663032919589854e-05,
+      "loss": 3.1791,
       "step": 80950
     },
     {
       "epoch": 8.733153638814017,
-      "grad_norm": 0.855089545249939,
-      "learning_rate": 7.629357798165137e-05,
-      "loss": 3.1893,
+      "grad_norm": 0.840469241142273,
+      "learning_rate": 7.630652995143011e-05,
+      "loss": 3.1937,
       "step": 81000
     },
     {
       "epoch": 8.733153638814017,
-      "eval_accuracy": 0.3925210523119107,
-      "eval_loss": 3.317495346069336,
-      "eval_runtime": 184.9457,
-      "eval_samples_per_second": 97.385,
-      "eval_steps_per_second": 6.088,
+      "eval_accuracy": 0.39170474316576054,
+      "eval_loss": 3.321425437927246,
+      "eval_runtime": 183.6292,
+      "eval_samples_per_second": 98.084,
+      "eval_steps_per_second": 6.132,
       "step": 81000
     },
     {
       "epoch": 8.738544474393532,
-      "grad_norm": 0.841830849647522,
-      "learning_rate": 7.596977873718294e-05,
-      "loss": 3.1854,
+      "grad_norm": 0.8486626148223877,
+      "learning_rate": 7.598273070696167e-05,
+      "loss": 3.1892,
       "step": 81050
     },
     {
       "epoch": 8.743935309973045,
-      "grad_norm": 0.8838796019554138,
-      "learning_rate": 7.564597949271451e-05,
-      "loss": 3.1856,
+      "grad_norm": 0.9279254078865051,
+      "learning_rate": 7.565893146249325e-05,
+      "loss": 3.1906,
       "step": 81100
     },
     {
       "epoch": 8.74932614555256,
-      "grad_norm": 0.8529191613197327,
-      "learning_rate": 7.532218024824608e-05,
-      "loss": 3.1826,
+      "grad_norm": 0.8810277581214905,
+      "learning_rate": 7.533513221802481e-05,
+      "loss": 3.1865,
       "step": 81150
     },
     {
       "epoch": 8.754716981132075,
-      "grad_norm": 0.8259932398796082,
-      "learning_rate": 7.499838100377764e-05,
-      "loss": 3.1631,
+      "grad_norm": 0.8444656729698181,
+      "learning_rate": 7.501133297355639e-05,
+      "loss": 3.1664,
       "step": 81200
     },
     {
       "epoch": 8.76010781671159,
-      "grad_norm": 0.9050982594490051,
-      "learning_rate": 7.467458175930922e-05,
-      "loss": 3.1785,
+      "grad_norm": 0.9161481261253357,
+      "learning_rate": 7.468753372908796e-05,
+      "loss": 3.1839,
       "step": 81250
     },
     {
       "epoch": 8.765498652291106,
-      "grad_norm": 0.8278824687004089,
-      "learning_rate": 7.435078251484078e-05,
-      "loss": 3.1802,
+      "grad_norm": 0.849538266658783,
+      "learning_rate": 7.436373448461953e-05,
+      "loss": 3.1841,
       "step": 81300
     },
     {
       "epoch": 8.77088948787062,
-      "grad_norm": 0.8377477526664734,
-      "learning_rate": 7.402698327037236e-05,
-      "loss": 3.1874,
+      "grad_norm": 0.8838067054748535,
+      "learning_rate": 7.40399352401511e-05,
+      "loss": 3.1931,
       "step": 81350
     },
     {
       "epoch": 8.776280323450134,
-      "grad_norm": 0.8223797678947449,
-      "learning_rate": 7.370318402590393e-05,
-      "loss": 3.1866,
+      "grad_norm": 0.8445566892623901,
+      "learning_rate": 7.371613599568268e-05,
+      "loss": 3.1901,
       "step": 81400
     },
     {
       "epoch": 8.78167115902965,
-      "grad_norm": 0.8468875288963318,
-      "learning_rate": 7.33793847814355e-05,
-      "loss": 3.1894,
+      "grad_norm": 0.8848838210105896,
+      "learning_rate": 7.339233675121424e-05,
+      "loss": 3.1938,
       "step": 81450
     },
     {
       "epoch": 8.787061994609164,
-      "grad_norm": 0.8402231931686401,
-      "learning_rate": 7.305558553696707e-05,
-      "loss": 3.198,
+      "grad_norm": 0.8396682739257812,
+      "learning_rate": 7.306853750674582e-05,
+      "loss": 3.2009,
       "step": 81500
     },
     {
       "epoch": 8.79245283018868,
-      "grad_norm": 0.965043842792511,
-      "learning_rate": 7.273178629249865e-05,
-      "loss": 3.1848,
+      "grad_norm": 0.9209688901901245,
+      "learning_rate": 7.274473826227738e-05,
+      "loss": 3.1914,
       "step": 81550
     },
     {
       "epoch": 8.797843665768195,
-      "grad_norm": 0.8373719453811646,
-      "learning_rate": 7.240798704803021e-05,
-      "loss": 3.1881,
+      "grad_norm": 0.8716294765472412,
+      "learning_rate": 7.242093901780896e-05,
+      "loss": 3.1908,
       "step": 81600
     },
     {
       "epoch": 8.80323450134771,
-      "grad_norm": 0.859524130821228,
-      "learning_rate": 7.208418780356179e-05,
-      "loss": 3.1844,
+      "grad_norm": 0.9350876808166504,
+      "learning_rate": 7.209713977334052e-05,
+      "loss": 3.188,
       "step": 81650
     },
     {
       "epoch": 8.808625336927223,
-      "grad_norm": 0.87333744764328,
-      "learning_rate": 7.176038855909337e-05,
-      "loss": 3.1909,
+      "grad_norm": 0.9141872525215149,
+      "learning_rate": 7.17733405288721e-05,
+      "loss": 3.1961,
       "step": 81700
     },
     {
       "epoch": 8.814016172506738,
-      "grad_norm": 0.8479647636413574,
-      "learning_rate": 7.143658931462493e-05,
-      "loss": 3.1834,
+      "grad_norm": 0.8961148858070374,
+      "learning_rate": 7.144954128440366e-05,
+      "loss": 3.1865,
       "step": 81750
     },
     {
       "epoch": 8.819407008086253,
-      "grad_norm": 0.8975802659988403,
-      "learning_rate": 7.11127900701565e-05,
-      "loss": 3.1805,
+      "grad_norm": 0.935304582118988,
+      "learning_rate": 7.112574203993523e-05,
+      "loss": 3.1843,
       "step": 81800
     },
     {
       "epoch": 8.824797843665769,
-      "grad_norm": 0.8739588260650635,
-      "learning_rate": 7.078899082568807e-05,
-      "loss": 3.1803,
+      "grad_norm": 0.9347729086875916,
+      "learning_rate": 7.08019427954668e-05,
+      "loss": 3.1842,
       "step": 81850
     },
     {
       "epoch": 8.830188679245284,
-      "grad_norm": 0.8485525250434875,
-      "learning_rate": 7.046519158121964e-05,
-      "loss": 3.1731,
+      "grad_norm": 0.8696584701538086,
+      "learning_rate": 7.047814355099837e-05,
+      "loss": 3.1796,
       "step": 81900
     },
     {
       "epoch": 8.835579514824797,
-      "grad_norm": 0.8152867555618286,
-      "learning_rate": 7.014139233675121e-05,
-      "loss": 3.2022,
+      "grad_norm": 0.8590584993362427,
+      "learning_rate": 7.015434430652993e-05,
+      "loss": 3.2031,
       "step": 81950
     },
     {
       "epoch": 8.840970350404312,
-      "grad_norm": 0.8222323060035706,
-      "learning_rate": 6.981759309228278e-05,
-      "loss": 3.1843,
+      "grad_norm": 0.8546836972236633,
+      "learning_rate": 6.983054506206151e-05,
+      "loss": 3.1875,
       "step": 82000
     },
     {
       "epoch": 8.840970350404312,
-      "eval_accuracy": 0.3926712106028703,
-      "eval_loss": 3.3136684894561768,
-      "eval_runtime": 184.9068,
-      "eval_samples_per_second": 97.406,
-      "eval_steps_per_second": 6.09,
+      "eval_accuracy": 0.39172136705759614,
+      "eval_loss": 3.318574905395508,
+      "eval_runtime": 182.7549,
+      "eval_samples_per_second": 98.553,
+      "eval_steps_per_second": 6.161,
       "step": 82000
     },
     {
       "epoch": 8.846361185983827,
-      "grad_norm": 0.8213887214660645,
-      "learning_rate": 6.949379384781434e-05,
-      "loss": 3.1873,
+      "grad_norm": 0.8496612906455994,
+      "learning_rate": 6.950674581759309e-05,
+      "loss": 3.1926,
       "step": 82050
     },
     {
       "epoch": 8.851752021563343,
-      "grad_norm": 0.8546909093856812,
-      "learning_rate": 6.916999460334592e-05,
-      "loss": 3.1781,
+      "grad_norm": 0.8842969536781311,
+      "learning_rate": 6.918294657312465e-05,
+      "loss": 3.1819,
       "step": 82100
     },
     {
       "epoch": 8.857142857142858,
-      "grad_norm": 0.8814563751220703,
-      "learning_rate": 6.884619535887748e-05,
-      "loss": 3.1857,
+      "grad_norm": 0.8467461466789246,
+      "learning_rate": 6.885914732865623e-05,
+      "loss": 3.1909,
       "step": 82150
     },
     {
       "epoch": 8.862533692722373,
-      "grad_norm": 0.8545170426368713,
-      "learning_rate": 6.852239611440906e-05,
-      "loss": 3.1847,
+      "grad_norm": Infinity,
+      "learning_rate": 6.854182406907717e-05,
+      "loss": 3.1884,
       "step": 82200
     },
     {
       "epoch": 8.867924528301886,
-      "grad_norm": 0.8242815732955933,
-      "learning_rate": 6.819859686994062e-05,
-      "loss": 3.1869,
+      "grad_norm": 0.8914542198181152,
+      "learning_rate": 6.821802482460873e-05,
+      "loss": 3.1907,
       "step": 82250
     },
     {
       "epoch": 8.873315363881401,
-      "grad_norm": 0.8673837184906006,
-      "learning_rate": 6.78747976254722e-05,
-      "loss": 3.1932,
+      "grad_norm": 0.8652558326721191,
+      "learning_rate": 6.78942255801403e-05,
+      "loss": 3.1959,
       "step": 82300
     },
     {
       "epoch": 8.878706199460916,
-      "grad_norm": 0.8401963114738464,
-      "learning_rate": 6.755747436589314e-05,
-      "loss": 3.1965,
+      "grad_norm": 0.8524869680404663,
+      "learning_rate": 6.757690232056125e-05,
+      "loss": 3.1992,
       "step": 82350
     },
     {
       "epoch": 8.884097035040432,
-      "grad_norm": 0.8353580236434937,
-      "learning_rate": 6.72336751214247e-05,
-      "loss": 3.172,
+      "grad_norm": 0.8906552791595459,
+      "learning_rate": 6.725310307609281e-05,
+      "loss": 3.1752,
       "step": 82400
     },
     {
       "epoch": 8.889487870619947,
-      "grad_norm": 0.8120338320732117,
-      "learning_rate": 6.690987587695628e-05,
-      "loss": 3.1932,
+      "grad_norm": 0.8732309341430664,
+      "learning_rate": 6.692930383162439e-05,
+      "loss": 3.1986,
       "step": 82450
     },
     {
       "epoch": 8.89487870619946,
-      "grad_norm": 0.8129428029060364,
-      "learning_rate": 6.658607663248785e-05,
-      "loss": 3.1704,
+      "grad_norm": 0.8735564351081848,
+      "learning_rate": 6.660550458715596e-05,
+      "loss": 3.1768,
       "step": 82500
     },
     {
       "epoch": 8.900269541778975,
-      "grad_norm": 0.8249984979629517,
-      "learning_rate": 6.626227738801942e-05,
-      "loss": 3.1927,
+      "grad_norm": 0.8268265724182129,
+      "learning_rate": 6.628170534268752e-05,
+      "loss": 3.1971,
       "step": 82550
     },
     {
       "epoch": 8.90566037735849,
-      "grad_norm": 0.8213885426521301,
-      "learning_rate": 6.593847814355099e-05,
-      "loss": 3.1605,
+      "grad_norm": 0.8611326813697815,
+      "learning_rate": 6.59579060982191e-05,
+      "loss": 3.1651,
       "step": 82600
     },
     {
       "epoch": 8.911051212938006,
-      "grad_norm": 0.8929885029792786,
-      "learning_rate": 6.561467889908257e-05,
-      "loss": 3.1752,
+      "grad_norm": 0.887067437171936,
+      "learning_rate": 6.563410685375066e-05,
+      "loss": 3.1799,
       "step": 82650
     },
     {
       "epoch": 8.91644204851752,
-      "grad_norm": 0.8327133655548096,
-      "learning_rate": 6.529087965461413e-05,
-      "loss": 3.1778,
+      "grad_norm": 0.8338527679443359,
+      "learning_rate": 6.531030760928224e-05,
+      "loss": 3.1839,
       "step": 82700
     },
     {
       "epoch": 8.921832884097036,
-      "grad_norm": 0.8178892731666565,
-      "learning_rate": 6.496708041014571e-05,
-      "loss": 3.1733,
+      "grad_norm": 0.8680648803710938,
+      "learning_rate": 6.49865083648138e-05,
+      "loss": 3.1786,
       "step": 82750
     },
     {
       "epoch": 8.92722371967655,
-      "grad_norm": 0.8787632584571838,
-      "learning_rate": 6.464328116567727e-05,
-      "loss": 3.1881,
+      "grad_norm": 0.8716435432434082,
+      "learning_rate": 6.466270912034538e-05,
+      "loss": 3.1926,
       "step": 82800
     },
     {
       "epoch": 8.932614555256064,
-      "grad_norm": 0.8857139945030212,
-      "learning_rate": 6.431948192120885e-05,
-      "loss": 3.1853,
+      "grad_norm": 0.8889674544334412,
+      "learning_rate": 6.433890987587695e-05,
+      "loss": 3.1891,
       "step": 82850
     },
     {
       "epoch": 8.93800539083558,
-      "grad_norm": 0.8443358540534973,
-      "learning_rate": 6.399568267674041e-05,
-      "loss": 3.1897,
+      "grad_norm": 0.8832100033760071,
+      "learning_rate": 6.401511063140852e-05,
+      "loss": 3.1935,
       "step": 82900
     },
     {
       "epoch": 8.943396226415095,
-      "grad_norm": 0.7980090379714966,
-      "learning_rate": 6.367188343227199e-05,
-      "loss": 3.1898,
+      "grad_norm": 0.8371535539627075,
+      "learning_rate": 6.369131138694009e-05,
+      "loss": 3.1934,
       "step": 82950
     },
     {
       "epoch": 8.94878706199461,
-      "grad_norm": 0.9037806391716003,
-      "learning_rate": 6.334808418780356e-05,
-      "loss": 3.1679,
+      "grad_norm": 0.8477622866630554,
+      "learning_rate": 6.336751214247167e-05,
+      "loss": 3.171,
       "step": 83000
     },
     {
       "epoch": 8.94878706199461,
-      "eval_accuracy": 0.3930123806706078,
-      "eval_loss": 3.31054425239563,
-      "eval_runtime": 184.7094,
-      "eval_samples_per_second": 97.51,
-      "eval_steps_per_second": 6.096,
+      "eval_accuracy": 0.39235111919536886,
+      "eval_loss": 3.314960479736328,
+      "eval_runtime": 182.7623,
+      "eval_samples_per_second": 98.549,
+      "eval_steps_per_second": 6.161,
       "step": 83000
     },
     {
       "epoch": 8.954177897574123,
-      "grad_norm": 0.8617427945137024,
-      "learning_rate": 6.302428494333512e-05,
-      "loss": 3.1946,
+      "grad_norm": 0.8831604719161987,
+      "learning_rate": 6.304371289800323e-05,
+      "loss": 3.1991,
       "step": 83050
     },
     {
       "epoch": 8.959568733153638,
-      "grad_norm": 0.8506762385368347,
-      "learning_rate": 6.27004856988667e-05,
-      "loss": 3.181,
+      "grad_norm": 0.8792138695716858,
+      "learning_rate": 6.271991365353481e-05,
+      "loss": 3.1855,
       "step": 83100
     },
     {
       "epoch": 8.964959568733153,
-      "grad_norm": 0.8383913040161133,
-      "learning_rate": 6.237668645439826e-05,
-      "loss": 3.1833,
+      "grad_norm": 0.886349081993103,
+      "learning_rate": 6.239611440906637e-05,
+      "loss": 3.1878,
       "step": 83150
     },
     {
       "epoch": 8.970350404312669,
-      "grad_norm": 0.8611315488815308,
-      "learning_rate": 6.205288720992984e-05,
-      "loss": 3.1827,
+      "grad_norm": 0.9038736820220947,
+      "learning_rate": 6.207231516459795e-05,
+      "loss": 3.1865,
       "step": 83200
     },
     {
       "epoch": 8.975741239892184,
-      "grad_norm": 0.8510012626647949,
-      "learning_rate": 6.17290879654614e-05,
-      "loss": 3.1884,
+      "grad_norm": 0.8625702857971191,
+      "learning_rate": 6.174851592012951e-05,
+      "loss": 3.1921,
       "step": 83250
     },
     {
       "epoch": 8.981132075471699,
-      "grad_norm": 0.826678991317749,
-      "learning_rate": 6.140528872099298e-05,
-      "loss": 3.1951,
+      "grad_norm": 0.8472139239311218,
+      "learning_rate": 6.142471667566109e-05,
+      "loss": 3.2,
       "step": 83300
     },
     {
       "epoch": 8.986522911051212,
-      "grad_norm": 0.8059219121932983,
-      "learning_rate": 6.108148947652455e-05,
-      "loss": 3.1925,
+      "grad_norm": 0.8829624652862549,
+      "learning_rate": 6.110091743119266e-05,
+      "loss": 3.1968,
       "step": 83350
     },
     {
       "epoch": 8.991913746630727,
-      "grad_norm": 0.8477720618247986,
-      "learning_rate": 6.075769023205612e-05,
-      "loss": 3.1788,
+      "grad_norm": 0.8627926111221313,
+      "learning_rate": 6.0777118186724226e-05,
+      "loss": 3.1819,
       "step": 83400
     },
     {
       "epoch": 8.997304582210242,
-      "grad_norm": 0.8244584798812866,
-      "learning_rate": 6.043389098758769e-05,
-      "loss": 3.1788,
+      "grad_norm": 0.887557864189148,
+      "learning_rate": 6.04533189422558e-05,
+      "loss": 3.1852,
       "step": 83450
     },
     {
       "epoch": 9.002695417789758,
-      "grad_norm": 0.8307106494903564,
-      "learning_rate": 6.0110091743119265e-05,
-      "loss": 3.155,
+      "grad_norm": 0.8692404627799988,
+      "learning_rate": 6.012951969778737e-05,
+      "loss": 3.159,
       "step": 83500
     },
     {
       "epoch": 9.008086253369273,
-      "grad_norm": 0.880568265914917,
-      "learning_rate": 5.978629249865083e-05,
-      "loss": 3.1189,
+      "grad_norm": 0.8924230337142944,
+      "learning_rate": 5.980572045331894e-05,
+      "loss": 3.1223,
       "step": 83550
     },
     {
       "epoch": 9.013477088948788,
-      "grad_norm": 0.8291214108467102,
-      "learning_rate": 5.94624932541824e-05,
-      "loss": 3.1419,
+      "grad_norm": 0.9057029485702515,
+      "learning_rate": 5.94819212088505e-05,
+      "loss": 3.1443,
       "step": 83600
     },
     {
       "epoch": 9.018867924528301,
-      "grad_norm": 0.814835250377655,
-      "learning_rate": 5.913869400971397e-05,
-      "loss": 3.1524,
+      "grad_norm": 0.8435297608375549,
+      "learning_rate": 5.915812196438208e-05,
+      "loss": 3.1564,
       "step": 83650
     },
     {
       "epoch": 9.024258760107816,
-      "grad_norm": 0.8376071453094482,
-      "learning_rate": 5.881489476524554e-05,
-      "loss": 3.1034,
+      "grad_norm": 0.8615345358848572,
+      "learning_rate": 5.8834322719913645e-05,
+      "loss": 3.1076,
       "step": 83700
     },
     {
       "epoch": 9.029649595687331,
-      "grad_norm": 0.8205394744873047,
-      "learning_rate": 5.849109552077711e-05,
-      "loss": 3.1062,
+      "grad_norm": 0.8578421473503113,
+      "learning_rate": 5.851052347544522e-05,
+      "loss": 3.1086,
       "step": 83750
     },
     {
       "epoch": 9.035040431266847,
-      "grad_norm": 0.8897798657417297,
-      "learning_rate": 5.8167296276308683e-05,
-      "loss": 3.1204,
+      "grad_norm": 0.9250162839889526,
+      "learning_rate": 5.8186724230976786e-05,
+      "loss": 3.1241,
       "step": 83800
     },
     {
       "epoch": 9.040431266846362,
-      "grad_norm": 0.8899309039115906,
-      "learning_rate": 5.784349703184026e-05,
-      "loss": 3.133,
+      "grad_norm": 0.9021680951118469,
+      "learning_rate": 5.786292498650836e-05,
+      "loss": 3.138,
       "step": 83850
     },
     {
       "epoch": 9.045822102425875,
-      "grad_norm": 0.8886575698852539,
-      "learning_rate": 5.7519697787371825e-05,
-      "loss": 3.1176,
+      "grad_norm": 0.9132741689682007,
+      "learning_rate": 5.753912574203993e-05,
+      "loss": 3.1225,
       "step": 83900
     },
     {
       "epoch": 9.05121293800539,
-      "grad_norm": 0.8485926389694214,
-      "learning_rate": 5.7195898542903396e-05,
-      "loss": 3.102,
+      "grad_norm": 0.8596984148025513,
+      "learning_rate": 5.72153264975715e-05,
+      "loss": 3.1047,
       "step": 83950
     },
     {
       "epoch": 9.056603773584905,
-      "grad_norm": 0.8616915941238403,
-      "learning_rate": 5.687209929843497e-05,
-      "loss": 3.1293,
+      "grad_norm": 0.87913578748703,
+      "learning_rate": 5.689152725310308e-05,
+      "loss": 3.133,
       "step": 84000
     },
     {
       "epoch": 9.056603773584905,
-      "eval_accuracy": 0.3930330247192925,
-      "eval_loss": 3.3129842281341553,
-      "eval_runtime": 184.7976,
-      "eval_samples_per_second": 97.463,
-      "eval_steps_per_second": 6.093,
+      "eval_accuracy": 0.3923710026738389,
+      "eval_loss": 3.317009925842285,
+      "eval_runtime": 182.5973,
+      "eval_samples_per_second": 98.638,
+      "eval_steps_per_second": 6.167,
       "step": 84000
     },
     {
       "epoch": 9.06199460916442,
-      "grad_norm": 0.8321478962898254,
-      "learning_rate": 5.654830005396654e-05,
-      "loss": 3.127,
+      "grad_norm": 0.8598724007606506,
+      "learning_rate": 5.656772800863464e-05,
+      "loss": 3.1318,
       "step": 84050
     },
     {
       "epoch": 9.067385444743936,
-      "grad_norm": 0.883215069770813,
-      "learning_rate": 5.62245008094981e-05,
-      "loss": 3.1293,
+      "grad_norm": 0.8952779173851013,
+      "learning_rate": 5.624392876416622e-05,
+      "loss": 3.1333,
       "step": 84100
     },
     {
       "epoch": 9.07277628032345,
-      "grad_norm": 0.8449257612228394,
-      "learning_rate": 5.590070156502968e-05,
-      "loss": 3.1404,
+      "grad_norm": 0.8883029222488403,
+      "learning_rate": 5.592012951969778e-05,
+      "loss": 3.1458,
       "step": 84150
     },
     {
       "epoch": 9.078167115902964,
-      "grad_norm": 0.8433477878570557,
-      "learning_rate": 5.5576902320561244e-05,
-      "loss": 3.117,
+      "grad_norm": 0.8887649178504944,
+      "learning_rate": 5.5596330275229353e-05,
+      "loss": 3.1202,
       "step": 84200
     },
     {
       "epoch": 9.08355795148248,
-      "grad_norm": 0.8466358780860901,
-      "learning_rate": 5.525310307609282e-05,
-      "loss": 3.1382,
+      "grad_norm": 0.8505600690841675,
+      "learning_rate": 5.5272531030760924e-05,
+      "loss": 3.1417,
       "step": 84250
     },
     {
       "epoch": 9.088948787061994,
-      "grad_norm": 0.8381649255752563,
-      "learning_rate": 5.4929303831624386e-05,
-      "loss": 3.1199,
+      "grad_norm": 0.8616210222244263,
+      "learning_rate": 5.4948731786292495e-05,
+      "loss": 3.1248,
       "step": 84300
     },
     {
       "epoch": 9.09433962264151,
-      "grad_norm": 0.840488076210022,
-      "learning_rate": 5.460550458715596e-05,
-      "loss": 3.1297,
+      "grad_norm": 0.9372165203094482,
+      "learning_rate": 5.462493254182406e-05,
+      "loss": 3.1337,
       "step": 84350
     },
     {
       "epoch": 9.099730458221025,
-      "grad_norm": 0.8325525522232056,
-      "learning_rate": 5.428170534268753e-05,
-      "loss": 3.0981,
+      "grad_norm": 0.8919275403022766,
+      "learning_rate": 5.430113329735564e-05,
+      "loss": 3.1015,
       "step": 84400
     },
     {
       "epoch": 9.10512129380054,
-      "grad_norm": 0.8756667971611023,
-      "learning_rate": 5.39579060982191e-05,
-      "loss": 3.111,
+      "grad_norm": 0.9315557479858398,
+      "learning_rate": 5.39773340528872e-05,
+      "loss": 3.1145,
       "step": 84450
     },
     {
       "epoch": 9.110512129380053,
-      "grad_norm": 0.8472950458526611,
-      "learning_rate": 5.363410685375067e-05,
-      "loss": 3.1322,
+      "grad_norm": 0.8863003849983215,
+      "learning_rate": 5.365353480841878e-05,
+      "loss": 3.137,
       "step": 84500
     },
     {
       "epoch": 9.115902964959568,
-      "grad_norm": 0.818108081817627,
-      "learning_rate": 5.331030760928224e-05,
-      "loss": 3.1101,
+      "grad_norm": 0.8823902606964111,
+      "learning_rate": 5.332973556395034e-05,
+      "loss": 3.114,
       "step": 84550
     },
     {
       "epoch": 9.121293800539084,
-      "grad_norm": 0.8659102916717529,
-      "learning_rate": 5.2986508364813804e-05,
-      "loss": 3.1253,
+      "grad_norm": 0.905902624130249,
+      "learning_rate": 5.300593631948192e-05,
+      "loss": 3.131,
       "step": 84600
     },
     {
       "epoch": 9.126684636118599,
-      "grad_norm": 0.8552064895629883,
-      "learning_rate": 5.266270912034538e-05,
-      "loss": 3.1252,
+      "grad_norm": 0.8999488949775696,
+      "learning_rate": 5.2682137075013485e-05,
+      "loss": 3.1285,
       "step": 84650
     },
     {
       "epoch": 9.132075471698114,
-      "grad_norm": 0.8596740365028381,
-      "learning_rate": 5.233890987587695e-05,
-      "loss": 3.1311,
+      "grad_norm": 0.8922767639160156,
+      "learning_rate": 5.2358337830545056e-05,
+      "loss": 3.137,
       "step": 84700
     },
     {
       "epoch": 9.137466307277627,
-      "grad_norm": 0.8284486532211304,
-      "learning_rate": 5.2015110631408524e-05,
-      "loss": 3.1365,
+      "grad_norm": 0.860727071762085,
+      "learning_rate": 5.2034538586076626e-05,
+      "loss": 3.1406,
       "step": 84750
     },
     {
       "epoch": 9.142857142857142,
-      "grad_norm": 0.8481780886650085,
-      "learning_rate": 5.1691311386940094e-05,
-      "loss": 3.131,
+      "grad_norm": 0.8779696226119995,
+      "learning_rate": 5.17107393416082e-05,
+      "loss": 3.1358,
       "step": 84800
     },
     {
       "epoch": 9.148247978436657,
-      "grad_norm": 0.824164092540741,
-      "learning_rate": 5.136751214247166e-05,
-      "loss": 3.1193,
+      "grad_norm": 0.8878582119941711,
+      "learning_rate": 5.1386940097139775e-05,
+      "loss": 3.1224,
       "step": 84850
     },
     {
       "epoch": 9.153638814016173,
-      "grad_norm": 0.851373016834259,
-      "learning_rate": 5.1050188882892604e-05,
-      "loss": 3.1248,
+      "grad_norm": 0.8620050549507141,
+      "learning_rate": 5.106961683756071e-05,
+      "loss": 3.1277,
       "step": 84900
     },
     {
       "epoch": 9.159029649595688,
-      "grad_norm": 0.8321095705032349,
-      "learning_rate": 5.072638963842417e-05,
-      "loss": 3.1121,
+      "grad_norm": 0.8673367500305176,
+      "learning_rate": 5.074581759309228e-05,
+      "loss": 3.1167,
       "step": 84950
     },
     {
       "epoch": 9.164420485175203,
-      "grad_norm": 0.7981374263763428,
-      "learning_rate": 5.0402590393955746e-05,
-      "loss": 3.1214,
+      "grad_norm": 0.8600957989692688,
+      "learning_rate": 5.042201834862385e-05,
+      "loss": 3.1247,
       "step": 85000
     },
     {
       "epoch": 9.164420485175203,
-      "eval_accuracy": 0.39314189491288265,
-      "eval_loss": 3.31256365776062,
-      "eval_runtime": 184.9292,
-      "eval_samples_per_second": 97.394,
-      "eval_steps_per_second": 6.089,
+      "eval_accuracy": 0.3923831717972741,
+      "eval_loss": 3.3169844150543213,
+      "eval_runtime": 182.6354,
+      "eval_samples_per_second": 98.617,
+      "eval_steps_per_second": 6.165,
       "step": 85000
     },
     {
       "epoch": 9.169811320754716,
-      "grad_norm": 0.8480749726295471,
-      "learning_rate": 5.007879114948731e-05,
-      "loss": 3.1318,
+      "grad_norm": 0.8784777522087097,
+      "learning_rate": 5.009821910415542e-05,
+      "loss": 3.1363,
       "step": 85050
     },
     {
       "epoch": 9.175202156334231,
-      "grad_norm": 0.8506357073783875,
-      "learning_rate": 4.975499190501889e-05,
-      "loss": 3.1311,
+      "grad_norm": 0.9139289259910583,
+      "learning_rate": 4.977441985968699e-05,
+      "loss": 3.1354,
       "step": 85100
     },
     {
       "epoch": 9.180592991913747,
-      "grad_norm": 0.864138662815094,
-      "learning_rate": 4.943119266055045e-05,
-      "loss": 3.1381,
+      "grad_norm": 0.8900482654571533,
+      "learning_rate": 4.945062061521856e-05,
+      "loss": 3.1426,
       "step": 85150
     },
     {
       "epoch": 9.185983827493262,
-      "grad_norm": 0.853118360042572,
-      "learning_rate": 4.910739341608202e-05,
-      "loss": 3.1358,
+      "grad_norm": 0.8851867914199829,
+      "learning_rate": 4.9126821370750125e-05,
+      "loss": 3.1369,
       "step": 85200
     },
     {
       "epoch": 9.191374663072777,
-      "grad_norm": 0.8572016358375549,
-      "learning_rate": 4.87835941716136e-05,
-      "loss": 3.1298,
+      "grad_norm": 0.8578564524650574,
+      "learning_rate": 4.88030221262817e-05,
+      "loss": 3.1338,
       "step": 85250
     },
     {
       "epoch": 9.19676549865229,
-      "grad_norm": 0.8740936517715454,
-      "learning_rate": 4.8459794927145164e-05,
-      "loss": 3.1435,
+      "grad_norm": 0.8799201846122742,
+      "learning_rate": 4.847922288181327e-05,
+      "loss": 3.1482,
       "step": 85300
     },
     {
       "epoch": 9.202156334231805,
-      "grad_norm": 0.8369245529174805,
-      "learning_rate": 4.813599568267674e-05,
-      "loss": 3.1389,
+      "grad_norm": 0.8776422142982483,
+      "learning_rate": 4.8155423637344845e-05,
+      "loss": 3.1422,
       "step": 85350
     },
     {
       "epoch": 9.20754716981132,
-      "grad_norm": 0.8236960768699646,
-      "learning_rate": 4.7812196438208306e-05,
-      "loss": 3.123,
+      "grad_norm": 0.8514833450317383,
+      "learning_rate": 4.7831624392876416e-05,
+      "loss": 3.1298,
       "step": 85400
     },
     {
       "epoch": 9.212938005390836,
-      "grad_norm": 0.8511966466903687,
-      "learning_rate": 4.748839719373988e-05,
-      "loss": 3.124,
+      "grad_norm": 0.8941630125045776,
+      "learning_rate": 4.750782514840798e-05,
+      "loss": 3.1265,
       "step": 85450
     },
     {
       "epoch": 9.21832884097035,
-      "grad_norm": 0.8575359582901001,
-      "learning_rate": 4.716459794927145e-05,
-      "loss": 3.1427,
+      "grad_norm": 0.9248002171516418,
+      "learning_rate": 4.718402590393956e-05,
+      "loss": 3.1475,
       "step": 85500
     },
     {
       "epoch": 9.223719676549866,
-      "grad_norm": 0.8404532670974731,
-      "learning_rate": 4.684079870480302e-05,
-      "loss": 3.1437,
+      "grad_norm": 0.8932766914367676,
+      "learning_rate": 4.686022665947112e-05,
+      "loss": 3.1478,
       "step": 85550
     },
     {
       "epoch": 9.22911051212938,
-      "grad_norm": 0.9211980104446411,
-      "learning_rate": 4.651699946033459e-05,
-      "loss": 3.1337,
+      "grad_norm": 0.8614445328712463,
+      "learning_rate": 4.65364274150027e-05,
+      "loss": 3.1373,
       "step": 85600
     },
     {
       "epoch": 9.234501347708894,
-      "grad_norm": 0.820880115032196,
-      "learning_rate": 4.619320021586616e-05,
-      "loss": 3.1308,
+      "grad_norm": 0.8718250393867493,
+      "learning_rate": 4.621262817053426e-05,
+      "loss": 3.1338,
       "step": 85650
     },
     {
       "epoch": 9.23989218328841,
-      "grad_norm": 0.8520514369010925,
-      "learning_rate": 4.5869400971397725e-05,
-      "loss": 3.1308,
+      "grad_norm": 0.8721039295196533,
+      "learning_rate": 4.588882892606584e-05,
+      "loss": 3.1366,
       "step": 85700
     },
     {
       "epoch": 9.245283018867925,
-      "grad_norm": 0.8518171906471252,
-      "learning_rate": 4.55456017269293e-05,
-      "loss": 3.1324,
+      "grad_norm": 0.8820124864578247,
+      "learning_rate": 4.5565029681597405e-05,
+      "loss": 3.1367,
       "step": 85750
     },
     {
       "epoch": 9.25067385444744,
-      "grad_norm": 0.8360389471054077,
-      "learning_rate": 4.5221802482460866e-05,
-      "loss": 3.1236,
+      "grad_norm": 0.924517810344696,
+      "learning_rate": 4.5241230437128976e-05,
+      "loss": 3.1285,
       "step": 85800
     },
     {
       "epoch": 9.256064690026955,
-      "grad_norm": 0.8417567610740662,
-      "learning_rate": 4.4898003237992444e-05,
-      "loss": 3.1152,
+      "grad_norm": 0.850003719329834,
+      "learning_rate": 4.491743119266055e-05,
+      "loss": 3.1198,
       "step": 85850
     },
     {
       "epoch": 9.261455525606468,
-      "grad_norm": 0.8565623164176941,
-      "learning_rate": 4.457420399352401e-05,
-      "loss": 3.1323,
+      "grad_norm": 0.9066160321235657,
+      "learning_rate": 4.459363194819212e-05,
+      "loss": 3.1366,
       "step": 85900
     },
     {
       "epoch": 9.266846361185983,
-      "grad_norm": 0.8710352182388306,
-      "learning_rate": 4.425040474905558e-05,
-      "loss": 3.1216,
+      "grad_norm": 0.8946210145950317,
+      "learning_rate": 4.426983270372368e-05,
+      "loss": 3.1262,
       "step": 85950
     },
     {
       "epoch": 9.272237196765499,
-      "grad_norm": 0.8659188151359558,
-      "learning_rate": 4.392660550458716e-05,
-      "loss": 3.1466,
+      "grad_norm": 0.8711512088775635,
+      "learning_rate": 4.394603345925526e-05,
+      "loss": 3.1504,
       "step": 86000
     },
     {
       "epoch": 9.272237196765499,
-      "eval_accuracy": 0.3936142090162122,
-      "eval_loss": 3.3102121353149414,
-      "eval_runtime": 184.7159,
-      "eval_samples_per_second": 97.507,
-      "eval_steps_per_second": 6.096,
+      "eval_accuracy": 0.3928059401837603,
+      "eval_loss": 3.314445972442627,
+      "eval_runtime": 182.6134,
+      "eval_samples_per_second": 98.629,
+      "eval_steps_per_second": 6.166,
       "step": 86000
     },
     {
       "epoch": 9.277628032345014,
-      "grad_norm": 0.8267226219177246,
-      "learning_rate": 4.360280626011872e-05,
-      "loss": 3.1328,
+      "grad_norm": 0.8891922235488892,
+      "learning_rate": 4.36287101996762e-05,
+      "loss": 3.1364,
       "step": 86050
     },
     {
       "epoch": 9.283018867924529,
-      "grad_norm": 0.848037838935852,
-      "learning_rate": 4.32790070156503e-05,
-      "loss": 3.1186,
+      "grad_norm": 0.9130899310112,
+      "learning_rate": 4.330491095520777e-05,
+      "loss": 3.1237,
       "step": 86100
     },
     {
       "epoch": 9.288409703504042,
-      "grad_norm": 0.841865062713623,
-      "learning_rate": 4.295520777118186e-05,
-      "loss": 3.1225,
+      "grad_norm": 0.9084102511405945,
+      "learning_rate": 4.298111171073934e-05,
+      "loss": 3.1262,
       "step": 86150
     },
     {
       "epoch": 9.293800539083557,
-      "grad_norm": 0.8285828232765198,
-      "learning_rate": 4.263788451160281e-05,
-      "loss": 3.1311,
+      "grad_norm": 0.8524343967437744,
+      "learning_rate": 4.2657312466270904e-05,
+      "loss": 3.1347,
       "step": 86200
     },
     {
       "epoch": 9.299191374663073,
-      "grad_norm": 0.8383690714836121,
-      "learning_rate": 4.231408526713437e-05,
-      "loss": 3.1298,
+      "grad_norm": 0.8612839579582214,
+      "learning_rate": 4.233351322180248e-05,
+      "loss": 3.1352,
       "step": 86250
     },
     {
       "epoch": 9.304582210242588,
-      "grad_norm": 0.8436566591262817,
-      "learning_rate": 4.199028602266594e-05,
-      "loss": 3.129,
+      "grad_norm": 0.9126859903335571,
+      "learning_rate": 4.2009713977334046e-05,
+      "loss": 3.1341,
       "step": 86300
     },
     {
       "epoch": 9.309973045822103,
-      "grad_norm": 0.8680644631385803,
-      "learning_rate": 4.1666486778197514e-05,
-      "loss": 3.1218,
+      "grad_norm": 0.8642739653587341,
+      "learning_rate": 4.1685914732865623e-05,
+      "loss": 3.1263,
       "step": 86350
     },
     {
       "epoch": 9.315363881401618,
-      "grad_norm": 0.8822382092475891,
-      "learning_rate": 4.1342687533729085e-05,
-      "loss": 3.1494,
+      "grad_norm": 0.8648379445075989,
+      "learning_rate": 4.136211548839719e-05,
+      "loss": 3.153,
       "step": 86400
     },
     {
       "epoch": 9.320754716981131,
-      "grad_norm": 0.8971915245056152,
-      "learning_rate": 4.101888828926065e-05,
-      "loss": 3.1238,
+      "grad_norm": 0.8947865962982178,
+      "learning_rate": 4.1038316243928765e-05,
+      "loss": 3.1298,
       "step": 86450
     },
     {
       "epoch": 9.326145552560646,
-      "grad_norm": 0.8807439208030701,
-      "learning_rate": 4.0695089044792226e-05,
-      "loss": 3.1397,
+      "grad_norm": 0.9128478169441223,
+      "learning_rate": 4.071451699946033e-05,
+      "loss": 3.1444,
       "step": 86500
     },
     {
       "epoch": 9.331536388140162,
-      "grad_norm": 0.8657439351081848,
-      "learning_rate": 4.037128980032379e-05,
-      "loss": 3.1451,
+      "grad_norm": 0.9041945934295654,
+      "learning_rate": 4.03907177549919e-05,
+      "loss": 3.1497,
       "step": 86550
     },
     {
       "epoch": 9.336927223719677,
-      "grad_norm": 0.8694212436676025,
-      "learning_rate": 4.004749055585537e-05,
-      "loss": 3.1326,
+      "grad_norm": 0.8801517486572266,
+      "learning_rate": 4.006691851052347e-05,
+      "loss": 3.1374,
       "step": 86600
     },
     {
       "epoch": 9.342318059299192,
-      "grad_norm": 0.868144690990448,
-      "learning_rate": 3.972369131138694e-05,
-      "loss": 3.1345,
+      "grad_norm": 0.893440306186676,
+      "learning_rate": 3.974311926605504e-05,
+      "loss": 3.1378,
       "step": 86650
     },
     {
       "epoch": 9.347708894878707,
-      "grad_norm": 0.9292327761650085,
-      "learning_rate": 3.93998920669185e-05,
-      "loss": 3.1448,
+      "grad_norm": 0.9336996674537659,
+      "learning_rate": 3.9419320021586606e-05,
+      "loss": 3.1495,
       "step": 86700
     },
     {
       "epoch": 9.35309973045822,
-      "grad_norm": 0.8244055509567261,
-      "learning_rate": 3.907609282245008e-05,
-      "loss": 3.1568,
+      "grad_norm": 0.8711035847663879,
+      "learning_rate": 3.9095520777118184e-05,
+      "loss": 3.1598,
       "step": 86750
     },
     {
       "epoch": 9.358490566037736,
-      "grad_norm": 0.8296806216239929,
-      "learning_rate": 3.8752293577981645e-05,
-      "loss": 3.1391,
+      "grad_norm": 0.8829390406608582,
+      "learning_rate": 3.8771721532649755e-05,
+      "loss": 3.1448,
       "step": 86800
     },
     {
       "epoch": 9.36388140161725,
-      "grad_norm": 0.8584550023078918,
-      "learning_rate": 3.842849433351322e-05,
-      "loss": 3.1285,
+      "grad_norm": 0.896020770072937,
+      "learning_rate": 3.8447922288181326e-05,
+      "loss": 3.1338,
       "step": 86850
     },
     {
       "epoch": 9.369272237196766,
-      "grad_norm": 0.936661422252655,
-      "learning_rate": 3.810469508904479e-05,
-      "loss": 3.1412,
+      "grad_norm": 0.9287421703338623,
+      "learning_rate": 3.8124123043712896e-05,
+      "loss": 3.1432,
       "step": 86900
     },
     {
       "epoch": 9.374663072776281,
-      "grad_norm": 0.8722965717315674,
-      "learning_rate": 3.7780895844576364e-05,
-      "loss": 3.1186,
+      "grad_norm": 0.917233407497406,
+      "learning_rate": 3.780032379924447e-05,
+      "loss": 3.1242,
       "step": 86950
     },
     {
       "epoch": 9.380053908355794,
-      "grad_norm": 0.8549427390098572,
-      "learning_rate": 3.745709660010793e-05,
-      "loss": 3.1247,
+      "grad_norm": 0.8801265358924866,
+      "learning_rate": 3.747652455477603e-05,
+      "loss": 3.1282,
       "step": 87000
     },
     {
       "epoch": 9.380053908355794,
-      "eval_accuracy": 0.3936190983961639,
-      "eval_loss": 3.3095529079437256,
-      "eval_runtime": 184.7197,
-      "eval_samples_per_second": 97.504,
-      "eval_steps_per_second": 6.096,
+      "eval_accuracy": 0.3929825011264588,
+      "eval_loss": 3.3134233951568604,
+      "eval_runtime": 182.9573,
+      "eval_samples_per_second": 98.444,
+      "eval_steps_per_second": 6.154,
       "step": 87000
     },
     {
       "epoch": 9.38544474393531,
-      "grad_norm": 0.8127095699310303,
-      "learning_rate": 3.71332973556395e-05,
-      "loss": 3.1343,
+      "grad_norm": 0.8843125700950623,
+      "learning_rate": 3.71527253103076e-05,
+      "loss": 3.1383,
       "step": 87050
     },
     {
       "epoch": 9.390835579514825,
-      "grad_norm": 0.8294982314109802,
-      "learning_rate": 3.680949811117107e-05,
-      "loss": 3.1216,
+      "grad_norm": 0.886641800403595,
+      "learning_rate": 3.682892606583917e-05,
+      "loss": 3.1272,
       "step": 87100
     },
     {
       "epoch": 9.39622641509434,
-      "grad_norm": 0.8491564393043518,
-      "learning_rate": 3.648569886670264e-05,
-      "loss": 3.1247,
+      "grad_norm": 0.8681601881980896,
+      "learning_rate": 3.650512682137075e-05,
+      "loss": 3.1291,
       "step": 87150
     },
     {
       "epoch": 9.401617250673855,
-      "grad_norm": 0.8749399781227112,
-      "learning_rate": 3.616189962223421e-05,
-      "loss": 3.1505,
+      "grad_norm": 0.8855504393577576,
+      "learning_rate": 3.618132757690232e-05,
+      "loss": 3.1544,
       "step": 87200
     },
     {
       "epoch": 9.40700808625337,
-      "grad_norm": 0.8547865152359009,
-      "learning_rate": 3.583810037776578e-05,
-      "loss": 3.1383,
+      "grad_norm": 0.8667643666267395,
+      "learning_rate": 3.585752833243389e-05,
+      "loss": 3.1417,
       "step": 87250
     },
     {
       "epoch": 9.412398921832883,
-      "grad_norm": 0.8318867683410645,
-      "learning_rate": 3.5514301133297354e-05,
-      "loss": 3.1398,
+      "grad_norm": 0.8815639615058899,
+      "learning_rate": 3.553372908796546e-05,
+      "loss": 3.145,
       "step": 87300
     },
     {
       "epoch": 9.417789757412399,
-      "grad_norm": 0.8504889607429504,
-      "learning_rate": 3.5190501888828925e-05,
-      "loss": 3.1407,
+      "grad_norm": 0.8683713674545288,
+      "learning_rate": 3.520992984349703e-05,
+      "loss": 3.1443,
       "step": 87350
     },
     {
       "epoch": 9.423180592991914,
-      "grad_norm": 0.8165703415870667,
-      "learning_rate": 3.4866702644360496e-05,
-      "loss": 3.1343,
+      "grad_norm": 0.8790931701660156,
+      "learning_rate": 3.48861305990286e-05,
+      "loss": 3.1394,
       "step": 87400
     },
     {
       "epoch": 9.428571428571429,
-      "grad_norm": 0.8537405729293823,
-      "learning_rate": 3.4542903399892067e-05,
-      "loss": 3.142,
+      "grad_norm": 0.8913049697875977,
+      "learning_rate": 3.456233135456017e-05,
+      "loss": 3.1472,
       "step": 87450
     },
     {
       "epoch": 9.433962264150944,
-      "grad_norm": 0.8769167065620422,
-      "learning_rate": 3.421910415542363e-05,
-      "loss": 3.1425,
+      "grad_norm": 0.8946883678436279,
+      "learning_rate": 3.423853211009174e-05,
+      "loss": 3.1464,
       "step": 87500
     },
     {
       "epoch": 9.439353099730457,
-      "grad_norm": 0.815278172492981,
-      "learning_rate": 3.38953049109552e-05,
-      "loss": 3.1374,
+      "grad_norm": 0.8821413516998291,
+      "learning_rate": 3.391473286562331e-05,
+      "loss": 3.1425,
       "step": 87550
     },
     {
       "epoch": 9.444743935309972,
-      "grad_norm": 0.8766458630561829,
-      "learning_rate": 3.357150566648677e-05,
-      "loss": 3.1123,
+      "grad_norm": 0.8774822950363159,
+      "learning_rate": 3.359093362115488e-05,
+      "loss": 3.1159,
       "step": 87600
     },
     {
       "epoch": 9.450134770889488,
-      "grad_norm": 0.8410953283309937,
-      "learning_rate": 3.324770642201834e-05,
-      "loss": 3.1393,
+      "grad_norm": 0.9348732829093933,
+      "learning_rate": 3.326713437668645e-05,
+      "loss": 3.1457,
       "step": 87650
     },
     {
       "epoch": 9.455525606469003,
-      "grad_norm": 0.8328889012336731,
-      "learning_rate": 3.2923907177549914e-05,
-      "loss": 3.134,
+      "grad_norm": 0.8729662895202637,
+      "learning_rate": 3.2943335132218024e-05,
+      "loss": 3.1372,
       "step": 87700
     },
     {
       "epoch": 9.460916442048518,
-      "grad_norm": 0.9297603368759155,
-      "learning_rate": 3.2600107933081485e-05,
-      "loss": 3.1313,
+      "grad_norm": 0.92522132396698,
+      "learning_rate": 3.261953588774959e-05,
+      "loss": 3.1338,
       "step": 87750
     },
     {
       "epoch": 9.466307277628033,
-      "grad_norm": 0.8465941548347473,
-      "learning_rate": 3.2276308688613056e-05,
-      "loss": 3.1189,
+      "grad_norm": 0.8727632164955139,
+      "learning_rate": 3.229573664328116e-05,
+      "loss": 3.1232,
       "step": 87800
     },
     {
       "epoch": 9.471698113207546,
-      "grad_norm": 0.8548228144645691,
-      "learning_rate": 3.195250944414463e-05,
-      "loss": 3.1317,
+      "grad_norm": 0.8773737549781799,
+      "learning_rate": 3.197193739881273e-05,
+      "loss": 3.136,
       "step": 87850
     },
     {
       "epoch": 9.477088948787062,
-      "grad_norm": 0.8719295263290405,
-      "learning_rate": 3.16287101996762e-05,
-      "loss": 3.119,
+      "grad_norm": 0.8866723775863647,
+      "learning_rate": 3.16481381543443e-05,
+      "loss": 3.1229,
       "step": 87900
     },
     {
       "epoch": 9.482479784366577,
-      "grad_norm": 0.8304482102394104,
-      "learning_rate": 3.130491095520777e-05,
-      "loss": 3.1184,
+      "grad_norm": 0.8770476579666138,
+      "learning_rate": 3.132433890987588e-05,
+      "loss": 3.1228,
       "step": 87950
     },
     {
       "epoch": 9.487870619946092,
-      "grad_norm": 0.8402472734451294,
-      "learning_rate": 3.098111171073934e-05,
-      "loss": 3.132,
+      "grad_norm": 0.87845379114151,
+      "learning_rate": 3.100053966540745e-05,
+      "loss": 3.1361,
       "step": 88000
     },
     {
       "epoch": 9.487870619946092,
-      "eval_accuracy": 0.3939127871519264,
-      "eval_loss": 3.3066344261169434,
-      "eval_runtime": 185.0557,
-      "eval_samples_per_second": 97.327,
-      "eval_steps_per_second": 6.085,
+      "eval_accuracy": 0.39317416482056355,
+      "eval_loss": 3.3103740215301514,
+      "eval_runtime": 182.6237,
+      "eval_samples_per_second": 98.624,
+      "eval_steps_per_second": 6.166,
       "step": 88000
     },
     {
       "epoch": 9.493261455525607,
-      "grad_norm": 0.8302611112594604,
-      "learning_rate": 3.065731246627091e-05,
-      "loss": 3.1315,
+      "grad_norm": 0.8504160642623901,
+      "learning_rate": 3.067674042093901e-05,
+      "loss": 3.1355,
       "step": 88050
     },
     {
       "epoch": 9.498652291105122,
-      "grad_norm": 0.8773887753486633,
-      "learning_rate": 3.033351322180248e-05,
-      "loss": 3.1278,
+      "grad_norm": 0.8972572684288025,
+      "learning_rate": 3.0352941176470588e-05,
+      "loss": 3.1312,
       "step": 88100
     },
     {
       "epoch": 9.504043126684635,
-      "grad_norm": 0.8220431804656982,
-      "learning_rate": 3.000971397733405e-05,
-      "loss": 3.1538,
+      "grad_norm": 0.847787082195282,
+      "learning_rate": 3.002914193200216e-05,
+      "loss": 3.1572,
       "step": 88150
     },
     {
       "epoch": 9.50943396226415,
-      "grad_norm": 0.8599470257759094,
-      "learning_rate": 2.968591473286562e-05,
-      "loss": 3.1354,
+      "grad_norm": 0.9112942218780518,
+      "learning_rate": 2.9705342687533726e-05,
+      "loss": 3.1363,
       "step": 88200
     },
     {
       "epoch": 9.514824797843666,
-      "grad_norm": 0.8439090251922607,
-      "learning_rate": 2.936211548839719e-05,
-      "loss": 3.1212,
+      "grad_norm": 0.8494869470596313,
+      "learning_rate": 2.9381543443065297e-05,
+      "loss": 3.1248,
       "step": 88250
     },
     {
       "epoch": 9.520215633423181,
-      "grad_norm": 0.9031540751457214,
-      "learning_rate": 2.903831624392876e-05,
-      "loss": 3.141,
+      "grad_norm": 0.8875943422317505,
+      "learning_rate": 2.9057744198596868e-05,
+      "loss": 3.1427,
       "step": 88300
     },
     {
       "epoch": 9.525606469002696,
-      "grad_norm": 0.8797536492347717,
-      "learning_rate": 2.8714516999460332e-05,
-      "loss": 3.1318,
+      "grad_norm": 0.8974742293357849,
+      "learning_rate": 2.873394495412844e-05,
+      "loss": 3.1375,
       "step": 88350
     },
     {
       "epoch": 9.530997304582211,
-      "grad_norm": 0.8907245397567749,
-      "learning_rate": 2.83907177549919e-05,
-      "loss": 3.1219,
+      "grad_norm": 0.8841229677200317,
+      "learning_rate": 2.841014570966001e-05,
+      "loss": 3.1253,
       "step": 88400
     },
     {
       "epoch": 9.536388140161725,
-      "grad_norm": 0.8620273470878601,
-      "learning_rate": 2.806691851052347e-05,
-      "loss": 3.1271,
+      "grad_norm": 0.893791139125824,
+      "learning_rate": 2.8086346465191577e-05,
+      "loss": 3.1306,
       "step": 88450
     },
     {
       "epoch": 9.54177897574124,
-      "grad_norm": 0.8356373310089111,
-      "learning_rate": 2.774311926605504e-05,
-      "loss": 3.1419,
+      "grad_norm": 0.86285001039505,
+      "learning_rate": 2.7762547220723148e-05,
+      "loss": 3.1451,
       "step": 88500
     },
     {
       "epoch": 9.547169811320755,
-      "grad_norm": 0.8548893928527832,
-      "learning_rate": 2.7419320021586613e-05,
-      "loss": 3.1212,
+      "grad_norm": 0.9095352292060852,
+      "learning_rate": 2.743874797625472e-05,
+      "loss": 3.1261,
       "step": 88550
     },
     {
       "epoch": 9.55256064690027,
-      "grad_norm": 0.8545761704444885,
-      "learning_rate": 2.7095520777118187e-05,
-      "loss": 3.1353,
+      "grad_norm": 0.88004070520401,
+      "learning_rate": 2.711494873178629e-05,
+      "loss": 3.1411,
       "step": 88600
     },
     {
       "epoch": 9.557951482479785,
-      "grad_norm": 0.8439458608627319,
-      "learning_rate": 2.6771721532649758e-05,
-      "loss": 3.1304,
+      "grad_norm": 0.8813946843147278,
+      "learning_rate": 2.679114948731786e-05,
+      "loss": 3.136,
       "step": 88650
     },
     {
       "epoch": 9.563342318059298,
-      "grad_norm": 0.8117093443870544,
-      "learning_rate": 2.6447922288181325e-05,
-      "loss": 3.1321,
+      "grad_norm": 0.8370325565338135,
+      "learning_rate": 2.6467350242849428e-05,
+      "loss": 3.135,
       "step": 88700
     },
     {
       "epoch": 9.568733153638814,
-      "grad_norm": 0.8481546640396118,
-      "learning_rate": 2.6124123043712896e-05,
-      "loss": 3.1397,
+      "grad_norm": 0.896294355392456,
+      "learning_rate": 2.6143550998381002e-05,
+      "loss": 3.1438,
       "step": 88750
     },
     {
       "epoch": 9.574123989218329,
-      "grad_norm": 0.8625066876411438,
-      "learning_rate": 2.5800323799244467e-05,
-      "loss": 3.1234,
+      "grad_norm": 0.8690078854560852,
+      "learning_rate": 2.5819751753912573e-05,
+      "loss": 3.1285,
       "step": 88800
     },
     {
       "epoch": 9.579514824797844,
-      "grad_norm": 0.8870933651924133,
-      "learning_rate": 2.5476524554776038e-05,
-      "loss": 3.132,
+      "grad_norm": 0.9460025429725647,
+      "learning_rate": 2.5495952509444144e-05,
+      "loss": 3.1365,
       "step": 88850
     },
     {
       "epoch": 9.584905660377359,
-      "grad_norm": 0.8924064636230469,
-      "learning_rate": 2.515272531030761e-05,
-      "loss": 3.156,
+      "grad_norm": 0.9100044965744019,
+      "learning_rate": 2.5172153264975715e-05,
+      "loss": 3.1596,
       "step": 88900
     },
     {
       "epoch": 9.590296495956874,
-      "grad_norm": 0.912037193775177,
-      "learning_rate": 2.4828926065839176e-05,
-      "loss": 3.1337,
+      "grad_norm": 0.9204899072647095,
+      "learning_rate": 2.4848354020507283e-05,
+      "loss": 3.1387,
       "step": 88950
     },
     {
       "epoch": 9.595687331536388,
-      "grad_norm": 0.8235248923301697,
-      "learning_rate": 2.4505126821370747e-05,
-      "loss": 3.1121,
+      "grad_norm": 0.853834331035614,
+      "learning_rate": 2.4524554776038853e-05,
+      "loss": 3.1146,
       "step": 89000
     },
     {
       "epoch": 9.595687331536388,
-      "eval_accuracy": 0.3942704724586117,
-      "eval_loss": 3.304278612136841,
-      "eval_runtime": 184.915,
-      "eval_samples_per_second": 97.402,
-      "eval_steps_per_second": 6.089,
+      "eval_accuracy": 0.39363930783329737,
+      "eval_loss": 3.3085296154022217,
+      "eval_runtime": 182.682,
+      "eval_samples_per_second": 98.592,
+      "eval_steps_per_second": 6.164,
       "step": 89000
     },
     {
       "epoch": 9.601078167115903,
-      "grad_norm": 0.8647744655609131,
-      "learning_rate": 2.4181327576902318e-05,
-      "loss": 3.1443,
+      "grad_norm": 0.8744606375694275,
+      "learning_rate": 2.4200755531570424e-05,
+      "loss": 3.1489,
       "step": 89050
     },
     {
       "epoch": 9.606469002695418,
-      "grad_norm": 0.879808783531189,
-      "learning_rate": 2.385752833243389e-05,
-      "loss": 3.1359,
+      "grad_norm": 0.9052774906158447,
+      "learning_rate": 2.3876956287101995e-05,
+      "loss": 3.1417,
       "step": 89100
     },
     {
       "epoch": 9.611859838274933,
-      "grad_norm": 0.891708254814148,
-      "learning_rate": 2.353372908796546e-05,
-      "loss": 3.1443,
+      "grad_norm": 0.8833246231079102,
+      "learning_rate": 2.3553157042633566e-05,
+      "loss": 3.1495,
       "step": 89150
     },
     {
       "epoch": 9.617250673854448,
-      "grad_norm": 0.9006338119506836,
-      "learning_rate": 2.3209929843497027e-05,
-      "loss": 3.1291,
+      "grad_norm": 0.8751738667488098,
+      "learning_rate": 2.3229357798165134e-05,
+      "loss": 3.1327,
       "step": 89200
     },
     {
       "epoch": 9.622641509433961,
-      "grad_norm": 0.854828417301178,
-      "learning_rate": 2.2886130599028598e-05,
-      "loss": 3.135,
+      "grad_norm": 0.9028359055519104,
+      "learning_rate": 2.2905558553696705e-05,
+      "loss": 3.1395,
       "step": 89250
     },
     {
       "epoch": 9.628032345013477,
-      "grad_norm": 0.8300639986991882,
-      "learning_rate": 2.256233135456017e-05,
-      "loss": 3.113,
+      "grad_norm": 0.8810246586799622,
+      "learning_rate": 2.2581759309228275e-05,
+      "loss": 3.1186,
       "step": 89300
     },
     {
       "epoch": 9.633423180592992,
-      "grad_norm": 0.815534234046936,
-      "learning_rate": 2.2238532110091743e-05,
-      "loss": 3.1478,
+      "grad_norm": 0.8861169219017029,
+      "learning_rate": 2.2257960064759846e-05,
+      "loss": 3.1503,
       "step": 89350
     },
     {
       "epoch": 9.638814016172507,
-      "grad_norm": 0.8253652453422546,
-      "learning_rate": 2.1914732865623314e-05,
-      "loss": 3.1443,
+      "grad_norm": 0.856515645980835,
+      "learning_rate": 2.1934160820291417e-05,
+      "loss": 3.149,
       "step": 89400
     },
     {
       "epoch": 9.644204851752022,
-      "grad_norm": 0.8685674667358398,
-      "learning_rate": 2.1590933621154885e-05,
-      "loss": 3.1428,
+      "grad_norm": 0.8759087324142456,
+      "learning_rate": 2.1610361575822985e-05,
+      "loss": 3.1465,
       "step": 89450
     },
     {
       "epoch": 9.649595687331537,
-      "grad_norm": 0.8638211488723755,
-      "learning_rate": 2.1267134376686453e-05,
-      "loss": 3.1376,
+      "grad_norm": 0.9285345673561096,
+      "learning_rate": 2.1286562331354556e-05,
+      "loss": 3.1421,
       "step": 89500
     },
     {
       "epoch": 9.65498652291105,
-      "grad_norm": 0.8781023621559143,
-      "learning_rate": 2.0943335132218024e-05,
-      "loss": 3.1339,
+      "grad_norm": 0.8611302971839905,
+      "learning_rate": 2.096276308688613e-05,
+      "loss": 3.1378,
       "step": 89550
     },
     {
       "epoch": 9.660377358490566,
-      "grad_norm": 0.8183006644248962,
-      "learning_rate": 2.0619535887749594e-05,
-      "loss": 3.1411,
+      "grad_norm": 0.8217964172363281,
+      "learning_rate": 2.06389638424177e-05,
+      "loss": 3.1429,
       "step": 89600
     },
     {
       "epoch": 9.66576819407008,
-      "grad_norm": 0.8766468167304993,
-      "learning_rate": 2.0295736643281165e-05,
-      "loss": 3.1244,
+      "grad_norm": 0.9399557113647461,
+      "learning_rate": 2.031516459794927e-05,
+      "loss": 3.1311,
       "step": 89650
     },
     {
       "epoch": 9.671159029649596,
-      "grad_norm": 0.8593804240226746,
-      "learning_rate": 1.9971937398812736e-05,
-      "loss": 3.1434,
+      "grad_norm": 0.9033957719802856,
+      "learning_rate": 1.9991365353480843e-05,
+      "loss": 3.1479,
       "step": 89700
     },
     {
       "epoch": 9.676549865229111,
-      "grad_norm": 0.8419660329818726,
-      "learning_rate": 1.9648138154344304e-05,
-      "loss": 3.1206,
+      "grad_norm": 0.86545729637146,
+      "learning_rate": 1.966756610901241e-05,
+      "loss": 3.1248,
       "step": 89750
     },
     {
       "epoch": 9.681940700808624,
-      "grad_norm": 0.8620070219039917,
-      "learning_rate": 1.9324338909875875e-05,
-      "loss": 3.1244,
+      "grad_norm": 0.8842559456825256,
+      "learning_rate": 1.934376686454398e-05,
+      "loss": 3.1265,
       "step": 89800
     },
     {
       "epoch": 9.68733153638814,
-      "grad_norm": 0.8452453017234802,
-      "learning_rate": 1.9000539665407446e-05,
-      "loss": 3.1357,
+      "grad_norm": 0.9108521342277527,
+      "learning_rate": 1.9019967620075552e-05,
+      "loss": 3.1399,
       "step": 89850
     },
     {
       "epoch": 9.692722371967655,
-      "grad_norm": 0.8461737036705017,
-      "learning_rate": 1.8676740420939016e-05,
-      "loss": 3.1241,
+      "grad_norm": 0.8824275135993958,
+      "learning_rate": 1.8696168375607123e-05,
+      "loss": 3.1304,
       "step": 89900
     },
     {
       "epoch": 9.69811320754717,
-      "grad_norm": 0.9039965867996216,
-      "learning_rate": 1.8352941176470587e-05,
-      "loss": 3.1415,
+      "grad_norm": 0.916059136390686,
+      "learning_rate": 1.8372369131138694e-05,
+      "loss": 3.146,
       "step": 89950
     },
     {
       "epoch": 9.703504043126685,
-      "grad_norm": 0.8599136471748352,
-      "learning_rate": 1.8029141932002158e-05,
-      "loss": 3.1384,
+      "grad_norm": 0.8686931133270264,
+      "learning_rate": 1.804856988667026e-05,
+      "loss": 3.1429,
       "step": 90000
     },
     {
       "epoch": 9.703504043126685,
-      "eval_accuracy": 0.3945016858038809,
-      "eval_loss": 3.3018898963928223,
-      "eval_runtime": 184.7082,
-      "eval_samples_per_second": 97.511,
-      "eval_steps_per_second": 6.096,
+      "eval_accuracy": 0.39395211949731523,
+      "eval_loss": 3.3053202629089355,
+      "eval_runtime": 182.6099,
+      "eval_samples_per_second": 98.631,
+      "eval_steps_per_second": 6.166,
       "step": 90000
     }
   ],