diff --git "a/trainer_state.json" "b/trainer_state.json"
deleted file mode 100644--- "a/trainer_state.json"
+++ /dev/null
@@ -1,5882 +0,0 @@
-{
-  "best_global_step": null,
-  "best_metric": null,
-  "best_model_checkpoint": null,
-  "epoch": 0.10179290234018888,
-  "eval_steps": 500,
-  "global_step": 68000,
-  "is_hyper_param_search": false,
-  "is_local_process_zero": true,
-  "is_world_process_zero": true,
-  "log_history": [
-    {
-      "epoch": 0.00014969544461792484,
-      "grad_norm": 1.1115108728408813,
-      "learning_rate": 9.900000000000001e-05,
-      "loss": 3.3276715087890625,
-      "step": 100
-    },
-    {
-      "epoch": 0.0002993908892358497,
-      "grad_norm": 1.3854551315307617,
-      "learning_rate": 0.000199,
-      "loss": 2.8366998291015624,
-      "step": 200
-    },
-    {
-      "epoch": 0.0004490863338537745,
-      "grad_norm": 1.645733118057251,
-      "learning_rate": 0.00019999011911457285,
-      "loss": 2.780883483886719,
-      "step": 300
-    },
-    {
-      "epoch": 0.0005987817784716994,
-      "grad_norm": 1.6934341192245483,
-      "learning_rate": 0.00019998013842222222,
-      "loss": 2.7250958251953126,
-      "step": 400
-    },
-    {
-      "epoch": 0.0007484772230896241,
-      "grad_norm": 2.9580564498901367,
-      "learning_rate": 0.0001999701577298716,
-      "loss": 2.734931640625,
-      "step": 500
-    },
-    {
-      "epoch": 0.0007484772230896241,
-      "eval_loss": 2.538707733154297,
-      "eval_runtime": 3197.8212,
-      "eval_samples_per_second": 185.688,
-      "eval_steps_per_second": 23.211,
-      "step": 500
-    },
-    {
-      "epoch": 0.000898172667707549,
-      "grad_norm": 1.6100478172302246,
-      "learning_rate": 0.00019996017703752093,
-      "loss": 2.6798614501953124,
-      "step": 600
-    },
-    {
-      "epoch": 0.0010478681123254738,
-      "grad_norm": 1.4200081825256348,
-      "learning_rate": 0.00019995019634517027,
-      "loss": 2.6447329711914063,
-      "step": 700
-    },
-    {
-      "epoch": 0.0011975635569433987,
-      "grad_norm": 1.532351016998291,
-      "learning_rate": 0.00019994021565281964,
-      "loss": 2.698541259765625,
-      "step": 800
-    },
-    {
-      "epoch": 0.0013472590015613235,
-      "grad_norm": 1.4328550100326538,
-      "learning_rate": 0.000199930234960469,
-      "loss": 2.5617575073242187,
-      "step": 900
-    },
-    {
-      "epoch": 0.0014969544461792483,
-      "grad_norm": 1.7113161087036133,
-      "learning_rate": 0.00019992025426811833,
-      "loss": 2.610606689453125,
-      "step": 1000
-    },
-    {
-      "epoch": 0.0014969544461792483,
-      "eval_loss": 2.4632270336151123,
-      "eval_runtime": 3199.5106,
-      "eval_samples_per_second": 185.59,
-      "eval_steps_per_second": 23.199,
-      "step": 1000
-    },
-    {
-      "epoch": 0.0016466498907971733,
-      "grad_norm": 1.4374314546585083,
-      "learning_rate": 0.00019991027357576767,
-      "loss": 2.5760433959960936,
-      "step": 1100
-    },
-    {
-      "epoch": 0.001796345335415098,
-      "grad_norm": 1.6222004890441895,
-      "learning_rate": 0.00019990029288341702,
-      "loss": 2.5517782592773437,
-      "step": 1200
-    },
-    {
-      "epoch": 0.0019460407800330228,
-      "grad_norm": 1.1975346803665161,
-      "learning_rate": 0.00019989031219106639,
-      "loss": 2.639231872558594,
-      "step": 1300
-    },
-    {
-      "epoch": 0.0020957362246509475,
-      "grad_norm": 1.5199350118637085,
-      "learning_rate": 0.00019988033149871576,
-      "loss": 2.6038848876953127,
-      "step": 1400
-    },
-    {
-      "epoch": 0.0022454316692688725,
-      "grad_norm": 1.8327330350875854,
-      "learning_rate": 0.0001998703508063651,
-      "loss": 2.5578712463378905,
-      "step": 1500
-    },
-    {
-      "epoch": 0.0022454316692688725,
-      "eval_loss": 2.401022434234619,
-      "eval_runtime": 3208.4297,
-      "eval_samples_per_second": 185.074,
-      "eval_steps_per_second": 23.134,
-      "step": 1500
-    },
-    {
-      "epoch": 0.0023951271138867975,
-      "grad_norm": 1.5121567249298096,
-      "learning_rate": 0.00019986037011401444,
-      "loss": 2.543453369140625,
-      "step": 1600
-    },
-    {
-      "epoch": 0.002544822558504722,
-      "grad_norm": 1.4418026208877563,
-      "learning_rate": 0.0001998503894216638,
-      "loss": 2.5406040954589844,
-      "step": 1700
-    },
-    {
-      "epoch": 0.002694518003122647,
-      "grad_norm": 1.3794381618499756,
-      "learning_rate": 0.00019984040872931316,
-      "loss": 2.5646536254882815,
-      "step": 1800
-    },
-    {
-      "epoch": 0.002844213447740572,
-      "grad_norm": 1.37217116355896,
-      "learning_rate": 0.0001998304280369625,
-      "loss": 2.503768005371094,
-      "step": 1900
-    },
-    {
-      "epoch": 0.0029939088923584965,
-      "grad_norm": 1.5015102624893188,
-      "learning_rate": 0.00019982044734461184,
-      "loss": 2.4893907165527343,
-      "step": 2000
-    },
-    {
-      "epoch": 0.0029939088923584965,
-      "eval_loss": 2.3696115016937256,
-      "eval_runtime": 3212.2817,
-      "eval_samples_per_second": 184.852,
-      "eval_steps_per_second": 23.107,
-      "step": 2000
-    },
-    {
-      "epoch": 0.0031436043369764215,
-      "grad_norm": 1.8465603590011597,
-      "learning_rate": 0.0001998104666522612,
-      "loss": 2.5316835021972657,
-      "step": 2100
-    },
-    {
-      "epoch": 0.0032932997815943465,
-      "grad_norm": 2.1077725887298584,
-      "learning_rate": 0.00019980048595991055,
-      "loss": 2.4913140869140626,
-      "step": 2200
-    },
-    {
-      "epoch": 0.003442995226212271,
-      "grad_norm": 1.8537485599517822,
-      "learning_rate": 0.00019979050526755992,
-      "loss": 2.5869821166992186,
-      "step": 2300
-    },
-    {
-      "epoch": 0.003592690670830196,
-      "grad_norm": 1.628941297531128,
-      "learning_rate": 0.00019978052457520927,
-      "loss": 2.5276083374023437,
-      "step": 2400
-    },
-    {
-      "epoch": 0.003742386115448121,
-      "grad_norm": 1.477745771408081,
-      "learning_rate": 0.00019977054388285864,
-      "loss": 2.5014384460449217,
-      "step": 2500
-    },
-    {
-      "epoch": 0.003742386115448121,
-      "eval_loss": 2.345714807510376,
-      "eval_runtime": 3214.6242,
-      "eval_samples_per_second": 184.718,
-      "eval_steps_per_second": 23.09,
-      "step": 2500
-    },
-    {
-      "epoch": 0.0038920815600660456,
-      "grad_norm": 1.3588407039642334,
-      "learning_rate": 0.00019976056319050798,
-      "loss": 2.457741546630859,
-      "step": 2600
-    },
-    {
-      "epoch": 0.00404177700468397,
-      "grad_norm": 1.5070505142211914,
-      "learning_rate": 0.00019975058249815732,
-      "loss": 2.4692388916015626,
-      "step": 2700
-    },
-    {
-      "epoch": 0.004191472449301895,
-      "grad_norm": 1.2107853889465332,
-      "learning_rate": 0.00019974060180580667,
-      "loss": 2.4899961853027346,
-      "step": 2800
-    },
-    {
-      "epoch": 0.00434116789391982,
-      "grad_norm": 1.3776215314865112,
-      "learning_rate": 0.000199730621113456,
-      "loss": 2.5087544250488283,
-      "step": 2900
-    },
-    {
-      "epoch": 0.004490863338537745,
-      "grad_norm": 1.4290128946304321,
-      "learning_rate": 0.00019972064042110538,
-      "loss": 2.516739959716797,
-      "step": 3000
-    },
-    {
-      "epoch": 0.004490863338537745,
-      "eval_loss": 2.320899724960327,
-      "eval_runtime": 3212.475,
-      "eval_samples_per_second": 184.841,
-      "eval_steps_per_second": 23.105,
-      "step": 3000
-    },
-    {
-      "epoch": 0.00464055878315567,
-      "grad_norm": 1.6739652156829834,
-      "learning_rate": 0.00019971065972875472,
-      "loss": 2.4414678955078126,
-      "step": 3100
-    },
-    {
-      "epoch": 0.004790254227773595,
-      "grad_norm": 1.5562019348144531,
-      "learning_rate": 0.0001997006790364041,
-      "loss": 2.500360870361328,
-      "step": 3200
-    },
-    {
-      "epoch": 0.004939949672391519,
-      "grad_norm": 1.6588985919952393,
-      "learning_rate": 0.00019969069834405344,
-      "loss": 2.4128042602539064,
-      "step": 3300
-    },
-    {
-      "epoch": 0.005089645117009444,
-      "grad_norm": 1.6403768062591553,
-      "learning_rate": 0.0001996807176517028,
-      "loss": 2.4562588500976563,
-      "step": 3400
-    },
-    {
-      "epoch": 0.005239340561627369,
-      "grad_norm": 1.383549690246582,
-      "learning_rate": 0.00019967073695935215,
-      "loss": 2.470595245361328,
-      "step": 3500
-    },
-    {
-      "epoch": 0.005239340561627369,
-      "eval_loss": 2.2999842166900635,
-      "eval_runtime": 3212.6656,
-      "eval_samples_per_second": 184.83,
-      "eval_steps_per_second": 23.104,
-      "step": 3500
-    },
-    {
-      "epoch": 0.005389036006245294,
-      "grad_norm": 1.445297360420227,
-      "learning_rate": 0.0001996607562670015,
-      "loss": 2.45697509765625,
-      "step": 3600
-    },
-    {
-      "epoch": 0.005538731450863219,
-      "grad_norm": 1.5831618309020996,
-      "learning_rate": 0.00019965077557465084,
-      "loss": 2.4563037109375,
-      "step": 3700
-    },
-    {
-      "epoch": 0.005688426895481144,
-      "grad_norm": 1.5653294324874878,
-      "learning_rate": 0.0001996407948823002,
-      "loss": 2.479073791503906,
-      "step": 3800
-    },
-    {
-      "epoch": 0.005838122340099068,
-      "grad_norm": 1.6026147603988647,
-      "learning_rate": 0.00019963081418994955,
-      "loss": 2.4385037231445312,
-      "step": 3900
-    },
-    {
-      "epoch": 0.005987817784716993,
-      "grad_norm": 1.4271267652511597,
-      "learning_rate": 0.0001996208334975989,
-      "loss": 2.4400352478027343,
-      "step": 4000
-    },
-    {
-      "epoch": 0.005987817784716993,
-      "eval_loss": 2.289179563522339,
-      "eval_runtime": 3217.2854,
-      "eval_samples_per_second": 184.565,
-      "eval_steps_per_second": 23.071,
-      "step": 4000
-    },
-    {
-      "epoch": 0.006137513229334918,
-      "grad_norm": 1.452820897102356,
-      "learning_rate": 0.00019961085280524826,
-      "loss": 2.440526428222656,
-      "step": 4100
-    },
-    {
-      "epoch": 0.006287208673952843,
-      "grad_norm": 1.6142672300338745,
-      "learning_rate": 0.00019960087211289763,
-      "loss": 2.406920623779297,
-      "step": 4200
-    },
-    {
-      "epoch": 0.006436904118570768,
-      "grad_norm": 1.7030407190322876,
-      "learning_rate": 0.00019959089142054697,
-      "loss": 2.431551055908203,
-      "step": 4300
-    },
-    {
-      "epoch": 0.006586599563188693,
-      "grad_norm": 1.6218478679656982,
-      "learning_rate": 0.00019958091072819632,
-      "loss": 2.4045729064941406,
-      "step": 4400
-    },
-    {
-      "epoch": 0.006736295007806617,
-      "grad_norm": 1.2772244215011597,
-      "learning_rate": 0.00019957093003584566,
-      "loss": 2.37245361328125,
-      "step": 4500
-    },
-    {
-      "epoch": 0.006736295007806617,
-      "eval_loss": 2.276613712310791,
-      "eval_runtime": 3208.9155,
-      "eval_samples_per_second": 185.046,
-      "eval_steps_per_second": 23.131,
-      "step": 4500
-    },
-    {
-      "epoch": 0.006885990452424542,
-      "grad_norm": 1.4533907175064087,
-      "learning_rate": 0.00019956094934349503,
-      "loss": 2.384718017578125,
-      "step": 4600
-    },
-    {
-      "epoch": 0.007035685897042467,
-      "grad_norm": 1.6579569578170776,
-      "learning_rate": 0.00019955096865114437,
-      "loss": 2.4032658386230468,
-      "step": 4700
-    },
-    {
-      "epoch": 0.007185381341660392,
-      "grad_norm": 1.6232718229293823,
-      "learning_rate": 0.00019954098795879372,
-      "loss": 2.4093026733398437,
-      "step": 4800
-    },
-    {
-      "epoch": 0.007335076786278317,
-      "grad_norm": 1.5691031217575073,
-      "learning_rate": 0.00019953100726644306,
-      "loss": 2.398963623046875,
-      "step": 4900
-    },
-    {
-      "epoch": 0.007484772230896242,
-      "grad_norm": 1.9691098928451538,
-      "learning_rate": 0.00019952102657409243,
-      "loss": 2.4387384033203126,
-      "step": 5000
-    },
-    {
-      "epoch": 0.007484772230896242,
-      "eval_loss": 2.264145851135254,
-      "eval_runtime": 3217.1692,
-      "eval_samples_per_second": 184.572,
-      "eval_steps_per_second": 23.072,
-      "step": 5000
-    },
-    {
-      "epoch": 0.007634467675514166,
-      "grad_norm": 1.7022979259490967,
-      "learning_rate": 0.0001995110458817418,
-      "loss": 2.4515452575683594,
-      "step": 5100
-    },
-    {
-      "epoch": 0.007784163120132091,
-      "grad_norm": 1.564897060394287,
-      "learning_rate": 0.00019950106518939114,
-      "loss": 2.408607940673828,
-      "step": 5200
-    },
-    {
-      "epoch": 0.007933858564750016,
-      "grad_norm": 1.201539158821106,
-      "learning_rate": 0.00019949108449704049,
-      "loss": 2.3884033203125,
-      "step": 5300
-    },
-    {
-      "epoch": 0.00808355400936794,
-      "grad_norm": 1.3706908226013184,
-      "learning_rate": 0.00019948110380468983,
-      "loss": 2.382029266357422,
-      "step": 5400
-    },
-    {
-      "epoch": 0.008233249453985866,
-      "grad_norm": 1.551080346107483,
-      "learning_rate": 0.0001994711231123392,
-      "loss": 2.367112731933594,
-      "step": 5500
-    },
-    {
-      "epoch": 0.008233249453985866,
-      "eval_loss": 2.2458696365356445,
-      "eval_runtime": 3223.0014,
-      "eval_samples_per_second": 184.238,
-      "eval_steps_per_second": 23.03,
-      "step": 5500
-    },
-    {
-      "epoch": 0.00838294489860379,
-      "grad_norm": 1.3347584009170532,
-      "learning_rate": 0.00019946114241998854,
-      "loss": 2.378773651123047,
-      "step": 5600
-    },
-    {
-      "epoch": 0.008532640343221716,
-      "grad_norm": 1.5166795253753662,
-      "learning_rate": 0.00019945116172763789,
-      "loss": 2.373096923828125,
-      "step": 5700
-    },
-    {
-      "epoch": 0.00868233578783964,
-      "grad_norm": 1.3144797086715698,
-      "learning_rate": 0.00019944118103528723,
-      "loss": 2.29150146484375,
-      "step": 5800
-    },
-    {
-      "epoch": 0.008832031232457564,
-      "grad_norm": 1.5164883136749268,
-      "learning_rate": 0.0001994312003429366,
-      "loss": 2.4362031555175783,
-      "step": 5900
-    },
-    {
-      "epoch": 0.00898172667707549,
-      "grad_norm": 1.8221055269241333,
-      "learning_rate": 0.00019942121965058597,
-      "loss": 2.3431991577148437,
-      "step": 6000
-    },
-    {
-      "epoch": 0.00898172667707549,
-      "eval_loss": 2.238246440887451,
-      "eval_runtime": 3225.6957,
-      "eval_samples_per_second": 184.084,
-      "eval_steps_per_second": 23.011,
-      "step": 6000
-    },
-    {
-      "epoch": 0.009131422121693414,
-      "grad_norm": 1.1972438097000122,
-      "learning_rate": 0.0001994112389582353,
-      "loss": 2.376190185546875,
-      "step": 6100
-    },
-    {
-      "epoch": 0.00928111756631134,
-      "grad_norm": 1.4378846883773804,
-      "learning_rate": 0.00019940125826588465,
-      "loss": 2.366408386230469,
-      "step": 6200
-    },
-    {
-      "epoch": 0.009430813010929264,
-      "grad_norm": 1.5107417106628418,
-      "learning_rate": 0.00019939127757353402,
-      "loss": 2.4345634460449217,
-      "step": 6300
-    },
-    {
-      "epoch": 0.00958050845554719,
-      "grad_norm": 1.4438202381134033,
-      "learning_rate": 0.00019938129688118337,
-      "loss": 2.3712887573242187,
-      "step": 6400
-    },
-    {
-      "epoch": 0.009730203900165114,
-      "grad_norm": 1.3664839267730713,
-      "learning_rate": 0.0001993713161888327,
-      "loss": 2.366647186279297,
-      "step": 6500
-    },
-    {
-      "epoch": 0.009730203900165114,
-      "eval_loss": 2.228607177734375,
-      "eval_runtime": 3226.9137,
-      "eval_samples_per_second": 184.014,
-      "eval_steps_per_second": 23.002,
-      "step": 6500
-    },
-    {
-      "epoch": 0.009879899344783038,
-      "grad_norm": 1.7177119255065918,
-      "learning_rate": 0.00019936133549648205,
-      "loss": 2.3463206481933594,
-      "step": 6600
-    },
-    {
-      "epoch": 0.010029594789400964,
-      "grad_norm": 1.2005643844604492,
-      "learning_rate": 0.0001993513548041314,
-      "loss": 2.415400390625,
-      "step": 6700
-    },
-    {
-      "epoch": 0.010179290234018888,
-      "grad_norm": 1.3096752166748047,
-      "learning_rate": 0.00019934137411178077,
-      "loss": 2.388221893310547,
-      "step": 6800
-    },
-    {
-      "epoch": 0.010328985678636814,
-      "grad_norm": 1.6408536434173584,
-      "learning_rate": 0.0001993313934194301,
-      "loss": 2.345827331542969,
-      "step": 6900
-    },
-    {
-      "epoch": 0.010478681123254738,
-      "grad_norm": 1.291214942932129,
-      "learning_rate": 0.00019932141272707948,
-      "loss": 2.2995399475097655,
-      "step": 7000
-    },
-    {
-      "epoch": 0.010478681123254738,
-      "eval_loss": 2.220872402191162,
-      "eval_runtime": 3226.9057,
-      "eval_samples_per_second": 184.015,
-      "eval_steps_per_second": 23.002,
-      "step": 7000
-    },
-    {
-      "epoch": 0.010628376567872662,
-      "grad_norm": 1.4510313272476196,
-      "learning_rate": 0.00019931143203472882,
-      "loss": 2.4538751220703126,
-      "step": 7100
-    },
-    {
-      "epoch": 0.010778072012490588,
-      "grad_norm": 1.4856514930725098,
-      "learning_rate": 0.0001993014513423782,
-      "loss": 2.3923910522460936,
-      "step": 7200
-    },
-    {
-      "epoch": 0.010927767457108512,
-      "grad_norm": 1.5649104118347168,
-      "learning_rate": 0.00019929147065002754,
-      "loss": 2.3877908325195314,
-      "step": 7300
-    },
-    {
-      "epoch": 0.011077462901726438,
-      "grad_norm": 1.5636347532272339,
-      "learning_rate": 0.00019928148995767688,
-      "loss": 2.307393035888672,
-      "step": 7400
-    },
-    {
-      "epoch": 0.011227158346344362,
-      "grad_norm": 1.433748483657837,
-      "learning_rate": 0.00019927150926532622,
-      "loss": 2.4335588073730468,
-      "step": 7500
-    },
-    {
-      "epoch": 0.011227158346344362,
-      "eval_loss": 2.21183180809021,
-      "eval_runtime": 3223.4234,
-      "eval_samples_per_second": 184.213,
-      "eval_steps_per_second": 23.027,
-      "step": 7500
-    },
-    {
-      "epoch": 0.011376853790962288,
-      "grad_norm": 1.5009076595306396,
-      "learning_rate": 0.0001992615285729756,
-      "loss": 2.361067352294922,
-      "step": 7600
-    },
-    {
-      "epoch": 0.011526549235580212,
-      "grad_norm": 1.614925742149353,
-      "learning_rate": 0.00019925154788062494,
-      "loss": 2.3444528198242187,
-      "step": 7700
-    },
-    {
-      "epoch": 0.011676244680198136,
-      "grad_norm": 1.5763992071151733,
-      "learning_rate": 0.00019924156718827428,
-      "loss": 2.393105773925781,
-      "step": 7800
-    },
-    {
-      "epoch": 0.011825940124816062,
-      "grad_norm": 1.141770362854004,
-      "learning_rate": 0.00019923158649592365,
-      "loss": 2.3527407836914063,
-      "step": 7900
-    },
-    {
-      "epoch": 0.011975635569433986,
-      "grad_norm": 1.4571629762649536,
-      "learning_rate": 0.00019922160580357302,
-      "loss": 2.3546812438964846,
-      "step": 8000
-    },
-    {
-      "epoch": 0.011975635569433986,
-      "eval_loss": 2.206197738647461,
-      "eval_runtime": 3230.9521,
-      "eval_samples_per_second": 183.784,
-      "eval_steps_per_second": 22.973,
-      "step": 8000
-    },
-    {
-      "epoch": 0.012125331014051912,
-      "grad_norm": 1.7330505847930908,
-      "learning_rate": 0.00019921162511122236,
-      "loss": 2.3846061706542967,
-      "step": 8100
-    },
-    {
-      "epoch": 0.012275026458669836,
-      "grad_norm": 1.382493495941162,
-      "learning_rate": 0.0001992016444188717,
-      "loss": 2.338687744140625,
-      "step": 8200
-    },
-    {
-      "epoch": 0.01242472190328776,
-      "grad_norm": 1.9080495834350586,
-      "learning_rate": 0.00019919166372652105,
-      "loss": 2.3646954345703124,
-      "step": 8300
-    },
-    {
-      "epoch": 0.012574417347905686,
-      "grad_norm": 1.598596215248108,
-      "learning_rate": 0.0001991816830341704,
-      "loss": 2.404818420410156,
-      "step": 8400
-    },
-    {
-      "epoch": 0.01272411279252361,
-      "grad_norm": 1.3987079858779907,
-      "learning_rate": 0.00019917170234181976,
-      "loss": 2.316581268310547,
-      "step": 8500
-    },
-    {
-      "epoch": 0.01272411279252361,
-      "eval_loss": 2.2019903659820557,
-      "eval_runtime": 3244.2805,
-      "eval_samples_per_second": 183.029,
-      "eval_steps_per_second": 22.879,
-      "step": 8500
-    },
-    {
-      "epoch": 0.012873808237141536,
-      "grad_norm": 1.7449572086334229,
-      "learning_rate": 0.0001991617216494691,
-      "loss": 2.3456092834472657,
-      "step": 8600
-    },
-    {
-      "epoch": 0.01302350368175946,
-      "grad_norm": 1.593859314918518,
-      "learning_rate": 0.00019915174095711845,
-      "loss": 2.3116250610351563,
-      "step": 8700
-    },
-    {
-      "epoch": 0.013173199126377386,
-      "grad_norm": 1.9670217037200928,
-      "learning_rate": 0.00019914176026476782,
-      "loss": 2.3385975646972654,
-      "step": 8800
-    },
-    {
-      "epoch": 0.01332289457099531,
-      "grad_norm": 1.2880886793136597,
-      "learning_rate": 0.0001991317795724172,
-      "loss": 2.3421080017089846,
-      "step": 8900
-    },
-    {
-      "epoch": 0.013472590015613234,
-      "grad_norm": 1.8144607543945312,
-      "learning_rate": 0.00019912179888006653,
-      "loss": 2.3279385375976562,
-      "step": 9000
-    },
-    {
-      "epoch": 0.013472590015613234,
-      "eval_loss": 2.1911423206329346,
-      "eval_runtime": 3235.6908,
-      "eval_samples_per_second": 183.515,
-      "eval_steps_per_second": 22.939,
-      "step": 9000
-    },
-    {
-      "epoch": 0.01362228546023116,
-      "grad_norm": 1.3210985660552979,
-      "learning_rate": 0.00019911181818771587,
-      "loss": 2.358592834472656,
-      "step": 9100
-    },
-    {
-      "epoch": 0.013771980904849084,
-      "grad_norm": 1.4912389516830444,
-      "learning_rate": 0.00019910183749536522,
-      "loss": 2.3899864196777343,
-      "step": 9200
-    },
-    {
-      "epoch": 0.01392167634946701,
-      "grad_norm": 1.5113123655319214,
-      "learning_rate": 0.0001990918568030146,
-      "loss": 2.3109588623046875,
-      "step": 9300
-    },
-    {
-      "epoch": 0.014071371794084934,
-      "grad_norm": 1.5983033180236816,
-      "learning_rate": 0.00019908187611066393,
-      "loss": 2.4011131286621095,
-      "step": 9400
-    },
-    {
-      "epoch": 0.014221067238702858,
-      "grad_norm": 1.591844916343689,
-      "learning_rate": 0.00019907189541831327,
-      "loss": 2.2923405456542967,
-      "step": 9500
-    },
-    {
-      "epoch": 0.014221067238702858,
-      "eval_loss": 2.1804819107055664,
-      "eval_runtime": 3240.6579,
-      "eval_samples_per_second": 183.234,
-      "eval_steps_per_second": 22.904,
-      "step": 9500
-    },
-    {
-      "epoch": 0.014370762683320784,
-      "grad_norm": 1.6611961126327515,
-      "learning_rate": 0.00019906191472596262,
-      "loss": 2.3377178955078124,
-      "step": 9600
-    },
-    {
-      "epoch": 0.014520458127938708,
-      "grad_norm": 1.3548938035964966,
-      "learning_rate": 0.00019905193403361199,
-      "loss": 2.284178924560547,
-      "step": 9700
-    },
-    {
-      "epoch": 0.014670153572556634,
-      "grad_norm": 1.5596938133239746,
-      "learning_rate": 0.00019904195334126136,
-      "loss": 2.37065185546875,
-      "step": 9800
-    },
-    {
-      "epoch": 0.014819849017174558,
-      "grad_norm": 1.8031177520751953,
-      "learning_rate": 0.0001990319726489107,
-      "loss": 2.327956237792969,
-      "step": 9900
-    },
-    {
-      "epoch": 0.014969544461792484,
-      "grad_norm": 1.4145318269729614,
-      "learning_rate": 0.00019902199195656004,
-      "loss": 2.295563201904297,
-      "step": 10000
-    },
-    {
-      "epoch": 0.014969544461792484,
-      "eval_loss": 2.178765296936035,
-      "eval_runtime": 3270.7576,
-      "eval_samples_per_second": 181.548,
-      "eval_steps_per_second": 22.694,
-      "step": 10000
-    },
-    {
-      "epoch": 0.015119239906410408,
-      "grad_norm": 1.6441259384155273,
-      "learning_rate": 0.00019901201126420938,
-      "loss": 2.2758912658691406,
-      "step": 10100
-    },
-    {
-      "epoch": 0.015268935351028332,
-      "grad_norm": 1.4381738901138306,
-      "learning_rate": 0.00019900203057185875,
-      "loss": 2.297754364013672,
-      "step": 10200
-    },
-    {
-      "epoch": 0.015418630795646258,
-      "grad_norm": 1.6563739776611328,
-      "learning_rate": 0.0001989920498795081,
-      "loss": 2.275712738037109,
-      "step": 10300
-    },
-    {
-      "epoch": 0.015568326240264182,
-      "grad_norm": 1.6161943674087524,
-      "learning_rate": 0.00019898206918715744,
-      "loss": 2.2912164306640626,
-      "step": 10400
-    },
-    {
-      "epoch": 0.015718021684882108,
-      "grad_norm": 1.5459507703781128,
-      "learning_rate": 0.00019897208849480678,
-      "loss": 2.3135377502441408,
-      "step": 10500
-    },
-    {
-      "epoch": 0.015718021684882108,
-      "eval_loss": 2.177732467651367,
-      "eval_runtime": 3255.6849,
-      "eval_samples_per_second": 182.388,
-      "eval_steps_per_second": 22.799,
-      "step": 10500
-    },
-    {
-      "epoch": 0.015867717129500032,
-      "grad_norm": 1.567585825920105,
-      "learning_rate": 0.00019896210780245615,
-      "loss": 2.3561952209472654,
-      "step": 10600
-    },
-    {
-      "epoch": 0.016017412574117956,
-      "grad_norm": 1.7034927606582642,
-      "learning_rate": 0.00019895212711010552,
-      "loss": 2.3578306579589845,
-      "step": 10700
-    },
-    {
-      "epoch": 0.01616710801873588,
-      "grad_norm": 1.6405142545700073,
-      "learning_rate": 0.00019894214641775487,
-      "loss": 2.3306727600097656,
-      "step": 10800
-    },
-    {
-      "epoch": 0.016316803463353808,
-      "grad_norm": 1.9965468645095825,
-      "learning_rate": 0.0001989321657254042,
-      "loss": 2.340437927246094,
-      "step": 10900
-    },
-    {
-      "epoch": 0.016466498907971732,
-      "grad_norm": 1.5367627143859863,
-      "learning_rate": 0.00019892218503305358,
-      "loss": 2.359449615478516,
-      "step": 11000
-    },
-    {
-      "epoch": 0.016466498907971732,
-      "eval_loss": 2.16288685798645,
-      "eval_runtime": 3255.9885,
-      "eval_samples_per_second": 182.371,
-      "eval_steps_per_second": 22.796,
-      "step": 11000
-    },
-    {
-      "epoch": 0.016616194352589656,
-      "grad_norm": 2.0562093257904053,
-      "learning_rate": 0.00019891220434070292,
-      "loss": 2.352833557128906,
-      "step": 11100
-    },
-    {
-      "epoch": 0.01676588979720758,
-      "grad_norm": 1.6274471282958984,
-      "learning_rate": 0.00019890222364835227,
-      "loss": 2.309020233154297,
-      "step": 11200
-    },
-    {
-      "epoch": 0.016915585241825504,
-      "grad_norm": 1.4910115003585815,
-      "learning_rate": 0.0001988922429560016,
-      "loss": 2.2657244873046873,
-      "step": 11300
-    },
-    {
-      "epoch": 0.017065280686443432,
-      "grad_norm": 1.563535213470459,
-      "learning_rate": 0.00019888226226365098,
-      "loss": 2.337887420654297,
-      "step": 11400
-    },
-    {
-      "epoch": 0.017214976131061356,
-      "grad_norm": 1.4178085327148438,
-      "learning_rate": 0.00019887228157130032,
-      "loss": 2.350980224609375,
-      "step": 11500
-    },
-    {
-      "epoch": 0.017214976131061356,
-      "eval_loss": 2.162635326385498,
-      "eval_runtime": 3273.721,
-      "eval_samples_per_second": 181.383,
-      "eval_steps_per_second": 22.673,
-      "step": 11500
-    },
-    {
-      "epoch": 0.01736467157567928,
-      "grad_norm": 1.3107730150222778,
-      "learning_rate": 0.0001988623008789497,
-      "loss": 2.2568296813964843,
-      "step": 11600
-    },
-    {
-      "epoch": 0.017514367020297204,
-      "grad_norm": 1.3061058521270752,
-      "learning_rate": 0.00019885232018659904,
-      "loss": 2.2581549072265625,
-      "step": 11700
-    },
-    {
-      "epoch": 0.01766406246491513,
-      "grad_norm": 1.6069859266281128,
-      "learning_rate": 0.00019884233949424838,
-      "loss": 2.3420458984375,
-      "step": 11800
-    },
-    {
-      "epoch": 0.017813757909533056,
-      "grad_norm": 1.3916401863098145,
-      "learning_rate": 0.00019883235880189775,
-      "loss": 2.311026306152344,
-      "step": 11900
-    },
-    {
-      "epoch": 0.01796345335415098,
-      "grad_norm": 1.0942401885986328,
-      "learning_rate": 0.0001988223781095471,
-      "loss": 2.3433848571777345,
-      "step": 12000
-    },
-    {
-      "epoch": 0.01796345335415098,
-      "eval_loss": 2.162487268447876,
-      "eval_runtime": 3276.4584,
-      "eval_samples_per_second": 181.232,
-      "eval_steps_per_second": 22.654,
-      "step": 12000
-    },
-    {
-      "epoch": 0.018113148798768904,
-      "grad_norm": 1.469489336013794,
-      "learning_rate": 0.00019881239741719643,
-      "loss": 2.3261029052734377,
-      "step": 12100
-    },
-    {
-      "epoch": 0.01826284424338683,
-      "grad_norm": 1.5263702869415283,
-      "learning_rate": 0.00019880241672484578,
-      "loss": 2.2827218627929686,
-      "step": 12200
-    },
-    {
-      "epoch": 0.018412539688004756,
-      "grad_norm": 1.849501132965088,
-      "learning_rate": 0.00019879243603249515,
-      "loss": 2.2822076416015626,
-      "step": 12300
-    },
-    {
-      "epoch": 0.01856223513262268,
-      "grad_norm": 1.294177770614624,
-      "learning_rate": 0.0001987824553401445,
-      "loss": 2.340493621826172,
-      "step": 12400
-    },
-    {
-      "epoch": 0.018711930577240604,
-      "grad_norm": 1.3849337100982666,
-      "learning_rate": 0.00019877247464779386,
-      "loss": 2.2688858032226564,
-      "step": 12500
-    },
-    {
-      "epoch": 0.018711930577240604,
-      "eval_loss": 2.15631103515625,
-      "eval_runtime": 3271.562,
-      "eval_samples_per_second": 181.503,
-      "eval_steps_per_second": 22.688,
-      "step": 12500
-    },
-    {
-      "epoch": 0.01886162602185853,
-      "grad_norm": 1.580287218093872,
-      "learning_rate": 0.0001987624939554432,
-      "loss": 2.37344970703125,
-      "step": 12600
-    },
-    {
-      "epoch": 0.019011321466476452,
-      "grad_norm": 1.4201791286468506,
-      "learning_rate": 0.00019875251326309257,
-      "loss": 2.2801904296875,
-      "step": 12700
-    },
-    {
-      "epoch": 0.01916101691109438,
-      "grad_norm": 1.5097748041152954,
-      "learning_rate": 0.00019874253257074192,
-      "loss": 2.2633958435058594,
-      "step": 12800
-    },
-    {
-      "epoch": 0.019310712355712304,
-      "grad_norm": 1.826706051826477,
-      "learning_rate": 0.00019873255187839126,
-      "loss": 2.2710801696777345,
-      "step": 12900
-    },
-    {
-      "epoch": 0.019460407800330228,
-      "grad_norm": 1.7749167680740356,
-      "learning_rate": 0.0001987225711860406,
-      "loss": 2.2632843017578126,
-      "step": 13000
-    },
-    {
-      "epoch": 0.019460407800330228,
-      "eval_loss": 2.1542553901672363,
-      "eval_runtime": 3286.5726,
-      "eval_samples_per_second": 180.674,
-      "eval_steps_per_second": 22.584,
-      "step": 13000
-    },
-    {
-      "epoch": 0.019610103244948152,
-      "grad_norm": 1.7070262432098389,
-      "learning_rate": 0.00019871259049368997,
-      "loss": 2.3246585083007814,
-      "step": 13100
-    },
-    {
-      "epoch": 0.019759798689566076,
-      "grad_norm": 1.848034381866455,
-      "learning_rate": 0.00019870260980133932,
-      "loss": 2.2999032592773436,
-      "step": 13200
-    },
-    {
-      "epoch": 0.019909494134184004,
-      "grad_norm": 1.549074411392212,
-      "learning_rate": 0.00019869262910898866,
-      "loss": 2.2773204040527344,
-      "step": 13300
-    },
-    {
-      "epoch": 0.020059189578801928,
-      "grad_norm": 1.699357032775879,
-      "learning_rate": 0.00019868264841663803,
-      "loss": 2.3312359619140626,
-      "step": 13400
-    },
-    {
-      "epoch": 0.020208885023419852,
-      "grad_norm": 1.3268290758132935,
-      "learning_rate": 0.00019867266772428737,
-      "loss": 2.240000762939453,
-      "step": 13500
-    },
-    {
-      "epoch": 0.020208885023419852,
-      "eval_loss": 2.150346040725708,
-      "eval_runtime": 3280.9844,
-      "eval_samples_per_second": 180.982,
-      "eval_steps_per_second": 22.623,
-      "step": 13500
-    },
-    {
-      "epoch": 0.020358580468037776,
-      "grad_norm": 1.7357659339904785,
-      "learning_rate": 0.00019866268703193674,
-      "loss": 2.322742919921875,
-      "step": 13600
-    },
-    {
-      "epoch": 0.0205082759126557,
-      "grad_norm": 1.465851068496704,
-      "learning_rate": 0.00019865270633958609,
-      "loss": 2.2868856811523437,
-      "step": 13700
-    },
-    {
-      "epoch": 0.020657971357273628,
-      "grad_norm": 1.5951443910598755,
-      "learning_rate": 0.00019864272564723543,
-      "loss": 2.2551199340820314,
-      "step": 13800
-    },
-    {
-      "epoch": 0.020807666801891552,
-      "grad_norm": 1.4535773992538452,
-      "learning_rate": 0.00019863274495488477,
-      "loss": 2.3290165710449218,
-      "step": 13900
-    },
-    {
-      "epoch": 0.020957362246509476,
-      "grad_norm": 1.3705129623413086,
-      "learning_rate": 0.00019862276426253414,
-      "loss": 2.278531799316406,
-      "step": 14000
-    },
-    {
-      "epoch": 0.020957362246509476,
-      "eval_loss": 2.1418333053588867,
-      "eval_runtime": 3296.6662,
-      "eval_samples_per_second": 180.121,
-      "eval_steps_per_second": 22.515,
-      "step": 14000
-    },
-    {
-      "epoch": 0.0211070576911274,
-      "grad_norm": 1.5736688375473022,
-      "learning_rate": 0.00019861278357018348,
-      "loss": 2.248696594238281,
-      "step": 14100
-    },
-    {
-      "epoch": 0.021256753135745324,
-      "grad_norm": 1.297442078590393,
-      "learning_rate": 0.00019860280287783283,
-      "loss": 2.316143798828125,
-      "step": 14200
-    },
-    {
-      "epoch": 0.021406448580363252,
-      "grad_norm": 1.5747723579406738,
-      "learning_rate": 0.0001985928221854822,
-      "loss": 2.273143615722656,
-      "step": 14300
-    },
-    {
-      "epoch": 0.021556144024981176,
-      "grad_norm": 1.3227375745773315,
-      "learning_rate": 0.00019858284149313157,
-      "loss": 2.294857635498047,
-      "step": 14400
-    },
-    {
-      "epoch": 0.0217058394695991,
-      "grad_norm": 1.4544196128845215,
-      "learning_rate": 0.0001985728608007809,
-      "loss": 2.3072767639160157,
-      "step": 14500
-    },
-    {
-      "epoch": 0.0217058394695991,
-      "eval_loss": 2.1366453170776367,
-      "eval_runtime": 3281.9912,
-      "eval_samples_per_second": 180.926,
-      "eval_steps_per_second": 22.616,
-      "step": 14500
-    },
-    {
-      "epoch": 0.021855534914217024,
-      "grad_norm": 1.3791955709457397,
-      "learning_rate": 0.00019856288010843025,
-      "loss": 2.3096559143066404,
-      "step": 14600
-    },
-    {
-      "epoch": 0.022005230358834952,
-      "grad_norm": 1.3311723470687866,
-      "learning_rate": 0.0001985528994160796,
-      "loss": 2.3028509521484377,
-      "step": 14700
-    },
-    {
-      "epoch": 0.022154925803452876,
-      "grad_norm": 1.7677994966506958,
-      "learning_rate": 0.00019854291872372897,
-      "loss": 2.2959837341308593,
-      "step": 14800
-    },
-    {
-      "epoch": 0.0223046212480708,
-      "grad_norm": 1.5415525436401367,
-      "learning_rate": 0.0001985329380313783,
-      "loss": 2.283103790283203,
-      "step": 14900
-    },
-    {
-      "epoch": 0.022454316692688724,
-      "grad_norm": 1.2594122886657715,
-      "learning_rate": 0.00019852295733902765,
-      "loss": 2.3055873107910156,
-      "step": 15000
-    },
-    {
-      "epoch": 0.022454316692688724,
-      "eval_loss": 2.138606548309326,
-      "eval_runtime": 3288.0122,
-      "eval_samples_per_second": 180.595,
-      "eval_steps_per_second": 22.574,
-      "step": 15000
-    },
-    {
-      "epoch": 0.02260401213730665,
-      "grad_norm": 1.474989891052246,
-      "learning_rate": 0.000198512976646677,
-      "loss": 2.261269836425781,
-      "step": 15100
-    },
-    {
-      "epoch": 0.022753707581924576,
-      "grad_norm": 1.6410372257232666,
-      "learning_rate": 0.00019850299595432637,
-      "loss": 2.3021755981445313,
-      "step": 15200
-    },
-    {
-      "epoch": 0.0229034030265425,
-      "grad_norm": 1.3945280313491821,
-      "learning_rate": 0.00019849301526197574,
-      "loss": 2.3544769287109375,
-      "step": 15300
-    },
-    {
-      "epoch": 0.023053098471160424,
-      "grad_norm": 1.4254323244094849,
-      "learning_rate": 0.00019848303456962508,
-      "loss": 2.3180072021484377,
-      "step": 15400
-    },
-    {
-      "epoch": 0.02320279391577835,
-      "grad_norm": 2.2179505825042725,
-      "learning_rate": 0.00019847305387727442,
-      "loss": 2.2615757751464844,
-      "step": 15500
-    },
-    {
-      "epoch": 0.02320279391577835,
-      "eval_loss": 2.1290035247802734,
-      "eval_runtime": 3300.9314,
-      "eval_samples_per_second": 179.888,
-      "eval_steps_per_second": 22.486,
-      "step": 15500
-    },
-    {
-      "epoch": 0.023352489360396272,
-      "grad_norm": 1.379185676574707,
-      "learning_rate": 0.00019846307318492377,
-      "loss": 2.222925109863281,
-      "step": 15600
-    },
-    {
-      "epoch": 0.0235021848050142,
-      "grad_norm": 1.447279691696167,
-      "learning_rate": 0.00019845309249257314,
-      "loss": 2.2476287841796876,
-      "step": 15700
-    },
-    {
-      "epoch": 0.023651880249632124,
-      "grad_norm": 1.5554574728012085,
-      "learning_rate": 0.00019844311180022248,
-      "loss": 2.267190246582031,
-      "step": 15800
-    },
-    {
-      "epoch": 0.023801575694250048,
-      "grad_norm": 1.4881314039230347,
-      "learning_rate": 0.00019843313110787182,
-      "loss": 2.3377020263671877,
-      "step": 15900
-    },
-    {
-      "epoch": 0.023951271138867972,
-      "grad_norm": 1.9818658828735352,
-      "learning_rate": 0.00019842315041552116,
-      "loss": 2.2596757507324217,
-      "step": 16000
-    },
-    {
-      "epoch": 0.023951271138867972,
-      "eval_loss": 2.12862491607666,
-      "eval_runtime": 3309.5861,
-      "eval_samples_per_second": 179.418,
-      "eval_steps_per_second": 22.427,
-      "step": 16000
-    },
-    {
-      "epoch": 0.024100966583485896,
-      "grad_norm": 1.7903109788894653,
-      "learning_rate": 0.00019841316972317053,
-      "loss": 2.293504638671875,
-      "step": 16100
-    },
-    {
-      "epoch": 0.024250662028103824,
-      "grad_norm": 1.6144092082977295,
-      "learning_rate": 0.0001984031890308199,
-      "loss": 2.289104309082031,
-      "step": 16200
-    },
-    {
-      "epoch": 0.024400357472721748,
-      "grad_norm": 1.3221566677093506,
-      "learning_rate": 0.00019839320833846925,
-      "loss": 2.3102418518066408,
-      "step": 16300
-    },
-    {
-      "epoch": 0.024550052917339672,
-      "grad_norm": 1.838915467262268,
-      "learning_rate": 0.0001983832276461186,
-      "loss": 2.279832305908203,
-      "step": 16400
-    },
-    {
-      "epoch": 0.024699748361957596,
-      "grad_norm": 1.371696949005127,
-      "learning_rate": 0.00019837324695376796,
-      "loss": 2.2707501220703126,
-      "step": 16500
-    },
-    {
-      "epoch": 0.024699748361957596,
-      "eval_loss": 2.1261136531829834,
-      "eval_runtime": 3300.5805,
-      "eval_samples_per_second": 179.907,
-      "eval_steps_per_second": 22.488,
-      "step": 16500
-    },
-    {
-      "epoch": 0.02484944380657552,
-      "grad_norm": 1.5465103387832642,
-      "learning_rate": 0.0001983632662614173,
-      "loss": 2.2787322998046875,
-      "step": 16600
-    },
-    {
-      "epoch": 0.024999139251193448,
-      "grad_norm": 1.8462467193603516,
-      "learning_rate": 0.00019835328556906665,
-      "loss": 2.286064453125,
-      "step": 16700
-    },
-    {
-      "epoch": 0.025148834695811372,
-      "grad_norm": 1.847586750984192,
-      "learning_rate": 0.000198343304876716,
-      "loss": 2.27947265625,
-      "step": 16800
-    },
-    {
-      "epoch": 0.025298530140429296,
-      "grad_norm": 1.333757996559143,
-      "learning_rate": 0.00019833332418436536,
-      "loss": 2.2753480529785155,
-      "step": 16900
-    },
-    {
-      "epoch": 0.02544822558504722,
-      "grad_norm": 1.5115796327590942,
-      "learning_rate": 0.0001983233434920147,
-      "loss": 2.3004881286621095,
-      "step": 17000
-    },
-    {
-      "epoch": 0.02544822558504722,
-      "eval_loss": 2.1196272373199463,
-      "eval_runtime": 3298.819,
-      "eval_samples_per_second": 180.003,
-      "eval_steps_per_second": 22.5,
-      "step": 17000
-    },
-    {
-      "epoch": 0.025597921029665148,
-      "grad_norm": 1.891655683517456,
-      "learning_rate": 0.00019831336279966405,
-      "loss": 2.2427667236328124,
-      "step": 17100
-    },
-    {
-      "epoch": 0.025747616474283072,
-      "grad_norm": 1.7352811098098755,
-      "learning_rate": 0.00019830338210731342,
-      "loss": 2.2087628173828127,
-      "step": 17200
-    },
-    {
-      "epoch": 0.025897311918900996,
-      "grad_norm": 1.553514003753662,
-      "learning_rate": 0.00019829340141496276,
-      "loss": 2.3139141845703124,
-      "step": 17300
-    },
-    {
-      "epoch": 0.02604700736351892,
-      "grad_norm": 1.2255268096923828,
-      "learning_rate": 0.00019828342072261213,
-      "loss": 2.267630310058594,
-      "step": 17400
-    },
-    {
-      "epoch": 0.026196702808136844,
-      "grad_norm": 1.707605004310608,
-      "learning_rate": 0.00019827344003026147,
-      "loss": 2.3676344299316407,
-      "step": 17500
-    },
-    {
-      "epoch": 0.026196702808136844,
-      "eval_loss": 2.11952805519104,
-      "eval_runtime": 3301.648,
-      "eval_samples_per_second": 179.849,
-      "eval_steps_per_second": 22.481,
-      "step": 17500
-    },
-    {
-      "epoch": 0.026346398252754772,
-      "grad_norm": 1.8780410289764404,
-      "learning_rate": 0.00019826345933791082,
-      "loss": 2.2455868530273437,
-      "step": 17600
-    },
-    {
-      "epoch": 0.026496093697372696,
-      "grad_norm": 1.2229410409927368,
-      "learning_rate": 0.00019825347864556016,
-      "loss": 2.280894470214844,
-      "step": 17700
-    },
-    {
-      "epoch": 0.02664578914199062,
-      "grad_norm": 2.0727083683013916,
-      "learning_rate": 0.00019824349795320953,
-      "loss": 2.2752020263671877,
-      "step": 17800
-    },
-    {
-      "epoch": 0.026795484586608544,
-      "grad_norm": 1.2752081155776978,
-      "learning_rate": 0.00019823351726085887,
-      "loss": 2.224423522949219,
-      "step": 17900
-    },
-    {
-      "epoch": 0.02694518003122647,
-      "grad_norm": 1.3691973686218262,
-      "learning_rate": 0.00019822353656850821,
-      "loss": 2.287049560546875,
-      "step": 18000
-    },
-    {
-      "epoch": 0.02694518003122647,
-      "eval_loss": 2.1210198402404785,
-      "eval_runtime": 3296.2323,
-      "eval_samples_per_second": 180.144,
-      "eval_steps_per_second": 22.518,
-      "step": 18000
-    },
-    {
-      "epoch": 0.027094875475844396,
-      "grad_norm": 1.6517150402069092,
-      "learning_rate": 0.00019821355587615758,
-      "loss": 2.211863708496094,
-      "step": 18100
-    },
-    {
-      "epoch": 0.02724457092046232,
-      "grad_norm": 1.7402405738830566,
-      "learning_rate": 0.00019820357518380696,
-      "loss": 2.2600927734375,
-      "step": 18200
-    },
-    {
-      "epoch": 0.027394266365080244,
-      "grad_norm": 1.452488899230957,
-      "learning_rate": 0.0001981935944914563,
-      "loss": 2.2501524353027342,
-      "step": 18300
-    },
-    {
-      "epoch": 0.02754396180969817,
-      "grad_norm": 1.4785183668136597,
-      "learning_rate": 0.00019818361379910564,
-      "loss": 2.2325625610351563,
-      "step": 18400
-    },
-    {
-      "epoch": 0.027693657254316092,
-      "grad_norm": 1.369954228401184,
-      "learning_rate": 0.00019817363310675498,
-      "loss": 2.286298828125,
-      "step": 18500
-    },
-    {
-      "epoch": 0.027693657254316092,
-      "eval_loss": 2.117324113845825,
-      "eval_runtime": 3303.6423,
-      "eval_samples_per_second": 179.74,
-      "eval_steps_per_second": 22.468,
-      "step": 18500
-    },
-    {
-      "epoch": 0.02784335269893402,
-      "grad_norm": 1.73832368850708,
-      "learning_rate": 0.00019816365241440435,
-      "loss": 2.3263778686523438,
-      "step": 18600
-    },
-    {
-      "epoch": 0.027993048143551944,
-      "grad_norm": 1.558114767074585,
-      "learning_rate": 0.0001981536717220537,
-      "loss": 2.3010546875,
-      "step": 18700
-    },
-    {
-      "epoch": 0.02814274358816987,
-      "grad_norm": 1.3523188829421997,
-      "learning_rate": 0.00019814369102970304,
-      "loss": 2.1968644714355468,
-      "step": 18800
-    },
-    {
-      "epoch": 0.028292439032787792,
-      "grad_norm": 1.4425495862960815,
-      "learning_rate": 0.00019813371033735238,
-      "loss": 2.221884307861328,
-      "step": 18900
-    },
-    {
-      "epoch": 0.028442134477405717,
-      "grad_norm": 1.9051581621170044,
-      "learning_rate": 0.00019812372964500175,
-      "loss": 2.2641307067871095,
-      "step": 19000
-    },
-    {
-      "epoch": 0.028442134477405717,
-      "eval_loss": 2.117173433303833,
-      "eval_runtime": 3296.9514,
-      "eval_samples_per_second": 180.105,
-      "eval_steps_per_second": 22.513,
-      "step": 19000
-    },
-    {
-      "epoch": 0.028591829922023644,
-      "grad_norm": 1.3613672256469727,
-      "learning_rate": 0.00019811374895265112,
-      "loss": 2.2960791015625,
-      "step": 19100
-    },
-    {
-      "epoch": 0.028741525366641568,
-      "grad_norm": 1.746584177017212,
-      "learning_rate": 0.00019810376826030047,
-      "loss": 2.2797586059570314,
-      "step": 19200
-    },
-    {
-      "epoch": 0.028891220811259492,
-      "grad_norm": 1.5967271327972412,
-      "learning_rate": 0.0001980937875679498,
-      "loss": 2.281904449462891,
-      "step": 19300
-    },
-    {
-      "epoch": 0.029040916255877416,
-      "grad_norm": 1.3153488636016846,
-      "learning_rate": 0.00019808380687559915,
-      "loss": 2.222927703857422,
-      "step": 19400
-    },
-    {
-      "epoch": 0.02919061170049534,
-      "grad_norm": 1.3665589094161987,
-      "learning_rate": 0.00019807382618324852,
-      "loss": 2.2356936645507814,
-      "step": 19500
-    },
-    {
-      "epoch": 0.02919061170049534,
-      "eval_loss": 2.1157138347625732,
-      "eval_runtime": 3300.1863,
-      "eval_samples_per_second": 179.929,
-      "eval_steps_per_second": 22.491,
-      "step": 19500
-    },
-    {
-      "epoch": 0.029340307145113268,
-      "grad_norm": 1.5521475076675415,
-      "learning_rate": 0.00019806384549089787,
-      "loss": 2.2691854858398437,
-      "step": 19600
-    },
-    {
-      "epoch": 0.029490002589731192,
-      "grad_norm": 1.5314592123031616,
-      "learning_rate": 0.0001980538647985472,
-      "loss": 2.277067108154297,
-      "step": 19700
-    },
-    {
-      "epoch": 0.029639698034349116,
-      "grad_norm": 1.3493821620941162,
-      "learning_rate": 0.00019804388410619655,
-      "loss": 2.2381459045410157,
-      "step": 19800
-    },
-    {
-      "epoch": 0.02978939347896704,
-      "grad_norm": 1.5802298784255981,
-      "learning_rate": 0.00019803390341384592,
-      "loss": 2.2472402954101565,
-      "step": 19900
-    },
-    {
-      "epoch": 0.029939088923584968,
-      "grad_norm": 1.704469919204712,
-      "learning_rate": 0.0001980239227214953,
-      "loss": 2.1897373962402344,
-      "step": 20000
-    },
-    {
-      "epoch": 0.029939088923584968,
-      "eval_loss": 2.1100823879241943,
-      "eval_runtime": 3310.1469,
-      "eval_samples_per_second": 179.387,
-      "eval_steps_per_second": 22.423,
-      "step": 20000
-    },
-    {
-      "epoch": 0.030088784368202892,
-      "grad_norm": 1.6501960754394531,
-      "learning_rate": 0.00019801394202914463,
-      "loss": 2.2383489990234375,
-      "step": 20100
-    },
-    {
-      "epoch": 0.030238479812820816,
-      "grad_norm": 1.4831558465957642,
-      "learning_rate": 0.00019800396133679398,
-      "loss": 2.261038818359375,
-      "step": 20200
-    },
-    {
-      "epoch": 0.03038817525743874,
-      "grad_norm": 1.4744017124176025,
-      "learning_rate": 0.00019799398064444335,
-      "loss": 2.317549133300781,
-      "step": 20300
-    },
-    {
-      "epoch": 0.030537870702056664,
-      "grad_norm": 1.3513352870941162,
-      "learning_rate": 0.0001979839999520927,
-      "loss": 2.2802613830566405,
-      "step": 20400
-    },
-    {
-      "epoch": 0.030687566146674592,
-      "grad_norm": 1.5280323028564453,
-      "learning_rate": 0.00019797401925974203,
-      "loss": 2.2517698669433592,
-      "step": 20500
-    },
-    {
-      "epoch": 0.030687566146674592,
-      "eval_loss": 2.1057777404785156,
-      "eval_runtime": 3308.3971,
-      "eval_samples_per_second": 179.482,
-      "eval_steps_per_second": 22.435,
-      "step": 20500
-    },
-    {
-      "epoch": 0.030837261591292516,
-      "grad_norm": 1.4214918613433838,
-      "learning_rate": 0.00019796403856739138,
-      "loss": 2.2889334106445314,
-      "step": 20600
-    },
-    {
-      "epoch": 0.03098695703591044,
-      "grad_norm": 1.8139195442199707,
-      "learning_rate": 0.00019795405787504072,
-      "loss": 2.2372776794433595,
-      "step": 20700
-    },
-    {
-      "epoch": 0.031136652480528364,
-      "grad_norm": 1.5987215042114258,
-      "learning_rate": 0.0001979440771826901,
-      "loss": 2.2489434814453126,
-      "step": 20800
-    },
-    {
-      "epoch": 0.03128634792514629,
-      "grad_norm": 1.6253299713134766,
-      "learning_rate": 0.00019793409649033946,
-      "loss": 2.2376019287109377,
-      "step": 20900
-    },
-    {
-      "epoch": 0.031436043369764216,
-      "grad_norm": 1.1986024379730225,
-      "learning_rate": 0.0001979241157979888,
-      "loss": 2.2178407287597657,
-      "step": 21000
-    },
-    {
-      "epoch": 0.031436043369764216,
-      "eval_loss": 2.1057324409484863,
-      "eval_runtime": 3309.5214,
-      "eval_samples_per_second": 179.421,
-      "eval_steps_per_second": 22.428,
-      "step": 21000
-    },
-    {
-      "epoch": 0.03158573881438214,
-      "grad_norm": 1.5893585681915283,
-      "learning_rate": 0.00019791413510563815,
-      "loss": 2.2059063720703125,
-      "step": 21100
-    },
-    {
-      "epoch": 0.031735434259000064,
-      "grad_norm": 1.3648179769515991,
-      "learning_rate": 0.00019790415441328752,
-      "loss": 2.281269836425781,
-      "step": 21200
-    },
-    {
-      "epoch": 0.03188512970361799,
-      "grad_norm": 1.362741231918335,
-      "learning_rate": 0.00019789417372093686,
-      "loss": 2.2103036499023436,
-      "step": 21300
-    },
-    {
-      "epoch": 0.03203482514823591,
-      "grad_norm": 1.9652200937271118,
-      "learning_rate": 0.0001978841930285862,
-      "loss": 2.2706588745117187,
-      "step": 21400
-    },
-    {
-      "epoch": 0.03218452059285384,
-      "grad_norm": 1.2960550785064697,
-      "learning_rate": 0.00019787421233623555,
-      "loss": 2.2664695739746095,
-      "step": 21500
-    },
-    {
-      "epoch": 0.03218452059285384,
-      "eval_loss": 2.104914903640747,
-      "eval_runtime": 3306.4149,
-      "eval_samples_per_second": 179.59,
-      "eval_steps_per_second": 22.449,
-      "step": 21500
-    },
-    {
-      "epoch": 0.03233421603747176,
-      "grad_norm": 1.4461086988449097,
-      "learning_rate": 0.00019786423164388492,
-      "loss": 2.2535964965820314,
-      "step": 21600
-    },
-    {
-      "epoch": 0.03248391148208969,
-      "grad_norm": 1.4941245317459106,
-      "learning_rate": 0.00019785425095153426,
-      "loss": 2.2836854553222654,
-      "step": 21700
-    },
-    {
-      "epoch": 0.032633606926707616,
-      "grad_norm": 1.626893162727356,
-      "learning_rate": 0.00019784427025918363,
-      "loss": 2.274701690673828,
-      "step": 21800
-    },
-    {
-      "epoch": 0.03278330237132554,
-      "grad_norm": 1.4712519645690918,
-      "learning_rate": 0.00019783428956683297,
-      "loss": 2.2501185607910155,
-      "step": 21900
-    },
-    {
-      "epoch": 0.032932997815943464,
-      "grad_norm": 1.7895413637161255,
-      "learning_rate": 0.00019782430887448234,
-      "loss": 2.25617919921875,
-      "step": 22000
-    },
-    {
-      "epoch": 0.032932997815943464,
-      "eval_loss": 2.1019809246063232,
-      "eval_runtime": 3305.9515,
-      "eval_samples_per_second": 179.615,
-      "eval_steps_per_second": 22.452,
-      "step": 22000
-    },
-    {
-      "epoch": 0.03308269326056139,
-      "grad_norm": 1.427451252937317,
-      "learning_rate": 0.00019781432818213168,
-      "loss": 2.2730235290527343,
-      "step": 22100
-    },
-    {
-      "epoch": 0.03323238870517931,
-      "grad_norm": 1.451992154121399,
-      "learning_rate": 0.00019780434748978103,
-      "loss": 2.1848146057128908,
-      "step": 22200
-    },
-    {
-      "epoch": 0.033382084149797236,
-      "grad_norm": 1.4783108234405518,
-      "learning_rate": 0.00019779436679743037,
-      "loss": 2.2585679626464845,
-      "step": 22300
-    },
-    {
-      "epoch": 0.03353177959441516,
-      "grad_norm": 1.6601756811141968,
-      "learning_rate": 0.00019778438610507971,
-      "loss": 2.289508056640625,
-      "step": 22400
-    },
-    {
-      "epoch": 0.033681475039033085,
-      "grad_norm": 1.4384208917617798,
-      "learning_rate": 0.00019777440541272908,
-      "loss": 2.214114532470703,
-      "step": 22500
-    },
-    {
-      "epoch": 0.033681475039033085,
-      "eval_loss": 2.098484754562378,
-      "eval_runtime": 3306.9235,
-      "eval_samples_per_second": 179.562,
-      "eval_steps_per_second": 22.445,
-      "step": 22500
-    },
-    {
-      "epoch": 0.03383117048365101,
-      "grad_norm": 1.5774037837982178,
-      "learning_rate": 0.00019776442472037843,
-      "loss": 2.276941223144531,
-      "step": 22600
-    },
-    {
-      "epoch": 0.03398086592826894,
-      "grad_norm": 1.5704011917114258,
-      "learning_rate": 0.0001977544440280278,
-      "loss": 2.2091140747070312,
-      "step": 22700
-    },
-    {
-      "epoch": 0.034130561372886864,
-      "grad_norm": 1.2489875555038452,
-      "learning_rate": 0.00019774446333567714,
-      "loss": 2.2158457946777346,
-      "step": 22800
-    },
-    {
-      "epoch": 0.03428025681750479,
-      "grad_norm": 1.4420257806777954,
-      "learning_rate": 0.0001977344826433265,
-      "loss": 2.1832150268554686,
-      "step": 22900
-    },
-    {
-      "epoch": 0.03442995226212271,
-      "grad_norm": 1.5147569179534912,
-      "learning_rate": 0.00019772450195097585,
-      "loss": 2.2175633239746095,
-      "step": 23000
-    },
-    {
-      "epoch": 0.03442995226212271,
-      "eval_loss": 2.099531412124634,
-      "eval_runtime": 3315.6022,
-      "eval_samples_per_second": 179.092,
-      "eval_steps_per_second": 22.387,
-      "step": 23000
-    },
-    {
-      "epoch": 0.034579647706740636,
-      "grad_norm": 1.6512202024459839,
-      "learning_rate": 0.0001977145212586252,
-      "loss": 2.23809814453125,
-      "step": 23100
-    },
-    {
-      "epoch": 0.03472934315135856,
-      "grad_norm": 1.255748987197876,
-      "learning_rate": 0.00019770454056627454,
-      "loss": 2.262901611328125,
-      "step": 23200
-    },
-    {
-      "epoch": 0.034879038595976485,
-      "grad_norm": 1.6600509881973267,
-      "learning_rate": 0.0001976945598739239,
-      "loss": 2.235970153808594,
-      "step": 23300
-    },
-    {
-      "epoch": 0.03502873404059441,
-      "grad_norm": 1.3919932842254639,
-      "learning_rate": 0.00019768457918157325,
-      "loss": 2.2053713989257813,
-      "step": 23400
-    },
-    {
-      "epoch": 0.03517842948521233,
-      "grad_norm": 1.5658373832702637,
-      "learning_rate": 0.0001976745984892226,
-      "loss": 2.218780059814453,
-      "step": 23500
-    },
-    {
-      "epoch": 0.03517842948521233,
-      "eval_loss": 2.094496965408325,
-      "eval_runtime": 3313.4641,
-      "eval_samples_per_second": 179.208,
-      "eval_steps_per_second": 22.401,
-      "step": 23500
-    },
-    {
-      "epoch": 0.03532812492983026,
-      "grad_norm": 2.070129156112671,
-      "learning_rate": 0.00019766461779687197,
-      "loss": 2.239718475341797,
-      "step": 23600
-    },
-    {
-      "epoch": 0.03547782037444819,
-      "grad_norm": 1.6429266929626465,
-      "learning_rate": 0.00019765463710452134,
-      "loss": 2.2072882080078124,
-      "step": 23700
-    },
-    {
-      "epoch": 0.03562751581906611,
-      "grad_norm": 1.4987503290176392,
-      "learning_rate": 0.00019764465641217068,
-      "loss": 2.281219024658203,
-      "step": 23800
-    },
-    {
-      "epoch": 0.035777211263684036,
-      "grad_norm": 1.4762109518051147,
-      "learning_rate": 0.00019763467571982002,
-      "loss": 2.2457525634765627,
-      "step": 23900
-    },
-    {
-      "epoch": 0.03592690670830196,
-      "grad_norm": 1.4080301523208618,
-      "learning_rate": 0.00019762469502746936,
-      "loss": 2.2922372436523437,
-      "step": 24000
-    },
-    {
-      "epoch": 0.03592690670830196,
-      "eval_loss": 2.0921194553375244,
-      "eval_runtime": 3316.1637,
-      "eval_samples_per_second": 179.062,
-      "eval_steps_per_second": 22.383,
-      "step": 24000
-    },
-    {
-      "epoch": 0.036076602152919884,
-      "grad_norm": 1.6255196332931519,
-      "learning_rate": 0.0001976147143351187,
-      "loss": 2.2189508056640626,
-      "step": 24100
-    },
-    {
-      "epoch": 0.03622629759753781,
-      "grad_norm": 1.4851337671279907,
-      "learning_rate": 0.00019760473364276808,
-      "loss": 2.2108830261230468,
-      "step": 24200
-    },
-    {
-      "epoch": 0.03637599304215573,
-      "grad_norm": 1.5986689329147339,
-      "learning_rate": 0.00019759475295041742,
-      "loss": 2.2327239990234373,
-      "step": 24300
-    },
-    {
-      "epoch": 0.03652568848677366,
-      "grad_norm": 1.5597631931304932,
-      "learning_rate": 0.00019758477225806676,
-      "loss": 2.2503005981445314,
-      "step": 24400
-    },
-    {
-      "epoch": 0.03667538393139158,
-      "grad_norm": 1.5710701942443848,
-      "learning_rate": 0.00019757479156571613,
-      "loss": 2.2456903076171875,
-      "step": 24500
-    },
-    {
-      "epoch": 0.03667538393139158,
-      "eval_loss": 2.085951566696167,
-      "eval_runtime": 3312.8181,
-      "eval_samples_per_second": 179.243,
-      "eval_steps_per_second": 22.405,
-      "step": 24500
-    },
-    {
-      "epoch": 0.03682507937600951,
-      "grad_norm": 1.4867461919784546,
-      "learning_rate": 0.0001975648108733655,
-      "loss": 2.258560791015625,
-      "step": 24600
-    },
-    {
-      "epoch": 0.036974774820627436,
-      "grad_norm": 1.8220168352127075,
-      "learning_rate": 0.00019755483018101485,
-      "loss": 2.2663095092773435,
-      "step": 24700
-    },
-    {
-      "epoch": 0.03712447026524536,
-      "grad_norm": 1.4194265604019165,
-      "learning_rate": 0.0001975448494886642,
-      "loss": 2.221974334716797,
-      "step": 24800
-    },
-    {
-      "epoch": 0.037274165709863284,
-      "grad_norm": 1.5951439142227173,
-      "learning_rate": 0.00019753486879631353,
-      "loss": 2.2663487243652343,
-      "step": 24900
-    },
-    {
-      "epoch": 0.03742386115448121,
-      "grad_norm": 1.4926362037658691,
-      "learning_rate": 0.0001975248881039629,
-      "loss": 2.2837535095214845,
-      "step": 25000
-    },
-    {
-      "epoch": 0.03742386115448121,
-      "eval_loss": 2.0880188941955566,
-      "eval_runtime": 3318.8379,
-      "eval_samples_per_second": 178.917,
-      "eval_steps_per_second": 22.365,
-      "step": 25000
-    },
-    {
-      "epoch": 0.03757355659909913,
-      "grad_norm": 1.4942559003829956,
-      "learning_rate": 0.00019751490741161225,
-      "loss": 2.2578759765625,
-      "step": 25100
-    },
-    {
-      "epoch": 0.03772325204371706,
-      "grad_norm": 1.400155782699585,
-      "learning_rate": 0.0001975049267192616,
-      "loss": 2.2389447021484377,
-      "step": 25200
-    },
-    {
-      "epoch": 0.03787294748833498,
-      "grad_norm": 1.688620924949646,
-      "learning_rate": 0.00019749494602691093,
-      "loss": 2.229586944580078,
-      "step": 25300
-    },
-    {
-      "epoch": 0.038022642932952905,
-      "grad_norm": 1.3628838062286377,
-      "learning_rate": 0.0001974849653345603,
-      "loss": 2.2383181762695314,
-      "step": 25400
-    },
-    {
-      "epoch": 0.03817233837757083,
-      "grad_norm": 1.4471839666366577,
-      "learning_rate": 0.00019747498464220967,
-      "loss": 2.1750228881835936,
-      "step": 25500
-    },
-    {
-      "epoch": 0.03817233837757083,
-      "eval_loss": 2.088946580886841,
-      "eval_runtime": 3316.0361,
-      "eval_samples_per_second": 179.069,
-      "eval_steps_per_second": 22.384,
-      "step": 25500
-    },
-    {
-      "epoch": 0.03832203382218876,
-      "grad_norm": 1.6881308555603027,
-      "learning_rate": 0.00019746500394985902,
-      "loss": 2.23192626953125,
-      "step": 25600
-    },
-    {
-      "epoch": 0.038471729266806684,
-      "grad_norm": 1.7420785427093506,
-      "learning_rate": 0.00019745502325750836,
-      "loss": 2.2621939086914065,
-      "step": 25700
-    },
-    {
-      "epoch": 0.03862142471142461,
-      "grad_norm": 1.331814169883728,
-      "learning_rate": 0.00019744504256515773,
-      "loss": 2.201291351318359,
-      "step": 25800
-    },
-    {
-      "epoch": 0.03877112015604253,
-      "grad_norm": 1.8634979724884033,
-      "learning_rate": 0.00019743506187280707,
-      "loss": 2.231629638671875,
-      "step": 25900
-    },
-    {
-      "epoch": 0.038920815600660456,
-      "grad_norm": 1.357436180114746,
-      "learning_rate": 0.00019742508118045641,
-      "loss": 2.3026466369628906,
-      "step": 26000
-    },
-    {
-      "epoch": 0.038920815600660456,
-      "eval_loss": 2.0866785049438477,
-      "eval_runtime": 3314.8783,
-      "eval_samples_per_second": 179.131,
-      "eval_steps_per_second": 22.391,
-      "step": 26000
-    },
-    {
-      "epoch": 0.03907051104527838,
-      "grad_norm": 1.4098141193389893,
-      "learning_rate": 0.00019741510048810576,
-      "loss": 2.2671484375,
-      "step": 26100
-    },
-    {
-      "epoch": 0.039220206489896305,
-      "grad_norm": 1.6442112922668457,
-      "learning_rate": 0.0001974051197957551,
-      "loss": 2.225537109375,
-      "step": 26200
-    },
-    {
-      "epoch": 0.03936990193451423,
-      "grad_norm": 1.4602155685424805,
-      "learning_rate": 0.00019739513910340447,
-      "loss": 2.225550537109375,
-      "step": 26300
-    },
-    {
-      "epoch": 0.03951959737913215,
-      "grad_norm": 1.529658317565918,
-      "learning_rate": 0.00019738515841105384,
-      "loss": 2.2010858154296873,
-      "step": 26400
-    },
-    {
-      "epoch": 0.039669292823750084,
-      "grad_norm": 1.3564810752868652,
-      "learning_rate": 0.00019737517771870318,
-      "loss": 2.2147760009765625,
-      "step": 26500
-    },
-    {
-      "epoch": 0.039669292823750084,
-      "eval_loss": 2.084024667739868,
-      "eval_runtime": 3313.9498,
-      "eval_samples_per_second": 179.181,
-      "eval_steps_per_second": 22.398,
-      "step": 26500
-    },
-    {
-      "epoch": 0.03981898826836801,
-      "grad_norm": 1.464477300643921,
-      "learning_rate": 0.00019736519702635253,
-      "loss": 2.2640489196777343,
-      "step": 26600
-    },
-    {
-      "epoch": 0.03996868371298593,
-      "grad_norm": 1.5118786096572876,
-      "learning_rate": 0.0001973552163340019,
-      "loss": 2.2367198181152346,
-      "step": 26700
-    },
-    {
-      "epoch": 0.040118379157603856,
-      "grad_norm": 1.5135878324508667,
-      "learning_rate": 0.00019734523564165124,
-      "loss": 2.2217457580566404,
-      "step": 26800
-    },
-    {
-      "epoch": 0.04026807460222178,
-      "grad_norm": 1.3223872184753418,
-      "learning_rate": 0.00019733525494930058,
-      "loss": 2.23560302734375,
-      "step": 26900
-    },
-    {
-      "epoch": 0.040417770046839704,
-      "grad_norm": 1.7223246097564697,
-      "learning_rate": 0.00019732527425694993,
-      "loss": 2.1723712158203123,
-      "step": 27000
-    },
-    {
-      "epoch": 0.040417770046839704,
-      "eval_loss": 2.0792603492736816,
-      "eval_runtime": 3316.8051,
-      "eval_samples_per_second": 179.027,
-      "eval_steps_per_second": 22.378,
-      "step": 27000
-    },
-    {
-      "epoch": 0.04056746549145763,
-      "grad_norm": 1.549896240234375,
-      "learning_rate": 0.0001973152935645993,
-      "loss": 2.1600308227539062,
-      "step": 27100
-    },
-    {
-      "epoch": 0.04071716093607555,
-      "grad_norm": 1.879260778427124,
-      "learning_rate": 0.00019730531287224864,
-      "loss": 2.202359924316406,
-      "step": 27200
-    },
-    {
-      "epoch": 0.04086685638069348,
-      "grad_norm": 1.318892002105713,
-      "learning_rate": 0.00019729533217989798,
-      "loss": 2.2444964599609376,
-      "step": 27300
-    },
-    {
-      "epoch": 0.0410165518253114,
-      "grad_norm": 1.480472207069397,
-      "learning_rate": 0.00019728535148754735,
-      "loss": 2.240170135498047,
-      "step": 27400
-    },
-    {
-      "epoch": 0.04116624726992933,
-      "grad_norm": 1.525627851486206,
-      "learning_rate": 0.00019727537079519672,
-      "loss": 2.216890411376953,
-      "step": 27500
-    },
-    {
-      "epoch": 0.04116624726992933,
-      "eval_loss": 2.0827136039733887,
-      "eval_runtime": 3315.9879,
-      "eval_samples_per_second": 179.071,
-      "eval_steps_per_second": 22.384,
-      "step": 27500
-    },
-    {
-      "epoch": 0.041315942714547256,
-      "grad_norm": 1.478674054145813,
-      "learning_rate": 0.00019726539010284607,
-      "loss": 2.220688781738281,
-      "step": 27600
-    },
-    {
-      "epoch": 0.04146563815916518,
-      "grad_norm": 1.5419797897338867,
-      "learning_rate": 0.0001972554094104954,
-      "loss": 2.2400482177734373,
-      "step": 27700
-    },
-    {
-      "epoch": 0.041615333603783104,
-      "grad_norm": 1.773439645767212,
-      "learning_rate": 0.00019724542871814475,
-      "loss": 2.2482081604003907,
-      "step": 27800
-    },
-    {
-      "epoch": 0.04176502904840103,
-      "grad_norm": 1.444304347038269,
-      "learning_rate": 0.0001972354480257941,
-      "loss": 2.172163391113281,
-      "step": 27900
-    },
-    {
-      "epoch": 0.04191472449301895,
-      "grad_norm": 1.4720896482467651,
-      "learning_rate": 0.00019722546733344346,
-      "loss": 2.234305114746094,
-      "step": 28000
-    },
-    {
-      "epoch": 0.04191472449301895,
-      "eval_loss": 2.078559160232544,
-      "eval_runtime": 3321.0744,
-      "eval_samples_per_second": 178.797,
-      "eval_steps_per_second": 22.35,
-      "step": 28000
-    },
-    {
-      "epoch": 0.04206441993763688,
-      "grad_norm": 1.649888515472412,
-      "learning_rate": 0.0001972154866410928,
-      "loss": 2.240731048583984,
-      "step": 28100
-    },
-    {
-      "epoch": 0.0422141153822548,
-      "grad_norm": 1.4914374351501465,
-      "learning_rate": 0.00019720550594874215,
-      "loss": 2.242912139892578,
-      "step": 28200
-    },
-    {
-      "epoch": 0.042363810826872725,
-      "grad_norm": 1.3904025554656982,
-      "learning_rate": 0.00019719552525639152,
-      "loss": 2.224173126220703,
-      "step": 28300
-    },
-    {
-      "epoch": 0.04251350627149065,
-      "grad_norm": 1.4036498069763184,
-      "learning_rate": 0.0001971855445640409,
-      "loss": 2.164631805419922,
-      "step": 28400
-    },
-    {
-      "epoch": 0.04266320171610858,
-      "grad_norm": 1.6599886417388916,
-      "learning_rate": 0.00019717556387169023,
-      "loss": 2.2452403259277345,
-      "step": 28500
-    },
-    {
-      "epoch": 0.04266320171610858,
-      "eval_loss": 2.077615976333618,
-      "eval_runtime": 3316.5879,
-      "eval_samples_per_second": 179.039,
-      "eval_steps_per_second": 22.38,
-      "step": 28500
-    },
-    {
-      "epoch": 0.042812897160726504,
-      "grad_norm": 1.4497243165969849,
-      "learning_rate": 0.00019716558317933958,
-      "loss": 2.20112548828125,
-      "step": 28600
-    },
-    {
-      "epoch": 0.04296259260534443,
-      "grad_norm": 1.3359723091125488,
-      "learning_rate": 0.00019715560248698892,
-      "loss": 2.2168283081054687,
-      "step": 28700
-    },
-    {
-      "epoch": 0.04311228804996235,
-      "grad_norm": 1.4851969480514526,
-      "learning_rate": 0.0001971456217946383,
-      "loss": 2.262620086669922,
-      "step": 28800
-    },
-    {
-      "epoch": 0.043261983494580276,
-      "grad_norm": 1.3208050727844238,
-      "learning_rate": 0.00019713564110228763,
-      "loss": 2.2391950988769533,
-      "step": 28900
-    },
-    {
-      "epoch": 0.0434116789391982,
-      "grad_norm": 1.3377376794815063,
-      "learning_rate": 0.00019712566040993698,
-      "loss": 2.2192486572265624,
-      "step": 29000
-    },
-    {
-      "epoch": 0.0434116789391982,
-      "eval_loss": 2.0744612216949463,
-      "eval_runtime": 3317.3085,
-      "eval_samples_per_second": 179.0,
-      "eval_steps_per_second": 22.375,
-      "step": 29000
-    },
-    {
-      "epoch": 0.043561374383816125,
-      "grad_norm": 1.5779460668563843,
-      "learning_rate": 0.00019711567971758632,
-      "loss": 2.228085784912109,
-      "step": 29100
-    },
-    {
-      "epoch": 0.04371106982843405,
-      "grad_norm": 1.4374401569366455,
-      "learning_rate": 0.0001971056990252357,
-      "loss": 2.2488969421386718,
-      "step": 29200
-    },
-    {
-      "epoch": 0.04386076527305197,
-      "grad_norm": 1.5538705587387085,
-      "learning_rate": 0.00019709571833288506,
-      "loss": 2.2609434509277344,
-      "step": 29300
-    },
-    {
-      "epoch": 0.044010460717669904,
-      "grad_norm": 1.36516535282135,
-      "learning_rate": 0.0001970857376405344,
-      "loss": 2.158708343505859,
-      "step": 29400
-    },
-    {
-      "epoch": 0.04416015616228783,
-      "grad_norm": 1.453946590423584,
-      "learning_rate": 0.00019707575694818375,
-      "loss": 2.187682800292969,
-      "step": 29500
-    },
-    {
-      "epoch": 0.04416015616228783,
-      "eval_loss": 2.0741775035858154,
-      "eval_runtime": 3290.1027,
-      "eval_samples_per_second": 180.48,
-      "eval_steps_per_second": 22.56,
-      "step": 29500
-    },
-    {
-      "epoch": 0.04430985160690575,
-      "grad_norm": 1.614549994468689,
-      "learning_rate": 0.0001970657762558331,
-      "loss": 2.237359924316406,
-      "step": 29600
-    },
-    {
-      "epoch": 0.044459547051523676,
-      "grad_norm": 1.2623584270477295,
-      "learning_rate": 0.00019705579556348246,
-      "loss": 2.243529052734375,
-      "step": 29700
-    },
-    {
-      "epoch": 0.0446092424961416,
-      "grad_norm": 1.5648853778839111,
-      "learning_rate": 0.0001970458148711318,
-      "loss": 2.267769470214844,
-      "step": 29800
-    },
-    {
-      "epoch": 0.044758937940759524,
-      "grad_norm": 1.7233803272247314,
-      "learning_rate": 0.00019703583417878114,
-      "loss": 2.24640625,
-      "step": 29900
-    },
-    {
-      "epoch": 0.04490863338537745,
-      "grad_norm": 1.3874558210372925,
-      "learning_rate": 0.0001970258534864305,
-      "loss": 2.267073211669922,
-      "step": 30000
-    },
-    {
-      "epoch": 0.04490863338537745,
-      "eval_loss": 2.075207233428955,
-      "eval_runtime": 3289.8229,
-      "eval_samples_per_second": 180.495,
-      "eval_steps_per_second": 22.562,
-      "step": 30000
-    },
-    {
-      "epoch": 0.04505832882999537,
-      "grad_norm": 1.4323738813400269,
-      "learning_rate": 0.00019701587279407986,
-      "loss": 2.2394744873046877,
-      "step": 30100
-    },
-    {
-      "epoch": 0.0452080242746133,
-      "grad_norm": 1.3241939544677734,
-      "learning_rate": 0.00019700589210172923,
-      "loss": 2.194419097900391,
-      "step": 30200
-    },
-    {
-      "epoch": 0.04535771971923122,
-      "grad_norm": 1.5629545450210571,
-      "learning_rate": 0.00019699591140937857,
-      "loss": 2.282225799560547,
-      "step": 30300
-    },
-    {
-      "epoch": 0.04550741516384915,
-      "grad_norm": 1.2985613346099854,
-      "learning_rate": 0.00019698593071702791,
-      "loss": 2.225314178466797,
-      "step": 30400
-    },
-    {
-      "epoch": 0.045657110608467076,
-      "grad_norm": 1.543164849281311,
-      "learning_rate": 0.00019697595002467728,
-      "loss": 2.217964935302734,
-      "step": 30500
-    },
-    {
-      "epoch": 0.045657110608467076,
-      "eval_loss": 2.0741989612579346,
-      "eval_runtime": 3289.7348,
-      "eval_samples_per_second": 180.5,
-      "eval_steps_per_second": 22.563,
-      "step": 30500
-    },
-    {
-      "epoch": 0.045806806053085,
-      "grad_norm": 1.6781221628189087,
-      "learning_rate": 0.00019696596933232663,
-      "loss": 2.2274684143066406,
-      "step": 30600
-    },
-    {
-      "epoch": 0.045956501497702924,
-      "grad_norm": 1.3499797582626343,
-      "learning_rate": 0.00019695598863997597,
-      "loss": 2.2298394775390626,
-      "step": 30700
-    },
-    {
-      "epoch": 0.04610619694232085,
-      "grad_norm": 1.2994440793991089,
-      "learning_rate": 0.0001969460079476253,
-      "loss": 2.2472613525390623,
-      "step": 30800
-    },
-    {
-      "epoch": 0.04625589238693877,
-      "grad_norm": 1.392971158027649,
-      "learning_rate": 0.00019693602725527468,
-      "loss": 2.2145263671875,
-      "step": 30900
-    },
-    {
-      "epoch": 0.0464055878315567,
-      "grad_norm": 1.3143973350524902,
-      "learning_rate": 0.00019692604656292403,
-      "loss": 2.200730743408203,
-      "step": 31000
-    },
-    {
-      "epoch": 0.0464055878315567,
-      "eval_loss": 2.070248603820801,
-      "eval_runtime": 3289.8227,
-      "eval_samples_per_second": 180.495,
-      "eval_steps_per_second": 22.562,
-      "step": 31000
-    },
-    {
-      "epoch": 0.04655528327617462,
-      "grad_norm": 1.292435646057129,
-      "learning_rate": 0.0001969160658705734,
-      "loss": 2.231234893798828,
-      "step": 31100
-    },
-    {
-      "epoch": 0.046704978720792545,
-      "grad_norm": 1.3820241689682007,
-      "learning_rate": 0.00019690608517822274,
-      "loss": 2.2168016052246093,
-      "step": 31200
-    },
-    {
-      "epoch": 0.04685467416541047,
-      "grad_norm": 1.63380765914917,
-      "learning_rate": 0.00019689610448587208,
-      "loss": 2.156103210449219,
-      "step": 31300
-    },
-    {
-      "epoch": 0.0470043696100284,
-      "grad_norm": 1.307085394859314,
-      "learning_rate": 0.00019688612379352145,
-      "loss": 2.2381263732910157,
-      "step": 31400
-    },
-    {
-      "epoch": 0.047154065054646324,
-      "grad_norm": 1.4193196296691895,
-      "learning_rate": 0.0001968761431011708,
-      "loss": 2.2343438720703124,
-      "step": 31500
-    },
-    {
-      "epoch": 0.047154065054646324,
-      "eval_loss": 2.0691416263580322,
-      "eval_runtime": 3290.019,
-      "eval_samples_per_second": 180.485,
-      "eval_steps_per_second": 22.561,
-      "step": 31500
-    },
-    {
-      "epoch": 0.04730376049926425,
-      "grad_norm": 1.3469774723052979,
-      "learning_rate": 0.00019686616240882014,
-      "loss": 2.227236633300781,
-      "step": 31600
-    },
-    {
-      "epoch": 0.04745345594388217,
-      "grad_norm": 1.5872033834457397,
-      "learning_rate": 0.00019685618171646948,
-      "loss": 2.2040524291992187,
-      "step": 31700
-    },
-    {
-      "epoch": 0.047603151388500096,
-      "grad_norm": 1.679634690284729,
-      "learning_rate": 0.00019684620102411885,
-      "loss": 2.2274920654296877,
-      "step": 31800
-    },
-    {
-      "epoch": 0.04775284683311802,
-      "grad_norm": 1.807590126991272,
-      "learning_rate": 0.0001968362203317682,
-      "loss": 2.199940185546875,
-      "step": 31900
-    },
-    {
-      "epoch": 0.047902542277735945,
-      "grad_norm": 1.4261001348495483,
-      "learning_rate": 0.00019682623963941756,
-      "loss": 2.2326353454589842,
-      "step": 32000
-    },
-    {
-      "epoch": 0.047902542277735945,
-      "eval_loss": 2.0701792240142822,
-      "eval_runtime": 3293.5776,
-      "eval_samples_per_second": 180.29,
-      "eval_steps_per_second": 22.536,
-      "step": 32000
-    },
-    {
-      "epoch": 0.04805223772235387,
-      "grad_norm": 1.294680118560791,
-      "learning_rate": 0.0001968162589470669,
-      "loss": 2.2357977294921874,
-      "step": 32100
-    },
-    {
-      "epoch": 0.04820193316697179,
-      "grad_norm": 1.7771323919296265,
-      "learning_rate": 0.00019680627825471628,
-      "loss": 2.250038299560547,
-      "step": 32200
-    },
-    {
-      "epoch": 0.048351628611589724,
-      "grad_norm": 1.513717532157898,
-      "learning_rate": 0.00019679629756236562,
-      "loss": 2.2489202880859374,
-      "step": 32300
-    },
-    {
-      "epoch": 0.04850132405620765,
-      "grad_norm": 1.8060063123703003,
-      "learning_rate": 0.00019678631687001496,
-      "loss": 2.2891302490234375,
-      "step": 32400
-    },
-    {
-      "epoch": 0.04865101950082557,
-      "grad_norm": 1.4588603973388672,
-      "learning_rate": 0.0001967763361776643,
-      "loss": 2.2351919555664064,
-      "step": 32500
-    },
-    {
-      "epoch": 0.04865101950082557,
-      "eval_loss": 2.060580253601074,
-      "eval_runtime": 3289.245,
-      "eval_samples_per_second": 180.527,
-      "eval_steps_per_second": 22.566,
-      "step": 32500
-    },
-    {
-      "epoch": 0.048800714945443496,
-      "grad_norm": 1.4652509689331055,
-      "learning_rate": 0.00019676635548531368,
-      "loss": 2.240154571533203,
-      "step": 32600
-    },
-    {
-      "epoch": 0.04895041039006142,
-      "grad_norm": 1.729806900024414,
-      "learning_rate": 0.00019675637479296302,
-      "loss": 2.2327191162109377,
-      "step": 32700
-    },
-    {
-      "epoch": 0.049100105834679345,
-      "grad_norm": 1.4504591226577759,
-      "learning_rate": 0.00019674639410061236,
-      "loss": 2.210146026611328,
-      "step": 32800
-    },
-    {
-      "epoch": 0.04924980127929727,
-      "grad_norm": 1.4042253494262695,
-      "learning_rate": 0.00019673641340826173,
-      "loss": 2.1929080200195314,
-      "step": 32900
-    },
-    {
-      "epoch": 0.04939949672391519,
-      "grad_norm": 1.3705559968948364,
-      "learning_rate": 0.00019672643271591108,
-      "loss": 2.264973449707031,
-      "step": 33000
-    },
-    {
-      "epoch": 0.04939949672391519,
-      "eval_loss": 2.0676751136779785,
-      "eval_runtime": 3289.463,
-      "eval_samples_per_second": 180.515,
-      "eval_steps_per_second": 22.564,
-      "step": 33000
-    },
-    {
-      "epoch": 0.04954919216853312,
-      "grad_norm": 1.5015199184417725,
-      "learning_rate": 0.00019671645202356045,
-      "loss": 2.152598114013672,
-      "step": 33100
-    },
-    {
-      "epoch": 0.04969888761315104,
-      "grad_norm": 1.5769134759902954,
-      "learning_rate": 0.0001967064713312098,
-      "loss": 2.2136253356933593,
-      "step": 33200
-    },
-    {
-      "epoch": 0.04984858305776897,
-      "grad_norm": 1.4534562826156616,
-      "learning_rate": 0.00019669649063885913,
-      "loss": 2.2250627136230468,
-      "step": 33300
-    },
-    {
-      "epoch": 0.049998278502386896,
-      "grad_norm": 1.591151475906372,
-      "learning_rate": 0.00019668650994650848,
-      "loss": 2.2143814086914064,
-      "step": 33400
-    },
-    {
-      "epoch": 0.05014797394700482,
-      "grad_norm": 1.4194039106369019,
-      "learning_rate": 0.00019667652925415785,
-      "loss": 2.166915283203125,
-      "step": 33500
-    },
-    {
-      "epoch": 0.05014797394700482,
-      "eval_loss": 2.0674703121185303,
-      "eval_runtime": 3292.5139,
-      "eval_samples_per_second": 180.348,
-      "eval_steps_per_second": 22.544,
-      "step": 33500
-    },
-    {
-      "epoch": 0.050297669391622744,
-      "grad_norm": 1.4517550468444824,
-      "learning_rate": 0.0001966665485618072,
-      "loss": 2.207733612060547,
-      "step": 33600
-    },
-    {
-      "epoch": 0.05044736483624067,
-      "grad_norm": 1.4861198663711548,
-      "learning_rate": 0.00019665656786945653,
-      "loss": 2.2044033813476562,
-      "step": 33700
-    },
-    {
-      "epoch": 0.05059706028085859,
-      "grad_norm": 1.1126517057418823,
-      "learning_rate": 0.0001966465871771059,
-      "loss": 2.2500875854492186,
-      "step": 33800
-    },
-    {
-      "epoch": 0.05074675572547652,
-      "grad_norm": 1.4116262197494507,
-      "learning_rate": 0.00019663660648475527,
-      "loss": 2.1853285217285157,
-      "step": 33900
-    },
-    {
-      "epoch": 0.05089645117009444,
-      "grad_norm": 1.9885811805725098,
-      "learning_rate": 0.00019662662579240461,
-      "loss": 2.236006774902344,
-      "step": 34000
-    },
-    {
-      "epoch": 0.05089645117009444,
-      "eval_loss": 2.0686213970184326,
-      "eval_runtime": 3306.2963,
-      "eval_samples_per_second": 179.596,
-      "eval_steps_per_second": 22.45,
-      "step": 34000
-    },
-    {
-      "epoch": 0.051046146614712365,
-      "grad_norm": 1.3172255754470825,
-      "learning_rate": 0.00019661664510005396,
-      "loss": 2.2278118896484376,
-      "step": 34100
-    },
-    {
-      "epoch": 0.051195842059330296,
-      "grad_norm": 1.2944632768630981,
-      "learning_rate": 0.0001966066644077033,
-      "loss": 2.21828125,
-      "step": 34200
-    },
-    {
-      "epoch": 0.05134553750394822,
-      "grad_norm": 1.3168495893478394,
-      "learning_rate": 0.00019659668371535267,
-      "loss": 2.211839599609375,
-      "step": 34300
-    },
-    {
-      "epoch": 0.051495232948566144,
-      "grad_norm": 1.8327419757843018,
-      "learning_rate": 0.00019658670302300201,
-      "loss": 2.256954803466797,
-      "step": 34400
-    },
-    {
-      "epoch": 0.05164492839318407,
-      "grad_norm": 1.3155707120895386,
-      "learning_rate": 0.00019657672233065136,
-      "loss": 2.173082733154297,
-      "step": 34500
-    },
-    {
-      "epoch": 0.05164492839318407,
-      "eval_loss": 2.0607357025146484,
-      "eval_runtime": 3313.6353,
-      "eval_samples_per_second": 179.198,
-      "eval_steps_per_second": 22.4,
-      "step": 34500
-    },
-    {
-      "epoch": 0.05179462383780199,
-      "grad_norm": 1.5257196426391602,
-      "learning_rate": 0.0001965667416383007,
-      "loss": 2.2291798400878906,
-      "step": 34600
-    },
-    {
-      "epoch": 0.051944319282419917,
-      "grad_norm": 1.3587309122085571,
-      "learning_rate": 0.00019655676094595007,
-      "loss": 2.211974639892578,
-      "step": 34700
-    },
-    {
-      "epoch": 0.05209401472703784,
-      "grad_norm": 1.444800853729248,
-      "learning_rate": 0.00019654678025359944,
-      "loss": 2.207247314453125,
-      "step": 34800
-    },
-    {
-      "epoch": 0.052243710171655765,
-      "grad_norm": 1.4456899166107178,
-      "learning_rate": 0.00019653679956124878,
-      "loss": 2.2471710205078126,
-      "step": 34900
-    },
-    {
-      "epoch": 0.05239340561627369,
-      "grad_norm": 1.9591150283813477,
-      "learning_rate": 0.00019652681886889813,
-      "loss": 2.190756378173828,
-      "step": 35000
-    },
-    {
-      "epoch": 0.05239340561627369,
-      "eval_loss": 2.061028003692627,
-      "eval_runtime": 3322.7537,
-      "eval_samples_per_second": 178.707,
-      "eval_steps_per_second": 22.338,
-      "step": 35000
-    },
-    {
-      "epoch": 0.05254310106089161,
-      "grad_norm": 1.5017095804214478,
-      "learning_rate": 0.00019651683817654747,
-      "loss": 2.245433349609375,
-      "step": 35100
-    },
-    {
-      "epoch": 0.052692796505509544,
-      "grad_norm": 1.4105480909347534,
-      "learning_rate": 0.00019650685748419684,
-      "loss": 2.1785232543945314,
-      "step": 35200
-    },
-    {
-      "epoch": 0.05284249195012747,
-      "grad_norm": 1.7517006397247314,
-      "learning_rate": 0.00019649687679184618,
-      "loss": 2.220989990234375,
-      "step": 35300
-    },
-    {
-      "epoch": 0.05299218739474539,
-      "grad_norm": 1.4039878845214844,
-      "learning_rate": 0.00019648689609949553,
-      "loss": 2.206803283691406,
-      "step": 35400
-    },
-    {
-      "epoch": 0.053141882839363316,
-      "grad_norm": 1.2501726150512695,
-      "learning_rate": 0.00019647691540714487,
-      "loss": 2.1815589904785155,
-      "step": 35500
-    },
-    {
-      "epoch": 0.053141882839363316,
-      "eval_loss": 2.0645787715911865,
-      "eval_runtime": 3323.6136,
-      "eval_samples_per_second": 178.66,
-      "eval_steps_per_second": 22.333,
-      "step": 35500
-    },
-    {
-      "epoch": 0.05329157828398124,
-      "grad_norm": 1.780024766921997,
-      "learning_rate": 0.00019646693471479424,
-      "loss": 2.302761993408203,
-      "step": 35600
-    },
-    {
-      "epoch": 0.053441273728599165,
-      "grad_norm": 1.519789457321167,
-      "learning_rate": 0.0001964569540224436,
-      "loss": 2.1987368774414064,
-      "step": 35700
-    },
-    {
-      "epoch": 0.05359096917321709,
-      "grad_norm": 1.4410274028778076,
-      "learning_rate": 0.00019644697333009295,
-      "loss": 2.228959197998047,
-      "step": 35800
-    },
-    {
-      "epoch": 0.05374066461783501,
-      "grad_norm": 1.6502699851989746,
-      "learning_rate": 0.0001964369926377423,
-      "loss": 2.2749424743652344,
-      "step": 35900
-    },
-    {
-      "epoch": 0.05389036006245294,
-      "grad_norm": 1.5358104705810547,
-      "learning_rate": 0.00019642701194539167,
-      "loss": 2.2157525634765625,
-      "step": 36000
-    },
-    {
-      "epoch": 0.05389036006245294,
-      "eval_loss": 2.059638500213623,
-      "eval_runtime": 3328.1344,
-      "eval_samples_per_second": 178.418,
-      "eval_steps_per_second": 22.302,
-      "step": 36000
-    },
-    {
-      "epoch": 0.05404005550707086,
-      "grad_norm": 1.4381723403930664,
-      "learning_rate": 0.000196417031253041,
-      "loss": 2.2542262268066406,
-      "step": 36100
-    },
-    {
-      "epoch": 0.05418975095168879,
-      "grad_norm": 1.619674801826477,
-      "learning_rate": 0.00019640705056069035,
-      "loss": 2.172146911621094,
-      "step": 36200
-    },
-    {
-      "epoch": 0.054339446396306716,
-      "grad_norm": 1.3287179470062256,
-      "learning_rate": 0.0001963970698683397,
-      "loss": 2.1780206298828126,
-      "step": 36300
-    },
-    {
-      "epoch": 0.05448914184092464,
-      "grad_norm": 1.912472128868103,
-      "learning_rate": 0.00019638708917598906,
-      "loss": 2.207314453125,
-      "step": 36400
-    },
-    {
-      "epoch": 0.054638837285542564,
-      "grad_norm": 1.3202009201049805,
-      "learning_rate": 0.0001963771084836384,
-      "loss": 2.2046878051757814,
-      "step": 36500
-    },
-    {
-      "epoch": 0.054638837285542564,
-      "eval_loss": 2.0594139099121094,
-      "eval_runtime": 3329.7448,
-      "eval_samples_per_second": 178.331,
-      "eval_steps_per_second": 22.291,
-      "step": 36500
-    },
-    {
-      "epoch": 0.05478853273016049,
-      "grad_norm": 1.199276089668274,
-      "learning_rate": 0.00019636712779128778,
-      "loss": 2.203477020263672,
-      "step": 36600
-    },
-    {
-      "epoch": 0.05493822817477841,
-      "grad_norm": 1.2738145589828491,
-      "learning_rate": 0.00019635714709893712,
-      "loss": 2.203280334472656,
-      "step": 36700
-    },
-    {
-      "epoch": 0.05508792361939634,
-      "grad_norm": 1.4286922216415405,
-      "learning_rate": 0.00019634716640658646,
-      "loss": 2.24751953125,
-      "step": 36800
-    },
-    {
-      "epoch": 0.05523761906401426,
-      "grad_norm": 1.6303517818450928,
-      "learning_rate": 0.00019633718571423583,
-      "loss": 2.2757583618164063,
-      "step": 36900
-    },
-    {
-      "epoch": 0.055387314508632185,
-      "grad_norm": 1.5175888538360596,
-      "learning_rate": 0.00019632720502188518,
-      "loss": 2.217519989013672,
-      "step": 37000
-    },
-    {
-      "epoch": 0.055387314508632185,
-      "eval_loss": 2.053313732147217,
-      "eval_runtime": 3333.5007,
-      "eval_samples_per_second": 178.13,
-      "eval_steps_per_second": 22.266,
-      "step": 37000
-    },
-    {
-      "epoch": 0.055537009953250116,
-      "grad_norm": 1.298179030418396,
-      "learning_rate": 0.00019631722432953452,
-      "loss": 2.1301605224609377,
-      "step": 37100
-    },
-    {
-      "epoch": 0.05568670539786804,
-      "grad_norm": 1.5428518056869507,
-      "learning_rate": 0.00019630724363718386,
-      "loss": 2.27997802734375,
-      "step": 37200
-    },
-    {
-      "epoch": 0.055836400842485964,
-      "grad_norm": 2.2962374687194824,
-      "learning_rate": 0.00019629726294483323,
-      "loss": 2.151377716064453,
-      "step": 37300
-    },
-    {
-      "epoch": 0.05598609628710389,
-      "grad_norm": 2.3104796409606934,
-      "learning_rate": 0.00019628728225248258,
-      "loss": 2.1969725036621095,
-      "step": 37400
-    },
-    {
-      "epoch": 0.05613579173172181,
-      "grad_norm": 1.3915497064590454,
-      "learning_rate": 0.00019627730156013192,
-      "loss": 2.1856144714355468,
-      "step": 37500
-    },
-    {
-      "epoch": 0.05613579173172181,
-      "eval_loss": 2.053218126296997,
-      "eval_runtime": 3326.8491,
-      "eval_samples_per_second": 178.487,
-      "eval_steps_per_second": 22.311,
-      "step": 37500
-    },
-    {
-      "epoch": 0.05628548717633974,
-      "grad_norm": 1.5029561519622803,
-      "learning_rate": 0.0001962673208677813,
-      "loss": 2.2035781860351564,
-      "step": 37600
-    },
-    {
-      "epoch": 0.05643518262095766,
-      "grad_norm": 1.7210547924041748,
-      "learning_rate": 0.00019625734017543066,
-      "loss": 2.20986083984375,
-      "step": 37700
-    },
-    {
-      "epoch": 0.056584878065575585,
-      "grad_norm": 1.545599341392517,
-      "learning_rate": 0.00019624735948308,
-      "loss": 2.1841116333007813,
-      "step": 37800
-    },
-    {
-      "epoch": 0.05673457351019351,
-      "grad_norm": 1.63248872756958,
-      "learning_rate": 0.00019623737879072934,
-      "loss": 2.1510467529296875,
-      "step": 37900
-    },
-    {
-      "epoch": 0.05688426895481143,
-      "grad_norm": 1.535138487815857,
-      "learning_rate": 0.0001962273980983787,
-      "loss": 2.2223579406738283,
-      "step": 38000
-    },
-    {
-      "epoch": 0.05688426895481143,
-      "eval_loss": 2.052565336227417,
-      "eval_runtime": 3328.4829,
-      "eval_samples_per_second": 178.399,
-      "eval_steps_per_second": 22.3,
-      "step": 38000
-    },
-    {
-      "epoch": 0.057033964399429364,
-      "grad_norm": 1.4007747173309326,
-      "learning_rate": 0.00019621741740602806,
-      "loss": 2.202989044189453,
-      "step": 38100
-    },
-    {
-      "epoch": 0.05718365984404729,
-      "grad_norm": 1.7556818723678589,
-      "learning_rate": 0.0001962074367136774,
-      "loss": 2.2285475158691406,
-      "step": 38200
-    },
-    {
-      "epoch": 0.05733335528866521,
-      "grad_norm": 1.6197268962860107,
-      "learning_rate": 0.00019619745602132674,
-      "loss": 2.177320556640625,
-      "step": 38300
-    },
-    {
-      "epoch": 0.057483050733283136,
-      "grad_norm": 1.3252156972885132,
-      "learning_rate": 0.0001961874753289761,
-      "loss": 2.2028353881835936,
-      "step": 38400
-    },
-    {
-      "epoch": 0.05763274617790106,
-      "grad_norm": 1.3603196144104004,
-      "learning_rate": 0.00019617749463662546,
-      "loss": 2.2576800537109376,
-      "step": 38500
-    },
-    {
-      "epoch": 0.05763274617790106,
-      "eval_loss": 2.049717903137207,
-      "eval_runtime": 3336.9559,
-      "eval_samples_per_second": 177.946,
-      "eval_steps_per_second": 22.243,
-      "step": 38500
-    },
-    {
-      "epoch": 0.057782441622518985,
-      "grad_norm": 1.6559139490127563,
-      "learning_rate": 0.00019616751394427483,
-      "loss": 2.199705047607422,
-      "step": 38600
-    },
-    {
-      "epoch": 0.05793213706713691,
-      "grad_norm": 2.397522449493408,
-      "learning_rate": 0.00019615753325192417,
-      "loss": 2.2045123291015627,
-      "step": 38700
-    },
-    {
-      "epoch": 0.05808183251175483,
-      "grad_norm": 1.3550915718078613,
-      "learning_rate": 0.0001961475525595735,
-      "loss": 2.205120849609375,
-      "step": 38800
-    },
-    {
-      "epoch": 0.05823152795637276,
-      "grad_norm": 1.2697638273239136,
-      "learning_rate": 0.00019613757186722286,
-      "loss": 2.209880065917969,
-      "step": 38900
-    },
-    {
-      "epoch": 0.05838122340099068,
-      "grad_norm": 1.2588802576065063,
-      "learning_rate": 0.00019612759117487223,
-      "loss": 2.1797088623046874,
-      "step": 39000
-    },
-    {
-      "epoch": 0.05838122340099068,
-      "eval_loss": 2.0538113117218018,
-      "eval_runtime": 3338.567,
-      "eval_samples_per_second": 177.86,
-      "eval_steps_per_second": 22.233,
-      "step": 39000
-    },
-    {
-      "epoch": 0.05853091884560861,
-      "grad_norm": 1.6260154247283936,
-      "learning_rate": 0.00019611761048252157,
-      "loss": 2.192071990966797,
-      "step": 39100
-    },
-    {
-      "epoch": 0.058680614290226536,
-      "grad_norm": 1.7198734283447266,
-      "learning_rate": 0.0001961076297901709,
-      "loss": 2.1906195068359375,
-      "step": 39200
-    },
-    {
-      "epoch": 0.05883030973484446,
-      "grad_norm": 1.6451668739318848,
-      "learning_rate": 0.00019609764909782026,
-      "loss": 2.2079037475585936,
-      "step": 39300
-    },
-    {
-      "epoch": 0.058980005179462384,
-      "grad_norm": 1.6980173587799072,
-      "learning_rate": 0.00019608766840546963,
-      "loss": 2.2185546875,
-      "step": 39400
-    },
-    {
-      "epoch": 0.05912970062408031,
-      "grad_norm": 1.511583685874939,
-      "learning_rate": 0.000196077687713119,
-      "loss": 2.1794978332519532,
-      "step": 39500
-    },
-    {
-      "epoch": 0.05912970062408031,
-      "eval_loss": 2.058593273162842,
-      "eval_runtime": 3331.7965,
-      "eval_samples_per_second": 178.222,
-      "eval_steps_per_second": 22.278,
-      "step": 39500
-    },
-    {
-      "epoch": 0.05927939606869823,
-      "grad_norm": 1.3856468200683594,
-      "learning_rate": 0.00019606770702076834,
-      "loss": 2.148001251220703,
-      "step": 39600
-    },
-    {
-      "epoch": 0.05942909151331616,
-      "grad_norm": 1.523389458656311,
-      "learning_rate": 0.00019605772632841768,
-      "loss": 2.2308888244628906,
-      "step": 39700
-    },
-    {
-      "epoch": 0.05957878695793408,
-      "grad_norm": 2.061485528945923,
-      "learning_rate": 0.00019604774563606705,
-      "loss": 2.21391357421875,
-      "step": 39800
-    },
-    {
-      "epoch": 0.059728482402552005,
-      "grad_norm": 1.5004793405532837,
-      "learning_rate": 0.0001960377649437164,
-      "loss": 2.1756239318847657,
-      "step": 39900
-    },
-    {
-      "epoch": 0.059878177847169936,
-      "grad_norm": 1.706130027770996,
-      "learning_rate": 0.00019602778425136574,
-      "loss": 2.2128558349609375,
-      "step": 40000
-    },
-    {
-      "epoch": 0.059878177847169936,
-      "eval_loss": 2.053236246109009,
-      "eval_runtime": 3339.1094,
-      "eval_samples_per_second": 177.831,
-      "eval_steps_per_second": 22.229,
-      "step": 40000
-    },
-    {
-      "epoch": 0.06002787329178786,
-      "grad_norm": 1.7956801652908325,
-      "learning_rate": 0.00019601780355901508,
-      "loss": 2.167734375,
-      "step": 40100
-    },
-    {
-      "epoch": 0.060177568736405784,
-      "grad_norm": 1.720413327217102,
-      "learning_rate": 0.00019600782286666442,
-      "loss": 2.2549983215332032,
-      "step": 40200
-    },
-    {
-      "epoch": 0.06032726418102371,
-      "grad_norm": 1.4791420698165894,
-      "learning_rate": 0.0001959978421743138,
-      "loss": 2.1937921142578123,
-      "step": 40300
-    },
-    {
-      "epoch": 0.06047695962564163,
-      "grad_norm": 1.5837610960006714,
-      "learning_rate": 0.00019598786148196316,
-      "loss": 2.2013475036621095,
-      "step": 40400
-    },
-    {
-      "epoch": 0.06062665507025956,
-      "grad_norm": 1.3171435594558716,
-      "learning_rate": 0.0001959778807896125,
-      "loss": 2.154152374267578,
-      "step": 40500
-    },
-    {
-      "epoch": 0.06062665507025956,
-      "eval_loss": 2.0521433353424072,
-      "eval_runtime": 3339.7904,
-      "eval_samples_per_second": 177.795,
-      "eval_steps_per_second": 22.224,
-      "step": 40500
-    },
-    {
-      "epoch": 0.06077635051487748,
-      "grad_norm": 1.6859766244888306,
-      "learning_rate": 0.00019596790009726185,
-      "loss": 2.237109375,
-      "step": 40600
-    },
-    {
-      "epoch": 0.060926045959495405,
-      "grad_norm": 1.6800007820129395,
-      "learning_rate": 0.00019595791940491122,
-      "loss": 2.2601802062988283,
-      "step": 40700
-    },
-    {
-      "epoch": 0.06107574140411333,
-      "grad_norm": 1.3269203901290894,
-      "learning_rate": 0.00019594793871256056,
-      "loss": 2.2489947509765624,
-      "step": 40800
-    },
-    {
-      "epoch": 0.06122543684873125,
-      "grad_norm": 1.71269690990448,
-      "learning_rate": 0.0001959379580202099,
-      "loss": 2.214556884765625,
-      "step": 40900
-    },
-    {
-      "epoch": 0.061375132293349184,
-      "grad_norm": 1.4304598569869995,
-      "learning_rate": 0.00019592797732785925,
-      "loss": 2.2044895935058593,
-      "step": 41000
-    },
-    {
-      "epoch": 0.061375132293349184,
-      "eval_loss": 2.051612138748169,
-      "eval_runtime": 3343.5742,
-      "eval_samples_per_second": 177.594,
-      "eval_steps_per_second": 22.199,
-      "step": 41000
-    },
-    {
-      "epoch": 0.06152482773796711,
-      "grad_norm": 1.458297848701477,
-      "learning_rate": 0.00019591799663550862,
-      "loss": 2.2664784240722655,
-      "step": 41100
-    },
-    {
-      "epoch": 0.06167452318258503,
-      "grad_norm": 1.4439661502838135,
-      "learning_rate": 0.00019590801594315796,
-      "loss": 2.210080108642578,
-      "step": 41200
-    },
-    {
-      "epoch": 0.061824218627202956,
-      "grad_norm": 1.6031644344329834,
-      "learning_rate": 0.00019589803525080733,
-      "loss": 2.194513244628906,
-      "step": 41300
-    },
-    {
-      "epoch": 0.06197391407182088,
-      "grad_norm": 1.5625102519989014,
-      "learning_rate": 0.00019588805455845668,
-      "loss": 2.206995086669922,
-      "step": 41400
-    },
-    {
-      "epoch": 0.062123609516438805,
-      "grad_norm": 1.6364563703536987,
-      "learning_rate": 0.00019587807386610605,
-      "loss": 2.209018096923828,
-      "step": 41500
-    },
-    {
-      "epoch": 0.062123609516438805,
-      "eval_loss": 2.044764280319214,
-      "eval_runtime": 3335.463,
-      "eval_samples_per_second": 178.026,
-      "eval_steps_per_second": 22.253,
-      "step": 41500
-    },
-    {
-      "epoch": 0.06227330496105673,
-      "grad_norm": 1.3473576307296753,
-      "learning_rate": 0.0001958680931737554,
-      "loss": 2.1620524597167967,
-      "step": 41600
-    },
-    {
-      "epoch": 0.06242300040567465,
-      "grad_norm": 1.9848613739013672,
-      "learning_rate": 0.00019585811248140473,
-      "loss": 2.207513732910156,
-      "step": 41700
-    },
-    {
-      "epoch": 0.06257269585029258,
-      "grad_norm": 1.4785856008529663,
-      "learning_rate": 0.00019584813178905407,
-      "loss": 2.2043218994140625,
-      "step": 41800
-    },
-    {
-      "epoch": 0.0627223912949105,
-      "grad_norm": 1.3069367408752441,
-      "learning_rate": 0.00019583815109670342,
-      "loss": 2.1929408264160157,
-      "step": 41900
-    },
-    {
-      "epoch": 0.06287208673952843,
-      "grad_norm": 1.4278662204742432,
-      "learning_rate": 0.0001958281704043528,
-      "loss": 2.226029052734375,
-      "step": 42000
-    },
-    {
-      "epoch": 0.06287208673952843,
-      "eval_loss": 2.044788122177124,
-      "eval_runtime": 3337.3419,
-      "eval_samples_per_second": 177.925,
-      "eval_steps_per_second": 22.241,
-      "step": 42000
-    },
-    {
-      "epoch": 0.06302178218414635,
-      "grad_norm": 1.6515027284622192,
-      "learning_rate": 0.00019581818971200213,
-      "loss": 2.2044723510742186,
-      "step": 42100
-    },
-    {
-      "epoch": 0.06317147762876428,
-      "grad_norm": 1.5539337396621704,
-      "learning_rate": 0.0001958082090196515,
-      "loss": 2.1681309509277344,
-      "step": 42200
-    },
-    {
-      "epoch": 0.0633211730733822,
-      "grad_norm": 1.8599845170974731,
-      "learning_rate": 0.00019579822832730084,
-      "loss": 2.2122256469726564,
-      "step": 42300
-    },
-    {
-      "epoch": 0.06347086851800013,
-      "grad_norm": 1.6595927476882935,
-      "learning_rate": 0.00019578824763495021,
-      "loss": 2.1658042907714843,
-      "step": 42400
-    },
-    {
-      "epoch": 0.06362056396261806,
-      "grad_norm": 1.2584881782531738,
-      "learning_rate": 0.00019577826694259956,
-      "loss": 2.206346282958984,
-      "step": 42500
-    },
-    {
-      "epoch": 0.06362056396261806,
-      "eval_loss": 2.051701307296753,
-      "eval_runtime": 3341.0355,
-      "eval_samples_per_second": 177.729,
-      "eval_steps_per_second": 22.216,
-      "step": 42500
-    },
-    {
-      "epoch": 0.06377025940723598,
-      "grad_norm": 1.4971222877502441,
-      "learning_rate": 0.0001957682862502489,
-      "loss": 2.176647033691406,
-      "step": 42600
-    },
-    {
-      "epoch": 0.06391995485185391,
-      "grad_norm": 1.5352060794830322,
-      "learning_rate": 0.00019575830555789824,
-      "loss": 2.2158883666992186,
-      "step": 42700
-    },
-    {
-      "epoch": 0.06406965029647183,
-      "grad_norm": 1.5945316553115845,
-      "learning_rate": 0.0001957483248655476,
-      "loss": 2.2141468811035154,
-      "step": 42800
-    },
-    {
-      "epoch": 0.06421934574108976,
-      "grad_norm": 1.598465919494629,
-      "learning_rate": 0.00019573834417319696,
-      "loss": 2.2058779907226564,
-      "step": 42900
-    },
-    {
-      "epoch": 0.06436904118570767,
-      "grad_norm": 1.5797964334487915,
-      "learning_rate": 0.0001957283634808463,
-      "loss": 2.2259124755859374,
-      "step": 43000
-    },
-    {
-      "epoch": 0.06436904118570767,
-      "eval_loss": 2.0464677810668945,
-      "eval_runtime": 3343.2683,
-      "eval_samples_per_second": 177.61,
-      "eval_steps_per_second": 22.201,
-      "step": 43000
-    },
-    {
-      "epoch": 0.0645187366303256,
-      "grad_norm": 1.6346147060394287,
-      "learning_rate": 0.00019571838278849567,
-      "loss": 2.164223480224609,
-      "step": 43100
-    },
-    {
-      "epoch": 0.06466843207494352,
-      "grad_norm": 1.610613465309143,
-      "learning_rate": 0.00019570840209614504,
-      "loss": 2.2035809326171876,
-      "step": 43200
-    },
-    {
-      "epoch": 0.06481812751956145,
-      "grad_norm": 1.5161710977554321,
-      "learning_rate": 0.00019569842140379438,
-      "loss": 2.2403134155273436,
-      "step": 43300
-    },
-    {
-      "epoch": 0.06496782296417938,
-      "grad_norm": 1.2547993659973145,
-      "learning_rate": 0.00019568844071144373,
-      "loss": 2.2077256774902345,
-      "step": 43400
-    },
-    {
-      "epoch": 0.0651175184087973,
-      "grad_norm": 1.3896839618682861,
-      "learning_rate": 0.00019567846001909307,
-      "loss": 2.195224609375,
-      "step": 43500
-    },
-    {
-      "epoch": 0.0651175184087973,
-      "eval_loss": 2.044255256652832,
-      "eval_runtime": 3353.1494,
-      "eval_samples_per_second": 177.087,
-      "eval_steps_per_second": 22.136,
-      "step": 43500
-    },
-    {
-      "epoch": 0.06526721385341523,
-      "grad_norm": 1.8882007598876953,
-      "learning_rate": 0.0001956684793267424,
-      "loss": 2.226784210205078,
-      "step": 43600
-    },
-    {
-      "epoch": 0.06541690929803315,
-      "grad_norm": 1.724118947982788,
-      "learning_rate": 0.00019565849863439178,
-      "loss": 2.2329942321777345,
-      "step": 43700
-    },
-    {
-      "epoch": 0.06556660474265108,
-      "grad_norm": 1.701139211654663,
-      "learning_rate": 0.00019564851794204112,
-      "loss": 2.197763214111328,
-      "step": 43800
-    },
-    {
-      "epoch": 0.065716300187269,
-      "grad_norm": 1.3662222623825073,
-      "learning_rate": 0.00019563853724969047,
-      "loss": 2.204658966064453,
-      "step": 43900
-    },
-    {
-      "epoch": 0.06586599563188693,
-      "grad_norm": 1.4381604194641113,
-      "learning_rate": 0.00019562855655733984,
-      "loss": 2.224393615722656,
-      "step": 44000
-    },
-    {
-      "epoch": 0.06586599563188693,
-      "eval_loss": 2.043705701828003,
-      "eval_runtime": 3348.7091,
-      "eval_samples_per_second": 177.321,
-      "eval_steps_per_second": 22.165,
-      "step": 44000
-    },
-    {
-      "epoch": 0.06601569107650485,
-      "grad_norm": 1.6465290784835815,
-      "learning_rate": 0.0001956185758649892,
-      "loss": 2.1961631774902344,
-      "step": 44100
-    },
-    {
-      "epoch": 0.06616538652112278,
-      "grad_norm": 1.525381088256836,
-      "learning_rate": 0.00019560859517263855,
-      "loss": 2.198580017089844,
-      "step": 44200
-    },
-    {
-      "epoch": 0.0663150819657407,
-      "grad_norm": 1.977921724319458,
-      "learning_rate": 0.0001955986144802879,
-      "loss": 2.221914978027344,
-      "step": 44300
-    },
-    {
-      "epoch": 0.06646477741035862,
-      "grad_norm": 1.5670602321624756,
-      "learning_rate": 0.00019558863378793724,
-      "loss": 2.217979736328125,
-      "step": 44400
-    },
-    {
-      "epoch": 0.06661447285497656,
-      "grad_norm": 1.8623807430267334,
-      "learning_rate": 0.0001955786530955866,
-      "loss": 2.2176544189453127,
-      "step": 44500
-    },
-    {
-      "epoch": 0.06661447285497656,
-      "eval_loss": 2.0461344718933105,
-      "eval_runtime": 3349.0038,
-      "eval_samples_per_second": 177.306,
-      "eval_steps_per_second": 22.163,
-      "step": 44500
-    },
-    {
-      "epoch": 0.06676416829959447,
-      "grad_norm": 1.60882568359375,
-      "learning_rate": 0.00019556867240323595,
-      "loss": 2.1415998840332033,
-      "step": 44600
-    },
-    {
-      "epoch": 0.0669138637442124,
-      "grad_norm": 1.4321738481521606,
-      "learning_rate": 0.0001955586917108853,
-      "loss": 2.2176394653320313,
-      "step": 44700
-    },
-    {
-      "epoch": 0.06706355918883032,
-      "grad_norm": 1.4379962682724,
-      "learning_rate": 0.00019554871101853464,
-      "loss": 2.1561395263671876,
-      "step": 44800
-    },
-    {
-      "epoch": 0.06721325463344825,
-      "grad_norm": 1.5312572717666626,
-      "learning_rate": 0.000195538730326184,
-      "loss": 2.210959167480469,
-      "step": 44900
-    },
-    {
-      "epoch": 0.06736295007806617,
-      "grad_norm": 1.5002049207687378,
-      "learning_rate": 0.00019552874963383338,
-      "loss": 2.159567565917969,
-      "step": 45000
-    },
-    {
-      "epoch": 0.06736295007806617,
-      "eval_loss": 2.042153835296631,
-      "eval_runtime": 3350.6062,
-      "eval_samples_per_second": 177.221,
-      "eval_steps_per_second": 22.153,
-      "step": 45000
-    },
-    {
-      "epoch": 0.0675126455226841,
-      "grad_norm": 1.2220619916915894,
-      "learning_rate": 0.00019551876894148272,
-      "loss": 2.218625793457031,
-      "step": 45100
-    },
-    {
-      "epoch": 0.06766234096730202,
-      "grad_norm": 1.63571035861969,
-      "learning_rate": 0.00019550878824913206,
-      "loss": 2.179343566894531,
-      "step": 45200
-    },
-    {
-      "epoch": 0.06781203641191995,
-      "grad_norm": 1.4634140729904175,
-      "learning_rate": 0.0001954988075567814,
-      "loss": 2.2385093688964846,
-      "step": 45300
-    },
-    {
-      "epoch": 0.06796173185653788,
-      "grad_norm": 1.603217601776123,
-      "learning_rate": 0.00019548882686443078,
-      "loss": 2.1959580993652343,
-      "step": 45400
-    },
-    {
-      "epoch": 0.0681114273011558,
-      "grad_norm": 1.5302245616912842,
-      "learning_rate": 0.00019547884617208012,
-      "loss": 2.1390216064453127,
-      "step": 45500
-    },
-    {
-      "epoch": 0.0681114273011558,
-      "eval_loss": 2.0404775142669678,
-      "eval_runtime": 3352.1312,
-      "eval_samples_per_second": 177.14,
-      "eval_steps_per_second": 22.143,
-      "step": 45500
-    },
-    {
-      "epoch": 0.06826112274577373,
-      "grad_norm": 1.4835774898529053,
-      "learning_rate": 0.00019546886547972946,
-      "loss": 2.2127622985839843,
-      "step": 45600
-    },
-    {
-      "epoch": 0.06841081819039165,
-      "grad_norm": 1.431746244430542,
-      "learning_rate": 0.0001954588847873788,
-      "loss": 2.1848431396484376,
-      "step": 45700
-    },
-    {
-      "epoch": 0.06856051363500958,
-      "grad_norm": 1.6252113580703735,
-      "learning_rate": 0.00019544890409502817,
-      "loss": 2.1899786376953125,
-      "step": 45800
-    },
-    {
-      "epoch": 0.0687102090796275,
-      "grad_norm": 1.6457818746566772,
-      "learning_rate": 0.00019543892340267755,
-      "loss": 2.21218017578125,
-      "step": 45900
-    },
-    {
-      "epoch": 0.06885990452424542,
-      "grad_norm": 1.7759894132614136,
-      "learning_rate": 0.0001954289427103269,
-      "loss": 2.2190863037109376,
-      "step": 46000
-    },
-    {
-      "epoch": 0.06885990452424542,
-      "eval_loss": 2.04162335395813,
-      "eval_runtime": 3348.7521,
-      "eval_samples_per_second": 177.319,
-      "eval_steps_per_second": 22.165,
-      "step": 46000
-    },
-    {
-      "epoch": 0.06900959996886334,
-      "grad_norm": 1.6779075860977173,
-      "learning_rate": 0.00019541896201797623,
-      "loss": 2.259061737060547,
-      "step": 46100
-    },
-    {
-      "epoch": 0.06915929541348127,
-      "grad_norm": 1.5484764575958252,
-      "learning_rate": 0.0001954089813256256,
-      "loss": 2.200908203125,
-      "step": 46200
-    },
-    {
-      "epoch": 0.0693089908580992,
-      "grad_norm": 1.517712950706482,
-      "learning_rate": 0.00019539900063327494,
-      "loss": 2.17314697265625,
-      "step": 46300
-    },
-    {
-      "epoch": 0.06945868630271712,
-      "grad_norm": 1.4977158308029175,
-      "learning_rate": 0.0001953890199409243,
-      "loss": 2.220084686279297,
-      "step": 46400
-    },
-    {
-      "epoch": 0.06960838174733505,
-      "grad_norm": 1.5509703159332275,
-      "learning_rate": 0.00019537903924857363,
-      "loss": 2.2040330505371095,
-      "step": 46500
-    },
-    {
-      "epoch": 0.06960838174733505,
-      "eval_loss": 2.0413084030151367,
-      "eval_runtime": 3355.4097,
-      "eval_samples_per_second": 176.967,
-      "eval_steps_per_second": 22.121,
-      "step": 46500
-    },
-    {
-      "epoch": 0.06975807719195297,
-      "grad_norm": 1.5994006395339966,
-      "learning_rate": 0.000195369058556223,
-      "loss": 2.179085388183594,
-      "step": 46600
-    },
-    {
-      "epoch": 0.0699077726365709,
-      "grad_norm": 1.8047237396240234,
-      "learning_rate": 0.00019535907786387234,
-      "loss": 2.213885955810547,
-      "step": 46700
-    },
-    {
-      "epoch": 0.07005746808118882,
-      "grad_norm": 1.3791507482528687,
-      "learning_rate": 0.0001953490971715217,
-      "loss": 2.160697784423828,
-      "step": 46800
-    },
-    {
-      "epoch": 0.07020716352580675,
-      "grad_norm": 1.4790291786193848,
-      "learning_rate": 0.00019533911647917106,
-      "loss": 2.1245072937011718,
-      "step": 46900
-    },
-    {
-      "epoch": 0.07035685897042467,
-      "grad_norm": 1.5957084894180298,
-      "learning_rate": 0.00019532913578682043,
-      "loss": 2.2148281860351564,
-      "step": 47000
-    },
-    {
-      "epoch": 0.07035685897042467,
-      "eval_loss": 2.04024338722229,
-      "eval_runtime": 3356.3679,
-      "eval_samples_per_second": 176.917,
-      "eval_steps_per_second": 22.115,
-      "step": 47000
-    },
-    {
-      "epoch": 0.0705065544150426,
-      "grad_norm": 1.7693493366241455,
-      "learning_rate": 0.00019531915509446977,
-      "loss": 2.199380645751953,
-      "step": 47100
-    },
-    {
-      "epoch": 0.07065624985966051,
-      "grad_norm": 1.5178700685501099,
-      "learning_rate": 0.0001953091744021191,
-      "loss": 2.2289695739746094,
-      "step": 47200
-    },
-    {
-      "epoch": 0.07080594530427844,
-      "grad_norm": 1.219441533088684,
-      "learning_rate": 0.00019529919370976846,
-      "loss": 2.1515988159179686,
-      "step": 47300
-    },
-    {
-      "epoch": 0.07095564074889638,
-      "grad_norm": 1.6566627025604248,
-      "learning_rate": 0.0001952892130174178,
-      "loss": 2.200910186767578,
-      "step": 47400
-    },
-    {
-      "epoch": 0.07110533619351429,
-      "grad_norm": 1.2569252252578735,
-      "learning_rate": 0.00019527923232506717,
-      "loss": 2.180551300048828,
-      "step": 47500
-    },
-    {
-      "epoch": 0.07110533619351429,
-      "eval_loss": 2.0403616428375244,
-      "eval_runtime": 3358.1702,
-      "eval_samples_per_second": 176.822,
-      "eval_steps_per_second": 22.103,
-      "step": 47500
-    },
-    {
-      "epoch": 0.07125503163813222,
-      "grad_norm": 1.5463579893112183,
-      "learning_rate": 0.0001952692516327165,
-      "loss": 2.1733158874511718,
-      "step": 47600
-    },
-    {
-      "epoch": 0.07140472708275014,
-      "grad_norm": 1.5307575464248657,
-      "learning_rate": 0.00019525927094036585,
-      "loss": 2.1878355407714842,
-      "step": 47700
-    },
-    {
-      "epoch": 0.07155442252736807,
-      "grad_norm": 1.491356611251831,
-      "learning_rate": 0.00019524929024801522,
-      "loss": 2.186581268310547,
-      "step": 47800
-    },
-    {
-      "epoch": 0.07170411797198599,
-      "grad_norm": 1.4641207456588745,
-      "learning_rate": 0.0001952393095556646,
-      "loss": 2.210890350341797,
-      "step": 47900
-    },
-    {
-      "epoch": 0.07185381341660392,
-      "grad_norm": 1.8117883205413818,
-      "learning_rate": 0.00019522932886331394,
-      "loss": 2.2362747192382812,
-      "step": 48000
-    },
-    {
-      "epoch": 0.07185381341660392,
-      "eval_loss": 2.0355639457702637,
-      "eval_runtime": 3356.7852,
-      "eval_samples_per_second": 176.895,
-      "eval_steps_per_second": 22.112,
-      "step": 48000
-    },
-    {
-      "epoch": 0.07200350886122184,
-      "grad_norm": 1.2147129774093628,
-      "learning_rate": 0.00019521934817096328,
-      "loss": 2.188218231201172,
-      "step": 48100
-    },
-    {
-      "epoch": 0.07215320430583977,
-      "grad_norm": 1.492746353149414,
-      "learning_rate": 0.00019520936747861262,
-      "loss": 2.1469573974609375,
-      "step": 48200
-    },
-    {
-      "epoch": 0.0723028997504577,
-      "grad_norm": 1.5810075998306274,
-      "learning_rate": 0.000195199386786262,
-      "loss": 2.2155390930175782,
-      "step": 48300
-    },
-    {
-      "epoch": 0.07245259519507562,
-      "grad_norm": 1.3962161540985107,
-      "learning_rate": 0.00019518940609391134,
-      "loss": 2.2096728515625,
-      "step": 48400
-    },
-    {
-      "epoch": 0.07260229063969355,
-      "grad_norm": 1.5837905406951904,
-      "learning_rate": 0.00019517942540156068,
-      "loss": 2.1907479858398435,
-      "step": 48500
-    },
-    {
-      "epoch": 0.07260229063969355,
-      "eval_loss": 2.033581256866455,
-      "eval_runtime": 3358.3425,
-      "eval_samples_per_second": 176.813,
-      "eval_steps_per_second": 22.102,
-      "step": 48500
-    },
-    {
-      "epoch": 0.07275198608431147,
-      "grad_norm": 1.6800642013549805,
-      "learning_rate": 0.00019516944470921002,
-      "loss": 2.236597900390625,
-      "step": 48600
-    },
-    {
-      "epoch": 0.0729016815289294,
-      "grad_norm": 1.3904973268508911,
-      "learning_rate": 0.0001951594640168594,
-      "loss": 2.176927032470703,
-      "step": 48700
-    },
-    {
-      "epoch": 0.07305137697354731,
-      "grad_norm": 1.318107008934021,
-      "learning_rate": 0.00019514948332450876,
-      "loss": 2.1542703247070314,
-      "step": 48800
-    },
-    {
-      "epoch": 0.07320107241816524,
-      "grad_norm": 1.5037977695465088,
-      "learning_rate": 0.0001951395026321581,
-      "loss": 2.1732440185546875,
-      "step": 48900
-    },
-    {
-      "epoch": 0.07335076786278316,
-      "grad_norm": 1.456257700920105,
-      "learning_rate": 0.00019512952193980745,
-      "loss": 2.1571356201171876,
-      "step": 49000
-    },
-    {
-      "epoch": 0.07335076786278316,
-      "eval_loss": 2.0312538146972656,
-      "eval_runtime": 3357.3883,
-      "eval_samples_per_second": 176.863,
-      "eval_steps_per_second": 22.108,
-      "step": 49000
-    },
-    {
-      "epoch": 0.07350046330740109,
-      "grad_norm": 1.4943408966064453,
-      "learning_rate": 0.0001951195412474568,
-      "loss": 2.1779771423339844,
-      "step": 49100
-    },
-    {
-      "epoch": 0.07365015875201902,
-      "grad_norm": 1.7443143129348755,
-      "learning_rate": 0.00019510956055510616,
-      "loss": 2.2168931579589843,
-      "step": 49200
-    },
-    {
-      "epoch": 0.07379985419663694,
-      "grad_norm": 1.4758964776992798,
-      "learning_rate": 0.0001950995798627555,
-      "loss": 2.2154759216308593,
-      "step": 49300
-    },
-    {
-      "epoch": 0.07394954964125487,
-      "grad_norm": 1.4002784490585327,
-      "learning_rate": 0.00019508959917040485,
-      "loss": 2.2180252075195312,
-      "step": 49400
-    },
-    {
-      "epoch": 0.07409924508587279,
-      "grad_norm": 2.2416744232177734,
-      "learning_rate": 0.0001950796184780542,
-      "loss": 2.166342315673828,
-      "step": 49500
-    },
-    {
-      "epoch": 0.07409924508587279,
-      "eval_loss": 2.0321226119995117,
-      "eval_runtime": 3354.3771,
-      "eval_samples_per_second": 177.022,
-      "eval_steps_per_second": 22.128,
-      "step": 49500
-    },
-    {
-      "epoch": 0.07424894053049072,
-      "grad_norm": 1.8986939191818237,
-      "learning_rate": 0.00019506963778570356,
-      "loss": 2.2061473083496095,
-      "step": 49600
-    },
-    {
-      "epoch": 0.07439863597510864,
-      "grad_norm": 1.4825496673583984,
-      "learning_rate": 0.00019505965709335293,
-      "loss": 2.2470991516113283,
-      "step": 49700
-    },
-    {
-      "epoch": 0.07454833141972657,
-      "grad_norm": 1.4881237745285034,
-      "learning_rate": 0.00019504967640100227,
-      "loss": 2.1634156799316404,
-      "step": 49800
-    },
-    {
-      "epoch": 0.07469802686434449,
-      "grad_norm": 1.707650065422058,
-      "learning_rate": 0.00019503969570865162,
-      "loss": 2.2187210083007813,
-      "step": 49900
-    },
-    {
-      "epoch": 0.07484772230896242,
-      "grad_norm": 1.6508225202560425,
-      "learning_rate": 0.000195029715016301,
-      "loss": 2.207003326416016,
-      "step": 50000
-    },
-    {
-      "epoch": 0.07484772230896242,
-      "eval_loss": 2.0348145961761475,
-      "eval_runtime": 3360.0667,
-      "eval_samples_per_second": 176.722,
-      "eval_steps_per_second": 22.09,
-      "step": 50000
-    },
-    {
-      "epoch": 0.07499741775358035,
-      "grad_norm": 1.630660057067871,
-      "learning_rate": 0.00019501973432395033,
-      "loss": 2.199310302734375,
-      "step": 50100
-    },
-    {
-      "epoch": 0.07514711319819826,
-      "grad_norm": 1.4535642862319946,
-      "learning_rate": 0.00019500975363159967,
-      "loss": 2.1751918029785156,
-      "step": 50200
-    },
-    {
-      "epoch": 0.0752968086428162,
-      "grad_norm": 1.446770429611206,
-      "learning_rate": 0.00019499977293924902,
-      "loss": 2.2047380065917968,
-      "step": 50300
-    },
-    {
-      "epoch": 0.07544650408743411,
-      "grad_norm": 1.3608453273773193,
-      "learning_rate": 0.0001949897922468984,
-      "loss": 2.210552978515625,
-      "step": 50400
-    },
-    {
-      "epoch": 0.07559619953205204,
-      "grad_norm": 1.2756270170211792,
-      "learning_rate": 0.00019497981155454773,
-      "loss": 2.2432513427734375,
-      "step": 50500
-    },
-    {
-      "epoch": 0.07559619953205204,
-      "eval_loss": 2.0331599712371826,
-      "eval_runtime": 3355.6567,
-      "eval_samples_per_second": 176.954,
-      "eval_steps_per_second": 22.119,
-      "step": 50500
-    },
-    {
-      "epoch": 0.07574589497666996,
-      "grad_norm": 1.1556885242462158,
-      "learning_rate": 0.0001949698308621971,
-      "loss": 2.2139823913574217,
-      "step": 50600
-    },
-    {
-      "epoch": 0.07589559042128789,
-      "grad_norm": 1.85573410987854,
-      "learning_rate": 0.00019495985016984644,
-      "loss": 2.179137878417969,
-      "step": 50700
-    },
-    {
-      "epoch": 0.07604528586590581,
-      "grad_norm": 1.9327566623687744,
-      "learning_rate": 0.00019494986947749579,
-      "loss": 2.1722471618652346,
-      "step": 50800
-    },
-    {
-      "epoch": 0.07619498131052374,
-      "grad_norm": 1.286449670791626,
-      "learning_rate": 0.00019493988878514516,
-      "loss": 2.2122433471679686,
-      "step": 50900
-    },
-    {
-      "epoch": 0.07634467675514166,
-      "grad_norm": 1.7619222402572632,
-      "learning_rate": 0.0001949299080927945,
-      "loss": 2.195132598876953,
-      "step": 51000
-    },
-    {
-      "epoch": 0.07634467675514166,
-      "eval_loss": 2.0331614017486572,
-      "eval_runtime": 3349.5701,
-      "eval_samples_per_second": 177.276,
-      "eval_steps_per_second": 22.16,
-      "step": 51000
-    },
-    {
-      "epoch": 0.07649437219975959,
-      "grad_norm": 1.5062743425369263,
-      "learning_rate": 0.00019491992740044384,
-      "loss": 2.1763368225097657,
-      "step": 51100
-    },
-    {
-      "epoch": 0.07664406764437752,
-      "grad_norm": 1.3758668899536133,
-      "learning_rate": 0.00019490994670809319,
-      "loss": 2.2346237182617186,
-      "step": 51200
-    },
-    {
-      "epoch": 0.07679376308899544,
-      "grad_norm": 2.2197225093841553,
-      "learning_rate": 0.00019489996601574256,
-      "loss": 2.183681640625,
-      "step": 51300
-    },
-    {
-      "epoch": 0.07694345853361337,
-      "grad_norm": 1.3164092302322388,
-      "learning_rate": 0.0001948899853233919,
-      "loss": 2.2316912841796874,
-      "step": 51400
-    },
-    {
-      "epoch": 0.07709315397823129,
-      "grad_norm": 1.9624909162521362,
-      "learning_rate": 0.00019488000463104127,
-      "loss": 2.181700897216797,
-      "step": 51500
-    },
-    {
-      "epoch": 0.07709315397823129,
-      "eval_loss": 2.0346908569335938,
-      "eval_runtime": 3355.3439,
-      "eval_samples_per_second": 176.971,
-      "eval_steps_per_second": 22.121,
-      "step": 51500
-    },
-    {
-      "epoch": 0.07724284942284922,
-      "grad_norm": 1.7334744930267334,
-      "learning_rate": 0.0001948700239386906,
-      "loss": 2.1632464599609373,
-      "step": 51600
-    },
-    {
-      "epoch": 0.07739254486746713,
-      "grad_norm": 1.539697527885437,
-      "learning_rate": 0.00019486004324633998,
-      "loss": 2.2661369323730467,
-      "step": 51700
-    },
-    {
-      "epoch": 0.07754224031208506,
-      "grad_norm": 1.5553861856460571,
-      "learning_rate": 0.00019485006255398932,
-      "loss": 2.156421661376953,
-      "step": 51800
-    },
-    {
-      "epoch": 0.07769193575670298,
-      "grad_norm": 1.3372256755828857,
-      "learning_rate": 0.00019484008186163867,
-      "loss": 2.143768005371094,
-      "step": 51900
-    },
-    {
-      "epoch": 0.07784163120132091,
-      "grad_norm": 1.6248444318771362,
-      "learning_rate": 0.000194830101169288,
-      "loss": 2.1935415649414063,
-      "step": 52000
-    },
-    {
-      "epoch": 0.07784163120132091,
-      "eval_loss": 2.032679796218872,
-      "eval_runtime": 3353.1637,
-      "eval_samples_per_second": 177.086,
-      "eval_steps_per_second": 22.136,
-      "step": 52000
-    },
-    {
-      "epoch": 0.07799132664593884,
-      "grad_norm": 1.4850828647613525,
-      "learning_rate": 0.00019482012047693738,
-      "loss": 2.183716278076172,
-      "step": 52100
-    },
-    {
-      "epoch": 0.07814102209055676,
-      "grad_norm": 1.494498610496521,
-      "learning_rate": 0.00019481013978458672,
-      "loss": 2.2453564453125,
-      "step": 52200
-    },
-    {
-      "epoch": 0.07829071753517469,
-      "grad_norm": 2.090491771697998,
-      "learning_rate": 0.00019480015909223607,
-      "loss": 2.1624659729003906,
-      "step": 52300
-    },
-    {
-      "epoch": 0.07844041297979261,
-      "grad_norm": 1.3629255294799805,
-      "learning_rate": 0.00019479017839988544,
-      "loss": 2.2663807678222656,
-      "step": 52400
-    },
-    {
-      "epoch": 0.07859010842441054,
-      "grad_norm": 1.759356141090393,
-      "learning_rate": 0.00019478019770753478,
-      "loss": 2.1764543151855467,
-      "step": 52500
-    },
-    {
-      "epoch": 0.07859010842441054,
-      "eval_loss": 2.0278491973876953,
-      "eval_runtime": 3357.4518,
-      "eval_samples_per_second": 176.86,
-      "eval_steps_per_second": 22.108,
-      "step": 52500
-    },
-    {
-      "epoch": 0.07873980386902846,
-      "grad_norm": 1.4570730924606323,
-      "learning_rate": 0.00019477021701518415,
-      "loss": 2.1704721069335937,
-      "step": 52600
-    },
-    {
-      "epoch": 0.07888949931364639,
-      "grad_norm": 1.59248685836792,
-      "learning_rate": 0.0001947602363228335,
-      "loss": 2.218486785888672,
-      "step": 52700
-    },
-    {
-      "epoch": 0.0790391947582643,
-      "grad_norm": 1.5800825357437134,
-      "learning_rate": 0.00019475025563048284,
-      "loss": 2.1960798645019532,
-      "step": 52800
-    },
-    {
-      "epoch": 0.07918889020288224,
-      "grad_norm": 1.7308518886566162,
-      "learning_rate": 0.00019474027493813218,
-      "loss": 2.175284729003906,
-      "step": 52900
-    },
-    {
-      "epoch": 0.07933858564750017,
-      "grad_norm": 1.424614667892456,
-      "learning_rate": 0.00019473029424578155,
-      "loss": 2.1659974670410156,
-      "step": 53000
-    },
-    {
-      "epoch": 0.07933858564750017,
-      "eval_loss": 2.0308055877685547,
-      "eval_runtime": 3351.4302,
-      "eval_samples_per_second": 177.177,
-      "eval_steps_per_second": 22.147,
-      "step": 53000
-    },
-    {
-      "epoch": 0.07948828109211808,
-      "grad_norm": 1.542346477508545,
-      "learning_rate": 0.0001947203135534309,
-      "loss": 2.1855616760253906,
-      "step": 53100
-    },
-    {
-      "epoch": 0.07963797653673602,
-      "grad_norm": 1.3285081386566162,
-      "learning_rate": 0.00019471033286108024,
-      "loss": 2.1977130126953126,
-      "step": 53200
-    },
-    {
-      "epoch": 0.07978767198135393,
-      "grad_norm": 1.7018537521362305,
-      "learning_rate": 0.0001947003521687296,
-      "loss": 2.2327279663085937,
-      "step": 53300
-    },
-    {
-      "epoch": 0.07993736742597186,
-      "grad_norm": 1.6820498704910278,
-      "learning_rate": 0.00019469037147637898,
-      "loss": 2.1948553466796876,
-      "step": 53400
-    },
-    {
-      "epoch": 0.08008706287058978,
-      "grad_norm": 1.2900291681289673,
-      "learning_rate": 0.00019468039078402832,
-      "loss": 2.238514862060547,
-      "step": 53500
-    },
-    {
-      "epoch": 0.08008706287058978,
-      "eval_loss": 2.027621269226074,
-      "eval_runtime": 3357.6072,
-      "eval_samples_per_second": 176.852,
-      "eval_steps_per_second": 22.107,
-      "step": 53500
-    },
-    {
-      "epoch": 0.08023675831520771,
-      "grad_norm": 1.5771751403808594,
-      "learning_rate": 0.00019467041009167766,
-      "loss": 2.2107664489746095,
-      "step": 53600
-    },
-    {
-      "epoch": 0.08038645375982563,
-      "grad_norm": 1.516222357749939,
-      "learning_rate": 0.000194660429399327,
-      "loss": 2.149069976806641,
-      "step": 53700
-    },
-    {
-      "epoch": 0.08053614920444356,
-      "grad_norm": 1.3445262908935547,
-      "learning_rate": 0.00019465044870697638,
-      "loss": 2.207548828125,
-      "step": 53800
-    },
-    {
-      "epoch": 0.08068584464906148,
-      "grad_norm": 1.5316623449325562,
-      "learning_rate": 0.00019464046801462572,
-      "loss": 2.1501731872558594,
-      "step": 53900
-    },
-    {
-      "epoch": 0.08083554009367941,
-      "grad_norm": 1.7762541770935059,
-      "learning_rate": 0.00019463048732227506,
-      "loss": 2.2503961181640624,
-      "step": 54000
-    },
-    {
-      "epoch": 0.08083554009367941,
-      "eval_loss": 2.027519941329956,
-      "eval_runtime": 3354.9989,
-      "eval_samples_per_second": 176.989,
-      "eval_steps_per_second": 22.124,
-      "step": 54000
-    },
-    {
-      "epoch": 0.08098523553829734,
-      "grad_norm": 1.4037964344024658,
-      "learning_rate": 0.0001946205066299244,
-      "loss": 2.214200897216797,
-      "step": 54100
-    },
-    {
-      "epoch": 0.08113493098291526,
-      "grad_norm": 1.4610317945480347,
-      "learning_rate": 0.00019461052593757377,
-      "loss": 2.1824208068847657,
-      "step": 54200
-    },
-    {
-      "epoch": 0.08128462642753319,
-      "grad_norm": 1.5150930881500244,
-      "learning_rate": 0.00019460054524522314,
-      "loss": 2.1996394348144532,
-      "step": 54300
-    },
-    {
-      "epoch": 0.0814343218721511,
-      "grad_norm": 1.5148564577102661,
-      "learning_rate": 0.0001945905645528725,
-      "loss": 2.164758605957031,
-      "step": 54400
-    },
-    {
-      "epoch": 0.08158401731676904,
-      "grad_norm": 1.5986486673355103,
-      "learning_rate": 0.00019458058386052183,
-      "loss": 2.185954132080078,
-      "step": 54500
-    },
-    {
-      "epoch": 0.08158401731676904,
-      "eval_loss": 2.0280683040618896,
-      "eval_runtime": 3349.7683,
-      "eval_samples_per_second": 177.265,
-      "eval_steps_per_second": 22.158,
-      "step": 54500
-    },
-    {
-      "epoch": 0.08173371276138695,
-      "grad_norm": 1.577016830444336,
-      "learning_rate": 0.00019457060316817117,
-      "loss": 2.1064523315429686,
-      "step": 54600
-    },
-    {
-      "epoch": 0.08188340820600488,
-      "grad_norm": 1.2722325325012207,
-      "learning_rate": 0.00019456062247582054,
-      "loss": 2.181499786376953,
-      "step": 54700
-    },
-    {
-      "epoch": 0.0820331036506228,
-      "grad_norm": 1.7814685106277466,
-      "learning_rate": 0.0001945506417834699,
-      "loss": 2.207163543701172,
-      "step": 54800
-    },
-    {
-      "epoch": 0.08218279909524073,
-      "grad_norm": 1.5722390413284302,
-      "learning_rate": 0.00019454066109111923,
-      "loss": 2.218402099609375,
-      "step": 54900
-    },
-    {
-      "epoch": 0.08233249453985866,
-      "grad_norm": 1.9530489444732666,
-      "learning_rate": 0.00019453068039876857,
-      "loss": 2.2142152404785156,
-      "step": 55000
-    },
-    {
-      "epoch": 0.08233249453985866,
-      "eval_loss": 2.0294041633605957,
-      "eval_runtime": 3348.0699,
-      "eval_samples_per_second": 177.355,
-      "eval_steps_per_second": 22.169,
-      "step": 55000
-    },
-    {
-      "epoch": 0.08248218998447658,
-      "grad_norm": 1.889075517654419,
-      "learning_rate": 0.00019452069970641794,
-      "loss": 2.2467337036132813,
-      "step": 55100
-    },
-    {
-      "epoch": 0.08263188542909451,
-      "grad_norm": 2.2500736713409424,
-      "learning_rate": 0.0001945107190140673,
-      "loss": 2.1397862243652344,
-      "step": 55200
-    },
-    {
-      "epoch": 0.08278158087371243,
-      "grad_norm": 1.6653536558151245,
-      "learning_rate": 0.00019450073832171666,
-      "loss": 2.2019688415527345,
-      "step": 55300
-    },
-    {
-      "epoch": 0.08293127631833036,
-      "grad_norm": 1.391103982925415,
-      "learning_rate": 0.000194490757629366,
-      "loss": 2.218947296142578,
-      "step": 55400
-    },
-    {
-      "epoch": 0.08308097176294828,
-      "grad_norm": 1.5539969205856323,
-      "learning_rate": 0.00019448077693701537,
-      "loss": 2.1757769775390625,
-      "step": 55500
-    },
-    {
-      "epoch": 0.08308097176294828,
-      "eval_loss": 2.027836799621582,
-      "eval_runtime": 3356.6236,
-      "eval_samples_per_second": 176.903,
-      "eval_steps_per_second": 22.113,
-      "step": 55500
-    },
-    {
-      "epoch": 0.08323066720756621,
-      "grad_norm": 1.478853702545166,
-      "learning_rate": 0.0001944707962446647,
-      "loss": 2.1783912658691404,
-      "step": 55600
-    },
-    {
-      "epoch": 0.08338036265218413,
-      "grad_norm": 1.7939350605010986,
-      "learning_rate": 0.00019446081555231405,
-      "loss": 2.1712496948242186,
-      "step": 55700
-    },
-    {
-      "epoch": 0.08353005809680206,
-      "grad_norm": 1.5582212209701538,
-      "learning_rate": 0.0001944508348599634,
-      "loss": 2.2127151489257812,
-      "step": 55800
-    },
-    {
-      "epoch": 0.08367975354141999,
-      "grad_norm": 1.3139976263046265,
-      "learning_rate": 0.00019444085416761274,
-      "loss": 2.1747758483886717,
-      "step": 55900
-    },
-    {
-      "epoch": 0.0838294489860379,
-      "grad_norm": 1.4519683122634888,
-      "learning_rate": 0.0001944308734752621,
-      "loss": 2.180957794189453,
-      "step": 56000
-    },
-    {
-      "epoch": 0.0838294489860379,
-      "eval_loss": 2.0259594917297363,
-      "eval_runtime": 3350.9526,
-      "eval_samples_per_second": 177.203,
-      "eval_steps_per_second": 22.15,
-      "step": 56000
-    },
-    {
-      "epoch": 0.08397914443065584,
-      "grad_norm": 1.7663406133651733,
-      "learning_rate": 0.00019442089278291148,
-      "loss": 2.250307769775391,
-      "step": 56100
-    },
-    {
-      "epoch": 0.08412883987527375,
-      "grad_norm": 1.2502931356430054,
-      "learning_rate": 0.00019441091209056082,
-      "loss": 2.2190484619140625,
-      "step": 56200
-    },
-    {
-      "epoch": 0.08427853531989168,
-      "grad_norm": 1.4392576217651367,
-      "learning_rate": 0.00019440093139821017,
-      "loss": 2.166448974609375,
-      "step": 56300
-    },
-    {
-      "epoch": 0.0844282307645096,
-      "grad_norm": 1.670168399810791,
-      "learning_rate": 0.00019439095070585954,
-      "loss": 2.239446716308594,
-      "step": 56400
-    },
-    {
-      "epoch": 0.08457792620912753,
-      "grad_norm": 1.5694557428359985,
-      "learning_rate": 0.00019438097001350888,
-      "loss": 2.1925311279296875,
-      "step": 56500
-    },
-    {
-      "epoch": 0.08457792620912753,
-      "eval_loss": 2.0273854732513428,
-      "eval_runtime": 3352.5641,
-      "eval_samples_per_second": 177.118,
-      "eval_steps_per_second": 22.14,
-      "step": 56500
-    },
-    {
-      "epoch": 0.08472762165374545,
-      "grad_norm": 1.4477951526641846,
-      "learning_rate": 0.00019437098932115822,
-      "loss": 2.1189813232421875,
-      "step": 56600
-    },
-    {
-      "epoch": 0.08487731709836338,
-      "grad_norm": 1.5075583457946777,
-      "learning_rate": 0.00019436100862880757,
-      "loss": 2.20123779296875,
-      "step": 56700
-    },
-    {
-      "epoch": 0.0850270125429813,
-      "grad_norm": 1.4937925338745117,
-      "learning_rate": 0.00019435102793645694,
-      "loss": 2.210856170654297,
-      "step": 56800
-    },
-    {
-      "epoch": 0.08517670798759923,
-      "grad_norm": 1.3712667226791382,
-      "learning_rate": 0.00019434104724410628,
-      "loss": 2.177626037597656,
-      "step": 56900
-    },
-    {
-      "epoch": 0.08532640343221716,
-      "grad_norm": 1.4159820079803467,
-      "learning_rate": 0.00019433106655175565,
-      "loss": 2.2161134338378905,
-      "step": 57000
-    },
-    {
-      "epoch": 0.08532640343221716,
-      "eval_loss": 2.0250601768493652,
-      "eval_runtime": 3355.9741,
-      "eval_samples_per_second": 176.938,
-      "eval_steps_per_second": 22.117,
-      "step": 57000
-    },
-    {
-      "epoch": 0.08547609887683508,
-      "grad_norm": 1.4781357049942017,
-      "learning_rate": 0.000194321085859405,
-      "loss": 2.137511749267578,
-      "step": 57100
-    },
-    {
-      "epoch": 0.08562579432145301,
-      "grad_norm": 1.6368396282196045,
-      "learning_rate": 0.00019431110516705436,
-      "loss": 2.2335321044921876,
-      "step": 57200
-    },
-    {
-      "epoch": 0.08577548976607093,
-      "grad_norm": 1.5657165050506592,
-      "learning_rate": 0.0001943011244747037,
-      "loss": 2.2050746154785155,
-      "step": 57300
-    },
-    {
-      "epoch": 0.08592518521068886,
-      "grad_norm": 2.005018472671509,
-      "learning_rate": 0.00019429114378235305,
-      "loss": 2.1914312744140627,
-      "step": 57400
-    },
-    {
-      "epoch": 0.08607488065530677,
-      "grad_norm": 1.4780051708221436,
-      "learning_rate": 0.0001942811630900024,
-      "loss": 2.1793414306640626,
-      "step": 57500
-    },
-    {
-      "epoch": 0.08607488065530677,
-      "eval_loss": 2.025092601776123,
-      "eval_runtime": 3351.157,
-      "eval_samples_per_second": 177.192,
-      "eval_steps_per_second": 22.149,
-      "step": 57500
-    },
-    {
-      "epoch": 0.0862245760999247,
-      "grad_norm": 1.6144280433654785,
-      "learning_rate": 0.00019427118239765176,
-      "loss": 2.2196609497070314,
-      "step": 57600
-    },
-    {
-      "epoch": 0.08637427154454262,
-      "grad_norm": 1.7048250436782837,
-      "learning_rate": 0.0001942612017053011,
-      "loss": 2.1954595947265627,
-      "step": 57700
-    },
-    {
-      "epoch": 0.08652396698916055,
-      "grad_norm": 2.0330028533935547,
-      "learning_rate": 0.00019425122101295045,
-      "loss": 2.1833831787109377,
-      "step": 57800
-    },
-    {
-      "epoch": 0.08667366243377848,
-      "grad_norm": 1.8549883365631104,
-      "learning_rate": 0.0001942412403205998,
-      "loss": 2.145687255859375,
-      "step": 57900
-    },
-    {
-      "epoch": 0.0868233578783964,
-      "grad_norm": 1.3814551830291748,
-      "learning_rate": 0.00019423125962824916,
-      "loss": 2.201437683105469,
-      "step": 58000
-    },
-    {
-      "epoch": 0.0868233578783964,
-      "eval_loss": 2.0211007595062256,
-      "eval_runtime": 3347.2097,
-      "eval_samples_per_second": 177.401,
-      "eval_steps_per_second": 22.175,
-      "step": 58000
-    },
-    {
-      "epoch": 0.08697305332301433,
-      "grad_norm": 1.6914571523666382,
-      "learning_rate": 0.00019422127893589853,
-      "loss": 2.1915217590332032,
-      "step": 58100
-    },
-    {
-      "epoch": 0.08712274876763225,
-      "grad_norm": 1.249216079711914,
-      "learning_rate": 0.00019421129824354787,
-      "loss": 2.202597198486328,
-      "step": 58200
-    },
-    {
-      "epoch": 0.08727244421225018,
-      "grad_norm": 1.7321069240570068,
-      "learning_rate": 0.00019420131755119722,
-      "loss": 2.1844943237304686,
-      "step": 58300
-    },
-    {
-      "epoch": 0.0874221396568681,
-      "grad_norm": 1.5633962154388428,
-      "learning_rate": 0.00019419133685884656,
-      "loss": 2.227957305908203,
-      "step": 58400
-    },
-    {
-      "epoch": 0.08757183510148603,
-      "grad_norm": 1.63968825340271,
-      "learning_rate": 0.00019418135616649593,
-      "loss": 2.2219532775878905,
-      "step": 58500
-    },
-    {
-      "epoch": 0.08757183510148603,
-      "eval_loss": 2.021097421646118,
-      "eval_runtime": 3347.2788,
-      "eval_samples_per_second": 177.397,
-      "eval_steps_per_second": 22.175,
-      "step": 58500
-    },
-    {
-      "epoch": 0.08772153054610395,
-      "grad_norm": 1.6275060176849365,
-      "learning_rate": 0.00019417137547414527,
-      "loss": 2.1754115295410155,
-      "step": 58600
-    },
-    {
-      "epoch": 0.08787122599072188,
-      "grad_norm": 1.605980396270752,
-      "learning_rate": 0.00019416139478179462,
-      "loss": 2.192578887939453,
-      "step": 58700
-    },
-    {
-      "epoch": 0.08802092143533981,
-      "grad_norm": Infinity,
-      "learning_rate": 0.00019415141408944396,
-      "loss": 2.1684124755859373,
-      "step": 58800
-    },
-    {
-      "epoch": 0.08817061687995773,
-      "grad_norm": 1.4836750030517578,
-      "learning_rate": 0.00019414143339709333,
-      "loss": 2.1549273681640626,
-      "step": 58900
-    },
-    {
-      "epoch": 0.08832031232457566,
-      "grad_norm": 1.578926920890808,
-      "learning_rate": 0.0001941314527047427,
-      "loss": 2.1298483276367186,
-      "step": 59000
-    },
-    {
-      "epoch": 0.08832031232457566,
-      "eval_loss": 2.0234227180480957,
-      "eval_runtime": 3355.2676,
-      "eval_samples_per_second": 176.975,
-      "eval_steps_per_second": 22.122,
-      "step": 59000
-    },
-    {
-      "epoch": 0.08847000776919357,
-      "grad_norm": 1.6871542930603027,
-      "learning_rate": 0.00019412147201239204,
-      "loss": 2.2164715576171874,
-      "step": 59100
-    },
-    {
-      "epoch": 0.0886197032138115,
-      "grad_norm": 1.3972798585891724,
-      "learning_rate": 0.00019411149132004139,
-      "loss": 2.1397531127929685,
-      "step": 59200
-    },
-    {
-      "epoch": 0.08876939865842942,
-      "grad_norm": 1.7074257135391235,
-      "learning_rate": 0.00019410151062769076,
-      "loss": 2.1401499938964843,
-      "step": 59300
-    },
-    {
-      "epoch": 0.08891909410304735,
-      "grad_norm": 1.5211334228515625,
-      "learning_rate": 0.0001940915299353401,
-      "loss": 2.1943243408203124,
-      "step": 59400
-    },
-    {
-      "epoch": 0.08906878954766527,
-      "grad_norm": 1.5690574645996094,
-      "learning_rate": 0.00019408154924298944,
-      "loss": 2.1882110595703126,
-      "step": 59500
-    },
-    {
-      "epoch": 0.08906878954766527,
-      "eval_loss": 2.019160509109497,
-      "eval_runtime": 3358.0244,
-      "eval_samples_per_second": 176.83,
-      "eval_steps_per_second": 22.104,
-      "step": 59500
-    },
-    {
-      "epoch": 0.0892184849922832,
-      "grad_norm": 1.3116379976272583,
-      "learning_rate": 0.00019407156855063878,
-      "loss": 2.231597442626953,
-      "step": 59600
-    },
-    {
-      "epoch": 0.08936818043690112,
-      "grad_norm": 1.5561062097549438,
-      "learning_rate": 0.00019406158785828813,
-      "loss": 2.1809584045410157,
-      "step": 59700
-    },
-    {
-      "epoch": 0.08951787588151905,
-      "grad_norm": 1.561083436012268,
-      "learning_rate": 0.0001940516071659375,
-      "loss": 2.208002166748047,
-      "step": 59800
-    },
-    {
-      "epoch": 0.08966757132613698,
-      "grad_norm": 1.44806706905365,
-      "learning_rate": 0.00019404162647358687,
-      "loss": 2.154641418457031,
-      "step": 59900
-    },
-    {
-      "epoch": 0.0898172667707549,
-      "grad_norm": 1.4026309251785278,
-      "learning_rate": 0.0001940316457812362,
-      "loss": 2.198057861328125,
-      "step": 60000
-    },
-    {
-      "epoch": 0.0898172667707549,
-      "eval_loss": 2.0184855461120605,
-      "eval_runtime": 3355.2151,
-      "eval_samples_per_second": 176.978,
-      "eval_steps_per_second": 22.122,
-      "step": 60000
-    },
-    {
-      "epoch": 0.08996696221537283,
-      "grad_norm": 2.0885839462280273,
-      "learning_rate": 0.00019402166508888555,
-      "loss": 2.144966735839844,
-      "step": 60100
-    },
-    {
-      "epoch": 0.09011665765999075,
-      "grad_norm": 1.751170039176941,
-      "learning_rate": 0.00019401168439653492,
-      "loss": 2.295005035400391,
-      "step": 60200
-    },
-    {
-      "epoch": 0.09026635310460868,
-      "grad_norm": 1.4370735883712769,
-      "learning_rate": 0.00019400170370418427,
-      "loss": 2.143782196044922,
-      "step": 60300
-    },
-    {
-      "epoch": 0.0904160485492266,
-      "grad_norm": 1.5014913082122803,
-      "learning_rate": 0.0001939917230118336,
-      "loss": 2.1681838989257813,
-      "step": 60400
-    },
-    {
-      "epoch": 0.09056574399384452,
-      "grad_norm": 1.5471678972244263,
-      "learning_rate": 0.00019398174231948295,
-      "loss": 2.185344696044922,
-      "step": 60500
-    },
-    {
-      "epoch": 0.09056574399384452,
-      "eval_loss": 2.0200774669647217,
-      "eval_runtime": 3355.9178,
-      "eval_samples_per_second": 176.941,
-      "eval_steps_per_second": 22.118,
-      "step": 60500
-    },
-    {
-      "epoch": 0.09071543943846244,
-      "grad_norm": 1.4113538265228271,
-      "learning_rate": 0.00019397176162713232,
-      "loss": 2.0933082580566404,
-      "step": 60600
-    },
-    {
-      "epoch": 0.09086513488308037,
-      "grad_norm": 1.5779986381530762,
-      "learning_rate": 0.00019396178093478167,
-      "loss": 2.203249664306641,
-      "step": 60700
-    },
-    {
-      "epoch": 0.0910148303276983,
-      "grad_norm": 1.5938165187835693,
-      "learning_rate": 0.00019395180024243104,
-      "loss": 2.211194610595703,
-      "step": 60800
-    },
-    {
-      "epoch": 0.09116452577231622,
-      "grad_norm": 1.427837610244751,
-      "learning_rate": 0.00019394181955008038,
-      "loss": 2.2026344299316407,
-      "step": 60900
-    },
-    {
-      "epoch": 0.09131422121693415,
-      "grad_norm": 1.9053469896316528,
-      "learning_rate": 0.00019393183885772975,
-      "loss": 2.2140348815917967,
-      "step": 61000
-    },
-    {
-      "epoch": 0.09131422121693415,
-      "eval_loss": 2.0231666564941406,
-      "eval_runtime": 3362.1817,
-      "eval_samples_per_second": 176.611,
-      "eval_steps_per_second": 22.076,
-      "step": 61000
-    },
-    {
-      "epoch": 0.09146391666155207,
-      "grad_norm": 1.4309881925582886,
-      "learning_rate": 0.0001939218581653791,
-      "loss": 2.1615617370605467,
-      "step": 61100
-    },
-    {
-      "epoch": 0.09161361210617,
-      "grad_norm": 1.6303349733352661,
-      "learning_rate": 0.00019391187747302844,
-      "loss": 2.1892765808105468,
-      "step": 61200
-    },
-    {
-      "epoch": 0.09176330755078792,
-      "grad_norm": 1.576392412185669,
-      "learning_rate": 0.00019390189678067778,
-      "loss": 2.1650485229492187,
-      "step": 61300
-    },
-    {
-      "epoch": 0.09191300299540585,
-      "grad_norm": 1.43353271484375,
-      "learning_rate": 0.00019389191608832712,
-      "loss": 2.2056204223632814,
-      "step": 61400
-    },
-    {
-      "epoch": 0.09206269844002377,
-      "grad_norm": 1.4109746217727661,
-      "learning_rate": 0.0001938819353959765,
-      "loss": 2.1762345886230468,
-      "step": 61500
-    },
-    {
-      "epoch": 0.09206269844002377,
-      "eval_loss": 2.0245656967163086,
-      "eval_runtime": 3362.2283,
-      "eval_samples_per_second": 176.608,
-      "eval_steps_per_second": 22.076,
-      "step": 61500
-    },
-    {
-      "epoch": 0.0922123938846417,
-      "grad_norm": 1.754417061805725,
-      "learning_rate": 0.00019387195470362583,
-      "loss": 2.22733154296875,
-      "step": 61600
-    },
-    {
-      "epoch": 0.09236208932925963,
-      "grad_norm": 1.8821182250976562,
-      "learning_rate": 0.0001938619740112752,
-      "loss": 2.239346466064453,
-      "step": 61700
-    },
-    {
-      "epoch": 0.09251178477387755,
-      "grad_norm": 1.3523200750350952,
-      "learning_rate": 0.00019385199331892455,
-      "loss": 2.200750274658203,
-      "step": 61800
-    },
-    {
-      "epoch": 0.09266148021849548,
-      "grad_norm": 1.8737268447875977,
-      "learning_rate": 0.00019384201262657392,
-      "loss": 2.137942352294922,
-      "step": 61900
-    },
-    {
-      "epoch": 0.0928111756631134,
-      "grad_norm": 1.3404455184936523,
-      "learning_rate": 0.00019383203193422326,
-      "loss": 2.2067385864257814,
-      "step": 62000
-    },
-    {
-      "epoch": 0.0928111756631134,
-      "eval_loss": 2.0173022747039795,
-      "eval_runtime": 3359.0659,
-      "eval_samples_per_second": 176.775,
-      "eval_steps_per_second": 22.097,
-      "step": 62000
-    },
-    {
-      "epoch": 0.09296087110773132,
-      "grad_norm": 1.465118169784546,
-      "learning_rate": 0.0001938220512418726,
-      "loss": 2.184667053222656,
-      "step": 62100
-    },
-    {
-      "epoch": 0.09311056655234924,
-      "grad_norm": 1.9017518758773804,
-      "learning_rate": 0.00019381207054952195,
-      "loss": 2.1828317260742187,
-      "step": 62200
-    },
-    {
-      "epoch": 0.09326026199696717,
-      "grad_norm": 1.7928073406219482,
-      "learning_rate": 0.00019380208985717132,
-      "loss": 2.22995849609375,
-      "step": 62300
-    },
-    {
-      "epoch": 0.09340995744158509,
-      "grad_norm": 1.316602349281311,
-      "learning_rate": 0.00019379210916482066,
-      "loss": 2.213928680419922,
-      "step": 62400
-    },
-    {
-      "epoch": 0.09355965288620302,
-      "grad_norm": 1.8527966737747192,
-      "learning_rate": 0.00019378212847247,
-      "loss": 2.196822509765625,
-      "step": 62500
-    },
-    {
-      "epoch": 0.09355965288620302,
-      "eval_loss": 2.0160346031188965,
-      "eval_runtime": 3363.0592,
-      "eval_samples_per_second": 176.565,
-      "eval_steps_per_second": 22.071,
-      "step": 62500
-    },
-    {
-      "epoch": 0.09370934833082094,
-      "grad_norm": 1.6928725242614746,
-      "learning_rate": 0.00019377214778011937,
-      "loss": 2.223183441162109,
-      "step": 62600
-    },
-    {
-      "epoch": 0.09385904377543887,
-      "grad_norm": 1.4706223011016846,
-      "learning_rate": 0.00019376216708776874,
-      "loss": 2.2284422302246094,
-      "step": 62700
-    },
-    {
-      "epoch": 0.0940087392200568,
-      "grad_norm": 1.4991801977157593,
-      "learning_rate": 0.0001937521863954181,
-      "loss": 2.204234619140625,
-      "step": 62800
-    },
-    {
-      "epoch": 0.09415843466467472,
-      "grad_norm": 1.5106202363967896,
-      "learning_rate": 0.00019374220570306743,
-      "loss": 2.1870407104492187,
-      "step": 62900
-    },
-    {
-      "epoch": 0.09430813010929265,
-      "grad_norm": 1.6304632425308228,
-      "learning_rate": 0.00019373222501071677,
-      "loss": 2.1809796142578124,
-      "step": 63000
-    },
-    {
-      "epoch": 0.09430813010929265,
-      "eval_loss": 2.018437147140503,
-      "eval_runtime": 3362.5241,
-      "eval_samples_per_second": 176.593,
-      "eval_steps_per_second": 22.074,
-      "step": 63000
-    },
-    {
-      "epoch": 0.09445782555391057,
-      "grad_norm": 1.8338884115219116,
-      "learning_rate": 0.00019372224431836612,
-      "loss": 2.186797637939453,
-      "step": 63100
-    },
-    {
-      "epoch": 0.0946075209985285,
-      "grad_norm": 1.3817193508148193,
-      "learning_rate": 0.00019371226362601549,
-      "loss": 2.2141650390625,
-      "step": 63200
-    },
-    {
-      "epoch": 0.09475721644314641,
-      "grad_norm": 1.872168779373169,
-      "learning_rate": 0.00019370228293366483,
-      "loss": 2.166869812011719,
-      "step": 63300
-    },
-    {
-      "epoch": 0.09490691188776434,
-      "grad_norm": 1.5168299674987793,
-      "learning_rate": 0.00019369230224131417,
-      "loss": 2.1360301208496093,
-      "step": 63400
-    },
-    {
-      "epoch": 0.09505660733238226,
-      "grad_norm": 1.8126424551010132,
-      "learning_rate": 0.00019368232154896354,
-      "loss": 2.198636932373047,
-      "step": 63500
-    },
-    {
-      "epoch": 0.09505660733238226,
-      "eval_loss": 2.0175743103027344,
-      "eval_runtime": 3362.4118,
-      "eval_samples_per_second": 176.599,
-      "eval_steps_per_second": 22.075,
-      "step": 63500
-    },
-    {
-      "epoch": 0.09520630277700019,
-      "grad_norm": 1.3551536798477173,
-      "learning_rate": 0.0001936723408566129,
-      "loss": 2.1685086059570313,
-      "step": 63600
-    },
-    {
-      "epoch": 0.09535599822161812,
-      "grad_norm": 1.4246097803115845,
-      "learning_rate": 0.00019366236016426226,
-      "loss": 2.188532409667969,
-      "step": 63700
-    },
-    {
-      "epoch": 0.09550569366623604,
-      "grad_norm": 1.7355008125305176,
-      "learning_rate": 0.0001936523794719116,
-      "loss": 2.139734344482422,
-      "step": 63800
-    },
-    {
-      "epoch": 0.09565538911085397,
-      "grad_norm": 2.054295539855957,
-      "learning_rate": 0.00019364239877956094,
-      "loss": 2.1252426147460937,
-      "step": 63900
-    },
-    {
-      "epoch": 0.09580508455547189,
-      "grad_norm": 1.4171401262283325,
-      "learning_rate": 0.0001936324180872103,
-      "loss": 2.1797625732421877,
-      "step": 64000
-    },
-    {
-      "epoch": 0.09580508455547189,
-      "eval_loss": 2.0230464935302734,
-      "eval_runtime": 3363.0242,
-      "eval_samples_per_second": 176.567,
-      "eval_steps_per_second": 22.071,
-      "step": 64000
-    },
-    {
-      "epoch": 0.09595478000008982,
-      "grad_norm": 1.447759985923767,
-      "learning_rate": 0.00019362243739485965,
-      "loss": 2.1533502197265624,
-      "step": 64100
-    },
-    {
-      "epoch": 0.09610447544470774,
-      "grad_norm": 1.6643582582473755,
-      "learning_rate": 0.000193612456702509,
-      "loss": 2.2082479858398436,
-      "step": 64200
-    },
-    {
-      "epoch": 0.09625417088932567,
-      "grad_norm": 1.3565841913223267,
-      "learning_rate": 0.00019360247601015834,
-      "loss": 2.1023699951171877,
-      "step": 64300
-    },
-    {
-      "epoch": 0.09640386633394359,
-      "grad_norm": 1.5387645959854126,
-      "learning_rate": 0.0001935924953178077,
-      "loss": 2.1815924072265624,
-      "step": 64400
-    },
-    {
-      "epoch": 0.09655356177856152,
-      "grad_norm": 1.5055195093154907,
-      "learning_rate": 0.00019358251462545708,
-      "loss": 2.1959104919433594,
-      "step": 64500
-    },
-    {
-      "epoch": 0.09655356177856152,
-      "eval_loss": 2.018390655517578,
-      "eval_runtime": 3365.7063,
-      "eval_samples_per_second": 176.426,
-      "eval_steps_per_second": 22.053,
-      "step": 64500
-    },
-    {
-      "epoch": 0.09670325722317945,
-      "grad_norm": 1.7348912954330444,
-      "learning_rate": 0.00019357253393310642,
-      "loss": 2.192291717529297,
-      "step": 64600
-    },
-    {
-      "epoch": 0.09685295266779737,
-      "grad_norm": 1.3179991245269775,
-      "learning_rate": 0.00019356255324075577,
-      "loss": 2.1625665283203124,
-      "step": 64700
-    },
-    {
-      "epoch": 0.0970026481124153,
-      "grad_norm": 1.4718928337097168,
-      "learning_rate": 0.0001935525725484051,
-      "loss": 2.1669488525390626,
-      "step": 64800
-    },
-    {
-      "epoch": 0.09715234355703321,
-      "grad_norm": 1.9540119171142578,
-      "learning_rate": 0.00019354259185605448,
-      "loss": 2.181168212890625,
-      "step": 64900
-    },
-    {
-      "epoch": 0.09730203900165114,
-      "grad_norm": 1.6663063764572144,
-      "learning_rate": 0.00019353261116370382,
-      "loss": 2.1752142333984374,
-      "step": 65000
-    },
-    {
-      "epoch": 0.09730203900165114,
-      "eval_loss": 2.01832914352417,
-      "eval_runtime": 3362.3824,
-      "eval_samples_per_second": 176.6,
-      "eval_steps_per_second": 22.075,
-      "step": 65000
-    },
-    {
-      "epoch": 0.09745173444626906,
-      "grad_norm": 1.6608840227127075,
-      "learning_rate": 0.00019352263047135317,
-      "loss": 2.1685086059570313,
-      "step": 65100
-    },
-    {
-      "epoch": 0.09760142989088699,
-      "grad_norm": 1.5451465845108032,
-      "learning_rate": 0.0001935126497790025,
-      "loss": 2.138538818359375,
-      "step": 65200
-    },
-    {
-      "epoch": 0.09775112533550491,
-      "grad_norm": 1.4013569355010986,
-      "learning_rate": 0.00019350266908665188,
-      "loss": 2.2300656127929686,
-      "step": 65300
-    },
-    {
-      "epoch": 0.09790082078012284,
-      "grad_norm": 1.406111717224121,
-      "learning_rate": 0.00019349268839430125,
-      "loss": 2.1768458557128905,
-      "step": 65400
-    },
-    {
-      "epoch": 0.09805051622474076,
-      "grad_norm": 1.5270110368728638,
-      "learning_rate": 0.0001934827077019506,
-      "loss": 2.182126007080078,
-      "step": 65500
-    },
-    {
-      "epoch": 0.09805051622474076,
-      "eval_loss": 2.015306234359741,
-      "eval_runtime": 3367.6022,
-      "eval_samples_per_second": 176.327,
-      "eval_steps_per_second": 22.041,
-      "step": 65500
-    },
-    {
-      "epoch": 0.09820021166935869,
-      "grad_norm": 1.5952726602554321,
-      "learning_rate": 0.00019347272700959993,
-      "loss": 2.182955780029297,
-      "step": 65600
-    },
-    {
-      "epoch": 0.09834990711397662,
-      "grad_norm": 1.4764333963394165,
-      "learning_rate": 0.0001934627463172493,
-      "loss": 2.21672607421875,
-      "step": 65700
-    },
-    {
-      "epoch": 0.09849960255859454,
-      "grad_norm": 1.7373528480529785,
-      "learning_rate": 0.00019345276562489865,
-      "loss": 2.202209777832031,
-      "step": 65800
-    },
-    {
-      "epoch": 0.09864929800321247,
-      "grad_norm": 1.446821928024292,
-      "learning_rate": 0.000193442784932548,
-      "loss": 2.2337907409667968,
-      "step": 65900
-    },
-    {
-      "epoch": 0.09879899344783039,
-      "grad_norm": 1.6569068431854248,
-      "learning_rate": 0.00019343280424019733,
-      "loss": 2.1957440185546875,
-      "step": 66000
-    },
-    {
-      "epoch": 0.09879899344783039,
-      "eval_loss": 2.0169997215270996,
-      "eval_runtime": 3370.4611,
-      "eval_samples_per_second": 176.177,
-      "eval_steps_per_second": 22.022,
-      "step": 66000
-    },
-    {
-      "epoch": 0.09894868889244832,
-      "grad_norm": 1.7657335996627808,
-      "learning_rate": 0.0001934228235478467,
-      "loss": 2.1630799865722654,
-      "step": 66100
-    },
-    {
-      "epoch": 0.09909838433706623,
-      "grad_norm": 1.484918475151062,
-      "learning_rate": 0.00019341284285549605,
-      "loss": 2.204174346923828,
-      "step": 66200
-    },
-    {
-      "epoch": 0.09924807978168416,
-      "grad_norm": 1.371679425239563,
-      "learning_rate": 0.00019340286216314542,
-      "loss": 2.153150177001953,
-      "step": 66300
-    },
-    {
-      "epoch": 0.09939777522630208,
-      "grad_norm": 1.7795679569244385,
-      "learning_rate": 0.00019339288147079476,
-      "loss": 2.2080335998535157,
-      "step": 66400
-    },
-    {
-      "epoch": 0.09954747067092001,
-      "grad_norm": 1.5002249479293823,
-      "learning_rate": 0.0001933829007784441,
-      "loss": 2.2401585388183594,
-      "step": 66500
-    },
-    {
-      "epoch": 0.09954747067092001,
-      "eval_loss": 2.016911506652832,
-      "eval_runtime": 3366.3972,
-      "eval_samples_per_second": 176.39,
-      "eval_steps_per_second": 22.049,
-      "step": 66500
-    },
-    {
-      "epoch": 0.09969716611553794,
-      "grad_norm": 1.292163372039795,
-      "learning_rate": 0.00019337292008609347,
-      "loss": 2.1654837036132815,
-      "step": 66600
-    },
-    {
-      "epoch": 0.09984686156015586,
-      "grad_norm": 1.4693008661270142,
-      "learning_rate": 0.00019336293939374282,
-      "loss": 2.1333607482910155,
-      "step": 66700
-    },
-    {
-      "epoch": 0.09999655700477379,
-      "grad_norm": 1.9843885898590088,
-      "learning_rate": 0.00019335295870139216,
-      "loss": 2.1326234436035154,
-      "step": 66800
-    },
-    {
-      "epoch": 0.10014625244939171,
-      "grad_norm": 1.5150692462921143,
-      "learning_rate": 0.0001933429780090415,
-      "loss": 2.1250564575195314,
-      "step": 66900
-    },
-    {
-      "epoch": 0.10029594789400964,
-      "grad_norm": 1.8549200296401978,
-      "learning_rate": 0.00019333299731669087,
-      "loss": 2.1179872131347657,
-      "step": 67000
-    },
-    {
-      "epoch": 0.10029594789400964,
-      "eval_loss": 2.017652750015259,
-      "eval_runtime": 3347.6159,
-      "eval_samples_per_second": 177.379,
-      "eval_steps_per_second": 22.172,
-      "step": 67000
-    },
-    {
-      "epoch": 0.10044564333862756,
-      "grad_norm": 2.3358261585235596,
-      "learning_rate": 0.00019332301662434022,
-      "loss": 2.2126824951171873,
-      "step": 67100
-    },
-    {
-      "epoch": 0.10059533878324549,
-      "grad_norm": 1.439832329750061,
-      "learning_rate": 0.00019331303593198959,
-      "loss": 2.177693328857422,
-      "step": 67200
-    },
-    {
-      "epoch": 0.1007450342278634,
-      "grad_norm": 1.4912147521972656,
-      "learning_rate": 0.00019330305523963893,
-      "loss": 2.148679504394531,
-      "step": 67300
-    },
-    {
-      "epoch": 0.10089472967248134,
-      "grad_norm": 2.269015073776245,
-      "learning_rate": 0.0001932930745472883,
-      "loss": 2.1965911865234373,
-      "step": 67400
-    },
-    {
-      "epoch": 0.10104442511709927,
-      "grad_norm": 1.5250779390335083,
-      "learning_rate": 0.00019328309385493764,
-      "loss": 2.157216033935547,
-      "step": 67500
-    },
-    {
-      "epoch": 0.10104442511709927,
-      "eval_loss": 2.020542621612549,
-      "eval_runtime": 3327.4542,
-      "eval_samples_per_second": 178.454,
-      "eval_steps_per_second": 22.307,
-      "step": 67500
-    },
-    {
-      "epoch": 0.10119412056171719,
-      "grad_norm": 2.0837090015411377,
-      "learning_rate": 0.00019327311316258698,
-      "loss": 2.1705401611328123,
-      "step": 67600
-    },
-    {
-      "epoch": 0.10134381600633512,
-      "grad_norm": 1.508073329925537,
-      "learning_rate": 0.00019326313247023633,
-      "loss": 2.185127410888672,
-      "step": 67700
-    },
-    {
-      "epoch": 0.10149351145095303,
-      "grad_norm": 1.6350743770599365,
-      "learning_rate": 0.0001932531517778857,
-      "loss": 2.1623744201660156,
-      "step": 67800
-    },
-    {
-      "epoch": 0.10164320689557096,
-      "grad_norm": 1.4622551202774048,
-      "learning_rate": 0.00019324317108553504,
-      "loss": 2.157512969970703,
-      "step": 67900
-    },
-    {
-      "epoch": 0.10179290234018888,
-      "grad_norm": 1.5243613719940186,
-      "learning_rate": 0.00019323319039318438,
-      "loss": 2.2052244567871093,
-      "step": 68000
-    },
-    {
-      "epoch": 0.10179290234018888,
-      "eval_loss": 2.0144357681274414,
-      "eval_runtime": 3311.8073,
-      "eval_samples_per_second": 179.297,
-      "eval_steps_per_second": 22.412,
-      "step": 68000
-    }
-  ],
-  "logging_steps": 100,
-  "max_steps": 2004069,
-  "num_input_tokens_seen": 0,
-  "num_train_epochs": 3,
-  "save_steps": 1000,
-  "stateful_callbacks": {
-    "TrainerControl": {
-      "args": {
-        "should_epoch_stop": false,
-        "should_evaluate": false,
-        "should_log": false,
-        "should_save": true,
-        "should_training_stop": false
-      },
-      "attributes": {}
-    }
-  },
-  "total_flos": 1.43676265660416e+17,
-  "train_batch_size": 8,
-  "trial_name": null,
-  "trial_params": null
-}