diff --git "a/checkpoint-41387/trainer_state.json" "b/checkpoint-41387/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-41387/trainer_state.json"
@@ -0,0 +1,5823 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 41387,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0012081307197438762,
+      "grad_norm": 2.0567336082458496,
+      "learning_rate": 2.94e-05,
+      "loss": 10.047689819335938,
+      "step": 50
+    },
+    {
+      "epoch": 0.0024162614394877525,
+      "grad_norm": 1.242946982383728,
+      "learning_rate": 5.94e-05,
+      "loss": 8.63542236328125,
+      "step": 100
+    },
+    {
+      "epoch": 0.003624392159231629,
+      "grad_norm": 1.0061999559402466,
+      "learning_rate": 8.939999999999999e-05,
+      "loss": 7.649925537109375,
+      "step": 150
+    },
+    {
+      "epoch": 0.004832522878975505,
+      "grad_norm": 0.8215569853782654,
+      "learning_rate": 0.0001194,
+      "loss": 7.332319946289062,
+      "step": 200
+    },
+    {
+      "epoch": 0.006040653598719381,
+      "grad_norm": 0.9846878051757812,
+      "learning_rate": 0.0001494,
+      "loss": 7.208408203125,
+      "step": 250
+    },
+    {
+      "epoch": 0.007248784318463258,
+      "grad_norm": 0.7860299348831177,
+      "learning_rate": 0.00017939999999999997,
+      "loss": 7.064036865234375,
+      "step": 300
+    },
+    {
+      "epoch": 0.008456915038207134,
+      "grad_norm": 1.1750568151474,
+      "learning_rate": 0.00020939999999999997,
+      "loss": 6.932866821289062,
+      "step": 350
+    },
+    {
+      "epoch": 0.00966504575795101,
+      "grad_norm": 1.1936959028244019,
+      "learning_rate": 0.0002394,
+      "loss": 6.835554809570312,
+      "step": 400
+    },
+    {
+      "epoch": 0.010873176477694887,
+      "grad_norm": 0.9108025431632996,
+      "learning_rate": 0.0002694,
+      "loss": 6.752925415039062,
+      "step": 450
+    },
+    {
+      "epoch": 0.012081307197438763,
+      "grad_norm": 0.9273257255554199,
+      "learning_rate": 0.00029939999999999996,
+      "loss": 6.665084228515625,
+      "step": 500
+    },
+    {
+      "epoch": 0.013289437917182638,
+      "grad_norm": 0.8685219883918762,
+      "learning_rate": 0.0002996404725218284,
+      "loss": 6.676280517578125,
+      "step": 550
+    },
+    {
+      "epoch": 0.014497568636926516,
+      "grad_norm": 0.8371725678443909,
+      "learning_rate": 0.000299273607748184,
+      "loss": 6.580692138671875,
+      "step": 600
+    },
+    {
+      "epoch": 0.01570569935667039,
+      "grad_norm": 0.8638702034950256,
+      "learning_rate": 0.00029890674297453957,
+      "loss": 6.545195922851563,
+      "step": 650
+    },
+    {
+      "epoch": 0.01691383007641427,
+      "grad_norm": 0.9655756950378418,
+      "learning_rate": 0.00029853987820089515,
+      "loss": 6.555042724609375,
+      "step": 700
+    },
+    {
+      "epoch": 0.018121960796158146,
+      "grad_norm": 0.9254677295684814,
+      "learning_rate": 0.0002981730134272507,
+      "loss": 6.496650390625,
+      "step": 750
+    },
+    {
+      "epoch": 0.01933009151590202,
+      "grad_norm": 1.0433354377746582,
+      "learning_rate": 0.00029780614865360626,
+      "loss": 6.45712158203125,
+      "step": 800
+    },
+    {
+      "epoch": 0.020538222235645897,
+      "grad_norm": 0.7390813827514648,
+      "learning_rate": 0.00029743928387996184,
+      "loss": 6.416035766601563,
+      "step": 850
+    },
+    {
+      "epoch": 0.021746352955389774,
+      "grad_norm": 0.8372873067855835,
+      "learning_rate": 0.0002970724191063174,
+      "loss": 6.3870001220703125,
+      "step": 900
+    },
+    {
+      "epoch": 0.022954483675133648,
+      "grad_norm": 0.9460623264312744,
+      "learning_rate": 0.00029670555433267294,
+      "loss": 6.375929565429687,
+      "step": 950
+    },
+    {
+      "epoch": 0.024162614394877525,
+      "grad_norm": 0.8223255276679993,
+      "learning_rate": 0.0002963386895590285,
+      "loss": 6.339353637695313,
+      "step": 1000
+    },
+    {
+      "epoch": 0.025370745114621403,
+      "grad_norm": 0.7255702018737793,
+      "learning_rate": 0.0002959718247853841,
+      "loss": 6.3057666015625,
+      "step": 1050
+    },
+    {
+      "epoch": 0.026578875834365277,
+      "grad_norm": 0.7963605523109436,
+      "learning_rate": 0.00029560496001173963,
+      "loss": 6.329375610351563,
+      "step": 1100
+    },
+    {
+      "epoch": 0.027787006554109154,
+      "grad_norm": 0.7842203974723816,
+      "learning_rate": 0.0002952380952380952,
+      "loss": 6.277288208007812,
+      "step": 1150
+    },
+    {
+      "epoch": 0.02899513727385303,
+      "grad_norm": 0.8361679315567017,
+      "learning_rate": 0.0002948712304644508,
+      "loss": 6.26784423828125,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03020326799359691,
+      "grad_norm": 0.7909688949584961,
+      "learning_rate": 0.00029450436569080637,
+      "loss": 6.240078125,
+      "step": 1250
+    },
+    {
+      "epoch": 0.03141139871334078,
+      "grad_norm": 0.7913162708282471,
+      "learning_rate": 0.0002941375009171619,
+      "loss": 6.185866088867187,
+      "step": 1300
+    },
+    {
+      "epoch": 0.03261952943308466,
+      "grad_norm": 0.8106042146682739,
+      "learning_rate": 0.0002937706361435175,
+      "loss": 6.174436645507813,
+      "step": 1350
+    },
+    {
+      "epoch": 0.03382766015282854,
+      "grad_norm": 0.8267247080802917,
+      "learning_rate": 0.00029340377136987306,
+      "loss": 6.1425341796875,
+      "step": 1400
+    },
+    {
+      "epoch": 0.035035790872572414,
+      "grad_norm": 0.778644859790802,
+      "learning_rate": 0.0002930369065962286,
+      "loss": 6.153001098632813,
+      "step": 1450
+    },
+    {
+      "epoch": 0.03624392159231629,
+      "grad_norm": 0.8770017027854919,
+      "learning_rate": 0.00029267004182258416,
+      "loss": 6.135545654296875,
+      "step": 1500
+    },
+    {
+      "epoch": 0.03745205231206016,
+      "grad_norm": 0.800643801689148,
+      "learning_rate": 0.00029230317704893974,
+      "loss": 6.1441357421875,
+      "step": 1550
+    },
+    {
+      "epoch": 0.03866018303180404,
+      "grad_norm": 0.8167304396629333,
+      "learning_rate": 0.00029193631227529527,
+      "loss": 6.084442749023437,
+      "step": 1600
+    },
+    {
+      "epoch": 0.03986831375154792,
+      "grad_norm": 0.7698378562927246,
+      "learning_rate": 0.00029156944750165085,
+      "loss": 6.058062744140625,
+      "step": 1650
+    },
+    {
+      "epoch": 0.041076444471291794,
+      "grad_norm": 0.893836259841919,
+      "learning_rate": 0.00029120258272800643,
+      "loss": 6.03717041015625,
+      "step": 1700
+    },
+    {
+      "epoch": 0.04228457519103567,
+      "grad_norm": 0.8040123581886292,
+      "learning_rate": 0.000290835717954362,
+      "loss": 6.0776129150390625,
+      "step": 1750
+    },
+    {
+      "epoch": 0.04349270591077955,
+      "grad_norm": 0.8429253101348877,
+      "learning_rate": 0.00029046885318071753,
+      "loss": 6.032772216796875,
+      "step": 1800
+    },
+    {
+      "epoch": 0.044700836630523426,
+      "grad_norm": 0.8081746101379395,
+      "learning_rate": 0.0002901019884070731,
+      "loss": 6.040414428710937,
+      "step": 1850
+    },
+    {
+      "epoch": 0.045908967350267296,
+      "grad_norm": 0.8269071578979492,
+      "learning_rate": 0.0002897351236334287,
+      "loss": 6.013905029296875,
+      "step": 1900
+    },
+    {
+      "epoch": 0.047117098070011174,
+      "grad_norm": 0.7277988195419312,
+      "learning_rate": 0.0002893682588597843,
+      "loss": 6.001159057617188,
+      "step": 1950
+    },
+    {
+      "epoch": 0.04832522878975505,
+      "grad_norm": 0.840861439704895,
+      "learning_rate": 0.0002890013940861398,
+      "loss": 5.982572021484375,
+      "step": 2000
+    },
+    {
+      "epoch": 0.04953335950949893,
+      "grad_norm": 0.9168997406959534,
+      "learning_rate": 0.0002886345293124954,
+      "loss": 5.953336791992188,
+      "step": 2050
+    },
+    {
+      "epoch": 0.050741490229242805,
+      "grad_norm": 0.7781474590301514,
+      "learning_rate": 0.00028826766453885096,
+      "loss": 5.9719677734375,
+      "step": 2100
+    },
+    {
+      "epoch": 0.05194962094898668,
+      "grad_norm": 0.7908787131309509,
+      "learning_rate": 0.00028790079976520654,
+      "loss": 5.920252075195313,
+      "step": 2150
+    },
+    {
+      "epoch": 0.05315775166873055,
+      "grad_norm": 0.7818727493286133,
+      "learning_rate": 0.00028753393499156207,
+      "loss": 5.925120239257812,
+      "step": 2200
+    },
+    {
+      "epoch": 0.05436588238847443,
+      "grad_norm": 0.8207040429115295,
+      "learning_rate": 0.00028716707021791765,
+      "loss": 5.943333740234375,
+      "step": 2250
+    },
+    {
+      "epoch": 0.05557401310821831,
+      "grad_norm": 0.8047415614128113,
+      "learning_rate": 0.0002868002054442732,
+      "loss": 5.859091796875,
+      "step": 2300
+    },
+    {
+      "epoch": 0.056782143827962185,
+      "grad_norm": 0.7661274671554565,
+      "learning_rate": 0.0002864333406706288,
+      "loss": 5.899588623046875,
+      "step": 2350
+    },
+    {
+      "epoch": 0.05799027454770606,
+      "grad_norm": 0.8919453620910645,
+      "learning_rate": 0.00028606647589698433,
+      "loss": 5.900999145507813,
+      "step": 2400
+    },
+    {
+      "epoch": 0.05919840526744994,
+      "grad_norm": 0.8325379490852356,
+      "learning_rate": 0.0002856996111233399,
+      "loss": 5.85697509765625,
+      "step": 2450
+    },
+    {
+      "epoch": 0.06040653598719382,
+      "grad_norm": 0.8285044431686401,
+      "learning_rate": 0.0002853327463496955,
+      "loss": 5.880800170898437,
+      "step": 2500
+    },
+    {
+      "epoch": 0.06161466670693769,
+      "grad_norm": 0.7765763998031616,
+      "learning_rate": 0.00028496588157605107,
+      "loss": 5.801648559570313,
+      "step": 2550
+    },
+    {
+      "epoch": 0.06282279742668156,
+      "grad_norm": 0.794620156288147,
+      "learning_rate": 0.0002845990168024066,
+      "loss": 5.81466552734375,
+      "step": 2600
+    },
+    {
+      "epoch": 0.06403092814642544,
+      "grad_norm": 0.7981798648834229,
+      "learning_rate": 0.0002842321520287622,
+      "loss": 5.792921142578125,
+      "step": 2650
+    },
+    {
+      "epoch": 0.06523905886616932,
+      "grad_norm": 0.7955827713012695,
+      "learning_rate": 0.00028386528725511776,
+      "loss": 5.778803100585938,
+      "step": 2700
+    },
+    {
+      "epoch": 0.0664471895859132,
+      "grad_norm": 0.810361921787262,
+      "learning_rate": 0.00028349842248147334,
+      "loss": 5.730283203125,
+      "step": 2750
+    },
+    {
+      "epoch": 0.06765532030565707,
+      "grad_norm": 0.8210492730140686,
+      "learning_rate": 0.00028313155770782886,
+      "loss": 5.771351928710938,
+      "step": 2800
+    },
+    {
+      "epoch": 0.06886345102540095,
+      "grad_norm": 0.8211035132408142,
+      "learning_rate": 0.00028276469293418444,
+      "loss": 5.741306762695313,
+      "step": 2850
+    },
+    {
+      "epoch": 0.07007158174514483,
+      "grad_norm": 0.8045103549957275,
+      "learning_rate": 0.00028239782816054,
+      "loss": 5.708994750976562,
+      "step": 2900
+    },
+    {
+      "epoch": 0.0712797124648887,
+      "grad_norm": 0.860809326171875,
+      "learning_rate": 0.0002820309633868956,
+      "loss": 5.729442749023438,
+      "step": 2950
+    },
+    {
+      "epoch": 0.07248784318463258,
+      "grad_norm": 0.7570837736129761,
+      "learning_rate": 0.00028166409861325113,
+      "loss": 5.683602294921875,
+      "step": 3000
+    },
+    {
+      "epoch": 0.07369597390437645,
+      "grad_norm": 0.7897762060165405,
+      "learning_rate": 0.0002812972338396067,
+      "loss": 5.693482055664062,
+      "step": 3050
+    },
+    {
+      "epoch": 0.07490410462412032,
+      "grad_norm": 0.8576429486274719,
+      "learning_rate": 0.0002809303690659623,
+      "loss": 5.674725952148438,
+      "step": 3100
+    },
+    {
+      "epoch": 0.0761122353438642,
+      "grad_norm": 0.7414172291755676,
+      "learning_rate": 0.00028056350429231787,
+      "loss": 5.6906787109375,
+      "step": 3150
+    },
+    {
+      "epoch": 0.07732036606360808,
+      "grad_norm": 0.7647370100021362,
+      "learning_rate": 0.0002801966395186734,
+      "loss": 5.68769775390625,
+      "step": 3200
+    },
+    {
+      "epoch": 0.07852849678335196,
+      "grad_norm": 0.7908104658126831,
+      "learning_rate": 0.000279829774745029,
+      "loss": 5.662813720703125,
+      "step": 3250
+    },
+    {
+      "epoch": 0.07973662750309583,
+      "grad_norm": 0.775360107421875,
+      "learning_rate": 0.00027946290997138456,
+      "loss": 5.618099365234375,
+      "step": 3300
+    },
+    {
+      "epoch": 0.08094475822283971,
+      "grad_norm": 0.7957372069358826,
+      "learning_rate": 0.0002790960451977401,
+      "loss": 5.621898803710938,
+      "step": 3350
+    },
+    {
+      "epoch": 0.08215288894258359,
+      "grad_norm": 0.7726836204528809,
+      "learning_rate": 0.00027872918042409566,
+      "loss": 5.626279296875,
+      "step": 3400
+    },
+    {
+      "epoch": 0.08336101966232747,
+      "grad_norm": 0.7751205563545227,
+      "learning_rate": 0.00027836231565045124,
+      "loss": 5.612183227539062,
+      "step": 3450
+    },
+    {
+      "epoch": 0.08456915038207134,
+      "grad_norm": 0.8598089218139648,
+      "learning_rate": 0.00027799545087680677,
+      "loss": 5.6166162109375,
+      "step": 3500
+    },
+    {
+      "epoch": 0.08577728110181522,
+      "grad_norm": 0.7791950106620789,
+      "learning_rate": 0.00027762858610316235,
+      "loss": 5.590696411132813,
+      "step": 3550
+    },
+    {
+      "epoch": 0.0869854118215591,
+      "grad_norm": 0.7699060440063477,
+      "learning_rate": 0.0002772617213295179,
+      "loss": 5.5623193359375,
+      "step": 3600
+    },
+    {
+      "epoch": 0.08819354254130297,
+      "grad_norm": 0.7668980360031128,
+      "learning_rate": 0.00027689485655587345,
+      "loss": 5.562638549804688,
+      "step": 3650
+    },
+    {
+      "epoch": 0.08940167326104685,
+      "grad_norm": 0.8171679973602295,
+      "learning_rate": 0.00027652799178222903,
+      "loss": 5.550632934570313,
+      "step": 3700
+    },
+    {
+      "epoch": 0.09060980398079072,
+      "grad_norm": 0.8103206753730774,
+      "learning_rate": 0.0002761611270085846,
+      "loss": 5.534771728515625,
+      "step": 3750
+    },
+    {
+      "epoch": 0.09181793470053459,
+      "grad_norm": 0.802372395992279,
+      "learning_rate": 0.0002757942622349402,
+      "loss": 5.537081298828125,
+      "step": 3800
+    },
+    {
+      "epoch": 0.09302606542027847,
+      "grad_norm": 0.781486451625824,
+      "learning_rate": 0.0002754273974612957,
+      "loss": 5.527772216796875,
+      "step": 3850
+    },
+    {
+      "epoch": 0.09423419614002235,
+      "grad_norm": 0.7913851141929626,
+      "learning_rate": 0.0002750605326876513,
+      "loss": 5.511227416992187,
+      "step": 3900
+    },
+    {
+      "epoch": 0.09544232685976622,
+      "grad_norm": 0.7766952514648438,
+      "learning_rate": 0.0002746936679140069,
+      "loss": 5.484191284179688,
+      "step": 3950
+    },
+    {
+      "epoch": 0.0966504575795101,
+      "grad_norm": 0.7719422578811646,
+      "learning_rate": 0.0002743268031403624,
+      "loss": 5.52441162109375,
+      "step": 4000
+    },
+    {
+      "epoch": 0.09785858829925398,
+      "grad_norm": 0.8148206472396851,
+      "learning_rate": 0.000273959938366718,
+      "loss": 5.476258544921875,
+      "step": 4050
+    },
+    {
+      "epoch": 0.09906671901899786,
+      "grad_norm": 0.8194459676742554,
+      "learning_rate": 0.00027359307359307356,
+      "loss": 5.489271240234375,
+      "step": 4100
+    },
+    {
+      "epoch": 0.10027484973874173,
+      "grad_norm": 0.7631927132606506,
+      "learning_rate": 0.00027322620881942914,
+      "loss": 5.490921630859375,
+      "step": 4150
+    },
+    {
+      "epoch": 0.10148298045848561,
+      "grad_norm": 0.7788265347480774,
+      "learning_rate": 0.00027285934404578467,
+      "loss": 5.4406396484375,
+      "step": 4200
+    },
+    {
+      "epoch": 0.10269111117822949,
+      "grad_norm": 0.7736191749572754,
+      "learning_rate": 0.00027249247927214025,
+      "loss": 5.464171142578125,
+      "step": 4250
+    },
+    {
+      "epoch": 0.10389924189797337,
+      "grad_norm": 0.9049032926559448,
+      "learning_rate": 0.00027212561449849583,
+      "loss": 5.465469970703125,
+      "step": 4300
+    },
+    {
+      "epoch": 0.10510737261771724,
+      "grad_norm": 0.8199710249900818,
+      "learning_rate": 0.0002717587497248514,
+      "loss": 5.46241943359375,
+      "step": 4350
+    },
+    {
+      "epoch": 0.1063155033374611,
+      "grad_norm": 0.7697764039039612,
+      "learning_rate": 0.00027139188495120694,
+      "loss": 5.39656005859375,
+      "step": 4400
+    },
+    {
+      "epoch": 0.10752363405720498,
+      "grad_norm": 0.9469720125198364,
+      "learning_rate": 0.0002710250201775625,
+      "loss": 5.419663696289063,
+      "step": 4450
+    },
+    {
+      "epoch": 0.10873176477694886,
+      "grad_norm": 0.7325762510299683,
+      "learning_rate": 0.0002706581554039181,
+      "loss": 5.414904174804687,
+      "step": 4500
+    },
+    {
+      "epoch": 0.10993989549669274,
+      "grad_norm": 0.8119387626647949,
+      "learning_rate": 0.0002702912906302737,
+      "loss": 5.421415405273438,
+      "step": 4550
+    },
+    {
+      "epoch": 0.11114802621643662,
+      "grad_norm": 0.7711684703826904,
+      "learning_rate": 0.0002699244258566292,
+      "loss": 5.374515991210938,
+      "step": 4600
+    },
+    {
+      "epoch": 0.11235615693618049,
+      "grad_norm": 0.7902948260307312,
+      "learning_rate": 0.0002695575610829848,
+      "loss": 5.389853515625,
+      "step": 4650
+    },
+    {
+      "epoch": 0.11356428765592437,
+      "grad_norm": 0.782555878162384,
+      "learning_rate": 0.00026919069630934036,
+      "loss": 5.413133544921875,
+      "step": 4700
+    },
+    {
+      "epoch": 0.11477241837566825,
+      "grad_norm": 0.7748803496360779,
+      "learning_rate": 0.00026882383153569594,
+      "loss": 5.3650360107421875,
+      "step": 4750
+    },
+    {
+      "epoch": 0.11598054909541212,
+      "grad_norm": 0.7870548963546753,
+      "learning_rate": 0.00026845696676205147,
+      "loss": 5.3870697021484375,
+      "step": 4800
+    },
+    {
+      "epoch": 0.117188679815156,
+      "grad_norm": 0.7400593757629395,
+      "learning_rate": 0.00026809010198840705,
+      "loss": 5.3786572265625,
+      "step": 4850
+    },
+    {
+      "epoch": 0.11839681053489988,
+      "grad_norm": 0.8056249618530273,
+      "learning_rate": 0.00026772323721476263,
+      "loss": 5.38482666015625,
+      "step": 4900
+    },
+    {
+      "epoch": 0.11960494125464376,
+      "grad_norm": 0.7575085163116455,
+      "learning_rate": 0.0002673563724411182,
+      "loss": 5.357495727539063,
+      "step": 4950
+    },
+    {
+      "epoch": 0.12081307197438763,
+      "grad_norm": 0.7345143556594849,
+      "learning_rate": 0.00026698950766747373,
+      "loss": 5.3505908203125,
+      "step": 5000
+    },
+    {
+      "epoch": 0.12202120269413151,
+      "grad_norm": 0.7637096047401428,
+      "learning_rate": 0.0002666226428938293,
+      "loss": 5.334730834960937,
+      "step": 5050
+    },
+    {
+      "epoch": 0.12322933341387537,
+      "grad_norm": 0.8016135096549988,
+      "learning_rate": 0.0002662557781201849,
+      "loss": 5.324992065429687,
+      "step": 5100
+    },
+    {
+      "epoch": 0.12443746413361925,
+      "grad_norm": 0.7238516211509705,
+      "learning_rate": 0.0002658889133465405,
+      "loss": 5.3167266845703125,
+      "step": 5150
+    },
+    {
+      "epoch": 0.12564559485336313,
+      "grad_norm": 0.7395594716072083,
+      "learning_rate": 0.000265522048572896,
+      "loss": 5.29600341796875,
+      "step": 5200
+    },
+    {
+      "epoch": 0.126853725573107,
+      "grad_norm": 0.7709106802940369,
+      "learning_rate": 0.0002651551837992516,
+      "loss": 5.281442260742187,
+      "step": 5250
+    },
+    {
+      "epoch": 0.12806185629285088,
+      "grad_norm": 0.7308861613273621,
+      "learning_rate": 0.00026478831902560716,
+      "loss": 5.3089666748046875,
+      "step": 5300
+    },
+    {
+      "epoch": 0.12926998701259476,
+      "grad_norm": 0.804020881652832,
+      "learning_rate": 0.00026442145425196274,
+      "loss": 5.290265502929688,
+      "step": 5350
+    },
+    {
+      "epoch": 0.13047811773233864,
+      "grad_norm": 0.7728541493415833,
+      "learning_rate": 0.00026405458947831827,
+      "loss": 5.272477416992188,
+      "step": 5400
+    },
+    {
+      "epoch": 0.13168624845208252,
+      "grad_norm": 0.7566307783126831,
+      "learning_rate": 0.00026368772470467385,
+      "loss": 5.312435302734375,
+      "step": 5450
+    },
+    {
+      "epoch": 0.1328943791718264,
+      "grad_norm": 0.785839319229126,
+      "learning_rate": 0.0002633208599310294,
+      "loss": 5.258353271484375,
+      "step": 5500
+    },
+    {
+      "epoch": 0.13410250989157027,
+      "grad_norm": 0.7946141362190247,
+      "learning_rate": 0.00026295399515738495,
+      "loss": 5.2605078125,
+      "step": 5550
+    },
+    {
+      "epoch": 0.13531064061131415,
+      "grad_norm": 0.7667028307914734,
+      "learning_rate": 0.00026258713038374053,
+      "loss": 5.247598876953125,
+      "step": 5600
+    },
+    {
+      "epoch": 0.13651877133105803,
+      "grad_norm": 0.7164963483810425,
+      "learning_rate": 0.0002622202656100961,
+      "loss": 5.27611328125,
+      "step": 5650
+    },
+    {
+      "epoch": 0.1377269020508019,
+      "grad_norm": 0.7562577128410339,
+      "learning_rate": 0.0002618534008364517,
+      "loss": 5.255584106445313,
+      "step": 5700
+    },
+    {
+      "epoch": 0.13893503277054578,
+      "grad_norm": 0.7422127723693848,
+      "learning_rate": 0.0002614865360628072,
+      "loss": 5.2476806640625,
+      "step": 5750
+    },
+    {
+      "epoch": 0.14014316349028966,
+      "grad_norm": 0.7244675755500793,
+      "learning_rate": 0.0002611196712891628,
+      "loss": 5.25503662109375,
+      "step": 5800
+    },
+    {
+      "epoch": 0.14135129421003353,
+      "grad_norm": 0.8102642893791199,
+      "learning_rate": 0.0002607528065155184,
+      "loss": 5.255423583984375,
+      "step": 5850
+    },
+    {
+      "epoch": 0.1425594249297774,
+      "grad_norm": 0.7501764893531799,
+      "learning_rate": 0.0002603859417418739,
+      "loss": 5.213814697265625,
+      "step": 5900
+    },
+    {
+      "epoch": 0.1437675556495213,
+      "grad_norm": 0.7393425107002258,
+      "learning_rate": 0.0002600190769682295,
+      "loss": 5.198919677734375,
+      "step": 5950
+    },
+    {
+      "epoch": 0.14497568636926517,
+      "grad_norm": 0.7690203189849854,
+      "learning_rate": 0.00025965221219458506,
+      "loss": 5.2246160888671875,
+      "step": 6000
+    },
+    {
+      "epoch": 0.14618381708900904,
+      "grad_norm": 0.7925234436988831,
+      "learning_rate": 0.0002592853474209406,
+      "loss": 5.181400756835938,
+      "step": 6050
+    },
+    {
+      "epoch": 0.1473919478087529,
+      "grad_norm": 0.7487099766731262,
+      "learning_rate": 0.00025891848264729617,
+      "loss": 5.193028564453125,
+      "step": 6100
+    },
+    {
+      "epoch": 0.14860007852849677,
+      "grad_norm": 0.7373976707458496,
+      "learning_rate": 0.00025855161787365175,
+      "loss": 5.18432373046875,
+      "step": 6150
+    },
+    {
+      "epoch": 0.14980820924824065,
+      "grad_norm": 0.7428847551345825,
+      "learning_rate": 0.0002581847531000073,
+      "loss": 5.147239990234375,
+      "step": 6200
+    },
+    {
+      "epoch": 0.15101633996798453,
+      "grad_norm": 0.7790428996086121,
+      "learning_rate": 0.00025781788832636286,
+      "loss": 5.17477783203125,
+      "step": 6250
+    },
+    {
+      "epoch": 0.1522244706877284,
+      "grad_norm": 0.768201470375061,
+      "learning_rate": 0.00025745102355271844,
+      "loss": 5.1685247802734375,
+      "step": 6300
+    },
+    {
+      "epoch": 0.15343260140747228,
+      "grad_norm": 0.724666953086853,
+      "learning_rate": 0.000257084158779074,
+      "loss": 5.172269897460938,
+      "step": 6350
+    },
+    {
+      "epoch": 0.15464073212721616,
+      "grad_norm": 0.8599744439125061,
+      "learning_rate": 0.00025671729400542954,
+      "loss": 5.1797418212890625,
+      "step": 6400
+    },
+    {
+      "epoch": 0.15584886284696003,
+      "grad_norm": 0.7066710591316223,
+      "learning_rate": 0.0002563504292317851,
+      "loss": 5.119457092285156,
+      "step": 6450
+    },
+    {
+      "epoch": 0.1570569935667039,
+      "grad_norm": 0.7463105916976929,
+      "learning_rate": 0.0002559835644581407,
+      "loss": 5.126748046875,
+      "step": 6500
+    },
+    {
+      "epoch": 0.1582651242864478,
+      "grad_norm": 0.7502253651618958,
+      "learning_rate": 0.0002556166996844963,
+      "loss": 5.127702026367188,
+      "step": 6550
+    },
+    {
+      "epoch": 0.15947325500619167,
+      "grad_norm": 0.7558940052986145,
+      "learning_rate": 0.0002552498349108518,
+      "loss": 5.148985595703125,
+      "step": 6600
+    },
+    {
+      "epoch": 0.16068138572593554,
+      "grad_norm": 0.774601936340332,
+      "learning_rate": 0.0002548829701372074,
+      "loss": 5.08835205078125,
+      "step": 6650
+    },
+    {
+      "epoch": 0.16188951644567942,
+      "grad_norm": 0.7979328036308289,
+      "learning_rate": 0.00025451610536356297,
+      "loss": 5.091900024414063,
+      "step": 6700
+    },
+    {
+      "epoch": 0.1630976471654233,
+      "grad_norm": 0.736694872379303,
+      "learning_rate": 0.00025414924058991855,
+      "loss": 5.161365966796875,
+      "step": 6750
+    },
+    {
+      "epoch": 0.16430577788516718,
+      "grad_norm": 0.7478530406951904,
+      "learning_rate": 0.0002537823758162741,
+      "loss": 5.156773071289063,
+      "step": 6800
+    },
+    {
+      "epoch": 0.16551390860491105,
+      "grad_norm": 0.7344756126403809,
+      "learning_rate": 0.00025341551104262965,
+      "loss": 5.077612609863281,
+      "step": 6850
+    },
+    {
+      "epoch": 0.16672203932465493,
+      "grad_norm": 0.7784936428070068,
+      "learning_rate": 0.00025304864626898523,
+      "loss": 5.097346496582031,
+      "step": 6900
+    },
+    {
+      "epoch": 0.1679301700443988,
+      "grad_norm": 0.8142541646957397,
+      "learning_rate": 0.0002526817814953408,
+      "loss": 5.088407897949219,
+      "step": 6950
+    },
+    {
+      "epoch": 0.16913830076414268,
+      "grad_norm": 0.7559444904327393,
+      "learning_rate": 0.00025231491672169634,
+      "loss": 5.092929992675781,
+      "step": 7000
+    },
+    {
+      "epoch": 0.17034643148388656,
+      "grad_norm": 0.8266547918319702,
+      "learning_rate": 0.0002519480519480519,
+      "loss": 5.090558471679688,
+      "step": 7050
+    },
+    {
+      "epoch": 0.17155456220363044,
+      "grad_norm": 0.7868309617042542,
+      "learning_rate": 0.0002515811871744075,
+      "loss": 5.059242248535156,
+      "step": 7100
+    },
+    {
+      "epoch": 0.17276269292337432,
+      "grad_norm": 0.7681360840797424,
+      "learning_rate": 0.0002512143224007631,
+      "loss": 5.053140869140625,
+      "step": 7150
+    },
+    {
+      "epoch": 0.1739708236431182,
+      "grad_norm": 0.7762099504470825,
+      "learning_rate": 0.0002508474576271186,
+      "loss": 5.044075927734375,
+      "step": 7200
+    },
+    {
+      "epoch": 0.17517895436286207,
+      "grad_norm": 0.7760724425315857,
+      "learning_rate": 0.0002504805928534742,
+      "loss": 5.042862548828125,
+      "step": 7250
+    },
+    {
+      "epoch": 0.17638708508260595,
+      "grad_norm": 0.9079201221466064,
+      "learning_rate": 0.00025011372807982977,
+      "loss": 5.0613796997070315,
+      "step": 7300
+    },
+    {
+      "epoch": 0.17759521580234983,
+      "grad_norm": 0.7784046530723572,
+      "learning_rate": 0.00024974686330618535,
+      "loss": 5.075418090820312,
+      "step": 7350
+    },
+    {
+      "epoch": 0.1788033465220937,
+      "grad_norm": 0.7922983765602112,
+      "learning_rate": 0.00024937999853254087,
+      "loss": 5.0171682739257815,
+      "step": 7400
+    },
+    {
+      "epoch": 0.18001147724183755,
+      "grad_norm": 0.7115756869316101,
+      "learning_rate": 0.00024901313375889645,
+      "loss": 5.027638549804688,
+      "step": 7450
+    },
+    {
+      "epoch": 0.18121960796158143,
+      "grad_norm": 0.8105598092079163,
+      "learning_rate": 0.00024864626898525203,
+      "loss": 4.98029296875,
+      "step": 7500
+    },
+    {
+      "epoch": 0.1824277386813253,
+      "grad_norm": 0.8051915764808655,
+      "learning_rate": 0.0002482794042116076,
+      "loss": 4.972913818359375,
+      "step": 7550
+    },
+    {
+      "epoch": 0.18363586940106918,
+      "grad_norm": 0.7506864070892334,
+      "learning_rate": 0.00024791253943796314,
+      "loss": 4.998473510742188,
+      "step": 7600
+    },
+    {
+      "epoch": 0.18484400012081306,
+      "grad_norm": 0.7961404323577881,
+      "learning_rate": 0.0002475456746643187,
+      "loss": 4.959234008789062,
+      "step": 7650
+    },
+    {
+      "epoch": 0.18605213084055694,
+      "grad_norm": 0.7458824515342712,
+      "learning_rate": 0.0002471788098906743,
+      "loss": 4.998105773925781,
+      "step": 7700
+    },
+    {
+      "epoch": 0.18726026156030082,
+      "grad_norm": 0.7325210571289062,
+      "learning_rate": 0.0002468119451170299,
+      "loss": 4.964185180664063,
+      "step": 7750
+    },
+    {
+      "epoch": 0.1884683922800447,
+      "grad_norm": 0.7937678694725037,
+      "learning_rate": 0.0002464450803433854,
+      "loss": 4.995216369628906,
+      "step": 7800
+    },
+    {
+      "epoch": 0.18967652299978857,
+      "grad_norm": 0.8441283702850342,
+      "learning_rate": 0.000246078215569741,
+      "loss": 4.9506619262695315,
+      "step": 7850
+    },
+    {
+      "epoch": 0.19088465371953245,
+      "grad_norm": 0.7642113566398621,
+      "learning_rate": 0.00024571135079609656,
+      "loss": 4.967312622070312,
+      "step": 7900
+    },
+    {
+      "epoch": 0.19209278443927633,
+      "grad_norm": 0.7840285897254944,
+      "learning_rate": 0.0002453444860224521,
+      "loss": 4.995877075195312,
+      "step": 7950
+    },
+    {
+      "epoch": 0.1933009151590202,
+      "grad_norm": 0.7469853758811951,
+      "learning_rate": 0.00024497762124880767,
+      "loss": 4.956955871582031,
+      "step": 8000
+    },
+    {
+      "epoch": 0.19450904587876408,
+      "grad_norm": 0.8073136806488037,
+      "learning_rate": 0.00024461075647516325,
+      "loss": 4.956441955566406,
+      "step": 8050
+    },
+    {
+      "epoch": 0.19571717659850796,
+      "grad_norm": 0.7915883660316467,
+      "learning_rate": 0.0002442438917015188,
+      "loss": 4.91659423828125,
+      "step": 8100
+    },
+    {
+      "epoch": 0.19692530731825184,
+      "grad_norm": 0.8093664050102234,
+      "learning_rate": 0.00024387702692787438,
+      "loss": 4.891084289550781,
+      "step": 8150
+    },
+    {
+      "epoch": 0.1981334380379957,
+      "grad_norm": 0.7481308579444885,
+      "learning_rate": 0.0002435101621542299,
+      "loss": 4.929019165039063,
+      "step": 8200
+    },
+    {
+      "epoch": 0.1993415687577396,
+      "grad_norm": 0.7783725261688232,
+      "learning_rate": 0.0002431432973805855,
+      "loss": 4.928171081542969,
+      "step": 8250
+    },
+    {
+      "epoch": 0.20054969947748347,
+      "grad_norm": 0.8197956085205078,
+      "learning_rate": 0.00024277643260694107,
+      "loss": 4.922730407714844,
+      "step": 8300
+    },
+    {
+      "epoch": 0.20175783019722734,
+      "grad_norm": 0.8039451837539673,
+      "learning_rate": 0.00024240956783329665,
+      "loss": 4.913426513671875,
+      "step": 8350
+    },
+    {
+      "epoch": 0.20296596091697122,
+      "grad_norm": 0.7451205849647522,
+      "learning_rate": 0.00024204270305965217,
+      "loss": 4.931348266601563,
+      "step": 8400
+    },
+    {
+      "epoch": 0.2041740916367151,
+      "grad_norm": 0.7971811294555664,
+      "learning_rate": 0.00024167583828600775,
+      "loss": 4.899661865234375,
+      "step": 8450
+    },
+    {
+      "epoch": 0.20538222235645898,
+      "grad_norm": 0.7709014415740967,
+      "learning_rate": 0.00024130897351236333,
+      "loss": 4.940385131835938,
+      "step": 8500
+    },
+    {
+      "epoch": 0.20659035307620285,
+      "grad_norm": 0.7686259150505066,
+      "learning_rate": 0.0002409421087387189,
+      "loss": 4.917569580078125,
+      "step": 8550
+    },
+    {
+      "epoch": 0.20779848379594673,
+      "grad_norm": 0.7921050190925598,
+      "learning_rate": 0.00024057524396507444,
+      "loss": 4.942485656738281,
+      "step": 8600
+    },
+    {
+      "epoch": 0.2090066145156906,
+      "grad_norm": 0.8189991116523743,
+      "learning_rate": 0.00024020837919143002,
+      "loss": 4.875235900878907,
+      "step": 8650
+    },
+    {
+      "epoch": 0.21021474523543449,
+      "grad_norm": 0.7947618365287781,
+      "learning_rate": 0.00023984151441778557,
+      "loss": 4.902606201171875,
+      "step": 8700
+    },
+    {
+      "epoch": 0.21142287595517836,
+      "grad_norm": 0.7725178599357605,
+      "learning_rate": 0.00023947464964414115,
+      "loss": 4.876233520507813,
+      "step": 8750
+    },
+    {
+      "epoch": 0.2126310066749222,
+      "grad_norm": 0.9121379852294922,
+      "learning_rate": 0.0002391077848704967,
+      "loss": 4.871437377929688,
+      "step": 8800
+    },
+    {
+      "epoch": 0.2138391373946661,
+      "grad_norm": 0.7896897196769714,
+      "learning_rate": 0.00023874092009685229,
+      "loss": 4.847481079101563,
+      "step": 8850
+    },
+    {
+      "epoch": 0.21504726811440997,
+      "grad_norm": 0.7763454914093018,
+      "learning_rate": 0.00023837405532320784,
+      "loss": 4.862617797851563,
+      "step": 8900
+    },
+    {
+      "epoch": 0.21625539883415384,
+      "grad_norm": 0.7935928106307983,
+      "learning_rate": 0.00023800719054956342,
+      "loss": 4.864496765136718,
+      "step": 8950
+    },
+    {
+      "epoch": 0.21746352955389772,
+      "grad_norm": 0.7815380096435547,
+      "learning_rate": 0.00023764032577591897,
+      "loss": 4.855505065917969,
+      "step": 9000
+    },
+    {
+      "epoch": 0.2186716602736416,
+      "grad_norm": 0.8099331259727478,
+      "learning_rate": 0.00023727346100227452,
+      "loss": 4.848117370605468,
+      "step": 9050
+    },
+    {
+      "epoch": 0.21987979099338548,
+      "grad_norm": 0.8041796684265137,
+      "learning_rate": 0.0002369065962286301,
+      "loss": 4.835195007324219,
+      "step": 9100
+    },
+    {
+      "epoch": 0.22108792171312935,
+      "grad_norm": 0.81269770860672,
+      "learning_rate": 0.00023653973145498569,
+      "loss": 4.860754089355469,
+      "step": 9150
+    },
+    {
+      "epoch": 0.22229605243287323,
+      "grad_norm": 0.7945086359977722,
+      "learning_rate": 0.0002361728666813412,
+      "loss": 4.817361450195312,
+      "step": 9200
+    },
+    {
+      "epoch": 0.2235041831526171,
+      "grad_norm": 0.7525965571403503,
+      "learning_rate": 0.0002358060019076968,
+      "loss": 4.852757263183594,
+      "step": 9250
+    },
+    {
+      "epoch": 0.22471231387236099,
+      "grad_norm": 0.8420506715774536,
+      "learning_rate": 0.00023543913713405237,
+      "loss": 4.873468017578125,
+      "step": 9300
+    },
+    {
+      "epoch": 0.22592044459210486,
+      "grad_norm": 0.7687504291534424,
+      "learning_rate": 0.00023507227236040795,
+      "loss": 4.81997802734375,
+      "step": 9350
+    },
+    {
+      "epoch": 0.22712857531184874,
+      "grad_norm": 0.8129599690437317,
+      "learning_rate": 0.00023470540758676348,
+      "loss": 4.82060546875,
+      "step": 9400
+    },
+    {
+      "epoch": 0.22833670603159262,
+      "grad_norm": 0.7765640616416931,
+      "learning_rate": 0.00023433854281311906,
+      "loss": 4.830821228027344,
+      "step": 9450
+    },
+    {
+      "epoch": 0.2295448367513365,
+      "grad_norm": 0.8525048494338989,
+      "learning_rate": 0.00023397167803947464,
+      "loss": 4.826491394042969,
+      "step": 9500
+    },
+    {
+      "epoch": 0.23075296747108037,
+      "grad_norm": 0.7599337100982666,
+      "learning_rate": 0.00023360481326583022,
+      "loss": 4.799618225097657,
+      "step": 9550
+    },
+    {
+      "epoch": 0.23196109819082425,
+      "grad_norm": 0.883025586605072,
+      "learning_rate": 0.00023323794849218574,
+      "loss": 4.815971984863281,
+      "step": 9600
+    },
+    {
+      "epoch": 0.23316922891056813,
+      "grad_norm": 0.8563060164451599,
+      "learning_rate": 0.00023287108371854132,
+      "loss": 4.799457092285156,
+      "step": 9650
+    },
+    {
+      "epoch": 0.234377359630312,
+      "grad_norm": 0.8175627589225769,
+      "learning_rate": 0.0002325042189448969,
+      "loss": 4.818922729492187,
+      "step": 9700
+    },
+    {
+      "epoch": 0.23558549035005588,
+      "grad_norm": 0.7504342198371887,
+      "learning_rate": 0.00023213735417125246,
+      "loss": 4.852285461425781,
+      "step": 9750
+    },
+    {
+      "epoch": 0.23679362106979976,
+      "grad_norm": 0.7624800205230713,
+      "learning_rate": 0.000231770489397608,
+      "loss": 4.757030334472656,
+      "step": 9800
+    },
+    {
+      "epoch": 0.23800175178954364,
+      "grad_norm": 0.7682377099990845,
+      "learning_rate": 0.0002314036246239636,
+      "loss": 4.7794973754882815,
+      "step": 9850
+    },
+    {
+      "epoch": 0.2392098825092875,
+      "grad_norm": 0.7975249290466309,
+      "learning_rate": 0.00023103675985031914,
+      "loss": 4.790357971191407,
+      "step": 9900
+    },
+    {
+      "epoch": 0.2404180132290314,
+      "grad_norm": 0.8084898591041565,
+      "learning_rate": 0.00023066989507667472,
+      "loss": 4.760804748535156,
+      "step": 9950
+    },
+    {
+      "epoch": 0.24162614394877527,
+      "grad_norm": 0.7954447865486145,
+      "learning_rate": 0.00023030303030303027,
+      "loss": 4.762905883789062,
+      "step": 10000
+    },
+    {
+      "epoch": 0.24283427466851915,
+      "grad_norm": 0.8145322799682617,
+      "learning_rate": 0.00022993616552938583,
+      "loss": 4.771827392578125,
+      "step": 10050
+    },
+    {
+      "epoch": 0.24404240538826302,
+      "grad_norm": 0.7774279117584229,
+      "learning_rate": 0.0002295693007557414,
+      "loss": 4.779819946289063,
+      "step": 10100
+    },
+    {
+      "epoch": 0.2452505361080069,
+      "grad_norm": 0.7880101799964905,
+      "learning_rate": 0.000229202435982097,
+      "loss": 4.778431091308594,
+      "step": 10150
+    },
+    {
+      "epoch": 0.24645866682775075,
+      "grad_norm": 0.7913017868995667,
+      "learning_rate": 0.00022883557120845254,
+      "loss": 4.729861145019531,
+      "step": 10200
+    },
+    {
+      "epoch": 0.24766679754749463,
+      "grad_norm": 0.7822272181510925,
+      "learning_rate": 0.0002284687064348081,
+      "loss": 4.7321142578125,
+      "step": 10250
+    },
+    {
+      "epoch": 0.2488749282672385,
+      "grad_norm": 0.7684321999549866,
+      "learning_rate": 0.00022810184166116367,
+      "loss": 4.733845520019531,
+      "step": 10300
+    },
+    {
+      "epoch": 0.2500830589869824,
+      "grad_norm": 0.805257260799408,
+      "learning_rate": 0.00022773497688751925,
+      "loss": 4.7672314453125,
+      "step": 10350
+    },
+    {
+      "epoch": 0.25129118970672626,
+      "grad_norm": 0.784733772277832,
+      "learning_rate": 0.00022736811211387478,
+      "loss": 4.761484680175781,
+      "step": 10400
+    },
+    {
+      "epoch": 0.25249932042647016,
+      "grad_norm": 0.7618144154548645,
+      "learning_rate": 0.00022700124734023036,
+      "loss": 4.746294555664062,
+      "step": 10450
+    },
+    {
+      "epoch": 0.253707451146214,
+      "grad_norm": 0.7694723010063171,
+      "learning_rate": 0.00022663438256658594,
+      "loss": 4.743426513671875,
+      "step": 10500
+    },
+    {
+      "epoch": 0.2549155818659579,
+      "grad_norm": 0.7867644429206848,
+      "learning_rate": 0.00022626751779294152,
+      "loss": 4.714345703125,
+      "step": 10550
+    },
+    {
+      "epoch": 0.25612371258570177,
+      "grad_norm": 0.7482685446739197,
+      "learning_rate": 0.00022590065301929705,
+      "loss": 4.770420227050781,
+      "step": 10600
+    },
+    {
+      "epoch": 0.2573318433054457,
+      "grad_norm": 0.811345100402832,
+      "learning_rate": 0.00022553378824565263,
+      "loss": 4.725180969238282,
+      "step": 10650
+    },
+    {
+      "epoch": 0.2585399740251895,
+      "grad_norm": 0.881449818611145,
+      "learning_rate": 0.0002251669234720082,
+      "loss": 4.748963928222656,
+      "step": 10700
+    },
+    {
+      "epoch": 0.25974810474493343,
+      "grad_norm": 0.8054546117782593,
+      "learning_rate": 0.00022480005869836376,
+      "loss": 4.7404483032226565,
+      "step": 10750
+    },
+    {
+      "epoch": 0.2609562354646773,
+      "grad_norm": 0.8301592469215393,
+      "learning_rate": 0.0002244331939247193,
+      "loss": 4.707667236328125,
+      "step": 10800
+    },
+    {
+      "epoch": 0.2621643661844211,
+      "grad_norm": 0.7749137878417969,
+      "learning_rate": 0.0002240663291510749,
+      "loss": 4.718084411621094,
+      "step": 10850
+    },
+    {
+      "epoch": 0.26337249690416503,
+      "grad_norm": 0.765621542930603,
+      "learning_rate": 0.00022369946437743047,
+      "loss": 4.70483642578125,
+      "step": 10900
+    },
+    {
+      "epoch": 0.2645806276239089,
+      "grad_norm": 0.8450783491134644,
+      "learning_rate": 0.00022333259960378602,
+      "loss": 4.702041015625,
+      "step": 10950
+    },
+    {
+      "epoch": 0.2657887583436528,
+      "grad_norm": 0.786019504070282,
+      "learning_rate": 0.0002229657348301416,
+      "loss": 4.73208251953125,
+      "step": 11000
+    },
+    {
+      "epoch": 0.26699688906339664,
+      "grad_norm": 0.844512403011322,
+      "learning_rate": 0.00022259887005649716,
+      "loss": 4.710789184570313,
+      "step": 11050
+    },
+    {
+      "epoch": 0.26820501978314054,
+      "grad_norm": 0.7994405627250671,
+      "learning_rate": 0.0002222320052828527,
+      "loss": 4.725503540039062,
+      "step": 11100
+    },
+    {
+      "epoch": 0.2694131505028844,
+      "grad_norm": 0.8023877143859863,
+      "learning_rate": 0.0002218651405092083,
+      "loss": 4.672783203125,
+      "step": 11150
+    },
+    {
+      "epoch": 0.2706212812226283,
+      "grad_norm": 0.7944827079772949,
+      "learning_rate": 0.00022149827573556387,
+      "loss": 4.7068939208984375,
+      "step": 11200
+    },
+    {
+      "epoch": 0.27182941194237215,
+      "grad_norm": 0.7936934232711792,
+      "learning_rate": 0.0002211314109619194,
+      "loss": 4.694465637207031,
+      "step": 11250
+    },
+    {
+      "epoch": 0.27303754266211605,
+      "grad_norm": 0.7808048725128174,
+      "learning_rate": 0.00022076454618827498,
+      "loss": 4.731701049804688,
+      "step": 11300
+    },
+    {
+      "epoch": 0.2742456733818599,
+      "grad_norm": 0.8140460848808289,
+      "learning_rate": 0.00022039768141463056,
+      "loss": 4.685305786132813,
+      "step": 11350
+    },
+    {
+      "epoch": 0.2754538041016038,
+      "grad_norm": 0.7861780524253845,
+      "learning_rate": 0.00022003081664098614,
+      "loss": 4.691961059570312,
+      "step": 11400
+    },
+    {
+      "epoch": 0.27666193482134765,
+      "grad_norm": 0.871821403503418,
+      "learning_rate": 0.00021966395186734166,
+      "loss": 4.664822692871094,
+      "step": 11450
+    },
+    {
+      "epoch": 0.27787006554109156,
+      "grad_norm": 0.8274059891700745,
+      "learning_rate": 0.00021929708709369724,
+      "loss": 4.716848449707031,
+      "step": 11500
+    },
+    {
+      "epoch": 0.2790781962608354,
+      "grad_norm": 0.7896090745925903,
+      "learning_rate": 0.00021893022232005282,
+      "loss": 4.687721557617188,
+      "step": 11550
+    },
+    {
+      "epoch": 0.2802863269805793,
+      "grad_norm": 0.8510223627090454,
+      "learning_rate": 0.0002185633575464084,
+      "loss": 4.675347900390625,
+      "step": 11600
+    },
+    {
+      "epoch": 0.28149445770032316,
+      "grad_norm": 0.8153030276298523,
+      "learning_rate": 0.00021819649277276393,
+      "loss": 4.6624560546875,
+      "step": 11650
+    },
+    {
+      "epoch": 0.28270258842006707,
+      "grad_norm": 0.8229708671569824,
+      "learning_rate": 0.0002178296279991195,
+      "loss": 4.6419219970703125,
+      "step": 11700
+    },
+    {
+      "epoch": 0.2839107191398109,
+      "grad_norm": 0.9006219506263733,
+      "learning_rate": 0.0002174627632254751,
+      "loss": 4.702740783691406,
+      "step": 11750
+    },
+    {
+      "epoch": 0.2851188498595548,
+      "grad_norm": 0.7775527238845825,
+      "learning_rate": 0.00021709589845183064,
+      "loss": 4.671443786621094,
+      "step": 11800
+    },
+    {
+      "epoch": 0.2863269805792987,
+      "grad_norm": 0.8033133149147034,
+      "learning_rate": 0.0002167290336781862,
+      "loss": 4.662550964355469,
+      "step": 11850
+    },
+    {
+      "epoch": 0.2875351112990426,
+      "grad_norm": 0.7958892583847046,
+      "learning_rate": 0.00021636216890454177,
+      "loss": 4.6466238403320315,
+      "step": 11900
+    },
+    {
+      "epoch": 0.28874324201878643,
+      "grad_norm": 0.8526051640510559,
+      "learning_rate": 0.00021599530413089733,
+      "loss": 4.6294076538085935,
+      "step": 11950
+    },
+    {
+      "epoch": 0.28995137273853033,
+      "grad_norm": 0.7941024899482727,
+      "learning_rate": 0.0002156284393572529,
+      "loss": 4.657857360839844,
+      "step": 12000
+    },
+    {
+      "epoch": 0.2911595034582742,
+      "grad_norm": 0.7991407513618469,
+      "learning_rate": 0.00021526157458360846,
+      "loss": 4.664903564453125,
+      "step": 12050
+    },
+    {
+      "epoch": 0.2923676341780181,
+      "grad_norm": 0.8050182461738586,
+      "learning_rate": 0.00021489470980996404,
+      "loss": 4.61813232421875,
+      "step": 12100
+    },
+    {
+      "epoch": 0.29357576489776194,
+      "grad_norm": 0.8054724931716919,
+      "learning_rate": 0.0002145278450363196,
+      "loss": 4.660469970703125,
+      "step": 12150
+    },
+    {
+      "epoch": 0.2947838956175058,
+      "grad_norm": 0.7791872024536133,
+      "learning_rate": 0.00021416098026267517,
+      "loss": 4.6317071533203125,
+      "step": 12200
+    },
+    {
+      "epoch": 0.2959920263372497,
+      "grad_norm": 0.8185362219810486,
+      "learning_rate": 0.00021379411548903073,
+      "loss": 4.635613098144531,
+      "step": 12250
+    },
+    {
+      "epoch": 0.29720015705699354,
+      "grad_norm": 0.7946044206619263,
+      "learning_rate": 0.00021342725071538628,
+      "loss": 4.653995056152343,
+      "step": 12300
+    },
+    {
+      "epoch": 0.29840828777673745,
+      "grad_norm": 0.7659194469451904,
+      "learning_rate": 0.00021306038594174186,
+      "loss": 4.647461853027344,
+      "step": 12350
+    },
+    {
+      "epoch": 0.2996164184964813,
+      "grad_norm": 0.8447695374488831,
+      "learning_rate": 0.00021269352116809744,
+      "loss": 4.663876037597657,
+      "step": 12400
+    },
+    {
+      "epoch": 0.3008245492162252,
+      "grad_norm": 0.832804799079895,
+      "learning_rate": 0.00021232665639445296,
+      "loss": 4.651454162597656,
+      "step": 12450
+    },
+    {
+      "epoch": 0.30203267993596905,
+      "grad_norm": 0.8120107650756836,
+      "learning_rate": 0.00021195979162080854,
+      "loss": 4.6463534545898435,
+      "step": 12500
+    },
+    {
+      "epoch": 0.30324081065571296,
+      "grad_norm": 0.8200873732566833,
+      "learning_rate": 0.00021159292684716412,
+      "loss": 4.666345825195313,
+      "step": 12550
+    },
+    {
+      "epoch": 0.3044489413754568,
+      "grad_norm": 0.7715031504631042,
+      "learning_rate": 0.0002112260620735197,
+      "loss": 4.62532470703125,
+      "step": 12600
+    },
+    {
+      "epoch": 0.3056570720952007,
+      "grad_norm": 0.8087065815925598,
+      "learning_rate": 0.00021085919729987523,
+      "loss": 4.599889221191407,
+      "step": 12650
+    },
+    {
+      "epoch": 0.30686520281494456,
+      "grad_norm": 0.8186317682266235,
+      "learning_rate": 0.0002104923325262308,
+      "loss": 4.607811279296875,
+      "step": 12700
+    },
+    {
+      "epoch": 0.30807333353468846,
+      "grad_norm": 0.7934784889221191,
+      "learning_rate": 0.0002101254677525864,
+      "loss": 4.617008056640625,
+      "step": 12750
+    },
+    {
+      "epoch": 0.3092814642544323,
+      "grad_norm": 0.8471022248268127,
+      "learning_rate": 0.00020975860297894197,
+      "loss": 4.6198193359375,
+      "step": 12800
+    },
+    {
+      "epoch": 0.3104895949741762,
+      "grad_norm": 0.8959652781486511,
+      "learning_rate": 0.0002093917382052975,
+      "loss": 4.609613342285156,
+      "step": 12850
+    },
+    {
+      "epoch": 0.31169772569392007,
+      "grad_norm": 0.8412113785743713,
+      "learning_rate": 0.00020902487343165308,
+      "loss": 4.567796325683593,
+      "step": 12900
+    },
+    {
+      "epoch": 0.312905856413664,
+      "grad_norm": 0.8049896955490112,
+      "learning_rate": 0.00020865800865800866,
+      "loss": 4.602123413085938,
+      "step": 12950
+    },
+    {
+      "epoch": 0.3141139871334078,
+      "grad_norm": 0.7926521897315979,
+      "learning_rate": 0.0002082911438843642,
+      "loss": 4.594961242675781,
+      "step": 13000
+    },
+    {
+      "epoch": 0.31532211785315173,
+      "grad_norm": 0.8098254799842834,
+      "learning_rate": 0.00020792427911071976,
+      "loss": 4.612647399902344,
+      "step": 13050
+    },
+    {
+      "epoch": 0.3165302485728956,
+      "grad_norm": 0.8844075798988342,
+      "learning_rate": 0.00020755741433707534,
+      "loss": 4.614940795898438,
+      "step": 13100
+    },
+    {
+      "epoch": 0.3177383792926395,
+      "grad_norm": 0.8263518214225769,
+      "learning_rate": 0.0002071905495634309,
+      "loss": 4.637500915527344,
+      "step": 13150
+    },
+    {
+      "epoch": 0.31894651001238333,
+      "grad_norm": 0.790128231048584,
+      "learning_rate": 0.00020682368478978648,
+      "loss": 4.628588562011719,
+      "step": 13200
+    },
+    {
+      "epoch": 0.32015464073212724,
+      "grad_norm": 0.7944300770759583,
+      "learning_rate": 0.00020645682001614203,
+      "loss": 4.620392456054687,
+      "step": 13250
+    },
+    {
+      "epoch": 0.3213627714518711,
+      "grad_norm": 0.7934144139289856,
+      "learning_rate": 0.0002060899552424976,
+      "loss": 4.576650695800781,
+      "step": 13300
+    },
+    {
+      "epoch": 0.322570902171615,
+      "grad_norm": 0.8359684944152832,
+      "learning_rate": 0.00020572309046885316,
+      "loss": 4.6265127563476565,
+      "step": 13350
+    },
+    {
+      "epoch": 0.32377903289135884,
+      "grad_norm": 0.8399538993835449,
+      "learning_rate": 0.00020535622569520874,
+      "loss": 4.605622253417969,
+      "step": 13400
+    },
+    {
+      "epoch": 0.32498716361110275,
+      "grad_norm": 0.8221124410629272,
+      "learning_rate": 0.0002049893609215643,
+      "loss": 4.578699340820313,
+      "step": 13450
+    },
+    {
+      "epoch": 0.3261952943308466,
+      "grad_norm": 0.7812410593032837,
+      "learning_rate": 0.00020462249614791985,
+      "loss": 4.581290588378907,
+      "step": 13500
+    },
+    {
+      "epoch": 0.32740342505059045,
+      "grad_norm": 0.7977650761604309,
+      "learning_rate": 0.00020425563137427543,
+      "loss": 4.543898010253907,
+      "step": 13550
+    },
+    {
+      "epoch": 0.32861155577033435,
+      "grad_norm": 0.8016378283500671,
+      "learning_rate": 0.000203888766600631,
+      "loss": 4.588157958984375,
+      "step": 13600
+    },
+    {
+      "epoch": 0.3298196864900782,
+      "grad_norm": 0.8839250206947327,
+      "learning_rate": 0.00020352190182698653,
+      "loss": 4.553284301757812,
+      "step": 13650
+    },
+    {
+      "epoch": 0.3310278172098221,
+      "grad_norm": 0.8013595938682556,
+      "learning_rate": 0.0002031550370533421,
+      "loss": 4.6255426025390625,
+      "step": 13700
+    },
+    {
+      "epoch": 0.33223594792956596,
+      "grad_norm": 0.8826407194137573,
+      "learning_rate": 0.0002027881722796977,
+      "loss": 4.547523803710938,
+      "step": 13750
+    },
+    {
+      "epoch": 0.33344407864930986,
+      "grad_norm": 0.8313543200492859,
+      "learning_rate": 0.00020242130750605327,
+      "loss": 4.5199212646484375,
+      "step": 13800
+    },
+    {
+      "epoch": 0.3346522093690537,
+      "grad_norm": 0.8718063235282898,
+      "learning_rate": 0.0002020544427324088,
+      "loss": 4.586356201171875,
+      "step": 13850
+    },
+    {
+      "epoch": 0.3358603400887976,
+      "grad_norm": 0.8381774425506592,
+      "learning_rate": 0.00020168757795876438,
+      "loss": 4.569805297851563,
+      "step": 13900
+    },
+    {
+      "epoch": 0.33706847080854146,
+      "grad_norm": 0.7955442070960999,
+      "learning_rate": 0.00020132071318511996,
+      "loss": 4.5325839233398435,
+      "step": 13950
+    },
+    {
+      "epoch": 0.33827660152828537,
+      "grad_norm": 0.8485741019248962,
+      "learning_rate": 0.00020095384841147554,
+      "loss": 4.533649291992187,
+      "step": 14000
+    },
+    {
+      "epoch": 0.3394847322480292,
+      "grad_norm": 0.9278120994567871,
+      "learning_rate": 0.00020058698363783107,
+      "loss": 4.553104248046875,
+      "step": 14050
+    },
+    {
+      "epoch": 0.3406928629677731,
+      "grad_norm": 0.833659827709198,
+      "learning_rate": 0.00020022011886418665,
+      "loss": 4.5760061645507815,
+      "step": 14100
+    },
+    {
+      "epoch": 0.341900993687517,
+      "grad_norm": 0.8239136338233948,
+      "learning_rate": 0.00019985325409054223,
+      "loss": 4.61402587890625,
+      "step": 14150
+    },
+    {
+      "epoch": 0.3431091244072609,
+      "grad_norm": 0.8305805325508118,
+      "learning_rate": 0.00019948638931689778,
+      "loss": 4.527126159667969,
+      "step": 14200
+    },
+    {
+      "epoch": 0.34431725512700473,
+      "grad_norm": 0.8342345952987671,
+      "learning_rate": 0.00019911952454325333,
+      "loss": 4.58315185546875,
+      "step": 14250
+    },
+    {
+      "epoch": 0.34552538584674863,
+      "grad_norm": 0.7768062353134155,
+      "learning_rate": 0.0001987526597696089,
+      "loss": 4.545763549804687,
+      "step": 14300
+    },
+    {
+      "epoch": 0.3467335165664925,
+      "grad_norm": 0.8638603687286377,
+      "learning_rate": 0.00019838579499596446,
+      "loss": 4.5633633422851565,
+      "step": 14350
+    },
+    {
+      "epoch": 0.3479416472862364,
+      "grad_norm": 0.8149457573890686,
+      "learning_rate": 0.00019801893022232004,
+      "loss": 4.527096862792969,
+      "step": 14400
+    },
+    {
+      "epoch": 0.34914977800598024,
+      "grad_norm": 0.7938598394393921,
+      "learning_rate": 0.0001976520654486756,
+      "loss": 4.532784118652343,
+      "step": 14450
+    },
+    {
+      "epoch": 0.35035790872572414,
+      "grad_norm": 0.8493742942810059,
+      "learning_rate": 0.00019728520067503115,
+      "loss": 4.554869384765625,
+      "step": 14500
+    },
+    {
+      "epoch": 0.351566039445468,
+      "grad_norm": 0.792421281337738,
+      "learning_rate": 0.00019691833590138673,
+      "loss": 4.560779113769531,
+      "step": 14550
+    },
+    {
+      "epoch": 0.3527741701652119,
+      "grad_norm": 0.8050044775009155,
+      "learning_rate": 0.0001965514711277423,
+      "loss": 4.5141604614257815,
+      "step": 14600
+    },
+    {
+      "epoch": 0.35398230088495575,
+      "grad_norm": 0.816679060459137,
+      "learning_rate": 0.00019618460635409786,
+      "loss": 4.5143798828125,
+      "step": 14650
+    },
+    {
+      "epoch": 0.35519043160469965,
+      "grad_norm": 0.8072745203971863,
+      "learning_rate": 0.00019581774158045342,
+      "loss": 4.527074890136719,
+      "step": 14700
+    },
+    {
+      "epoch": 0.3563985623244435,
+      "grad_norm": 0.8425279259681702,
+      "learning_rate": 0.000195450876806809,
+      "loss": 4.540933532714844,
+      "step": 14750
+    },
+    {
+      "epoch": 0.3576066930441874,
+      "grad_norm": 0.8636919856071472,
+      "learning_rate": 0.00019508401203316458,
+      "loss": 4.550223388671875,
+      "step": 14800
+    },
+    {
+      "epoch": 0.35881482376393126,
+      "grad_norm": 0.7965017557144165,
+      "learning_rate": 0.0001947171472595201,
+      "loss": 4.53455078125,
+      "step": 14850
+    },
+    {
+      "epoch": 0.3600229544836751,
+      "grad_norm": 0.8195152878761292,
+      "learning_rate": 0.00019435028248587568,
+      "loss": 4.525997619628907,
+      "step": 14900
+    },
+    {
+      "epoch": 0.361231085203419,
+      "grad_norm": 0.8306058645248413,
+      "learning_rate": 0.00019398341771223126,
+      "loss": 4.519093627929688,
+      "step": 14950
+    },
+    {
+      "epoch": 0.36243921592316286,
+      "grad_norm": 0.803713321685791,
+      "learning_rate": 0.00019361655293858684,
+      "loss": 4.549564819335938,
+      "step": 15000
+    },
+    {
+      "epoch": 0.36364734664290677,
+      "grad_norm": 0.835669994354248,
+      "learning_rate": 0.00019324968816494237,
+      "loss": 4.529533996582031,
+      "step": 15050
+    },
+    {
+      "epoch": 0.3648554773626506,
+      "grad_norm": 0.977713942527771,
+      "learning_rate": 0.00019288282339129795,
+      "loss": 4.527159729003906,
+      "step": 15100
+    },
+    {
+      "epoch": 0.3660636080823945,
+      "grad_norm": 0.7918193936347961,
+      "learning_rate": 0.00019251595861765353,
+      "loss": 4.523968200683594,
+      "step": 15150
+    },
+    {
+      "epoch": 0.36727173880213837,
+      "grad_norm": 0.8662403225898743,
+      "learning_rate": 0.00019214909384400908,
+      "loss": 4.535425720214843,
+      "step": 15200
+    },
+    {
+      "epoch": 0.3684798695218823,
+      "grad_norm": 0.8149556517601013,
+      "learning_rate": 0.00019178222907036463,
+      "loss": 4.518607482910157,
+      "step": 15250
+    },
+    {
+      "epoch": 0.3696880002416261,
+      "grad_norm": 0.827381432056427,
+      "learning_rate": 0.00019141536429672021,
+      "loss": 4.498069458007812,
+      "step": 15300
+    },
+    {
+      "epoch": 0.37089613096137003,
+      "grad_norm": 0.7827855944633484,
+      "learning_rate": 0.0001910484995230758,
+      "loss": 4.5646533203125,
+      "step": 15350
+    },
+    {
+      "epoch": 0.3721042616811139,
+      "grad_norm": 0.8546699285507202,
+      "learning_rate": 0.00019068163474943135,
+      "loss": 4.506566772460937,
+      "step": 15400
+    },
+    {
+      "epoch": 0.3733123924008578,
+      "grad_norm": 0.8151622414588928,
+      "learning_rate": 0.0001903147699757869,
+      "loss": 4.526211242675782,
+      "step": 15450
+    },
+    {
+      "epoch": 0.37452052312060163,
+      "grad_norm": 0.8364620208740234,
+      "learning_rate": 0.00018994790520214248,
+      "loss": 4.5639035034179685,
+      "step": 15500
+    },
+    {
+      "epoch": 0.37572865384034554,
+      "grad_norm": 0.8212669491767883,
+      "learning_rate": 0.00018958104042849803,
+      "loss": 4.533548889160156,
+      "step": 15550
+    },
+    {
+      "epoch": 0.3769367845600894,
+      "grad_norm": 0.8324586153030396,
+      "learning_rate": 0.0001892141756548536,
+      "loss": 4.502703857421875,
+      "step": 15600
+    },
+    {
+      "epoch": 0.3781449152798333,
+      "grad_norm": 0.8079342246055603,
+      "learning_rate": 0.00018884731088120917,
+      "loss": 4.5140142822265625,
+      "step": 15650
+    },
+    {
+      "epoch": 0.37935304599957714,
+      "grad_norm": 0.8679006695747375,
+      "learning_rate": 0.00018848044610756472,
+      "loss": 4.466272888183593,
+      "step": 15700
+    },
+    {
+      "epoch": 0.38056117671932105,
+      "grad_norm": 0.845909833908081,
+      "learning_rate": 0.0001881135813339203,
+      "loss": 4.510520629882812,
+      "step": 15750
+    },
+    {
+      "epoch": 0.3817693074390649,
+      "grad_norm": 0.830926775932312,
+      "learning_rate": 0.00018774671656027588,
+      "loss": 4.5115673828125,
+      "step": 15800
+    },
+    {
+      "epoch": 0.3829774381588088,
+      "grad_norm": 0.8161680698394775,
+      "learning_rate": 0.00018737985178663143,
+      "loss": 4.543225402832031,
+      "step": 15850
+    },
+    {
+      "epoch": 0.38418556887855265,
+      "grad_norm": 0.8474006056785583,
+      "learning_rate": 0.00018701298701298698,
+      "loss": 4.485232238769531,
+      "step": 15900
+    },
+    {
+      "epoch": 0.38539369959829656,
+      "grad_norm": 0.8017637729644775,
+      "learning_rate": 0.00018664612223934256,
+      "loss": 4.5208154296875,
+      "step": 15950
+    },
+    {
+      "epoch": 0.3866018303180404,
+      "grad_norm": 0.8082236051559448,
+      "learning_rate": 0.00018627925746569814,
+      "loss": 4.4930859375,
+      "step": 16000
+    },
+    {
+      "epoch": 0.3878099610377843,
+      "grad_norm": 0.8038812279701233,
+      "learning_rate": 0.00018591239269205367,
+      "loss": 4.503486633300781,
+      "step": 16050
+    },
+    {
+      "epoch": 0.38901809175752816,
+      "grad_norm": 0.8439401388168335,
+      "learning_rate": 0.00018554552791840925,
+      "loss": 4.538678588867188,
+      "step": 16100
+    },
+    {
+      "epoch": 0.39022622247727207,
+      "grad_norm": 0.8262456059455872,
+      "learning_rate": 0.00018517866314476483,
+      "loss": 4.465330505371094,
+      "step": 16150
+    },
+    {
+      "epoch": 0.3914343531970159,
+      "grad_norm": 0.9444901943206787,
+      "learning_rate": 0.0001848117983711204,
+      "loss": 4.457840576171875,
+      "step": 16200
+    },
+    {
+      "epoch": 0.39264248391675977,
+      "grad_norm": 0.8073329329490662,
+      "learning_rate": 0.00018444493359747594,
+      "loss": 4.469413452148437,
+      "step": 16250
+    },
+    {
+      "epoch": 0.39385061463650367,
+      "grad_norm": 0.8293718695640564,
+      "learning_rate": 0.00018407806882383152,
+      "loss": 4.467626647949219,
+      "step": 16300
+    },
+    {
+      "epoch": 0.3950587453562475,
+      "grad_norm": 0.9100238680839539,
+      "learning_rate": 0.0001837112040501871,
+      "loss": 4.480698852539063,
+      "step": 16350
+    },
+    {
+      "epoch": 0.3962668760759914,
+      "grad_norm": 0.8253241181373596,
+      "learning_rate": 0.00018334433927654265,
+      "loss": 4.495014953613281,
+      "step": 16400
+    },
+    {
+      "epoch": 0.3974750067957353,
+      "grad_norm": 0.8466793298721313,
+      "learning_rate": 0.0001829774745028982,
+      "loss": 4.456530151367187,
+      "step": 16450
+    },
+    {
+      "epoch": 0.3986831375154792,
+      "grad_norm": 0.8435772657394409,
+      "learning_rate": 0.00018261060972925378,
+      "loss": 4.511859741210937,
+      "step": 16500
+    },
+    {
+      "epoch": 0.39989126823522303,
+      "grad_norm": 0.8680708408355713,
+      "learning_rate": 0.00018224374495560936,
+      "loss": 4.481240234375,
+      "step": 16550
+    },
+    {
+      "epoch": 0.40109939895496693,
+      "grad_norm": 0.809312641620636,
+      "learning_rate": 0.00018187688018196492,
+      "loss": 4.455697631835937,
+      "step": 16600
+    },
+    {
+      "epoch": 0.4023075296747108,
+      "grad_norm": 0.8194457292556763,
+      "learning_rate": 0.00018151001540832047,
+      "loss": 4.4838330078125,
+      "step": 16650
+    },
+    {
+      "epoch": 0.4035156603944547,
+      "grad_norm": 0.8517265915870667,
+      "learning_rate": 0.00018114315063467605,
+      "loss": 4.53313232421875,
+      "step": 16700
+    },
+    {
+      "epoch": 0.40472379111419854,
+      "grad_norm": 0.8272504210472107,
+      "learning_rate": 0.0001807762858610316,
+      "loss": 4.4739532470703125,
+      "step": 16750
+    },
+    {
+      "epoch": 0.40593192183394244,
+      "grad_norm": 0.8620533347129822,
+      "learning_rate": 0.00018040942108738718,
+      "loss": 4.480051574707031,
+      "step": 16800
+    },
+    {
+      "epoch": 0.4071400525536863,
+      "grad_norm": 0.8758191466331482,
+      "learning_rate": 0.00018004255631374273,
+      "loss": 4.451490783691407,
+      "step": 16850
+    },
+    {
+      "epoch": 0.4083481832734302,
+      "grad_norm": 0.8606282472610474,
+      "learning_rate": 0.0001796756915400983,
+      "loss": 4.490790710449219,
+      "step": 16900
+    },
+    {
+      "epoch": 0.40955631399317405,
+      "grad_norm": 0.8265689611434937,
+      "learning_rate": 0.00017930882676645387,
+      "loss": 4.438824768066406,
+      "step": 16950
+    },
+    {
+      "epoch": 0.41076444471291795,
+      "grad_norm": 0.8095406889915466,
+      "learning_rate": 0.00017894196199280945,
+      "loss": 4.500664367675781,
+      "step": 17000
+    },
+    {
+      "epoch": 0.4119725754326618,
+      "grad_norm": 0.8660163283348083,
+      "learning_rate": 0.00017857509721916497,
+      "loss": 4.419149780273438,
+      "step": 17050
+    },
+    {
+      "epoch": 0.4131807061524057,
+      "grad_norm": 0.8833030462265015,
+      "learning_rate": 0.00017820823244552055,
+      "loss": 4.445498046875,
+      "step": 17100
+    },
+    {
+      "epoch": 0.41438883687214956,
+      "grad_norm": 0.8525263667106628,
+      "learning_rate": 0.00017784136767187613,
+      "loss": 4.4429083251953125,
+      "step": 17150
+    },
+    {
+      "epoch": 0.41559696759189346,
+      "grad_norm": 0.8065162301063538,
+      "learning_rate": 0.0001774745028982317,
+      "loss": 4.415712585449219,
+      "step": 17200
+    },
+    {
+      "epoch": 0.4168050983116373,
+      "grad_norm": 0.8518642783164978,
+      "learning_rate": 0.00017710763812458724,
+      "loss": 4.463846130371094,
+      "step": 17250
+    },
+    {
+      "epoch": 0.4180132290313812,
+      "grad_norm": 0.8767589926719666,
+      "learning_rate": 0.00017674077335094282,
+      "loss": 4.470376586914062,
+      "step": 17300
+    },
+    {
+      "epoch": 0.41922135975112507,
+      "grad_norm": 0.8096307516098022,
+      "learning_rate": 0.0001763739085772984,
+      "loss": 4.446529235839844,
+      "step": 17350
+    },
+    {
+      "epoch": 0.42042949047086897,
+      "grad_norm": 0.873613178730011,
+      "learning_rate": 0.00017600704380365398,
+      "loss": 4.452382507324219,
+      "step": 17400
+    },
+    {
+      "epoch": 0.4216376211906128,
+      "grad_norm": 0.8206058740615845,
+      "learning_rate": 0.0001756401790300095,
+      "loss": 4.3969580078125,
+      "step": 17450
+    },
+    {
+      "epoch": 0.4228457519103567,
+      "grad_norm": 0.833097517490387,
+      "learning_rate": 0.00017527331425636509,
+      "loss": 4.4554660034179685,
+      "step": 17500
+    },
+    {
+      "epoch": 0.4240538826301006,
+      "grad_norm": 0.8413265347480774,
+      "learning_rate": 0.00017490644948272067,
+      "loss": 4.430065612792969,
+      "step": 17550
+    },
+    {
+      "epoch": 0.4252620133498444,
+      "grad_norm": 0.8387306332588196,
+      "learning_rate": 0.00017453958470907622,
+      "loss": 4.449215393066407,
+      "step": 17600
+    },
+    {
+      "epoch": 0.42647014406958833,
+      "grad_norm": 0.8804054260253906,
+      "learning_rate": 0.00017417271993543177,
+      "loss": 4.462908630371094,
+      "step": 17650
+    },
+    {
+      "epoch": 0.4276782747893322,
+      "grad_norm": 0.8563447594642639,
+      "learning_rate": 0.00017380585516178735,
+      "loss": 4.459098205566407,
+      "step": 17700
+    },
+    {
+      "epoch": 0.4288864055090761,
+      "grad_norm": 0.8126643300056458,
+      "learning_rate": 0.00017343899038814293,
+      "loss": 4.444090576171875,
+      "step": 17750
+    },
+    {
+      "epoch": 0.43009453622881993,
+      "grad_norm": 0.8239515423774719,
+      "learning_rate": 0.00017307212561449848,
+      "loss": 4.454663696289063,
+      "step": 17800
+    },
+    {
+      "epoch": 0.43130266694856384,
+      "grad_norm": 0.8403202891349792,
+      "learning_rate": 0.00017270526084085404,
+      "loss": 4.42758056640625,
+      "step": 17850
+    },
+    {
+      "epoch": 0.4325107976683077,
+      "grad_norm": 0.8469553589820862,
+      "learning_rate": 0.00017233839606720962,
+      "loss": 4.432910766601562,
+      "step": 17900
+    },
+    {
+      "epoch": 0.4337189283880516,
+      "grad_norm": 0.8169428706169128,
+      "learning_rate": 0.00017197153129356517,
+      "loss": 4.442397155761719,
+      "step": 17950
+    },
+    {
+      "epoch": 0.43492705910779544,
+      "grad_norm": 0.8393473625183105,
+      "learning_rate": 0.00017160466651992075,
+      "loss": 4.471981506347657,
+      "step": 18000
+    },
+    {
+      "epoch": 0.43613518982753935,
+      "grad_norm": 0.8713968396186829,
+      "learning_rate": 0.0001712378017462763,
+      "loss": 4.4335018920898435,
+      "step": 18050
+    },
+    {
+      "epoch": 0.4373433205472832,
+      "grad_norm": 0.8308825492858887,
+      "learning_rate": 0.00017087093697263186,
+      "loss": 4.431024475097656,
+      "step": 18100
+    },
+    {
+      "epoch": 0.4385514512670271,
+      "grad_norm": 0.8060277104377747,
+      "learning_rate": 0.00017050407219898744,
+      "loss": 4.439822998046875,
+      "step": 18150
+    },
+    {
+      "epoch": 0.43975958198677095,
+      "grad_norm": 0.8310439586639404,
+      "learning_rate": 0.00017013720742534302,
+      "loss": 4.433624267578125,
+      "step": 18200
+    },
+    {
+      "epoch": 0.44096771270651486,
+      "grad_norm": 0.8447884917259216,
+      "learning_rate": 0.00016977034265169854,
+      "loss": 4.415485229492187,
+      "step": 18250
+    },
+    {
+      "epoch": 0.4421758434262587,
+      "grad_norm": 0.8541338443756104,
+      "learning_rate": 0.00016940347787805412,
+      "loss": 4.440605163574219,
+      "step": 18300
+    },
+    {
+      "epoch": 0.4433839741460026,
+      "grad_norm": 0.8574206233024597,
+      "learning_rate": 0.0001690366131044097,
+      "loss": 4.44079833984375,
+      "step": 18350
+    },
+    {
+      "epoch": 0.44459210486574646,
+      "grad_norm": 0.8827728033065796,
+      "learning_rate": 0.00016866974833076528,
+      "loss": 4.451456909179687,
+      "step": 18400
+    },
+    {
+      "epoch": 0.44580023558549037,
+      "grad_norm": 0.8566647171974182,
+      "learning_rate": 0.0001683028835571208,
+      "loss": 4.4345159912109375,
+      "step": 18450
+    },
+    {
+      "epoch": 0.4470083663052342,
+      "grad_norm": 0.833660900592804,
+      "learning_rate": 0.0001679360187834764,
+      "loss": 4.464471740722656,
+      "step": 18500
+    },
+    {
+      "epoch": 0.4482164970249781,
+      "grad_norm": 0.8761898279190063,
+      "learning_rate": 0.00016756915400983197,
+      "loss": 4.437816772460938,
+      "step": 18550
+    },
+    {
+      "epoch": 0.44942462774472197,
+      "grad_norm": 0.9449495673179626,
+      "learning_rate": 0.00016720228923618755,
+      "loss": 4.44222900390625,
+      "step": 18600
+    },
+    {
+      "epoch": 0.4506327584644659,
+      "grad_norm": 0.8865442872047424,
+      "learning_rate": 0.00016683542446254307,
+      "loss": 4.4382781982421875,
+      "step": 18650
+    },
+    {
+      "epoch": 0.4518408891842097,
+      "grad_norm": 0.846558690071106,
+      "learning_rate": 0.00016646855968889865,
+      "loss": 4.434262390136719,
+      "step": 18700
+    },
+    {
+      "epoch": 0.45304901990395363,
+      "grad_norm": 0.818831741809845,
+      "learning_rate": 0.00016610169491525423,
+      "loss": 4.389546813964844,
+      "step": 18750
+    },
+    {
+      "epoch": 0.4542571506236975,
+      "grad_norm": 0.9020473957061768,
+      "learning_rate": 0.0001657348301416098,
+      "loss": 4.451723022460937,
+      "step": 18800
+    },
+    {
+      "epoch": 0.4554652813434414,
+      "grad_norm": 0.8355647921562195,
+      "learning_rate": 0.00016536796536796534,
+      "loss": 4.402284545898437,
+      "step": 18850
+    },
+    {
+      "epoch": 0.45667341206318524,
+      "grad_norm": 0.8465541005134583,
+      "learning_rate": 0.00016500110059432092,
+      "loss": 4.409869689941406,
+      "step": 18900
+    },
+    {
+      "epoch": 0.45788154278292914,
+      "grad_norm": 0.8707362413406372,
+      "learning_rate": 0.00016463423582067647,
+      "loss": 4.373127136230469,
+      "step": 18950
+    },
+    {
+      "epoch": 0.459089673502673,
+      "grad_norm": 0.8290424942970276,
+      "learning_rate": 0.00016426737104703205,
+      "loss": 4.406386108398437,
+      "step": 19000
+    },
+    {
+      "epoch": 0.46029780422241684,
+      "grad_norm": 0.8582199215888977,
+      "learning_rate": 0.0001639005062733876,
+      "loss": 4.4494845581054685,
+      "step": 19050
+    },
+    {
+      "epoch": 0.46150593494216074,
+      "grad_norm": 0.7977620363235474,
+      "learning_rate": 0.00016353364149974319,
+      "loss": 4.400769653320313,
+      "step": 19100
+    },
+    {
+      "epoch": 0.4627140656619046,
+      "grad_norm": 0.8718297481536865,
+      "learning_rate": 0.00016316677672609874,
+      "loss": 4.4187710571289065,
+      "step": 19150
+    },
+    {
+      "epoch": 0.4639221963816485,
+      "grad_norm": 0.8437783718109131,
+      "learning_rate": 0.00016279991195245432,
+      "loss": 4.414367065429688,
+      "step": 19200
+    },
+    {
+      "epoch": 0.46513032710139235,
+      "grad_norm": 0.8435456156730652,
+      "learning_rate": 0.00016243304717880987,
+      "loss": 4.395423583984375,
+      "step": 19250
+    },
+    {
+      "epoch": 0.46633845782113625,
+      "grad_norm": 0.8388056755065918,
+      "learning_rate": 0.00016206618240516542,
+      "loss": 4.4262393188476565,
+      "step": 19300
+    },
+    {
+      "epoch": 0.4675465885408801,
+      "grad_norm": 0.9124335646629333,
+      "learning_rate": 0.000161699317631521,
+      "loss": 4.411950988769531,
+      "step": 19350
+    },
+    {
+      "epoch": 0.468754719260624,
+      "grad_norm": 0.9084233045578003,
+      "learning_rate": 0.00016133245285787658,
+      "loss": 4.402586059570313,
+      "step": 19400
+    },
+    {
+      "epoch": 0.46996284998036786,
+      "grad_norm": 0.8846220970153809,
+      "learning_rate": 0.0001609655880842321,
+      "loss": 4.387697143554687,
+      "step": 19450
+    },
+    {
+      "epoch": 0.47117098070011176,
+      "grad_norm": 0.8609309792518616,
+      "learning_rate": 0.0001605987233105877,
+      "loss": 4.379610290527344,
+      "step": 19500
+    },
+    {
+      "epoch": 0.4723791114198556,
+      "grad_norm": 0.9807800650596619,
+      "learning_rate": 0.00016023185853694327,
+      "loss": 4.344981079101562,
+      "step": 19550
+    },
+    {
+      "epoch": 0.4735872421395995,
+      "grad_norm": 0.8327066898345947,
+      "learning_rate": 0.00015986499376329885,
+      "loss": 4.413209228515625,
+      "step": 19600
+    },
+    {
+      "epoch": 0.47479537285934337,
+      "grad_norm": 0.8389619588851929,
+      "learning_rate": 0.00015949812898965438,
+      "loss": 4.385924682617188,
+      "step": 19650
+    },
+    {
+      "epoch": 0.4760035035790873,
+      "grad_norm": 0.8684993386268616,
+      "learning_rate": 0.00015913126421600996,
+      "loss": 4.3818408203125,
+      "step": 19700
+    },
+    {
+      "epoch": 0.4772116342988311,
+      "grad_norm": 0.8499497771263123,
+      "learning_rate": 0.00015876439944236554,
+      "loss": 4.399507446289062,
+      "step": 19750
+    },
+    {
+      "epoch": 0.478419765018575,
+      "grad_norm": 0.8550466895103455,
+      "learning_rate": 0.00015839753466872112,
+      "loss": 4.411488342285156,
+      "step": 19800
+    },
+    {
+      "epoch": 0.4796278957383189,
+      "grad_norm": 0.8694366812705994,
+      "learning_rate": 0.00015803066989507664,
+      "loss": 4.403375854492188,
+      "step": 19850
+    },
+    {
+      "epoch": 0.4808360264580628,
+      "grad_norm": 0.8916079998016357,
+      "learning_rate": 0.00015766380512143222,
+      "loss": 4.3569097900390625,
+      "step": 19900
+    },
+    {
+      "epoch": 0.48204415717780663,
+      "grad_norm": 0.8819521069526672,
+      "learning_rate": 0.0001572969403477878,
+      "loss": 4.365809020996093,
+      "step": 19950
+    },
+    {
+      "epoch": 0.48325228789755054,
+      "grad_norm": 0.876788854598999,
+      "learning_rate": 0.00015693007557414336,
+      "loss": 4.359242553710938,
+      "step": 20000
+    },
+    {
+      "epoch": 0.4844604186172944,
+      "grad_norm": 0.8509864211082458,
+      "learning_rate": 0.0001565632108004989,
+      "loss": 4.407760314941406,
+      "step": 20050
+    },
+    {
+      "epoch": 0.4856685493370383,
+      "grad_norm": 0.863982081413269,
+      "learning_rate": 0.0001561963460268545,
+      "loss": 4.41439453125,
+      "step": 20100
+    },
+    {
+      "epoch": 0.48687668005678214,
+      "grad_norm": 0.8627960085868835,
+      "learning_rate": 0.00015582948125321004,
+      "loss": 4.319361572265625,
+      "step": 20150
+    },
+    {
+      "epoch": 0.48808481077652605,
+      "grad_norm": 0.8824877142906189,
+      "learning_rate": 0.00015546261647956562,
+      "loss": 4.390765686035156,
+      "step": 20200
+    },
+    {
+      "epoch": 0.4892929414962699,
+      "grad_norm": 0.9343333840370178,
+      "learning_rate": 0.00015509575170592117,
+      "loss": 4.369144287109375,
+      "step": 20250
+    },
+    {
+      "epoch": 0.4905010722160138,
+      "grad_norm": 0.8687631487846375,
+      "learning_rate": 0.00015472888693227675,
+      "loss": 4.374679260253906,
+      "step": 20300
+    },
+    {
+      "epoch": 0.49170920293575765,
+      "grad_norm": 0.9157174229621887,
+      "learning_rate": 0.0001543620221586323,
+      "loss": 4.381021423339844,
+      "step": 20350
+    },
+    {
+      "epoch": 0.4929173336555015,
+      "grad_norm": 0.8637880086898804,
+      "learning_rate": 0.0001539951573849879,
+      "loss": 4.369039306640625,
+      "step": 20400
+    },
+    {
+      "epoch": 0.4941254643752454,
+      "grad_norm": 0.8694828748703003,
+      "learning_rate": 0.00015362829261134344,
+      "loss": 4.380176086425781,
+      "step": 20450
+    },
+    {
+      "epoch": 0.49533359509498925,
+      "grad_norm": 0.8386253118515015,
+      "learning_rate": 0.000153261427837699,
+      "loss": 4.365638427734375,
+      "step": 20500
+    },
+    {
+      "epoch": 0.49654172581473316,
+      "grad_norm": 0.8944020867347717,
+      "learning_rate": 0.00015289456306405457,
+      "loss": 4.341180419921875,
+      "step": 20550
+    },
+    {
+      "epoch": 0.497749856534477,
+      "grad_norm": 0.8739315271377563,
+      "learning_rate": 0.00015252769829041015,
+      "loss": 4.378220520019531,
+      "step": 20600
+    },
+    {
+      "epoch": 0.4989579872542209,
+      "grad_norm": 0.8651403784751892,
+      "learning_rate": 0.00015216083351676573,
+      "loss": 4.367572937011719,
+      "step": 20650
+    },
+    {
+      "epoch": 0.5001661179739648,
+      "grad_norm": 0.9506304860115051,
+      "learning_rate": 0.00015179396874312126,
+      "loss": 4.372146911621094,
+      "step": 20700
+    },
+    {
+      "epoch": 0.5013742486937086,
+      "grad_norm": 0.8832495212554932,
+      "learning_rate": 0.00015142710396947684,
+      "loss": 4.365708312988281,
+      "step": 20750
+    },
+    {
+      "epoch": 0.5025823794134525,
+      "grad_norm": 0.8573395609855652,
+      "learning_rate": 0.00015106023919583242,
+      "loss": 4.3646456909179685,
+      "step": 20800
+    },
+    {
+      "epoch": 0.5037905101331964,
+      "grad_norm": 0.8642771244049072,
+      "learning_rate": 0.00015069337442218797,
+      "loss": 4.332890625,
+      "step": 20850
+    },
+    {
+      "epoch": 0.5049986408529403,
+      "grad_norm": 0.8693931698799133,
+      "learning_rate": 0.00015032650964854352,
+      "loss": 4.364405517578125,
+      "step": 20900
+    },
+    {
+      "epoch": 0.5062067715726841,
+      "grad_norm": 0.9043340086936951,
+      "learning_rate": 0.0001499596448748991,
+      "loss": 4.378132629394531,
+      "step": 20950
+    },
+    {
+      "epoch": 0.507414902292428,
+      "grad_norm": 0.8317884206771851,
+      "learning_rate": 0.00014959278010125469,
+      "loss": 4.356962585449219,
+      "step": 21000
+    },
+    {
+      "epoch": 0.5086230330121719,
+      "grad_norm": 0.8647122383117676,
+      "learning_rate": 0.00014922591532761024,
+      "loss": 4.366798706054688,
+      "step": 21050
+    },
+    {
+      "epoch": 0.5098311637319158,
+      "grad_norm": 0.8674430251121521,
+      "learning_rate": 0.0001488590505539658,
+      "loss": 4.357047424316407,
+      "step": 21100
+    },
+    {
+      "epoch": 0.5110392944516596,
+      "grad_norm": 0.8647886514663696,
+      "learning_rate": 0.00014849218578032137,
+      "loss": 4.377456970214844,
+      "step": 21150
+    },
+    {
+      "epoch": 0.5122474251714035,
+      "grad_norm": 0.8840771913528442,
+      "learning_rate": 0.00014812532100667692,
+      "loss": 4.363984069824219,
+      "step": 21200
+    },
+    {
+      "epoch": 0.5134555558911474,
+      "grad_norm": 0.8642849326133728,
+      "learning_rate": 0.0001477584562330325,
+      "loss": 4.369249877929687,
+      "step": 21250
+    },
+    {
+      "epoch": 0.5146636866108913,
+      "grad_norm": 0.8466240763664246,
+      "learning_rate": 0.00014739159145938806,
+      "loss": 4.38531005859375,
+      "step": 21300
+    },
+    {
+      "epoch": 0.5158718173306351,
+      "grad_norm": 0.8386222124099731,
+      "learning_rate": 0.0001470247266857436,
+      "loss": 4.360840759277344,
+      "step": 21350
+    },
+    {
+      "epoch": 0.517079948050379,
+      "grad_norm": 0.8661811351776123,
+      "learning_rate": 0.0001466578619120992,
+      "loss": 4.354701232910156,
+      "step": 21400
+    },
+    {
+      "epoch": 0.518288078770123,
+      "grad_norm": 0.8432431817054749,
+      "learning_rate": 0.00014629099713845474,
+      "loss": 4.383183898925782,
+      "step": 21450
+    },
+    {
+      "epoch": 0.5194962094898669,
+      "grad_norm": 0.9352067112922668,
+      "learning_rate": 0.00014592413236481032,
+      "loss": 4.379555053710938,
+      "step": 21500
+    },
+    {
+      "epoch": 0.5207043402096106,
+      "grad_norm": 0.8690420985221863,
+      "learning_rate": 0.00014555726759116588,
+      "loss": 4.354761047363281,
+      "step": 21550
+    },
+    {
+      "epoch": 0.5219124709293546,
+      "grad_norm": 0.8794584274291992,
+      "learning_rate": 0.00014519040281752146,
+      "loss": 4.359398498535156,
+      "step": 21600
+    },
+    {
+      "epoch": 0.5231206016490985,
+      "grad_norm": 0.8569353818893433,
+      "learning_rate": 0.000144823538043877,
+      "loss": 4.306293029785156,
+      "step": 21650
+    },
+    {
+      "epoch": 0.5243287323688423,
+      "grad_norm": 0.8656736016273499,
+      "learning_rate": 0.0001444566732702326,
+      "loss": 4.3668692016601565,
+      "step": 21700
+    },
+    {
+      "epoch": 0.5255368630885862,
+      "grad_norm": 0.9000532031059265,
+      "learning_rate": 0.00014408980849658814,
+      "loss": 4.354334106445313,
+      "step": 21750
+    },
+    {
+      "epoch": 0.5267449938083301,
+      "grad_norm": 0.8964462280273438,
+      "learning_rate": 0.00014372294372294372,
+      "loss": 4.371478576660156,
+      "step": 21800
+    },
+    {
+      "epoch": 0.527953124528074,
+      "grad_norm": 0.8937968611717224,
+      "learning_rate": 0.00014335607894929927,
+      "loss": 4.370023193359375,
+      "step": 21850
+    },
+    {
+      "epoch": 0.5291612552478178,
+      "grad_norm": 0.8516099452972412,
+      "learning_rate": 0.00014298921417565485,
+      "loss": 4.306471862792969,
+      "step": 21900
+    },
+    {
+      "epoch": 0.5303693859675617,
+      "grad_norm": 0.9718441963195801,
+      "learning_rate": 0.0001426223494020104,
+      "loss": 4.345325927734375,
+      "step": 21950
+    },
+    {
+      "epoch": 0.5315775166873056,
+      "grad_norm": 0.9068713784217834,
+      "learning_rate": 0.000142255484628366,
+      "loss": 4.349406127929687,
+      "step": 22000
+    },
+    {
+      "epoch": 0.5327856474070495,
+      "grad_norm": 0.8823951482772827,
+      "learning_rate": 0.00014188861985472154,
+      "loss": 4.273417663574219,
+      "step": 22050
+    },
+    {
+      "epoch": 0.5339937781267933,
+      "grad_norm": 0.8769415020942688,
+      "learning_rate": 0.00014152175508107712,
+      "loss": 4.353021850585938,
+      "step": 22100
+    },
+    {
+      "epoch": 0.5352019088465372,
+      "grad_norm": 0.8678697347640991,
+      "learning_rate": 0.00014115489030743267,
+      "loss": 4.307740173339844,
+      "step": 22150
+    },
+    {
+      "epoch": 0.5364100395662811,
+      "grad_norm": 0.8746253848075867,
+      "learning_rate": 0.00014078802553378825,
+      "loss": 4.313048400878906,
+      "step": 22200
+    },
+    {
+      "epoch": 0.537618170286025,
+      "grad_norm": 0.9007458686828613,
+      "learning_rate": 0.0001404211607601438,
+      "loss": 4.3306298828125,
+      "step": 22250
+    },
+    {
+      "epoch": 0.5388263010057688,
+      "grad_norm": 0.8988893628120422,
+      "learning_rate": 0.00014005429598649936,
+      "loss": 4.333839111328125,
+      "step": 22300
+    },
+    {
+      "epoch": 0.5400344317255127,
+      "grad_norm": 0.9225552082061768,
+      "learning_rate": 0.00013968743121285494,
+      "loss": 4.3047552490234375,
+      "step": 22350
+    },
+    {
+      "epoch": 0.5412425624452566,
+      "grad_norm": 0.9349094033241272,
+      "learning_rate": 0.0001393205664392105,
+      "loss": 4.318995361328125,
+      "step": 22400
+    },
+    {
+      "epoch": 0.5424506931650005,
+      "grad_norm": 0.860418438911438,
+      "learning_rate": 0.00013895370166556605,
+      "loss": 4.326046142578125,
+      "step": 22450
+    },
+    {
+      "epoch": 0.5436588238847443,
+      "grad_norm": 0.9154239892959595,
+      "learning_rate": 0.00013858683689192163,
+      "loss": 4.322186889648438,
+      "step": 22500
+    },
+    {
+      "epoch": 0.5448669546044882,
+      "grad_norm": 0.8727108836174011,
+      "learning_rate": 0.00013821997211827718,
+      "loss": 4.342572326660156,
+      "step": 22550
+    },
+    {
+      "epoch": 0.5460750853242321,
+      "grad_norm": 0.8912889361381531,
+      "learning_rate": 0.00013785310734463276,
+      "loss": 4.3360684204101565,
+      "step": 22600
+    },
+    {
+      "epoch": 0.547283216043976,
+      "grad_norm": 0.9072328805923462,
+      "learning_rate": 0.0001374862425709883,
+      "loss": 4.319748840332031,
+      "step": 22650
+    },
+    {
+      "epoch": 0.5484913467637198,
+      "grad_norm": 0.8966073393821716,
+      "learning_rate": 0.0001371193777973439,
+      "loss": 4.348955688476562,
+      "step": 22700
+    },
+    {
+      "epoch": 0.5496994774834637,
+      "grad_norm": 0.9064639210700989,
+      "learning_rate": 0.00013675251302369944,
+      "loss": 4.293034362792969,
+      "step": 22750
+    },
+    {
+      "epoch": 0.5509076082032076,
+      "grad_norm": 0.9258774518966675,
+      "learning_rate": 0.00013638564825005502,
+      "loss": 4.3505160522460935,
+      "step": 22800
+    },
+    {
+      "epoch": 0.5521157389229515,
+      "grad_norm": 0.8719062805175781,
+      "learning_rate": 0.00013601878347641058,
+      "loss": 4.332078857421875,
+      "step": 22850
+    },
+    {
+      "epoch": 0.5533238696426953,
+      "grad_norm": 0.9730004072189331,
+      "learning_rate": 0.00013565191870276616,
+      "loss": 4.349375305175781,
+      "step": 22900
+    },
+    {
+      "epoch": 0.5545320003624392,
+      "grad_norm": 0.8471688628196716,
+      "learning_rate": 0.0001352850539291217,
+      "loss": 4.327649536132813,
+      "step": 22950
+    },
+    {
+      "epoch": 0.5557401310821831,
+      "grad_norm": 0.8453298807144165,
+      "learning_rate": 0.0001349181891554773,
+      "loss": 4.337743225097657,
+      "step": 23000
+    },
+    {
+      "epoch": 0.5569482618019269,
+      "grad_norm": 0.9087938070297241,
+      "learning_rate": 0.00013455132438183284,
+      "loss": 4.318193054199218,
+      "step": 23050
+    },
+    {
+      "epoch": 0.5581563925216708,
+      "grad_norm": 0.8770619630813599,
+      "learning_rate": 0.00013418445960818842,
+      "loss": 4.321719970703125,
+      "step": 23100
+    },
+    {
+      "epoch": 0.5593645232414147,
+      "grad_norm": 0.9098190665245056,
+      "learning_rate": 0.00013381759483454398,
+      "loss": 4.297068786621094,
+      "step": 23150
+    },
+    {
+      "epoch": 0.5605726539611586,
+      "grad_norm": 0.8891416192054749,
+      "learning_rate": 0.00013345073006089956,
+      "loss": 4.331806335449219,
+      "step": 23200
+    },
+    {
+      "epoch": 0.5617807846809024,
+      "grad_norm": 0.8606537580490112,
+      "learning_rate": 0.0001330838652872551,
+      "loss": 4.327531433105468,
+      "step": 23250
+    },
+    {
+      "epoch": 0.5629889154006463,
+      "grad_norm": 0.8917016386985779,
+      "learning_rate": 0.0001327170005136107,
+      "loss": 4.336094665527344,
+      "step": 23300
+    },
+    {
+      "epoch": 0.5641970461203902,
+      "grad_norm": 0.8686886429786682,
+      "learning_rate": 0.00013235013573996624,
+      "loss": 4.316971130371094,
+      "step": 23350
+    },
+    {
+      "epoch": 0.5654051768401341,
+      "grad_norm": 0.9273053407669067,
+      "learning_rate": 0.0001319832709663218,
+      "loss": 4.301628723144531,
+      "step": 23400
+    },
+    {
+      "epoch": 0.5666133075598779,
+      "grad_norm": 0.897447943687439,
+      "learning_rate": 0.00013161640619267738,
+      "loss": 4.3230984497070315,
+      "step": 23450
+    },
+    {
+      "epoch": 0.5678214382796218,
+      "grad_norm": 0.8964062333106995,
+      "learning_rate": 0.00013124954141903293,
+      "loss": 4.295698852539062,
+      "step": 23500
+    },
+    {
+      "epoch": 0.5690295689993657,
+      "grad_norm": 0.8870056867599487,
+      "learning_rate": 0.0001308826766453885,
+      "loss": 4.306682739257813,
+      "step": 23550
+    },
+    {
+      "epoch": 0.5702376997191096,
+      "grad_norm": 0.8741028904914856,
+      "learning_rate": 0.00013051581187174406,
+      "loss": 4.333457641601562,
+      "step": 23600
+    },
+    {
+      "epoch": 0.5714458304388534,
+      "grad_norm": 0.8987818360328674,
+      "learning_rate": 0.00013014894709809961,
+      "loss": 4.318333435058594,
+      "step": 23650
+    },
+    {
+      "epoch": 0.5726539611585973,
+      "grad_norm": 0.9215242266654968,
+      "learning_rate": 0.0001297820823244552,
+      "loss": 4.304813537597656,
+      "step": 23700
+    },
+    {
+      "epoch": 0.5738620918783413,
+      "grad_norm": 0.8973007202148438,
+      "learning_rate": 0.00012941521755081075,
+      "loss": 4.286683044433594,
+      "step": 23750
+    },
+    {
+      "epoch": 0.5750702225980852,
+      "grad_norm": 0.9218378663063049,
+      "learning_rate": 0.00012904835277716633,
+      "loss": 4.300123596191407,
+      "step": 23800
+    },
+    {
+      "epoch": 0.576278353317829,
+      "grad_norm": 0.8990265130996704,
+      "learning_rate": 0.00012868148800352188,
+      "loss": 4.277065124511719,
+      "step": 23850
+    },
+    {
+      "epoch": 0.5774864840375729,
+      "grad_norm": 0.8685271739959717,
+      "learning_rate": 0.00012831462322987746,
+      "loss": 4.3058645629882815,
+      "step": 23900
+    },
+    {
+      "epoch": 0.5786946147573168,
+      "grad_norm": 0.8670499920845032,
+      "learning_rate": 0.000127947758456233,
+      "loss": 4.253737487792969,
+      "step": 23950
+    },
+    {
+      "epoch": 0.5799027454770607,
+      "grad_norm": 0.8900935053825378,
+      "learning_rate": 0.0001275808936825886,
+      "loss": 4.316374816894531,
+      "step": 24000
+    },
+    {
+      "epoch": 0.5811108761968045,
+      "grad_norm": 0.8673965334892273,
+      "learning_rate": 0.00012721402890894415,
+      "loss": 4.285259399414063,
+      "step": 24050
+    },
+    {
+      "epoch": 0.5823190069165484,
+      "grad_norm": 0.9063747525215149,
+      "learning_rate": 0.00012684716413529973,
+      "loss": 4.283208312988282,
+      "step": 24100
+    },
+    {
+      "epoch": 0.5835271376362923,
+      "grad_norm": 0.9064856171607971,
+      "learning_rate": 0.00012648029936165528,
+      "loss": 4.248049011230469,
+      "step": 24150
+    },
+    {
+      "epoch": 0.5847352683560362,
+      "grad_norm": 0.9232784509658813,
+      "learning_rate": 0.00012611343458801086,
+      "loss": 4.295502624511719,
+      "step": 24200
+    },
+    {
+      "epoch": 0.58594339907578,
+      "grad_norm": 0.8695572018623352,
+      "learning_rate": 0.0001257465698143664,
+      "loss": 4.2591073608398435,
+      "step": 24250
+    },
+    {
+      "epoch": 0.5871515297955239,
+      "grad_norm": 0.9018963575363159,
+      "learning_rate": 0.000125379705040722,
+      "loss": 4.319864807128906,
+      "step": 24300
+    },
+    {
+      "epoch": 0.5883596605152678,
+      "grad_norm": 0.9171856641769409,
+      "learning_rate": 0.00012501284026707754,
+      "loss": 4.265516967773437,
+      "step": 24350
+    },
+    {
+      "epoch": 0.5895677912350116,
+      "grad_norm": 0.9028427004814148,
+      "learning_rate": 0.00012464597549343312,
+      "loss": 4.275965270996093,
+      "step": 24400
+    },
+    {
+      "epoch": 0.5907759219547555,
+      "grad_norm": 0.9330050349235535,
+      "learning_rate": 0.00012427911071978868,
+      "loss": 4.2545404052734375,
+      "step": 24450
+    },
+    {
+      "epoch": 0.5919840526744994,
+      "grad_norm": 0.9791463017463684,
+      "learning_rate": 0.00012391224594614426,
+      "loss": 4.2765435791015625,
+      "step": 24500
+    },
+    {
+      "epoch": 0.5931921833942433,
+      "grad_norm": 0.8853760957717896,
+      "learning_rate": 0.0001235453811724998,
+      "loss": 4.308526916503906,
+      "step": 24550
+    },
+    {
+      "epoch": 0.5944003141139871,
+      "grad_norm": 0.9191492795944214,
+      "learning_rate": 0.00012317851639885536,
+      "loss": 4.285240173339844,
+      "step": 24600
+    },
+    {
+      "epoch": 0.595608444833731,
+      "grad_norm": 0.8623924851417542,
+      "learning_rate": 0.00012281165162521094,
+      "loss": 4.318269958496094,
+      "step": 24650
+    },
+    {
+      "epoch": 0.5968165755534749,
+      "grad_norm": 0.8917416930198669,
+      "learning_rate": 0.0001224447868515665,
+      "loss": 4.269252624511719,
+      "step": 24700
+    },
+    {
+      "epoch": 0.5980247062732188,
+      "grad_norm": 0.8966076374053955,
+      "learning_rate": 0.00012207792207792208,
+      "loss": 4.2773611450195315,
+      "step": 24750
+    },
+    {
+      "epoch": 0.5992328369929626,
+      "grad_norm": 0.8976031541824341,
+      "learning_rate": 0.00012171105730427764,
+      "loss": 4.257319946289062,
+      "step": 24800
+    },
+    {
+      "epoch": 0.6004409677127065,
+      "grad_norm": 0.8931077718734741,
+      "learning_rate": 0.0001213441925306332,
+      "loss": 4.29817626953125,
+      "step": 24850
+    },
+    {
+      "epoch": 0.6016490984324504,
+      "grad_norm": 0.8995993137359619,
+      "learning_rate": 0.00012097732775698876,
+      "loss": 4.291929931640625,
+      "step": 24900
+    },
+    {
+      "epoch": 0.6028572291521943,
+      "grad_norm": 0.9027631282806396,
+      "learning_rate": 0.00012061046298334433,
+      "loss": 4.288674926757812,
+      "step": 24950
+    },
+    {
+      "epoch": 0.6040653598719381,
+      "grad_norm": 0.8671455383300781,
+      "learning_rate": 0.0001202435982096999,
+      "loss": 4.288687133789063,
+      "step": 25000
+    },
+    {
+      "epoch": 0.605273490591682,
+      "grad_norm": 0.9251123070716858,
+      "learning_rate": 0.00011987673343605546,
+      "loss": 4.312415466308594,
+      "step": 25050
+    },
+    {
+      "epoch": 0.6064816213114259,
+      "grad_norm": 0.8921682238578796,
+      "learning_rate": 0.00011950986866241103,
+      "loss": 4.271250915527344,
+      "step": 25100
+    },
+    {
+      "epoch": 0.6076897520311698,
+      "grad_norm": 0.9173095226287842,
+      "learning_rate": 0.00011914300388876658,
+      "loss": 4.278268432617187,
+      "step": 25150
+    },
+    {
+      "epoch": 0.6088978827509136,
+      "grad_norm": 0.9297692775726318,
+      "learning_rate": 0.00011877613911512216,
+      "loss": 4.289117431640625,
+      "step": 25200
+    },
+    {
+      "epoch": 0.6101060134706575,
+      "grad_norm": 0.8938084840774536,
+      "learning_rate": 0.00011840927434147771,
+      "loss": 4.273968811035156,
+      "step": 25250
+    },
+    {
+      "epoch": 0.6113141441904014,
+      "grad_norm": 0.9352951049804688,
+      "learning_rate": 0.0001180424095678333,
+      "loss": 4.271760864257812,
+      "step": 25300
+    },
+    {
+      "epoch": 0.6125222749101453,
+      "grad_norm": 0.8771641850471497,
+      "learning_rate": 0.00011767554479418885,
+      "loss": 4.278330383300781,
+      "step": 25350
+    },
+    {
+      "epoch": 0.6137304056298891,
+      "grad_norm": 0.9419417977333069,
+      "learning_rate": 0.00011730868002054443,
+      "loss": 4.285042724609375,
+      "step": 25400
+    },
+    {
+      "epoch": 0.614938536349633,
+      "grad_norm": 0.874875545501709,
+      "learning_rate": 0.00011694181524689998,
+      "loss": 4.263815002441406,
+      "step": 25450
+    },
+    {
+      "epoch": 0.6161466670693769,
+      "grad_norm": 0.9196811318397522,
+      "learning_rate": 0.00011657495047325555,
+      "loss": 4.271156005859375,
+      "step": 25500
+    },
+    {
+      "epoch": 0.6173547977891208,
+      "grad_norm": 0.9265231490135193,
+      "learning_rate": 0.00011620808569961111,
+      "loss": 4.263125915527343,
+      "step": 25550
+    },
+    {
+      "epoch": 0.6185629285088646,
+      "grad_norm": 0.9503073692321777,
+      "learning_rate": 0.00011584122092596668,
+      "loss": 4.272619018554687,
+      "step": 25600
+    },
+    {
+      "epoch": 0.6197710592286085,
+      "grad_norm": 0.917907178401947,
+      "learning_rate": 0.00011547435615232223,
+      "loss": 4.2632080078125,
+      "step": 25650
+    },
+    {
+      "epoch": 0.6209791899483524,
+      "grad_norm": 1.0107134580612183,
+      "learning_rate": 0.00011510749137867781,
+      "loss": 4.285297241210937,
+      "step": 25700
+    },
+    {
+      "epoch": 0.6221873206680962,
+      "grad_norm": 0.8880563378334045,
+      "learning_rate": 0.00011474062660503337,
+      "loss": 4.268618469238281,
+      "step": 25750
+    },
+    {
+      "epoch": 0.6233954513878401,
+      "grad_norm": 0.8972634673118591,
+      "learning_rate": 0.00011437376183138895,
+      "loss": 4.260936889648438,
+      "step": 25800
+    },
+    {
+      "epoch": 0.624603582107584,
+      "grad_norm": 0.9134317636489868,
+      "learning_rate": 0.0001140068970577445,
+      "loss": 4.260354919433594,
+      "step": 25850
+    },
+    {
+      "epoch": 0.625811712827328,
+      "grad_norm": 0.8854179978370667,
+      "learning_rate": 0.00011364003228410008,
+      "loss": 4.266827392578125,
+      "step": 25900
+    },
+    {
+      "epoch": 0.6270198435470717,
+      "grad_norm": 0.9477384686470032,
+      "learning_rate": 0.00011327316751045563,
+      "loss": 4.264967651367187,
+      "step": 25950
+    },
+    {
+      "epoch": 0.6282279742668156,
+      "grad_norm": 0.9076700210571289,
+      "learning_rate": 0.0001129063027368112,
+      "loss": 4.2312619018554685,
+      "step": 26000
+    },
+    {
+      "epoch": 0.6294361049865596,
+      "grad_norm": 0.9770967364311218,
+      "learning_rate": 0.00011253943796316676,
+      "loss": 4.221799621582031,
+      "step": 26050
+    },
+    {
+      "epoch": 0.6306442357063035,
+      "grad_norm": 0.9270659685134888,
+      "learning_rate": 0.00011217257318952233,
+      "loss": 4.246094665527344,
+      "step": 26100
+    },
+    {
+      "epoch": 0.6318523664260473,
+      "grad_norm": 0.9425565004348755,
+      "learning_rate": 0.0001118057084158779,
+      "loss": 4.2824365234375,
+      "step": 26150
+    },
+    {
+      "epoch": 0.6330604971457912,
+      "grad_norm": 0.901681661605835,
+      "learning_rate": 0.00011143884364223346,
+      "loss": 4.247665710449219,
+      "step": 26200
+    },
+    {
+      "epoch": 0.6342686278655351,
+      "grad_norm": 0.9279155135154724,
+      "learning_rate": 0.00011107197886858902,
+      "loss": 4.291918640136719,
+      "step": 26250
+    },
+    {
+      "epoch": 0.635476758585279,
+      "grad_norm": 0.9004014730453491,
+      "learning_rate": 0.0001107051140949446,
+      "loss": 4.260052490234375,
+      "step": 26300
+    },
+    {
+      "epoch": 0.6366848893050228,
+      "grad_norm": 0.9324517846107483,
+      "learning_rate": 0.00011033824932130015,
+      "loss": 4.238099975585937,
+      "step": 26350
+    },
+    {
+      "epoch": 0.6378930200247667,
+      "grad_norm": 0.9364690780639648,
+      "learning_rate": 0.00010997138454765573,
+      "loss": 4.297799682617187,
+      "step": 26400
+    },
+    {
+      "epoch": 0.6391011507445106,
+      "grad_norm": 0.9201944470405579,
+      "learning_rate": 0.00010960451977401128,
+      "loss": 4.2686245727539065,
+      "step": 26450
+    },
+    {
+      "epoch": 0.6403092814642545,
+      "grad_norm": 0.9261684417724609,
+      "learning_rate": 0.00010923765500036686,
+      "loss": 4.273673095703125,
+      "step": 26500
+    },
+    {
+      "epoch": 0.6415174121839983,
+      "grad_norm": 0.9427759051322937,
+      "learning_rate": 0.00010887079022672242,
+      "loss": 4.257789916992188,
+      "step": 26550
+    },
+    {
+      "epoch": 0.6427255429037422,
+      "grad_norm": 0.9197338819503784,
+      "learning_rate": 0.00010850392545307798,
+      "loss": 4.258461608886718,
+      "step": 26600
+    },
+    {
+      "epoch": 0.6439336736234861,
+      "grad_norm": 0.9081806540489197,
+      "learning_rate": 0.00010813706067943355,
+      "loss": 4.234719848632812,
+      "step": 26650
+    },
+    {
+      "epoch": 0.64514180434323,
+      "grad_norm": 0.9609241485595703,
+      "learning_rate": 0.00010777019590578912,
+      "loss": 4.216001281738281,
+      "step": 26700
+    },
+    {
+      "epoch": 0.6463499350629738,
+      "grad_norm": 0.9191202521324158,
+      "learning_rate": 0.00010740333113214468,
+      "loss": 4.223638000488282,
+      "step": 26750
+    },
+    {
+      "epoch": 0.6475580657827177,
+      "grad_norm": 0.9228413701057434,
+      "learning_rate": 0.00010703646635850025,
+      "loss": 4.244529418945312,
+      "step": 26800
+    },
+    {
+      "epoch": 0.6487661965024616,
+      "grad_norm": 0.9351758360862732,
+      "learning_rate": 0.0001066696015848558,
+      "loss": 4.269735717773438,
+      "step": 26850
+    },
+    {
+      "epoch": 0.6499743272222055,
+      "grad_norm": 0.9101845026016235,
+      "learning_rate": 0.00010630273681121138,
+      "loss": 4.2333837890625,
+      "step": 26900
+    },
+    {
+      "epoch": 0.6511824579419493,
+      "grad_norm": 0.9198936223983765,
+      "learning_rate": 0.00010593587203756693,
+      "loss": 4.247688293457031,
+      "step": 26950
+    },
+    {
+      "epoch": 0.6523905886616932,
+      "grad_norm": 0.9287570714950562,
+      "learning_rate": 0.00010556900726392251,
+      "loss": 4.210073547363281,
+      "step": 27000
+    },
+    {
+      "epoch": 0.6535987193814371,
+      "grad_norm": 0.9351828694343567,
+      "learning_rate": 0.00010520214249027807,
+      "loss": 4.213340454101562,
+      "step": 27050
+    },
+    {
+      "epoch": 0.6548068501011809,
+      "grad_norm": 0.8970820307731628,
+      "learning_rate": 0.00010483527771663365,
+      "loss": 4.231409912109375,
+      "step": 27100
+    },
+    {
+      "epoch": 0.6560149808209248,
+      "grad_norm": 0.9209808707237244,
+      "learning_rate": 0.0001044684129429892,
+      "loss": 4.220360412597656,
+      "step": 27150
+    },
+    {
+      "epoch": 0.6572231115406687,
+      "grad_norm": 0.9852465987205505,
+      "learning_rate": 0.00010410154816934477,
+      "loss": 4.246997680664062,
+      "step": 27200
+    },
+    {
+      "epoch": 0.6584312422604126,
+      "grad_norm": 0.8992114663124084,
+      "learning_rate": 0.00010373468339570033,
+      "loss": 4.247727966308593,
+      "step": 27250
+    },
+    {
+      "epoch": 0.6596393729801564,
+      "grad_norm": 0.9355093836784363,
+      "learning_rate": 0.0001033678186220559,
+      "loss": 4.242572631835937,
+      "step": 27300
+    },
+    {
+      "epoch": 0.6608475036999003,
+      "grad_norm": 0.9295455813407898,
+      "learning_rate": 0.00010300095384841147,
+      "loss": 4.209480895996093,
+      "step": 27350
+    },
+    {
+      "epoch": 0.6620556344196442,
+      "grad_norm": 0.9523081183433533,
+      "learning_rate": 0.00010263408907476703,
+      "loss": 4.256694030761719,
+      "step": 27400
+    },
+    {
+      "epoch": 0.6632637651393881,
+      "grad_norm": 0.9379644393920898,
+      "learning_rate": 0.00010226722430112259,
+      "loss": 4.248100280761719,
+      "step": 27450
+    },
+    {
+      "epoch": 0.6644718958591319,
+      "grad_norm": 1.01943039894104,
+      "learning_rate": 0.00010190035952747817,
+      "loss": 4.224146728515625,
+      "step": 27500
+    },
+    {
+      "epoch": 0.6656800265788758,
+      "grad_norm": 0.9667420387268066,
+      "learning_rate": 0.00010153349475383372,
+      "loss": 4.240595092773438,
+      "step": 27550
+    },
+    {
+      "epoch": 0.6668881572986197,
+      "grad_norm": 0.9696133136749268,
+      "learning_rate": 0.0001011666299801893,
+      "loss": 4.215619506835938,
+      "step": 27600
+    },
+    {
+      "epoch": 0.6680962880183636,
+      "grad_norm": 0.9353596568107605,
+      "learning_rate": 0.00010079976520654485,
+      "loss": 4.216428527832031,
+      "step": 27650
+    },
+    {
+      "epoch": 0.6693044187381074,
+      "grad_norm": 0.9524815082550049,
+      "learning_rate": 0.00010043290043290043,
+      "loss": 4.189766845703125,
+      "step": 27700
+    },
+    {
+      "epoch": 0.6705125494578513,
+      "grad_norm": 0.912409245967865,
+      "learning_rate": 0.00010006603565925598,
+      "loss": 4.190501708984375,
+      "step": 27750
+    },
+    {
+      "epoch": 0.6717206801775952,
+      "grad_norm": 0.9484263062477112,
+      "learning_rate": 9.969917088561155e-05,
+      "loss": 4.253775939941407,
+      "step": 27800
+    },
+    {
+      "epoch": 0.6729288108973391,
+      "grad_norm": 0.9141276478767395,
+      "learning_rate": 9.933230611196712e-05,
+      "loss": 4.215238647460938,
+      "step": 27850
+    },
+    {
+      "epoch": 0.6741369416170829,
+      "grad_norm": 0.9494786858558655,
+      "learning_rate": 9.896544133832268e-05,
+      "loss": 4.268603820800781,
+      "step": 27900
+    },
+    {
+      "epoch": 0.6753450723368268,
+      "grad_norm": 1.0391219854354858,
+      "learning_rate": 9.859857656467825e-05,
+      "loss": 4.174599304199218,
+      "step": 27950
+    },
+    {
+      "epoch": 0.6765532030565707,
+      "grad_norm": 0.9204246401786804,
+      "learning_rate": 9.823171179103382e-05,
+      "loss": 4.2752001953125,
+      "step": 28000
+    },
+    {
+      "epoch": 0.6777613337763146,
+      "grad_norm": 0.9134758710861206,
+      "learning_rate": 9.786484701738937e-05,
+      "loss": 4.255004272460938,
+      "step": 28050
+    },
+    {
+      "epoch": 0.6789694644960584,
+      "grad_norm": 0.9489421844482422,
+      "learning_rate": 9.749798224374495e-05,
+      "loss": 4.211203002929688,
+      "step": 28100
+    },
+    {
+      "epoch": 0.6801775952158023,
+      "grad_norm": 0.9262958765029907,
+      "learning_rate": 9.71311174701005e-05,
+      "loss": 4.26359375,
+      "step": 28150
+    },
+    {
+      "epoch": 0.6813857259355462,
+      "grad_norm": 0.8936835527420044,
+      "learning_rate": 9.676425269645608e-05,
+      "loss": 4.22978271484375,
+      "step": 28200
+    },
+    {
+      "epoch": 0.6825938566552902,
+      "grad_norm": 0.8912099003791809,
+      "learning_rate": 9.639738792281164e-05,
+      "loss": 4.213511352539062,
+      "step": 28250
+    },
+    {
+      "epoch": 0.683801987375034,
+      "grad_norm": 0.9310536980628967,
+      "learning_rate": 9.603052314916722e-05,
+      "loss": 4.227381591796875,
+      "step": 28300
+    },
+    {
+      "epoch": 0.6850101180947779,
+      "grad_norm": 0.9319400787353516,
+      "learning_rate": 9.566365837552277e-05,
+      "loss": 4.246932678222656,
+      "step": 28350
+    },
+    {
+      "epoch": 0.6862182488145218,
+      "grad_norm": 0.9386299252510071,
+      "learning_rate": 9.529679360187834e-05,
+      "loss": 4.208887939453125,
+      "step": 28400
+    },
+    {
+      "epoch": 0.6874263795342656,
+      "grad_norm": 0.9500228762626648,
+      "learning_rate": 9.49299288282339e-05,
+      "loss": 4.23306640625,
+      "step": 28450
+    },
+    {
+      "epoch": 0.6886345102540095,
+      "grad_norm": 0.9140240550041199,
+      "learning_rate": 9.456306405458947e-05,
+      "loss": 4.227235412597656,
+      "step": 28500
+    },
+    {
+      "epoch": 0.6898426409737534,
+      "grad_norm": 0.9192743897438049,
+      "learning_rate": 9.419619928094503e-05,
+      "loss": 4.211682739257813,
+      "step": 28550
+    },
+    {
+      "epoch": 0.6910507716934973,
+      "grad_norm": 0.9977185130119324,
+      "learning_rate": 9.38293345073006e-05,
+      "loss": 4.194341735839844,
+      "step": 28600
+    },
+    {
+      "epoch": 0.6922589024132411,
+      "grad_norm": 0.9499483108520508,
+      "learning_rate": 9.346246973365615e-05,
+      "loss": 4.202852783203125,
+      "step": 28650
+    },
+    {
+      "epoch": 0.693467033132985,
+      "grad_norm": 1.0312954187393188,
+      "learning_rate": 9.309560496001173e-05,
+      "loss": 4.21665283203125,
+      "step": 28700
+    },
+    {
+      "epoch": 0.6946751638527289,
+      "grad_norm": 0.9412572979927063,
+      "learning_rate": 9.272874018636729e-05,
+      "loss": 4.199068908691406,
+      "step": 28750
+    },
+    {
+      "epoch": 0.6958832945724728,
+      "grad_norm": 0.9228193759918213,
+      "learning_rate": 9.236187541272287e-05,
+      "loss": 4.205174865722657,
+      "step": 28800
+    },
+    {
+      "epoch": 0.6970914252922166,
+      "grad_norm": 0.9229955673217773,
+      "learning_rate": 9.199501063907842e-05,
+      "loss": 4.192348022460937,
+      "step": 28850
+    },
+    {
+      "epoch": 0.6982995560119605,
+      "grad_norm": 0.9405125975608826,
+      "learning_rate": 9.1628145865434e-05,
+      "loss": 4.2276678466796875,
+      "step": 28900
+    },
+    {
+      "epoch": 0.6995076867317044,
+      "grad_norm": 0.9145158529281616,
+      "learning_rate": 9.126128109178955e-05,
+      "loss": 4.226159973144531,
+      "step": 28950
+    },
+    {
+      "epoch": 0.7007158174514483,
+      "grad_norm": 0.9723268151283264,
+      "learning_rate": 9.089441631814512e-05,
+      "loss": 4.23314208984375,
+      "step": 29000
+    },
+    {
+      "epoch": 0.7019239481711921,
+      "grad_norm": 0.9130745530128479,
+      "learning_rate": 9.052755154450069e-05,
+      "loss": 4.199472961425781,
+      "step": 29050
+    },
+    {
+      "epoch": 0.703132078890936,
+      "grad_norm": 0.9244306087493896,
+      "learning_rate": 9.016068677085625e-05,
+      "loss": 4.225984802246094,
+      "step": 29100
+    },
+    {
+      "epoch": 0.7043402096106799,
+      "grad_norm": 1.0370877981185913,
+      "learning_rate": 8.97938219972118e-05,
+      "loss": 4.185047607421875,
+      "step": 29150
+    },
+    {
+      "epoch": 0.7055483403304238,
+      "grad_norm": 1.0177369117736816,
+      "learning_rate": 8.942695722356739e-05,
+      "loss": 4.176769409179688,
+      "step": 29200
+    },
+    {
+      "epoch": 0.7067564710501676,
+      "grad_norm": 0.9620659947395325,
+      "learning_rate": 8.906009244992294e-05,
+      "loss": 4.223794555664062,
+      "step": 29250
+    },
+    {
+      "epoch": 0.7079646017699115,
+      "grad_norm": 0.9300894737243652,
+      "learning_rate": 8.869322767627852e-05,
+      "loss": 4.1740328979492185,
+      "step": 29300
+    },
+    {
+      "epoch": 0.7091727324896554,
+      "grad_norm": 0.9292963147163391,
+      "learning_rate": 8.832636290263407e-05,
+      "loss": 4.1931396484375,
+      "step": 29350
+    },
+    {
+      "epoch": 0.7103808632093993,
+      "grad_norm": 0.9343336224555969,
+      "learning_rate": 8.795949812898965e-05,
+      "loss": 4.19296630859375,
+      "step": 29400
+    },
+    {
+      "epoch": 0.7115889939291431,
+      "grad_norm": 0.9427083134651184,
+      "learning_rate": 8.75926333553452e-05,
+      "loss": 4.19119873046875,
+      "step": 29450
+    },
+    {
+      "epoch": 0.712797124648887,
+      "grad_norm": 0.931380569934845,
+      "learning_rate": 8.722576858170078e-05,
+      "loss": 4.181403503417969,
+      "step": 29500
+    },
+    {
+      "epoch": 0.7140052553686309,
+      "grad_norm": 0.9445347785949707,
+      "learning_rate": 8.685890380805634e-05,
+      "loss": 4.217647399902344,
+      "step": 29550
+    },
+    {
+      "epoch": 0.7152133860883748,
+      "grad_norm": 0.9495642185211182,
+      "learning_rate": 8.64920390344119e-05,
+      "loss": 4.191078796386718,
+      "step": 29600
+    },
+    {
+      "epoch": 0.7164215168081186,
+      "grad_norm": 0.914910614490509,
+      "learning_rate": 8.612517426076747e-05,
+      "loss": 4.20900146484375,
+      "step": 29650
+    },
+    {
+      "epoch": 0.7176296475278625,
+      "grad_norm": 0.9568190574645996,
+      "learning_rate": 8.575830948712304e-05,
+      "loss": 4.174463500976563,
+      "step": 29700
+    },
+    {
+      "epoch": 0.7188377782476064,
+      "grad_norm": 0.9742776155471802,
+      "learning_rate": 8.539144471347859e-05,
+      "loss": 4.226783142089844,
+      "step": 29750
+    },
+    {
+      "epoch": 0.7200459089673502,
+      "grad_norm": 0.9690802693367004,
+      "learning_rate": 8.502457993983417e-05,
+      "loss": 4.190617065429688,
+      "step": 29800
+    },
+    {
+      "epoch": 0.7212540396870941,
+      "grad_norm": 0.9639431834220886,
+      "learning_rate": 8.465771516618972e-05,
+      "loss": 4.173596496582031,
+      "step": 29850
+    },
+    {
+      "epoch": 0.722462170406838,
+      "grad_norm": 0.9556230306625366,
+      "learning_rate": 8.42908503925453e-05,
+      "loss": 4.185955505371094,
+      "step": 29900
+    },
+    {
+      "epoch": 0.7236703011265819,
+      "grad_norm": 0.9598275423049927,
+      "learning_rate": 8.392398561890086e-05,
+      "loss": 4.182140808105469,
+      "step": 29950
+    },
+    {
+      "epoch": 0.7248784318463257,
+      "grad_norm": 0.9849634170532227,
+      "learning_rate": 8.355712084525644e-05,
+      "loss": 4.184024658203125,
+      "step": 30000
+    },
+    {
+      "epoch": 0.7260865625660696,
+      "grad_norm": 0.9694328308105469,
+      "learning_rate": 8.3190256071612e-05,
+      "loss": 4.2082366943359375,
+      "step": 30050
+    },
+    {
+      "epoch": 0.7272946932858135,
+      "grad_norm": 0.9504693150520325,
+      "learning_rate": 8.282339129796756e-05,
+      "loss": 4.172368469238282,
+      "step": 30100
+    },
+    {
+      "epoch": 0.7285028240055574,
+      "grad_norm": 0.9747318029403687,
+      "learning_rate": 8.245652652432314e-05,
+      "loss": 4.211484375,
+      "step": 30150
+    },
+    {
+      "epoch": 0.7297109547253012,
+      "grad_norm": 0.941775918006897,
+      "learning_rate": 8.208966175067869e-05,
+      "loss": 4.200064697265625,
+      "step": 30200
+    },
+    {
+      "epoch": 0.7309190854450451,
+      "grad_norm": 0.9301289916038513,
+      "learning_rate": 8.172279697703427e-05,
+      "loss": 4.198236389160156,
+      "step": 30250
+    },
+    {
+      "epoch": 0.732127216164789,
+      "grad_norm": 0.9551920294761658,
+      "learning_rate": 8.135593220338982e-05,
+      "loss": 4.211250305175781,
+      "step": 30300
+    },
+    {
+      "epoch": 0.733335346884533,
+      "grad_norm": 0.925805389881134,
+      "learning_rate": 8.09890674297454e-05,
+      "loss": 4.180384521484375,
+      "step": 30350
+    },
+    {
+      "epoch": 0.7345434776042767,
+      "grad_norm": 0.9729126691818237,
+      "learning_rate": 8.062220265610095e-05,
+      "loss": 4.183777770996094,
+      "step": 30400
+    },
+    {
+      "epoch": 0.7357516083240206,
+      "grad_norm": 0.9727151393890381,
+      "learning_rate": 8.025533788245652e-05,
+      "loss": 4.210162658691406,
+      "step": 30450
+    },
+    {
+      "epoch": 0.7369597390437646,
+      "grad_norm": 0.9462403059005737,
+      "learning_rate": 7.988847310881209e-05,
+      "loss": 4.168313903808594,
+      "step": 30500
+    },
+    {
+      "epoch": 0.7381678697635085,
+      "grad_norm": 0.9376530051231384,
+      "learning_rate": 7.952160833516765e-05,
+      "loss": 4.186539306640625,
+      "step": 30550
+    },
+    {
+      "epoch": 0.7393760004832522,
+      "grad_norm": 0.9415252804756165,
+      "learning_rate": 7.915474356152322e-05,
+      "loss": 4.179319763183594,
+      "step": 30600
+    },
+    {
+      "epoch": 0.7405841312029962,
+      "grad_norm": 0.9394069910049438,
+      "learning_rate": 7.878787878787879e-05,
+      "loss": 4.16141357421875,
+      "step": 30650
+    },
+    {
+      "epoch": 0.7417922619227401,
+      "grad_norm": 0.9953026175498962,
+      "learning_rate": 7.842101401423434e-05,
+      "loss": 4.202671813964844,
+      "step": 30700
+    },
+    {
+      "epoch": 0.743000392642484,
+      "grad_norm": 0.9346957802772522,
+      "learning_rate": 7.805414924058992e-05,
+      "loss": 4.178117370605468,
+      "step": 30750
+    },
+    {
+      "epoch": 0.7442085233622278,
+      "grad_norm": 1.0032802820205688,
+      "learning_rate": 7.768728446694547e-05,
+      "loss": 4.1957550048828125,
+      "step": 30800
+    },
+    {
+      "epoch": 0.7454166540819717,
+      "grad_norm": 0.906671941280365,
+      "learning_rate": 7.732041969330105e-05,
+      "loss": 4.175614013671875,
+      "step": 30850
+    },
+    {
+      "epoch": 0.7466247848017156,
+      "grad_norm": 0.9167490601539612,
+      "learning_rate": 7.69535549196566e-05,
+      "loss": 4.1725070190429685,
+      "step": 30900
+    },
+    {
+      "epoch": 0.7478329155214595,
+      "grad_norm": 0.981493353843689,
+      "learning_rate": 7.658669014601219e-05,
+      "loss": 4.173250427246094,
+      "step": 30950
+    },
+    {
+      "epoch": 0.7490410462412033,
+      "grad_norm": 0.9785531759262085,
+      "learning_rate": 7.621982537236774e-05,
+      "loss": 4.168323364257812,
+      "step": 31000
+    },
+    {
+      "epoch": 0.7502491769609472,
+      "grad_norm": 0.9385384321212769,
+      "learning_rate": 7.58529605987233e-05,
+      "loss": 4.175681457519532,
+      "step": 31050
+    },
+    {
+      "epoch": 0.7514573076806911,
+      "grad_norm": 0.975518524646759,
+      "learning_rate": 7.548609582507887e-05,
+      "loss": 4.191498413085937,
+      "step": 31100
+    },
+    {
+      "epoch": 0.7526654384004349,
+      "grad_norm": 0.9347544312477112,
+      "learning_rate": 7.511923105143444e-05,
+      "loss": 4.206516418457031,
+      "step": 31150
+    },
+    {
+      "epoch": 0.7538735691201788,
+      "grad_norm": 0.9503827095031738,
+      "learning_rate": 7.475236627779e-05,
+      "loss": 4.166010131835938,
+      "step": 31200
+    },
+    {
+      "epoch": 0.7550816998399227,
+      "grad_norm": 1.0107334852218628,
+      "learning_rate": 7.438550150414557e-05,
+      "loss": 4.12983642578125,
+      "step": 31250
+    },
+    {
+      "epoch": 0.7562898305596666,
+      "grad_norm": 1.0273202657699585,
+      "learning_rate": 7.401863673050112e-05,
+      "loss": 4.160785827636719,
+      "step": 31300
+    },
+    {
+      "epoch": 0.7574979612794104,
+      "grad_norm": 0.9567047357559204,
+      "learning_rate": 7.365177195685669e-05,
+      "loss": 4.200793151855469,
+      "step": 31350
+    },
+    {
+      "epoch": 0.7587060919991543,
+      "grad_norm": 0.9638941884040833,
+      "learning_rate": 7.328490718321226e-05,
+      "loss": 4.171089782714843,
+      "step": 31400
+    },
+    {
+      "epoch": 0.7599142227188982,
+      "grad_norm": 0.9351734519004822,
+      "learning_rate": 7.291804240956782e-05,
+      "loss": 4.153325805664062,
+      "step": 31450
+    },
+    {
+      "epoch": 0.7611223534386421,
+      "grad_norm": 0.9753860235214233,
+      "learning_rate": 7.255117763592339e-05,
+      "loss": 4.165850524902344,
+      "step": 31500
+    },
+    {
+      "epoch": 0.7623304841583859,
+      "grad_norm": 0.9597043395042419,
+      "learning_rate": 7.218431286227896e-05,
+      "loss": 4.159142761230469,
+      "step": 31550
+    },
+    {
+      "epoch": 0.7635386148781298,
+      "grad_norm": 0.9583253264427185,
+      "learning_rate": 7.181744808863452e-05,
+      "loss": 4.168012084960938,
+      "step": 31600
+    },
+    {
+      "epoch": 0.7647467455978737,
+      "grad_norm": 1.0249325037002563,
+      "learning_rate": 7.145058331499009e-05,
+      "loss": 4.178051147460938,
+      "step": 31650
+    },
+    {
+      "epoch": 0.7659548763176176,
+      "grad_norm": 0.96359783411026,
+      "learning_rate": 7.108371854134566e-05,
+      "loss": 4.195538940429688,
+      "step": 31700
+    },
+    {
+      "epoch": 0.7671630070373614,
+      "grad_norm": 0.9502347707748413,
+      "learning_rate": 7.071685376770122e-05,
+      "loss": 4.153787536621094,
+      "step": 31750
+    },
+    {
+      "epoch": 0.7683711377571053,
+      "grad_norm": 0.9883692264556885,
+      "learning_rate": 7.034998899405679e-05,
+      "loss": 4.150211181640625,
+      "step": 31800
+    },
+    {
+      "epoch": 0.7695792684768492,
+      "grad_norm": 0.9689139127731323,
+      "learning_rate": 6.998312422041234e-05,
+      "loss": 4.193338012695312,
+      "step": 31850
+    },
+    {
+      "epoch": 0.7707873991965931,
+      "grad_norm": 0.9628708362579346,
+      "learning_rate": 6.961625944676791e-05,
+      "loss": 4.143995666503907,
+      "step": 31900
+    },
+    {
+      "epoch": 0.7719955299163369,
+      "grad_norm": 1.1675163507461548,
+      "learning_rate": 6.924939467312347e-05,
+      "loss": 4.174693603515625,
+      "step": 31950
+    },
+    {
+      "epoch": 0.7732036606360808,
+      "grad_norm": 0.9363202452659607,
+      "learning_rate": 6.888252989947904e-05,
+      "loss": 4.18224365234375,
+      "step": 32000
+    },
+    {
+      "epoch": 0.7744117913558247,
+      "grad_norm": 0.9869855046272278,
+      "learning_rate": 6.851566512583461e-05,
+      "loss": 4.189596252441406,
+      "step": 32050
+    },
+    {
+      "epoch": 0.7756199220755686,
+      "grad_norm": 0.9943952560424805,
+      "learning_rate": 6.814880035219017e-05,
+      "loss": 4.158417358398437,
+      "step": 32100
+    },
+    {
+      "epoch": 0.7768280527953124,
+      "grad_norm": 0.9464765787124634,
+      "learning_rate": 6.778193557854574e-05,
+      "loss": 4.1605908203125,
+      "step": 32150
+    },
+    {
+      "epoch": 0.7780361835150563,
+      "grad_norm": 0.9998800754547119,
+      "learning_rate": 6.741507080490131e-05,
+      "loss": 4.180458068847656,
+      "step": 32200
+    },
+    {
+      "epoch": 0.7792443142348002,
+      "grad_norm": 0.9495953321456909,
+      "learning_rate": 6.704820603125687e-05,
+      "loss": 4.177342529296875,
+      "step": 32250
+    },
+    {
+      "epoch": 0.7804524449545441,
+      "grad_norm": 0.9588845372200012,
+      "learning_rate": 6.668134125761244e-05,
+      "loss": 4.1812451171875,
+      "step": 32300
+    },
+    {
+      "epoch": 0.7816605756742879,
+      "grad_norm": 0.9747833013534546,
+      "learning_rate": 6.6314476483968e-05,
+      "loss": 4.142763061523437,
+      "step": 32350
+    },
+    {
+      "epoch": 0.7828687063940318,
+      "grad_norm": 1.0218557119369507,
+      "learning_rate": 6.594761171032357e-05,
+      "loss": 4.154402465820312,
+      "step": 32400
+    },
+    {
+      "epoch": 0.7840768371137757,
+      "grad_norm": 0.9641887545585632,
+      "learning_rate": 6.558074693667913e-05,
+      "loss": 4.1369412231445315,
+      "step": 32450
+    },
+    {
+      "epoch": 0.7852849678335195,
+      "grad_norm": 0.9768737554550171,
+      "learning_rate": 6.521388216303469e-05,
+      "loss": 4.162682189941406,
+      "step": 32500
+    },
+    {
+      "epoch": 0.7864930985532634,
+      "grad_norm": 0.9474012851715088,
+      "learning_rate": 6.484701738939026e-05,
+      "loss": 4.156160888671875,
+      "step": 32550
+    },
+    {
+      "epoch": 0.7877012292730073,
+      "grad_norm": 0.9589933753013611,
+      "learning_rate": 6.448015261574583e-05,
+      "loss": 4.144935913085938,
+      "step": 32600
+    },
+    {
+      "epoch": 0.7889093599927512,
+      "grad_norm": 0.9754806756973267,
+      "learning_rate": 6.411328784210139e-05,
+      "loss": 4.155699462890625,
+      "step": 32650
+    },
+    {
+      "epoch": 0.790117490712495,
+      "grad_norm": 0.9701508283615112,
+      "learning_rate": 6.374642306845696e-05,
+      "loss": 4.161666259765625,
+      "step": 32700
+    },
+    {
+      "epoch": 0.791325621432239,
+      "grad_norm": 1.0167360305786133,
+      "learning_rate": 6.337955829481252e-05,
+      "loss": 4.1516961669921875,
+      "step": 32750
+    },
+    {
+      "epoch": 0.7925337521519829,
+      "grad_norm": 0.9813135862350464,
+      "learning_rate": 6.301269352116809e-05,
+      "loss": 4.134502868652344,
+      "step": 32800
+    },
+    {
+      "epoch": 0.7937418828717268,
+      "grad_norm": 1.0000736713409424,
+      "learning_rate": 6.264582874752366e-05,
+      "loss": 4.175084228515625,
+      "step": 32850
+    },
+    {
+      "epoch": 0.7949500135914706,
+      "grad_norm": 0.9590751528739929,
+      "learning_rate": 6.227896397387922e-05,
+      "loss": 4.197709655761718,
+      "step": 32900
+    },
+    {
+      "epoch": 0.7961581443112145,
+      "grad_norm": 0.9502274990081787,
+      "learning_rate": 6.191209920023479e-05,
+      "loss": 4.1478402709960935,
+      "step": 32950
+    },
+    {
+      "epoch": 0.7973662750309584,
+      "grad_norm": 0.9672371745109558,
+      "learning_rate": 6.154523442659036e-05,
+      "loss": 4.179851379394531,
+      "step": 33000
+    },
+    {
+      "epoch": 0.7985744057507023,
+      "grad_norm": 0.9944890141487122,
+      "learning_rate": 6.117836965294591e-05,
+      "loss": 4.172831726074219,
+      "step": 33050
+    },
+    {
+      "epoch": 0.7997825364704461,
+      "grad_norm": 0.9638657569885254,
+      "learning_rate": 6.0811504879301484e-05,
+      "loss": 4.179300231933594,
+      "step": 33100
+    },
+    {
+      "epoch": 0.80099066719019,
+      "grad_norm": 1.0092177391052246,
+      "learning_rate": 6.044464010565705e-05,
+      "loss": 4.1206817626953125,
+      "step": 33150
+    },
+    {
+      "epoch": 0.8021987979099339,
+      "grad_norm": 0.9667877554893494,
+      "learning_rate": 6.007777533201261e-05,
+      "loss": 4.129392700195313,
+      "step": 33200
+    },
+    {
+      "epoch": 0.8034069286296778,
+      "grad_norm": 0.9979503750801086,
+      "learning_rate": 5.9710910558368176e-05,
+      "loss": 4.135091247558594,
+      "step": 33250
+    },
+    {
+      "epoch": 0.8046150593494216,
+      "grad_norm": 0.9955921769142151,
+      "learning_rate": 5.934404578472374e-05,
+      "loss": 4.1466259765625,
+      "step": 33300
+    },
+    {
+      "epoch": 0.8058231900691655,
+      "grad_norm": 0.9967898726463318,
+      "learning_rate": 5.897718101107931e-05,
+      "loss": 4.1172265625,
+      "step": 33350
+    },
+    {
+      "epoch": 0.8070313207889094,
+      "grad_norm": 0.9824782013893127,
+      "learning_rate": 5.8610316237434876e-05,
+      "loss": 4.183871154785156,
+      "step": 33400
+    },
+    {
+      "epoch": 0.8082394515086533,
+      "grad_norm": 0.9574515223503113,
+      "learning_rate": 5.8243451463790435e-05,
+      "loss": 4.141428527832031,
+      "step": 33450
+    },
+    {
+      "epoch": 0.8094475822283971,
+      "grad_norm": 0.948474645614624,
+      "learning_rate": 5.7876586690146e-05,
+      "loss": 4.155045776367188,
+      "step": 33500
+    },
+    {
+      "epoch": 0.810655712948141,
+      "grad_norm": 0.983756959438324,
+      "learning_rate": 5.750972191650157e-05,
+      "loss": 4.114388427734375,
+      "step": 33550
+    },
+    {
+      "epoch": 0.8118638436678849,
+      "grad_norm": 1.0394560098648071,
+      "learning_rate": 5.7142857142857135e-05,
+      "loss": 4.130142517089844,
+      "step": 33600
+    },
+    {
+      "epoch": 0.8130719743876288,
+      "grad_norm": 0.9778473973274231,
+      "learning_rate": 5.67759923692127e-05,
+      "loss": 4.118342590332031,
+      "step": 33650
+    },
+    {
+      "epoch": 0.8142801051073726,
+      "grad_norm": 0.9543645977973938,
+      "learning_rate": 5.640912759556827e-05,
+      "loss": 4.14258544921875,
+      "step": 33700
+    },
+    {
+      "epoch": 0.8154882358271165,
+      "grad_norm": 0.9755384922027588,
+      "learning_rate": 5.604226282192383e-05,
+      "loss": 4.172442932128906,
+      "step": 33750
+    },
+    {
+      "epoch": 0.8166963665468604,
+      "grad_norm": 0.9590539336204529,
+      "learning_rate": 5.5675398048279394e-05,
+      "loss": 4.129637451171875,
+      "step": 33800
+    },
+    {
+      "epoch": 0.8179044972666042,
+      "grad_norm": 0.9638984203338623,
+      "learning_rate": 5.530853327463496e-05,
+      "loss": 4.155792236328125,
+      "step": 33850
+    },
+    {
+      "epoch": 0.8191126279863481,
+      "grad_norm": 0.9844355583190918,
+      "learning_rate": 5.494166850099053e-05,
+      "loss": 4.122502136230469,
+      "step": 33900
+    },
+    {
+      "epoch": 0.820320758706092,
+      "grad_norm": 0.9721888899803162,
+      "learning_rate": 5.4574803727346094e-05,
+      "loss": 4.122838439941407,
+      "step": 33950
+    },
+    {
+      "epoch": 0.8215288894258359,
+      "grad_norm": 0.9714981317520142,
+      "learning_rate": 5.420793895370166e-05,
+      "loss": 4.1767581176757815,
+      "step": 34000
+    },
+    {
+      "epoch": 0.8227370201455797,
+      "grad_norm": 0.9962683320045471,
+      "learning_rate": 5.384107418005722e-05,
+      "loss": 4.127580261230468,
+      "step": 34050
+    },
+    {
+      "epoch": 0.8239451508653236,
+      "grad_norm": 0.9760364890098572,
+      "learning_rate": 5.3474209406412786e-05,
+      "loss": 4.136761779785156,
+      "step": 34100
+    },
+    {
+      "epoch": 0.8251532815850675,
+      "grad_norm": 0.9831131100654602,
+      "learning_rate": 5.310734463276835e-05,
+      "loss": 4.1412725830078125,
+      "step": 34150
+    },
+    {
+      "epoch": 0.8263614123048114,
+      "grad_norm": 0.9481918811798096,
+      "learning_rate": 5.274047985912392e-05,
+      "loss": 4.08984130859375,
+      "step": 34200
+    },
+    {
+      "epoch": 0.8275695430245552,
+      "grad_norm": 1.0539478063583374,
+      "learning_rate": 5.2373615085479486e-05,
+      "loss": 4.127510375976563,
+      "step": 34250
+    },
+    {
+      "epoch": 0.8287776737442991,
+      "grad_norm": 0.9744389653205872,
+      "learning_rate": 5.200675031183505e-05,
+      "loss": 4.149533081054687,
+      "step": 34300
+    },
+    {
+      "epoch": 0.829985804464043,
+      "grad_norm": 0.952286958694458,
+      "learning_rate": 5.163988553819061e-05,
+      "loss": 4.157025451660156,
+      "step": 34350
+    },
+    {
+      "epoch": 0.8311939351837869,
+      "grad_norm": 1.062190055847168,
+      "learning_rate": 5.127302076454618e-05,
+      "loss": 4.137975463867187,
+      "step": 34400
+    },
+    {
+      "epoch": 0.8324020659035307,
+      "grad_norm": 0.9755569696426392,
+      "learning_rate": 5.0906155990901745e-05,
+      "loss": 4.145784912109375,
+      "step": 34450
+    },
+    {
+      "epoch": 0.8336101966232746,
+      "grad_norm": 1.0148369073867798,
+      "learning_rate": 5.053929121725731e-05,
+      "loss": 4.11475830078125,
+      "step": 34500
+    },
+    {
+      "epoch": 0.8348183273430185,
+      "grad_norm": 1.028790831565857,
+      "learning_rate": 5.017242644361288e-05,
+      "loss": 4.156703796386719,
+      "step": 34550
+    },
+    {
+      "epoch": 0.8360264580627624,
+      "grad_norm": 1.038443922996521,
+      "learning_rate": 4.9805561669968444e-05,
+      "loss": 4.138312683105469,
+      "step": 34600
+    },
+    {
+      "epoch": 0.8372345887825062,
+      "grad_norm": 1.0102113485336304,
+      "learning_rate": 4.9438696896324004e-05,
+      "loss": 4.146728515625,
+      "step": 34650
+    },
+    {
+      "epoch": 0.8384427195022501,
+      "grad_norm": 0.9652780890464783,
+      "learning_rate": 4.907183212267957e-05,
+      "loss": 4.143882446289062,
+      "step": 34700
+    },
+    {
+      "epoch": 0.839650850221994,
+      "grad_norm": 1.0120545625686646,
+      "learning_rate": 4.8704967349035144e-05,
+      "loss": 4.131054992675781,
+      "step": 34750
+    },
+    {
+      "epoch": 0.8408589809417379,
+      "grad_norm": 0.9618927240371704,
+      "learning_rate": 4.833810257539071e-05,
+      "loss": 4.1066891479492185,
+      "step": 34800
+    },
+    {
+      "epoch": 0.8420671116614817,
+      "grad_norm": 0.976310670375824,
+      "learning_rate": 4.797123780174628e-05,
+      "loss": 4.1329037475585935,
+      "step": 34850
+    },
+    {
+      "epoch": 0.8432752423812256,
+      "grad_norm": 0.9385315179824829,
+      "learning_rate": 4.760437302810184e-05,
+      "loss": 4.14763427734375,
+      "step": 34900
+    },
+    {
+      "epoch": 0.8444833731009695,
+      "grad_norm": 1.0614820718765259,
+      "learning_rate": 4.723750825445741e-05,
+      "loss": 4.1223211669921875,
+      "step": 34950
+    },
+    {
+      "epoch": 0.8456915038207135,
+      "grad_norm": 1.0392005443572998,
+      "learning_rate": 4.687064348081297e-05,
+      "loss": 4.104436340332032,
+      "step": 35000
+    },
+    {
+      "epoch": 0.8468996345404572,
+      "grad_norm": 0.9826388955116272,
+      "learning_rate": 4.6503778707168536e-05,
+      "loss": 4.130885009765625,
+      "step": 35050
+    },
+    {
+      "epoch": 0.8481077652602012,
+      "grad_norm": 0.9898689389228821,
+      "learning_rate": 4.61369139335241e-05,
+      "loss": 4.113179626464844,
+      "step": 35100
+    },
+    {
+      "epoch": 0.8493158959799451,
+      "grad_norm": 0.9709085822105408,
+      "learning_rate": 4.577004915987967e-05,
+      "loss": 4.136431274414062,
+      "step": 35150
+    },
+    {
+      "epoch": 0.8505240266996889,
+      "grad_norm": 1.0114762783050537,
+      "learning_rate": 4.5403184386235235e-05,
+      "loss": 4.141319885253906,
+      "step": 35200
+    },
+    {
+      "epoch": 0.8517321574194328,
+      "grad_norm": 0.9931055307388306,
+      "learning_rate": 4.50363196125908e-05,
+      "loss": 4.1039950561523435,
+      "step": 35250
+    },
+    {
+      "epoch": 0.8529402881391767,
+      "grad_norm": 0.988318920135498,
+      "learning_rate": 4.466945483894636e-05,
+      "loss": 4.108782653808594,
+      "step": 35300
+    },
+    {
+      "epoch": 0.8541484188589206,
+      "grad_norm": 0.9901145696640015,
+      "learning_rate": 4.430259006530193e-05,
+      "loss": 4.1287109375,
+      "step": 35350
+    },
+    {
+      "epoch": 0.8553565495786644,
+      "grad_norm": 1.0645872354507446,
+      "learning_rate": 4.3935725291657495e-05,
+      "loss": 4.0877291870117185,
+      "step": 35400
+    },
+    {
+      "epoch": 0.8565646802984083,
+      "grad_norm": 0.9934980869293213,
+      "learning_rate": 4.356886051801306e-05,
+      "loss": 4.142991027832031,
+      "step": 35450
+    },
+    {
+      "epoch": 0.8577728110181522,
+      "grad_norm": 1.0430258512496948,
+      "learning_rate": 4.320199574436863e-05,
+      "loss": 4.068301086425781,
+      "step": 35500
+    },
+    {
+      "epoch": 0.8589809417378961,
+      "grad_norm": 0.9746304154396057,
+      "learning_rate": 4.2835130970724194e-05,
+      "loss": 4.128489685058594,
+      "step": 35550
+    },
+    {
+      "epoch": 0.8601890724576399,
+      "grad_norm": 0.9641109704971313,
+      "learning_rate": 4.2468266197079754e-05,
+      "loss": 4.121369018554687,
+      "step": 35600
+    },
+    {
+      "epoch": 0.8613972031773838,
+      "grad_norm": 0.9532524347305298,
+      "learning_rate": 4.210140142343532e-05,
+      "loss": 4.090971374511719,
+      "step": 35650
+    },
+    {
+      "epoch": 0.8626053338971277,
+      "grad_norm": 0.9789423942565918,
+      "learning_rate": 4.173453664979089e-05,
+      "loss": 4.1288214111328125,
+      "step": 35700
+    },
+    {
+      "epoch": 0.8638134646168716,
+      "grad_norm": 1.0037946701049805,
+      "learning_rate": 4.136767187614645e-05,
+      "loss": 4.139686889648438,
+      "step": 35750
+    },
+    {
+      "epoch": 0.8650215953366154,
+      "grad_norm": 1.01668119430542,
+      "learning_rate": 4.100080710250202e-05,
+      "loss": 4.092429809570312,
+      "step": 35800
+    },
+    {
+      "epoch": 0.8662297260563593,
+      "grad_norm": 0.9829847812652588,
+      "learning_rate": 4.063394232885758e-05,
+      "loss": 4.10031494140625,
+      "step": 35850
+    },
+    {
+      "epoch": 0.8674378567761032,
+      "grad_norm": 0.95753014087677,
+      "learning_rate": 4.0267077555213146e-05,
+      "loss": 4.117217407226563,
+      "step": 35900
+    },
+    {
+      "epoch": 0.8686459874958471,
+      "grad_norm": 1.0203341245651245,
+      "learning_rate": 3.990021278156871e-05,
+      "loss": 4.094928894042969,
+      "step": 35950
+    },
+    {
+      "epoch": 0.8698541182155909,
+      "grad_norm": 0.982297420501709,
+      "learning_rate": 3.953334800792428e-05,
+      "loss": 4.105053100585938,
+      "step": 36000
+    },
+    {
+      "epoch": 0.8710622489353348,
+      "grad_norm": 0.9894450902938843,
+      "learning_rate": 3.9166483234279845e-05,
+      "loss": 4.102446899414063,
+      "step": 36050
+    },
+    {
+      "epoch": 0.8722703796550787,
+      "grad_norm": 0.9681161046028137,
+      "learning_rate": 3.879961846063541e-05,
+      "loss": 4.127907409667968,
+      "step": 36100
+    },
+    {
+      "epoch": 0.8734785103748226,
+      "grad_norm": 1.0139074325561523,
+      "learning_rate": 3.843275368699097e-05,
+      "loss": 4.11950927734375,
+      "step": 36150
+    },
+    {
+      "epoch": 0.8746866410945664,
+      "grad_norm": 0.9912289381027222,
+      "learning_rate": 3.806588891334654e-05,
+      "loss": 4.113659057617188,
+      "step": 36200
+    },
+    {
+      "epoch": 0.8758947718143103,
+      "grad_norm": 1.002201795578003,
+      "learning_rate": 3.7699024139702105e-05,
+      "loss": 4.124761657714844,
+      "step": 36250
+    },
+    {
+      "epoch": 0.8771029025340542,
+      "grad_norm": 0.9588595628738403,
+      "learning_rate": 3.7332159366057664e-05,
+      "loss": 4.082340087890625,
+      "step": 36300
+    },
+    {
+      "epoch": 0.8783110332537981,
+      "grad_norm": 0.990828812122345,
+      "learning_rate": 3.696529459241323e-05,
+      "loss": 4.108609008789062,
+      "step": 36350
+    },
+    {
+      "epoch": 0.8795191639735419,
+      "grad_norm": 0.9968988299369812,
+      "learning_rate": 3.65984298187688e-05,
+      "loss": 4.090320434570312,
+      "step": 36400
+    },
+    {
+      "epoch": 0.8807272946932858,
+      "grad_norm": 0.9899400472640991,
+      "learning_rate": 3.6231565045124364e-05,
+      "loss": 4.083380432128906,
+      "step": 36450
+    },
+    {
+      "epoch": 0.8819354254130297,
+      "grad_norm": 0.9858762621879578,
+      "learning_rate": 3.586470027147993e-05,
+      "loss": 4.098477783203125,
+      "step": 36500
+    },
+    {
+      "epoch": 0.8831435561327735,
+      "grad_norm": 0.9726820588111877,
+      "learning_rate": 3.549783549783549e-05,
+      "loss": 4.107604064941406,
+      "step": 36550
+    },
+    {
+      "epoch": 0.8843516868525174,
+      "grad_norm": 1.0214866399765015,
+      "learning_rate": 3.5130970724191056e-05,
+      "loss": 4.104823303222656,
+      "step": 36600
+    },
+    {
+      "epoch": 0.8855598175722613,
+      "grad_norm": 1.0095114707946777,
+      "learning_rate": 3.476410595054662e-05,
+      "loss": 4.057843627929688,
+      "step": 36650
+    },
+    {
+      "epoch": 0.8867679482920052,
+      "grad_norm": 1.138928771018982,
+      "learning_rate": 3.439724117690219e-05,
+      "loss": 4.099917907714843,
+      "step": 36700
+    },
+    {
+      "epoch": 0.887976079011749,
+      "grad_norm": 1.0221952199935913,
+      "learning_rate": 3.4030376403257756e-05,
+      "loss": 4.100775146484375,
+      "step": 36750
+    },
+    {
+      "epoch": 0.8891842097314929,
+      "grad_norm": 1.0162906646728516,
+      "learning_rate": 3.366351162961332e-05,
+      "loss": 4.074209289550781,
+      "step": 36800
+    },
+    {
+      "epoch": 0.8903923404512368,
+      "grad_norm": 1.0013428926467896,
+      "learning_rate": 3.329664685596888e-05,
+      "loss": 4.1091915893554685,
+      "step": 36850
+    },
+    {
+      "epoch": 0.8916004711709807,
+      "grad_norm": 0.9851410984992981,
+      "learning_rate": 3.292978208232445e-05,
+      "loss": 4.087139587402344,
+      "step": 36900
+    },
+    {
+      "epoch": 0.8928086018907245,
+      "grad_norm": 1.0403225421905518,
+      "learning_rate": 3.2562917308680015e-05,
+      "loss": 4.0806448364257815,
+      "step": 36950
+    },
+    {
+      "epoch": 0.8940167326104684,
+      "grad_norm": 0.9999246597290039,
+      "learning_rate": 3.219605253503558e-05,
+      "loss": 4.085538940429688,
+      "step": 37000
+    },
+    {
+      "epoch": 0.8952248633302123,
+      "grad_norm": 1.0281506776809692,
+      "learning_rate": 3.182918776139115e-05,
+      "loss": 4.089646301269531,
+      "step": 37050
+    },
+    {
+      "epoch": 0.8964329940499562,
+      "grad_norm": 0.9889469742774963,
+      "learning_rate": 3.1462322987746715e-05,
+      "loss": 4.114851989746094,
+      "step": 37100
+    },
+    {
+      "epoch": 0.8976411247697,
+      "grad_norm": 1.0321831703186035,
+      "learning_rate": 3.109545821410228e-05,
+      "loss": 4.092791442871094,
+      "step": 37150
+    },
+    {
+      "epoch": 0.8988492554894439,
+      "grad_norm": 0.9642487168312073,
+      "learning_rate": 3.072859344045785e-05,
+      "loss": 4.125544128417968,
+      "step": 37200
+    },
+    {
+      "epoch": 0.9000573862091878,
+      "grad_norm": 0.9732397198677063,
+      "learning_rate": 3.036172866681341e-05,
+      "loss": 4.122996826171875,
+      "step": 37250
+    },
+    {
+      "epoch": 0.9012655169289318,
+      "grad_norm": 1.0595879554748535,
+      "learning_rate": 2.9994863893168977e-05,
+      "loss": 4.096632385253907,
+      "step": 37300
+    },
+    {
+      "epoch": 0.9024736476486755,
+      "grad_norm": 1.0211713314056396,
+      "learning_rate": 2.9627999119524544e-05,
+      "loss": 4.123179016113281,
+      "step": 37350
+    },
+    {
+      "epoch": 0.9036817783684195,
+      "grad_norm": 1.020198106765747,
+      "learning_rate": 2.9261134345880107e-05,
+      "loss": 4.076532287597656,
+      "step": 37400
+    },
+    {
+      "epoch": 0.9048899090881634,
+      "grad_norm": 1.010294795036316,
+      "learning_rate": 2.8894269572235673e-05,
+      "loss": 4.050249938964844,
+      "step": 37450
+    },
+    {
+      "epoch": 0.9060980398079073,
+      "grad_norm": 1.0190550088882446,
+      "learning_rate": 2.852740479859124e-05,
+      "loss": 4.101957092285156,
+      "step": 37500
+    },
+    {
+      "epoch": 0.9073061705276511,
+      "grad_norm": 0.9936062693595886,
+      "learning_rate": 2.8160540024946803e-05,
+      "loss": 4.103214111328125,
+      "step": 37550
+    },
+    {
+      "epoch": 0.908514301247395,
+      "grad_norm": 0.9958685636520386,
+      "learning_rate": 2.779367525130237e-05,
+      "loss": 4.139706726074219,
+      "step": 37600
+    },
+    {
+      "epoch": 0.9097224319671389,
+      "grad_norm": 1.0311979055404663,
+      "learning_rate": 2.7426810477657936e-05,
+      "loss": 4.103955078125,
+      "step": 37650
+    },
+    {
+      "epoch": 0.9109305626868828,
+      "grad_norm": 1.0025954246520996,
+      "learning_rate": 2.70599457040135e-05,
+      "loss": 4.091174926757812,
+      "step": 37700
+    },
+    {
+      "epoch": 0.9121386934066266,
+      "grad_norm": 0.9823068380355835,
+      "learning_rate": 2.6693080930369065e-05,
+      "loss": 4.085905456542969,
+      "step": 37750
+    },
+    {
+      "epoch": 0.9133468241263705,
+      "grad_norm": 1.0165550708770752,
+      "learning_rate": 2.6326216156724632e-05,
+      "loss": 4.102291259765625,
+      "step": 37800
+    },
+    {
+      "epoch": 0.9145549548461144,
+      "grad_norm": 0.9877409934997559,
+      "learning_rate": 2.5959351383080195e-05,
+      "loss": 4.085850524902344,
+      "step": 37850
+    },
+    {
+      "epoch": 0.9157630855658583,
+      "grad_norm": 1.037433385848999,
+      "learning_rate": 2.559248660943576e-05,
+      "loss": 4.140678405761719,
+      "step": 37900
+    },
+    {
+      "epoch": 0.9169712162856021,
+      "grad_norm": 1.0109459161758423,
+      "learning_rate": 2.5225621835791328e-05,
+      "loss": 4.086475830078125,
+      "step": 37950
+    },
+    {
+      "epoch": 0.918179347005346,
+      "grad_norm": 0.9636803865432739,
+      "learning_rate": 2.485875706214689e-05,
+      "loss": 4.078011169433593,
+      "step": 38000
+    },
+    {
+      "epoch": 0.9193874777250899,
+      "grad_norm": 1.0318667888641357,
+      "learning_rate": 2.4491892288502457e-05,
+      "loss": 4.090683898925781,
+      "step": 38050
+    },
+    {
+      "epoch": 0.9205956084448337,
+      "grad_norm": 1.012874960899353,
+      "learning_rate": 2.4125027514858024e-05,
+      "loss": 4.112925720214844,
+      "step": 38100
+    },
+    {
+      "epoch": 0.9218037391645776,
+      "grad_norm": 0.9845221042633057,
+      "learning_rate": 2.3758162741213587e-05,
+      "loss": 4.061522827148438,
+      "step": 38150
+    },
+    {
+      "epoch": 0.9230118698843215,
+      "grad_norm": 0.9903792142868042,
+      "learning_rate": 2.3391297967569154e-05,
+      "loss": 4.118177490234375,
+      "step": 38200
+    },
+    {
+      "epoch": 0.9242200006040654,
+      "grad_norm": 0.9599031209945679,
+      "learning_rate": 2.302443319392472e-05,
+      "loss": 4.081640014648437,
+      "step": 38250
+    },
+    {
+      "epoch": 0.9254281313238092,
+      "grad_norm": 1.003709316253662,
+      "learning_rate": 2.2657568420280283e-05,
+      "loss": 4.081379699707031,
+      "step": 38300
+    },
+    {
+      "epoch": 0.9266362620435531,
+      "grad_norm": 1.0040228366851807,
+      "learning_rate": 2.229070364663585e-05,
+      "loss": 4.043332214355469,
+      "step": 38350
+    },
+    {
+      "epoch": 0.927844392763297,
+      "grad_norm": 1.0042468309402466,
+      "learning_rate": 2.1923838872991413e-05,
+      "loss": 4.0926971435546875,
+      "step": 38400
+    },
+    {
+      "epoch": 0.9290525234830409,
+      "grad_norm": 1.0174452066421509,
+      "learning_rate": 2.155697409934698e-05,
+      "loss": 4.10265625,
+      "step": 38450
+    },
+    {
+      "epoch": 0.9302606542027847,
+      "grad_norm": 1.0567253828048706,
+      "learning_rate": 2.1190109325702546e-05,
+      "loss": 4.095572509765625,
+      "step": 38500
+    },
+    {
+      "epoch": 0.9314687849225286,
+      "grad_norm": 1.0051237344741821,
+      "learning_rate": 2.082324455205811e-05,
+      "loss": 4.063082580566406,
+      "step": 38550
+    },
+    {
+      "epoch": 0.9326769156422725,
+      "grad_norm": 1.0048378705978394,
+      "learning_rate": 2.0456379778413675e-05,
+      "loss": 4.085982360839844,
+      "step": 38600
+    },
+    {
+      "epoch": 0.9338850463620164,
+      "grad_norm": 0.9908283352851868,
+      "learning_rate": 2.0089515004769242e-05,
+      "loss": 4.082131958007812,
+      "step": 38650
+    },
+    {
+      "epoch": 0.9350931770817602,
+      "grad_norm": 1.0110317468643188,
+      "learning_rate": 1.9722650231124805e-05,
+      "loss": 4.047686157226562,
+      "step": 38700
+    },
+    {
+      "epoch": 0.9363013078015041,
+      "grad_norm": 0.9872307181358337,
+      "learning_rate": 1.935578545748037e-05,
+      "loss": 4.056953430175781,
+      "step": 38750
+    },
+    {
+      "epoch": 0.937509438521248,
+      "grad_norm": 0.9820250868797302,
+      "learning_rate": 1.8988920683835938e-05,
+      "loss": 4.067868041992187,
+      "step": 38800
+    },
+    {
+      "epoch": 0.9387175692409919,
+      "grad_norm": 1.0473401546478271,
+      "learning_rate": 1.86220559101915e-05,
+      "loss": 4.0692291259765625,
+      "step": 38850
+    },
+    {
+      "epoch": 0.9399256999607357,
+      "grad_norm": 0.9992564916610718,
+      "learning_rate": 1.8255191136547067e-05,
+      "loss": 4.076083984375,
+      "step": 38900
+    },
+    {
+      "epoch": 0.9411338306804796,
+      "grad_norm": 1.022905707359314,
+      "learning_rate": 1.7888326362902634e-05,
+      "loss": 4.084729309082031,
+      "step": 38950
+    },
+    {
+      "epoch": 0.9423419614002235,
+      "grad_norm": 1.004463791847229,
+      "learning_rate": 1.7521461589258197e-05,
+      "loss": 4.108402099609375,
+      "step": 39000
+    },
+    {
+      "epoch": 0.9435500921199674,
+      "grad_norm": 1.0129356384277344,
+      "learning_rate": 1.7154596815613763e-05,
+      "loss": 4.074634399414062,
+      "step": 39050
+    },
+    {
+      "epoch": 0.9447582228397112,
+      "grad_norm": 1.049232840538025,
+      "learning_rate": 1.678773204196933e-05,
+      "loss": 4.068507080078125,
+      "step": 39100
+    },
+    {
+      "epoch": 0.9459663535594551,
+      "grad_norm": 1.0615679025650024,
+      "learning_rate": 1.6420867268324893e-05,
+      "loss": 4.053023376464844,
+      "step": 39150
+    },
+    {
+      "epoch": 0.947174484279199,
+      "grad_norm": 1.0288054943084717,
+      "learning_rate": 1.605400249468046e-05,
+      "loss": 4.06638671875,
+      "step": 39200
+    },
+    {
+      "epoch": 0.9483826149989429,
+      "grad_norm": 1.0288102626800537,
+      "learning_rate": 1.5687137721036026e-05,
+      "loss": 4.061174011230468,
+      "step": 39250
+    },
+    {
+      "epoch": 0.9495907457186867,
+      "grad_norm": 0.9671359658241272,
+      "learning_rate": 1.532027294739159e-05,
+      "loss": 4.080897216796875,
+      "step": 39300
+    },
+    {
+      "epoch": 0.9507988764384306,
+      "grad_norm": 0.9935935735702515,
+      "learning_rate": 1.4953408173747156e-05,
+      "loss": 4.092217712402344,
+      "step": 39350
+    },
+    {
+      "epoch": 0.9520070071581745,
+      "grad_norm": 0.9811704158782959,
+      "learning_rate": 1.458654340010272e-05,
+      "loss": 4.102410888671875,
+      "step": 39400
+    },
+    {
+      "epoch": 0.9532151378779183,
+      "grad_norm": 1.0112156867980957,
+      "learning_rate": 1.4219678626458287e-05,
+      "loss": 4.043121337890625,
+      "step": 39450
+    },
+    {
+      "epoch": 0.9544232685976622,
+      "grad_norm": 0.9941301941871643,
+      "learning_rate": 1.3852813852813852e-05,
+      "loss": 4.0836181640625,
+      "step": 39500
+    },
+    {
+      "epoch": 0.9556313993174061,
+      "grad_norm": 1.0051082372665405,
+      "learning_rate": 1.3485949079169417e-05,
+      "loss": 4.0856561279296875,
+      "step": 39550
+    },
+    {
+      "epoch": 0.95683953003715,
+      "grad_norm": 0.9955536127090454,
+      "learning_rate": 1.3119084305524983e-05,
+      "loss": 4.072062683105469,
+      "step": 39600
+    },
+    {
+      "epoch": 0.9580476607568938,
+      "grad_norm": 0.9985010623931885,
+      "learning_rate": 1.2752219531880548e-05,
+      "loss": 4.084420776367187,
+      "step": 39650
+    },
+    {
+      "epoch": 0.9592557914766378,
+      "grad_norm": 0.9988077282905579,
+      "learning_rate": 1.2385354758236113e-05,
+      "loss": 4.090346374511719,
+      "step": 39700
+    },
+    {
+      "epoch": 0.9604639221963817,
+      "grad_norm": 0.9799109101295471,
+      "learning_rate": 1.2018489984591677e-05,
+      "loss": 4.089367980957031,
+      "step": 39750
+    },
+    {
+      "epoch": 0.9616720529161256,
+      "grad_norm": 1.001146912574768,
+      "learning_rate": 1.1651625210947244e-05,
+      "loss": 4.0670556640625,
+      "step": 39800
+    },
+    {
+      "epoch": 0.9628801836358694,
+      "grad_norm": 1.009992003440857,
+      "learning_rate": 1.1284760437302809e-05,
+      "loss": 4.115885925292969,
+      "step": 39850
+    },
+    {
+      "epoch": 0.9640883143556133,
+      "grad_norm": 0.9880353808403015,
+      "learning_rate": 1.0917895663658373e-05,
+      "loss": 4.062857971191407,
+      "step": 39900
+    },
+    {
+      "epoch": 0.9652964450753572,
+      "grad_norm": 0.9826756715774536,
+      "learning_rate": 1.055103089001394e-05,
+      "loss": 4.114837646484375,
+      "step": 39950
+    },
+    {
+      "epoch": 0.9665045757951011,
+      "grad_norm": 1.0152478218078613,
+      "learning_rate": 1.0184166116369505e-05,
+      "loss": 4.068152160644531,
+      "step": 40000
+    },
+    {
+      "epoch": 0.9677127065148449,
+      "grad_norm": 0.9987122416496277,
+      "learning_rate": 9.81730134272507e-06,
+      "loss": 4.095503234863282,
+      "step": 40050
+    },
+    {
+      "epoch": 0.9689208372345888,
+      "grad_norm": 0.9974833726882935,
+      "learning_rate": 9.450436569080636e-06,
+      "loss": 4.070238037109375,
+      "step": 40100
+    },
+    {
+      "epoch": 0.9701289679543327,
+      "grad_norm": 1.0077698230743408,
+      "learning_rate": 9.083571795436203e-06,
+      "loss": 4.094546203613281,
+      "step": 40150
+    },
+    {
+      "epoch": 0.9713370986740766,
+      "grad_norm": 1.018665075302124,
+      "learning_rate": 8.716707021791767e-06,
+      "loss": 4.067680053710937,
+      "step": 40200
+    },
+    {
+      "epoch": 0.9725452293938204,
+      "grad_norm": 1.073243260383606,
+      "learning_rate": 8.349842248147332e-06,
+      "loss": 4.066005859375,
+      "step": 40250
+    },
+    {
+      "epoch": 0.9737533601135643,
+      "grad_norm": 1.0599485635757446,
+      "learning_rate": 7.982977474502899e-06,
+      "loss": 4.073214111328125,
+      "step": 40300
+    },
+    {
+      "epoch": 0.9749614908333082,
+      "grad_norm": 1.0399075746536255,
+      "learning_rate": 7.616112700858463e-06,
+      "loss": 4.0739743041992185,
+      "step": 40350
+    },
+    {
+      "epoch": 0.9761696215530521,
+      "grad_norm": 1.0088541507720947,
+      "learning_rate": 7.249247927214028e-06,
+      "loss": 4.066875915527344,
+      "step": 40400
+    },
+    {
+      "epoch": 0.9773777522727959,
+      "grad_norm": 1.0096062421798706,
+      "learning_rate": 6.882383153569594e-06,
+      "loss": 4.093489990234375,
+      "step": 40450
+    },
+    {
+      "epoch": 0.9785858829925398,
+      "grad_norm": 1.022308349609375,
+      "learning_rate": 6.5155183799251594e-06,
+      "loss": 4.046855773925781,
+      "step": 40500
+    },
+    {
+      "epoch": 0.9797940137122837,
+      "grad_norm": 0.9825040698051453,
+      "learning_rate": 6.148653606280724e-06,
+      "loss": 4.074610595703125,
+      "step": 40550
+    },
+    {
+      "epoch": 0.9810021444320276,
+      "grad_norm": Infinity,
+      "learning_rate": 5.78178883263629e-06,
+      "loss": 4.078878784179688,
+      "step": 40600
+    },
+    {
+      "epoch": 0.9822102751517714,
+      "grad_norm": 1.0147684812545776,
+      "learning_rate": 5.4149240589918555e-06,
+      "loss": 4.105263366699218,
+      "step": 40650
+    },
+    {
+      "epoch": 0.9834184058715153,
+      "grad_norm": 0.9963952302932739,
+      "learning_rate": 5.04805928534742e-06,
+      "loss": 4.120052795410157,
+      "step": 40700
+    },
+    {
+      "epoch": 0.9846265365912592,
+      "grad_norm": 0.995857298374176,
+      "learning_rate": 4.681194511702986e-06,
+      "loss": 4.0579986572265625,
+      "step": 40750
+    },
+    {
+      "epoch": 0.985834667311003,
+      "grad_norm": 1.0292693376541138,
+      "learning_rate": 4.314329738058551e-06,
+      "loss": 4.134707946777343,
+      "step": 40800
+    },
+    {
+      "epoch": 0.9870427980307469,
+      "grad_norm": 0.9979900121688843,
+      "learning_rate": 3.947464964414116e-06,
+      "loss": 4.068620300292968,
+      "step": 40850
+    },
+    {
+      "epoch": 0.9882509287504908,
+      "grad_norm": 0.9745947122573853,
+      "learning_rate": 3.5806001907696816e-06,
+      "loss": 4.043538818359375,
+      "step": 40900
+    },
+    {
+      "epoch": 0.9894590594702347,
+      "grad_norm": 0.9747163653373718,
+      "learning_rate": 3.2137354171252472e-06,
+      "loss": 4.085929260253907,
+      "step": 40950
+    },
+    {
+      "epoch": 0.9906671901899785,
+      "grad_norm": 1.0071576833724976,
+      "learning_rate": 2.846870643480813e-06,
+      "loss": 4.030368041992188,
+      "step": 41000
+    },
+    {
+      "epoch": 0.9918753209097224,
+      "grad_norm": 1.0262352228164673,
+      "learning_rate": 2.480005869836378e-06,
+      "loss": 4.061216430664063,
+      "step": 41050
+    },
+    {
+      "epoch": 0.9930834516294663,
+      "grad_norm": 1.0081514120101929,
+      "learning_rate": 2.1131410961919433e-06,
+      "loss": 4.090030212402343,
+      "step": 41100
+    },
+    {
+      "epoch": 0.9942915823492102,
+      "grad_norm": 1.001246452331543,
+      "learning_rate": 1.746276322547509e-06,
+      "loss": 4.078215637207031,
+      "step": 41150
+    },
+    {
+      "epoch": 0.995499713068954,
+      "grad_norm": 0.9826595187187195,
+      "learning_rate": 1.3794115489030742e-06,
+      "loss": 4.048282470703125,
+      "step": 41200
+    },
+    {
+      "epoch": 0.9967078437886979,
+      "grad_norm": 1.0446892976760864,
+      "learning_rate": 1.0125467752586396e-06,
+      "loss": 4.058524169921875,
+      "step": 41250
+    },
+    {
+      "epoch": 0.9979159745084418,
+      "grad_norm": 1.0285199880599976,
+      "learning_rate": 6.456820016142049e-07,
+      "loss": 4.080808410644531,
+      "step": 41300
+    },
+    {
+      "epoch": 0.9991241052281857,
+      "grad_norm": 1.0168330669403076,
+      "learning_rate": 2.788172279697703e-07,
+      "loss": 4.036456298828125,
+      "step": 41350
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 41387,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.6510917582848e+16,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}