{
  "best_global_step": 4600,
  "best_metric": 0.7991446852684021,
  "best_model_checkpoint": "lfm2.5-350m-multilang-code-math/checkpoint-4600",
  "epoch": 2.0,
  "eval_steps": 200,
  "global_step": 5778,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "entropy": 1.2043220043182372,
      "epoch": 0.008655011251514627,
      "grad_norm": 23.5,
      "learning_rate": 2.7586206896551725e-06,
      "loss": 3.0389138793945314,
      "mean_token_accuracy": 0.5589780753850937,
      "num_tokens": 151155.0,
      "step": 25
    },
    {
      "entropy": 1.11516796708107,
      "epoch": 0.017310022503029255,
      "grad_norm": 7.34375,
      "learning_rate": 5.6321839080459775e-06,
      "loss": 1.8928765869140625,
      "mean_token_accuracy": 0.6762128543853759,
      "num_tokens": 310957.0,
      "step": 50
    },
    {
      "entropy": 1.2031074380874633,
      "epoch": 0.02596503375454388,
      "grad_norm": 3.78125,
      "learning_rate": 8.505747126436782e-06,
      "loss": 1.2897039794921874,
      "mean_token_accuracy": 0.7282541489601135,
      "num_tokens": 461068.0,
      "step": 75
    },
    {
      "entropy": 1.0323714792728425,
      "epoch": 0.03462004500605851,
      "grad_norm": 3.71875,
      "learning_rate": 1.1379310344827587e-05,
      "loss": 1.0183975982666016,
      "mean_token_accuracy": 0.757768098115921,
      "num_tokens": 615118.0,
      "step": 100
    },
    {
      "entropy": 0.9247142732143402,
      "epoch": 0.043275056257573136,
      "grad_norm": 2.578125,
      "learning_rate": 1.4252873563218392e-05,
      "loss": 0.8757442474365235,
      "mean_token_accuracy": 0.7792941844463348,
      "num_tokens": 762906.0,
      "step": 125
    },
    {
      "entropy": 0.9190187740325928,
      "epoch": 0.05193006750908776,
      "grad_norm": 2.296875,
      "learning_rate": 1.7126436781609197e-05,
      "loss": 0.891803970336914,
      "mean_token_accuracy": 0.7753661525249481,
      "num_tokens": 921356.0,
      "step": 150
    },
    {
      "entropy": 0.886636551618576,
      "epoch": 0.06058507876060239,
      "grad_norm": 2.296875,
      "learning_rate": 2e-05,
      "loss": 0.861456298828125,
      "mean_token_accuracy": 0.7826532816886902,
      "num_tokens": 1076183.0,
      "step": 175
    },
    {
      "entropy": 0.9015009069442749,
      "epoch": 0.06924009001211702,
      "grad_norm": 2.5625,
      "learning_rate": 1.9999017921043748e-05,
      "loss": 0.8748048400878906,
      "mean_token_accuracy": 0.7792416322231293,
      "num_tokens": 1229170.0,
      "step": 200
    },
    {
      "epoch": 0.06924009001211702,
      "eval_entropy": 0.8920288547140653,
      "eval_loss": 0.8666635155677795,
      "eval_mean_token_accuracy": 0.7797926357535065,
      "eval_num_tokens": 1229170.0,
      "eval_runtime": 16.3134,
      "eval_samples_per_second": 298.222,
      "eval_steps_per_second": 18.696,
      "step": 200
    },
    {
      "entropy": 0.9062747704982758,
      "epoch": 0.07789510126363164,
      "grad_norm": 2.40625,
      "learning_rate": 1.999607187707079e-05,
      "loss": 0.8789938354492187,
      "mean_token_accuracy": 0.7778702569007874,
      "num_tokens": 1381380.0,
      "step": 225
    },
    {
      "entropy": 0.8776320171356201,
      "epoch": 0.08655011251514627,
      "grad_norm": 2.46875,
      "learning_rate": 1.9991162446730694e-05,
      "loss": 0.8517158508300782,
      "mean_token_accuracy": 0.7819743084907532,
      "num_tokens": 1538186.0,
      "step": 250
    },
    {
      "entropy": 0.8657179689407348,
      "epoch": 0.09520512376666089,
      "grad_norm": 2.25,
      "learning_rate": 1.9984290594313102e-05,
      "loss": 0.8408647155761719,
      "mean_token_accuracy": 0.7840179252624512,
      "num_tokens": 1690067.0,
      "step": 275
    },
    {
      "entropy": 0.8955290937423706,
      "epoch": 0.10386013501817552,
      "grad_norm": 2.328125,
      "learning_rate": 1.9975457669558346e-05,
      "loss": 0.8683735656738282,
      "mean_token_accuracy": 0.7763688516616821,
      "num_tokens": 1845254.0,
      "step": 300
    },
    {
      "entropy": 0.8744900381565094,
      "epoch": 0.11251514626969016,
      "grad_norm": 2.34375,
      "learning_rate": 1.996466540739233e-05,
      "loss": 0.8470248413085938,
      "mean_token_accuracy": 0.7828968441486359,
      "num_tokens": 2000855.0,
      "step": 325
    },
    {
      "entropy": 0.8875207018852234,
      "epoch": 0.12117015752120477,
      "grad_norm": 2.34375,
      "learning_rate": 1.9951915927585762e-05,
      "loss": 0.8633122253417969,
      "mean_token_accuracy": 0.7807920384407043,
      "num_tokens": 2159255.0,
      "step": 350
    },
    {
      "entropy": 0.8727393555641174,
      "epoch": 0.1298251687727194,
      "grad_norm": 2.40625,
      "learning_rate": 1.9937211734337812e-05,
      "loss": 0.8484575653076172,
      "mean_token_accuracy": 0.7856597316265106,
      "num_tokens": 2313980.0,
      "step": 375
    },
    {
      "entropy": 0.8450514304637909,
      "epoch": 0.13848018002423404,
      "grad_norm": 2.390625,
      "learning_rate": 1.992055571578423e-05,
      "loss": 0.8231178283691406,
      "mean_token_accuracy": 0.7884540009498596,
      "num_tokens": 2468392.0,
      "step": 400
    },
    {
      "epoch": 0.13848018002423404,
      "eval_entropy": 0.8820633643963298,
      "eval_loss": 0.8462674021720886,
      "eval_mean_token_accuracy": 0.7830698802822926,
      "eval_num_tokens": 2468392.0,
      "eval_runtime": 16.2902,
      "eval_samples_per_second": 298.645,
      "eval_steps_per_second": 18.723,
      "step": 400
    },
    {
      "entropy": 0.8852161896228791,
      "epoch": 0.14713519127574864,
      "grad_norm": 2.25,
      "learning_rate": 1.990195114343008e-05,
      "loss": 0.8604666137695313,
      "mean_token_accuracy": 0.7796429479122162,
      "num_tokens": 2626504.0,
      "step": 425
    },
    {
      "entropy": 0.8595891773700715,
      "epoch": 0.15579020252726328,
      "grad_norm": 2.109375,
      "learning_rate": 1.988140167150716e-05,
      "loss": 0.8375725555419922,
      "mean_token_accuracy": 0.7837563192844391,
      "num_tokens": 2785992.0,
      "step": 450
    },
    {
      "entropy": 0.8787437427043915,
      "epoch": 0.1644452137787779,
      "grad_norm": 2.21875,
      "learning_rate": 1.9858911336256258e-05,
      "loss": 0.8602984619140625,
      "mean_token_accuracy": 0.7801311421394348,
      "num_tokens": 2943389.0,
      "step": 475
    },
    {
      "entropy": 0.8887673962116241,
      "epoch": 0.17310022503029254,
      "grad_norm": 2.34375,
      "learning_rate": 1.983448455513437e-05,
      "loss": 0.8559275054931641,
      "mean_token_accuracy": 0.7799410057067871,
      "num_tokens": 3098910.0,
      "step": 500
    },
    {
      "entropy": 0.8632562351226807,
      "epoch": 0.18175523628180718,
      "grad_norm": 2.359375,
      "learning_rate": 1.9808126125947035e-05,
      "loss": 0.8414566040039062,
      "mean_token_accuracy": 0.7838216722011566,
      "num_tokens": 3259790.0,
      "step": 525
    },
    {
      "entropy": 0.8468562078475952,
      "epoch": 0.19041024753332178,
      "grad_norm": 2.921875,
      "learning_rate": 1.977984122590598e-05,
      "loss": 0.8214312744140625,
      "mean_token_accuracy": 0.7904518485069275,
      "num_tokens": 3408921.0,
      "step": 550
    },
    {
      "entropy": 0.862267234325409,
      "epoch": 0.1990652587848364,
      "grad_norm": 2.21875,
      "learning_rate": 1.9749635410612222e-05,
      "loss": 0.8381607818603516,
      "mean_token_accuracy": 0.7850737714767456,
      "num_tokens": 3564021.0,
      "step": 575
    },
    {
      "entropy": 0.8673117184638977,
      "epoch": 0.20772027003635105,
      "grad_norm": 2.28125,
      "learning_rate": 1.971751461296488e-05,
      "loss": 0.8422865295410156,
      "mean_token_accuracy": 0.7839160299301148,
      "num_tokens": 3720222.0,
      "step": 600
    },
    {
      "epoch": 0.20772027003635105,
      "eval_entropy": 0.8581681211463741,
      "eval_loss": 0.8380036950111389,
      "eval_mean_token_accuracy": 0.7849426593936858,
      "eval_num_tokens": 3720222.0,
      "eval_runtime": 16.2659,
      "eval_samples_per_second": 299.091,
      "eval_steps_per_second": 18.751,
      "step": 600
    },
    {
      "entropy": 0.896509393453598,
      "epoch": 0.21637528128786568,
      "grad_norm": 2.25,
      "learning_rate": 1.9683485141995832e-05,
      "loss": 0.8751831817626953,
      "mean_token_accuracy": 0.778223477602005,
      "num_tokens": 3868288.0,
      "step": 625
    },
    {
      "entropy": 0.8858666527271271,
      "epoch": 0.2250302925393803,
      "grad_norm": 2.4375,
      "learning_rate": 1.964755368163055e-05,
      "loss": 0.8597821807861328,
      "mean_token_accuracy": 0.7796040976047516,
      "num_tokens": 4020387.0,
      "step": 650
    },
    {
      "entropy": 0.8457015335559845,
      "epoch": 0.23368530379089492,
      "grad_norm": 2.0625,
      "learning_rate": 1.9609727289375254e-05,
      "loss": 0.8197212982177734,
      "mean_token_accuracy": 0.7878750622272491,
      "num_tokens": 4175769.0,
      "step": 675
    },
    {
      "entropy": 0.8535351765155792,
      "epoch": 0.24234031504240955,
      "grad_norm": 2.328125,
      "learning_rate": 1.9570013394930704e-05,
      "loss": 0.8300620269775391,
      "mean_token_accuracy": 0.786250706911087,
      "num_tokens": 4331114.0,
      "step": 700
    },
    {
      "entropy": 0.8614680600166321,
      "epoch": 0.2509953262939242,
      "grad_norm": 2.0625,
      "learning_rate": 1.9528419798732902e-05,
      "loss": 0.8407713317871094,
      "mean_token_accuracy": 0.7833027255535125,
      "num_tokens": 4488237.0,
      "step": 725
    },
    {
      "entropy": 0.8501963889598847,
      "epoch": 0.2596503375454388,
      "grad_norm": 2.328125,
      "learning_rate": 1.948495467042095e-05,
      "loss": 0.8218775177001953,
      "mean_token_accuracy": 0.788744968175888,
      "num_tokens": 4642068.0,
      "step": 750
    },
    {
      "entropy": 0.8383999621868133,
      "epoch": 0.26830534879695345,
      "grad_norm": 2.296875,
      "learning_rate": 1.9439626547232433e-05,
      "loss": 0.80978515625,
      "mean_token_accuracy": 0.7903480684757233,
      "num_tokens": 4788314.0,
      "step": 775
    },
    {
      "entropy": 0.8568655729293824,
      "epoch": 0.2769603600484681,
      "grad_norm": 2.53125,
      "learning_rate": 1.939244433232652e-05,
      "loss": 0.8343601226806641,
      "mean_token_accuracy": 0.7855056190490722,
      "num_tokens": 4941131.0,
      "step": 800
    },
    {
      "epoch": 0.2769603600484681,
      "eval_entropy": 0.859573404906226,
      "eval_loss": 0.8307726383209229,
      "eval_mean_token_accuracy": 0.7864908138259512,
      "eval_num_tokens": 4941131.0,
      "eval_runtime": 16.2516,
      "eval_samples_per_second": 299.355,
      "eval_steps_per_second": 18.767,
      "step": 800
    },
    {
      "entropy": 0.8670335674285888,
      "epoch": 0.2856153712999827,
      "grad_norm": 1.9765625,
      "learning_rate": 1.934341729303529e-05,
      "loss": 0.8394245910644531,
      "mean_token_accuracy": 0.7841019070148468,
      "num_tokens": 5096940.0,
      "step": 825
    },
    {
      "entropy": 0.867160736322403,
      "epoch": 0.2942703825514973,
      "grad_norm": 2.21875,
      "learning_rate": 1.9292555059043457e-05,
      "loss": 0.8409418487548828,
      "mean_token_accuracy": 0.7830280113220215,
      "num_tokens": 5250616.0,
      "step": 850
    },
    {
      "entropy": 0.8582142984867096,
      "epoch": 0.3029253938030119,
      "grad_norm": 2.15625,
      "learning_rate": 1.9239867620496952e-05,
      "loss": 0.8346880340576172,
      "mean_token_accuracy": 0.783586630821228,
      "num_tokens": 5408253.0,
      "step": 875
    },
    {
      "entropy": 0.84959406375885,
      "epoch": 0.31158040505452655,
      "grad_norm": 2.1875,
      "learning_rate": 1.9185365326040713e-05,
      "loss": 0.826163330078125,
      "mean_token_accuracy": 0.7881129395961761,
      "num_tokens": 5564921.0,
      "step": 900
    },
    {
      "entropy": 0.8895779585838318,
      "epoch": 0.3202354163060412,
      "grad_norm": 2.1875,
      "learning_rate": 1.9129058880786026e-05,
      "loss": 0.866309585571289,
      "mean_token_accuracy": 0.7782393181324005,
      "num_tokens": 5722704.0,
      "step": 925
    },
    {
      "entropy": 0.8372803354263305,
      "epoch": 0.3288904275575558,
      "grad_norm": 1.890625,
      "learning_rate": 1.907095934420789e-05,
      "loss": 0.8083054351806641,
      "mean_token_accuracy": 0.7903852713108063,
      "num_tokens": 5881153.0,
      "step": 950
    },
    {
      "entropy": 0.8762626206874847,
      "epoch": 0.33754543880907045,
      "grad_norm": 1.96875,
      "learning_rate": 1.9011078127972748e-05,
      "loss": 0.8495908355712891,
      "mean_token_accuracy": 0.7841223049163818,
      "num_tokens": 6039323.0,
      "step": 975
    },
    {
      "entropy": 0.8670505321025849,
      "epoch": 0.3462004500605851,
      "grad_norm": 2.25,
      "learning_rate": 1.8949426993697074e-05,
      "loss": 0.8448488616943359,
      "mean_token_accuracy": 0.785959769487381,
      "num_tokens": 6197683.0,
      "step": 1000
    },
    {
      "epoch": 0.3462004500605851,
      "eval_entropy": 0.8536441159052927,
      "eval_loss": 0.8251546025276184,
      "eval_mean_token_accuracy": 0.7873384729760592,
      "eval_num_tokens": 6197683.0,
      "eval_runtime": 16.2591,
      "eval_samples_per_second": 299.217,
      "eval_steps_per_second": 18.759,
      "step": 1000
    },
    {
      "entropy": 0.8527129173278809,
      "epoch": 0.3548554613120997,
      "grad_norm": 2.171875,
      "learning_rate": 1.888601805063719e-05,
      "loss": 0.8221394348144532,
      "mean_token_accuracy": 0.7878592002391815,
      "num_tokens": 6353591.0,
      "step": 1025
    },
    {
      "entropy": 0.8309027600288391,
      "epoch": 0.36351047256361435,
      "grad_norm": 2.1875,
      "learning_rate": 1.8820863753310812e-05,
      "loss": 0.804949951171875,
      "mean_token_accuracy": 0.7907659840583802,
      "num_tokens": 6503522.0,
      "step": 1050
    },
    {
      "entropy": 0.8664905667304993,
      "epoch": 0.372165483815129,
      "grad_norm": 2.234375,
      "learning_rate": 1.875397689905081e-05,
      "loss": 0.8386592102050782,
      "mean_token_accuracy": 0.783461629152298,
      "num_tokens": 6657038.0,
      "step": 1075
    },
    {
      "entropy": 0.8521995890140533,
      "epoch": 0.38082049506664356,
      "grad_norm": 2.140625,
      "learning_rate": 1.8685370625491582e-05,
      "loss": 0.8274839782714843,
      "mean_token_accuracy": 0.7857646262645721,
      "num_tokens": 6813403.0,
      "step": 1100
    },
    {
      "entropy": 0.85324840426445,
      "epoch": 0.3894755063181582,
      "grad_norm": 2.03125,
      "learning_rate": 1.861505840798864e-05,
      "loss": 0.8251450347900391,
      "mean_token_accuracy": 0.7872388279438018,
      "num_tokens": 6969337.0,
      "step": 1125
    },
    {
      "entropy": 0.8578063642978668,
      "epoch": 0.3981305175696728,
      "grad_norm": 2.09375,
      "learning_rate": 1.8543054056971815e-05,
      "loss": 0.834039077758789,
      "mean_token_accuracy": 0.7848770570755005,
      "num_tokens": 7124334.0,
      "step": 1150
    },
    {
      "entropy": 0.8491312241554261,
      "epoch": 0.40678552882118746,
      "grad_norm": 2.0625,
      "learning_rate": 1.8469371715232686e-05,
      "loss": 0.8220222473144532,
      "mean_token_accuracy": 0.7871887254714965,
      "num_tokens": 7277475.0,
      "step": 1175
    },
    {
      "entropy": 0.8443890881538391,
      "epoch": 0.4154405400727021,
      "grad_norm": 2.203125,
      "learning_rate": 1.839402585514671e-05,
      "loss": 0.8166568756103516,
      "mean_token_accuracy": 0.7897883796691895,
      "num_tokens": 7428616.0,
      "step": 1200
    },
    {
      "epoch": 0.4154405400727021,
      "eval_entropy": 0.8290919256014903,
      "eval_loss": 0.8211944103240967,
      "eval_mean_token_accuracy": 0.788363546621604,
      "eval_num_tokens": 7428616.0,
      "eval_runtime": 16.2451,
      "eval_samples_per_second": 299.474,
      "eval_steps_per_second": 18.775,
      "step": 1200
    },
    {
      "entropy": 0.8553554141521453,
      "epoch": 0.4240955513242167,
      "grad_norm": 2.140625,
      "learning_rate": 1.8317031275830607e-05,
      "loss": 0.8335649108886719,
      "mean_token_accuracy": 0.7848791778087616,
      "num_tokens": 7586653.0,
      "step": 1225
    },
    {
      "entropy": 0.8513217878341675,
      "epoch": 0.43275056257573136,
      "grad_norm": 2.078125,
      "learning_rate": 1.8238403100235602e-05,
      "loss": 0.8226930999755859,
      "mean_token_accuracy": 0.7892848730087281,
      "num_tokens": 7736619.0,
      "step": 1250
    },
    {
      "entropy": 0.8567298305034637,
      "epoch": 0.441405573827246,
      "grad_norm": 2.25,
      "learning_rate": 1.8158156772177015e-05,
      "loss": 0.8301739501953125,
      "mean_token_accuracy": 0.7857687485218048,
      "num_tokens": 7895674.0,
      "step": 1275
    },
    {
      "entropy": 0.8383160126209259,
      "epoch": 0.4500605850787606,
      "grad_norm": 2.28125,
      "learning_rate": 1.807630805330087e-05,
      "loss": 0.8143462371826172,
      "mean_token_accuracy": 0.7904347920417786,
      "num_tokens": 8052368.0,
      "step": 1300
    },
    {
      "entropy": 0.858016027212143,
      "epoch": 0.45871559633027525,
      "grad_norm": 2.328125,
      "learning_rate": 1.7992873019988044e-05,
      "loss": 0.8361105346679687,
      "mean_token_accuracy": 0.7837494373321533,
      "num_tokens": 8209037.0,
      "step": 1325
    },
    {
      "entropy": 0.8485116994380951,
      "epoch": 0.46737060758178983,
      "grad_norm": 2.078125,
      "learning_rate": 1.7907868060196627e-05,
      "loss": 0.8189002990722656,
      "mean_token_accuracy": 0.7860724258422852,
      "num_tokens": 8366536.0,
      "step": 1350
    },
    {
      "entropy": 0.8689926111698151,
      "epoch": 0.47602561883330446,
      "grad_norm": 2.28125,
      "learning_rate": 1.7821309870243056e-05,
      "loss": 0.8450764465332031,
      "mean_token_accuracy": 0.7817527639865876,
      "num_tokens": 8524698.0,
      "step": 1375
    },
    {
      "entropy": 0.8510251712799072,
      "epoch": 0.4846806300848191,
      "grad_norm": 2.03125,
      "learning_rate": 1.77332154515227e-05,
      "loss": 0.8352996063232422,
      "mean_token_accuracy": 0.786122510433197,
      "num_tokens": 8680719.0,
      "step": 1400
    },
    {
      "epoch": 0.4846806300848191,
      "eval_entropy": 0.8555903234442727,
      "eval_loss": 0.8170819878578186,
      "eval_mean_token_accuracy": 0.7890570513537672,
      "eval_num_tokens": 8680719.0,
      "eval_runtime": 16.2534,
      "eval_samples_per_second": 299.322,
      "eval_steps_per_second": 18.765,
      "step": 1400
    },
    {
      "entropy": 0.8443090248107911,
      "epoch": 0.49333564133633373,
      "grad_norm": 2.34375,
      "learning_rate": 1.764360210717051e-05,
      "loss": 0.8160585784912109,
      "mean_token_accuracy": 0.7888865935802459,
      "num_tokens": 8839183.0,
      "step": 1425
    },
    {
      "entropy": 0.8397502684593201,
      "epoch": 0.5019906525878484,
      "grad_norm": 2.015625,
      "learning_rate": 1.7552487438662432e-05,
      "loss": 0.8148053741455078,
      "mean_token_accuracy": 0.7885977244377136,
      "num_tokens": 8995586.0,
      "step": 1450
    },
    {
      "entropy": 0.8441075325012207,
      "epoch": 0.5106456638393629,
      "grad_norm": 2.34375,
      "learning_rate": 1.7459889342358176e-05,
      "loss": 0.8217481994628906,
      "mean_token_accuracy": 0.7888561117649079,
      "num_tokens": 9146595.0,
      "step": 1475
    },
    {
      "entropy": 0.8394048309326172,
      "epoch": 0.5193006750908776,
      "grad_norm": 2.296875,
      "learning_rate": 1.736582600598609e-05,
      "loss": 0.8105857849121094,
      "mean_token_accuracy": 0.7910154783725738,
      "num_tokens": 9297789.0,
      "step": 1500
    },
    {
      "entropy": 0.8256420779228211,
      "epoch": 0.5279556863423922,
      "grad_norm": 2.140625,
      "learning_rate": 1.7270315905070822e-05,
      "loss": 0.7967173004150391,
      "mean_token_accuracy": 0.7917508590221405,
      "num_tokens": 9450532.0,
      "step": 1525
    },
    {
      "entropy": 0.817123122215271,
      "epoch": 0.5366106975939069,
      "grad_norm": 1.8828125,
      "learning_rate": 1.717337779930441e-05,
      "loss": 0.7964045715332031,
      "mean_token_accuracy": 0.793145546913147,
      "num_tokens": 9605286.0,
      "step": 1550
    },
    {
      "entropy": 0.8218163585662842,
      "epoch": 0.5452657088454215,
      "grad_norm": 2.25,
      "learning_rate": 1.7075030728861602e-05,
      "loss": 0.7938800048828125,
      "mean_token_accuracy": 0.7939867222309113,
      "num_tokens": 9755402.0,
      "step": 1575
    },
    {
      "entropy": 0.8473324692249298,
      "epoch": 0.5539207200969362,
      "grad_norm": 2.171875,
      "learning_rate": 1.697529401066006e-05,
      "loss": 0.8210481262207031,
      "mean_token_accuracy": 0.7879379904270172,
      "num_tokens": 9915397.0,
      "step": 1600
    },
    {
      "epoch": 0.5539207200969362,
      "eval_entropy": 0.8279588664164309,
      "eval_loss": 0.8136925101280212,
      "eval_mean_token_accuracy": 0.7897968792524495,
      "eval_num_tokens": 9915397.0,
      "eval_runtime": 16.2401,
      "eval_samples_per_second": 299.567,
      "eval_steps_per_second": 18.781,
      "step": 1600
    },
    {
      "entropy": 0.83288947224617,
      "epoch": 0.5625757313484507,
      "grad_norm": 2.125,
      "learning_rate": 1.6874187234566198e-05,
      "loss": 0.8104652404785156,
      "mean_token_accuracy": 0.7881759464740753,
      "num_tokens": 10066176.0,
      "step": 1625
    },
    {
      "entropy": 0.8406240797042847,
      "epoch": 0.5712307425999654,
      "grad_norm": 2.21875,
      "learning_rate": 1.677173025954745e-05,
      "loss": 0.8155978393554687,
      "mean_token_accuracy": 0.7885338914394379,
      "num_tokens": 10216630.0,
      "step": 1650
    },
    {
      "entropy": 0.8361802196502686,
      "epoch": 0.57988575385148,
      "grad_norm": 2.078125,
      "learning_rate": 1.6667943209771634e-05,
      "loss": 0.8124593353271484,
      "mean_token_accuracy": 0.7909375095367431,
      "num_tokens": 10368775.0,
      "step": 1675
    },
    {
      "entropy": 0.8546354687213897,
      "epoch": 0.5885407651029946,
      "grad_norm": 2.390625,
      "learning_rate": 1.6562846470654253e-05,
      "loss": 0.8320878601074219,
      "mean_token_accuracy": 0.7862278306484223,
      "num_tokens": 10522306.0,
      "step": 1700
    },
    {
      "entropy": 0.8271597623825073,
      "epoch": 0.5971957763545093,
      "grad_norm": 2.359375,
      "learning_rate": 1.6456460684854474e-05,
      "loss": 0.802950210571289,
      "mean_token_accuracy": 0.7914960932731628,
      "num_tokens": 10677278.0,
      "step": 1725
    },
    {
      "entropy": 0.8680684280395508,
      "epoch": 0.6058507876060238,
      "grad_norm": 2.21875,
      "learning_rate": 1.6348806748220596e-05,
      "loss": 0.8468040466308594,
      "mean_token_accuracy": 0.7825022864341736,
      "num_tokens": 10832759.0,
      "step": 1750
    },
    {
      "entropy": 0.8460379076004029,
      "epoch": 0.6145057988575385,
      "grad_norm": 1.90625,
      "learning_rate": 1.6239905805685766e-05,
      "loss": 0.81728515625,
      "mean_token_accuracy": 0.7885628819465638,
      "num_tokens": 10987982.0,
      "step": 1775
    },
    {
      "entropy": 0.8258041095733643,
      "epoch": 0.6231608101090531,
      "grad_norm": 1.9609375,
      "learning_rate": 1.6129779247114776e-05,
      "loss": 0.8052913665771484,
      "mean_token_accuracy": 0.7923011314868927,
      "num_tokens": 11142409.0,
      "step": 1800
    },
    {
      "epoch": 0.6231608101090531,
      "eval_entropy": 0.8428774887421092,
      "eval_loss": 0.8103926181793213,
      "eval_mean_token_accuracy": 0.790385250967057,
      "eval_num_tokens": 11142409.0,
      "eval_runtime": 16.2616,
      "eval_samples_per_second": 299.17,
      "eval_steps_per_second": 18.756,
      "step": 1800
    },
    {
      "entropy": 0.8314215159416198,
      "epoch": 0.6318158213605678,
      "grad_norm": 2.09375,
      "learning_rate": 1.601844870310277e-05,
      "loss": 0.8011739349365234,
      "mean_token_accuracy": 0.7905190980434418,
      "num_tokens": 11295629.0,
      "step": 1825
    },
    {
      "entropy": 0.8174971842765808,
      "epoch": 0.6404708326120824,
      "grad_norm": 2.03125,
      "learning_rate": 1.5905936040726636e-05,
      "loss": 0.7869554138183594,
      "mean_token_accuracy": 0.7940086150169372,
      "num_tokens": 11449328.0,
      "step": 1850
    },
    {
      "entropy": 0.8386989748477935,
      "epoch": 0.6491258438635971,
      "grad_norm": 2.03125,
      "learning_rate": 1.579226335924999e-05,
      "loss": 0.8210865783691407,
      "mean_token_accuracy": 0.7893707096576691,
      "num_tokens": 11605530.0,
      "step": 1875
    },
    {
      "entropy": 0.8110369992256165,
      "epoch": 0.6577808551151116,
      "grad_norm": 2.3125,
      "learning_rate": 1.5677452985782493e-05,
      "loss": 0.7894587707519531,
      "mean_token_accuracy": 0.7973484849929809,
      "num_tokens": 11760213.0,
      "step": 1900
    },
    {
      "entropy": 0.8280953919887543,
      "epoch": 0.6664358663666263,
      "grad_norm": 2.203125,
      "learning_rate": 1.5561527470894503e-05,
      "loss": 0.7997794342041016,
      "mean_token_accuracy": 0.7926359629631042,
      "num_tokens": 11913380.0,
      "step": 1925
    },
    {
      "entropy": 0.8196328341960907,
      "epoch": 0.6750908776181409,
      "grad_norm": 2.171875,
      "learning_rate": 1.544450958418775e-05,
      "loss": 0.7956825256347656,
      "mean_token_accuracy": 0.7946672177314759,
      "num_tokens": 12065793.0,
      "step": 1950
    },
    {
      "entropy": 0.8513234782218934,
      "epoch": 0.6837458888696555,
      "grad_norm": 1.8671875,
      "learning_rate": 1.5326422309823046e-05,
      "loss": 0.8301435089111329,
      "mean_token_accuracy": 0.7851076364517212,
      "num_tokens": 12229815.0,
      "step": 1975
    },
    {
      "entropy": 0.8259239780902863,
      "epoch": 0.6924009001211702,
      "grad_norm": 2.078125,
      "learning_rate": 1.5207288842005813e-05,
      "loss": 0.8005783081054687,
      "mean_token_accuracy": 0.79151895403862,
      "num_tokens": 12391391.0,
      "step": 2000
    },
    {
      "epoch": 0.6924009001211702,
      "eval_entropy": 0.828711828149733,
      "eval_loss": 0.8071080446243286,
      "eval_mean_token_accuracy": 0.7911858433582744,
      "eval_num_tokens": 12391391.0,
      "eval_runtime": 16.2568,
      "eval_samples_per_second": 299.259,
      "eval_steps_per_second": 18.761,
      "step": 2000
    },
    {
      "entropy": 0.8361215782165528,
      "epoch": 0.7010559113726847,
      "grad_norm": 2.109375,
      "learning_rate": 1.508713258043041e-05,
      "loss": 0.8149121856689453,
      "mean_token_accuracy": 0.7890836715698242,
      "num_tokens": 12548530.0,
      "step": 2025
    },
    {
      "entropy": 0.8404046738147736,
      "epoch": 0.7097109226241994,
      "grad_norm": 2.0,
      "learning_rate": 1.4965977125684018e-05,
      "loss": 0.814455337524414,
      "mean_token_accuracy": 0.7903144693374634,
      "num_tokens": 12703495.0,
      "step": 2050
    },
    {
      "entropy": 0.8286141097545624,
      "epoch": 0.718365933875714,
      "grad_norm": 2.296875,
      "learning_rate": 1.484384627461115e-05,
      "loss": 0.8082135009765625,
      "mean_token_accuracy": 0.7908019495010375,
      "num_tokens": 12856585.0,
      "step": 2075
    },
    {
      "entropy": 0.8540136122703552,
      "epoch": 0.7270209451272287,
      "grad_norm": 2.09375,
      "learning_rate": 1.4720764015639555e-05,
      "loss": 0.8297102355957031,
      "mean_token_accuracy": 0.7875827658176422,
      "num_tokens": 13011641.0,
      "step": 2100
    },
    {
      "entropy": 0.8180252170562744,
      "epoch": 0.7356759563787433,
      "grad_norm": 2.171875,
      "learning_rate": 1.4596754524068515e-05,
      "loss": 0.791978988647461,
      "mean_token_accuracy": 0.7945426023006439,
      "num_tokens": 13167258.0,
      "step": 2125
    },
    {
      "entropy": 0.8130170261859894,
      "epoch": 0.744330967630258,
      "grad_norm": 2.1875,
      "learning_rate": 1.4471842157320443e-05,
      "loss": 0.7898551940917968,
      "mean_token_accuracy": 0.7971277523040772,
      "num_tokens": 13315266.0,
      "step": 2150
    },
    {
      "entropy": 0.8508910453319549,
      "epoch": 0.7529859788817725,
      "grad_norm": 2.140625,
      "learning_rate": 1.4346051450156689e-05,
      "loss": 0.8204324340820313,
      "mean_token_accuracy": 0.7856499540805817,
      "num_tokens": 13473686.0,
      "step": 2175
    },
    {
      "entropy": 0.8430261933803558,
      "epoch": 0.7616409901332871,
      "grad_norm": 2.125,
      "learning_rate": 1.4219407109858532e-05,
      "loss": 0.8170342254638672,
      "mean_token_accuracy": 0.7883759355545044,
      "num_tokens": 13627766.0,
      "step": 2200
    },
    {
      "epoch": 0.7616409901332871,
      "eval_entropy": 0.8279429495334625,
      "eval_loss": 0.8048078417778015,
      "eval_mean_token_accuracy": 0.7913855454960808,
      "eval_num_tokens": 13627766.0,
      "eval_runtime": 16.2774,
      "eval_samples_per_second": 298.88,
      "eval_steps_per_second": 18.738,
      "step": 2200
    },
    {
      "entropy": 0.8365470349788666,
      "epoch": 0.7702960013848018,
      "grad_norm": 2.3125,
      "learning_rate": 1.4091934011374281e-05,
      "loss": 0.8083656311035157,
      "mean_token_accuracy": 0.7891746735572815,
      "num_tokens": 13781809.0,
      "step": 2225
    },
    {
      "entropy": 0.8262781822681426,
      "epoch": 0.7789510126363164,
      "grad_norm": 2.265625,
      "learning_rate": 1.396365719243344e-05,
      "loss": 0.8065071868896484,
      "mean_token_accuracy": 0.7918438529968261,
      "num_tokens": 13937124.0,
      "step": 2250
    },
    {
      "entropy": 0.8354354929924012,
      "epoch": 0.7876060238878311,
      "grad_norm": 2.03125,
      "learning_rate": 1.3834601848628897e-05,
      "loss": 0.8083207702636719,
      "mean_token_accuracy": 0.7916932463645935,
      "num_tokens": 14088345.0,
      "step": 2275
    },
    {
      "entropy": 0.8217170929908753,
      "epoch": 0.7962610351393457,
      "grad_norm": 2.203125,
      "learning_rate": 1.3704793328468122e-05,
      "loss": 0.7991712951660156,
      "mean_token_accuracy": 0.7924394917488098,
      "num_tokens": 14241089.0,
      "step": 2300
    },
    {
      "entropy": 0.8267798566818237,
      "epoch": 0.8049160463908603,
      "grad_norm": 2.09375,
      "learning_rate": 1.3574257128394315e-05,
      "loss": 0.8018778228759765,
      "mean_token_accuracy": 0.7922985780239106,
      "num_tokens": 14392337.0,
      "step": 2325
    },
    {
      "entropy": 0.8261617541313171,
      "epoch": 0.8135710576423749,
      "grad_norm": 2.171875,
      "learning_rate": 1.3443018887778494e-05,
      "loss": 0.8028769683837891,
      "mean_token_accuracy": 0.7926994669437408,
      "num_tokens": 14544382.0,
      "step": 2350
    },
    {
      "entropy": 0.8380094921588898,
      "epoch": 0.8222260688938896,
      "grad_norm": 2.3125,
      "learning_rate": 1.3311104383883537e-05,
      "loss": 0.8226547241210938,
      "mean_token_accuracy": 0.7886342179775238,
      "num_tokens": 14703460.0,
      "step": 2375
    },
    {
      "entropy": 0.8281042277812958,
      "epoch": 0.8308810801454042,
      "grad_norm": 2.296875,
      "learning_rate": 1.3178539526801106e-05,
      "loss": 0.8025864410400391,
      "mean_token_accuracy": 0.7910430765151978,
      "num_tokens": 14860294.0,
      "step": 2400
    },
    {
      "epoch": 0.8308810801454042,
      "eval_entropy": 0.8327872855741469,
      "eval_loss": 0.8026647567749023,
      "eval_mean_token_accuracy": 0.7919463300314106,
      "eval_num_tokens": 14860294.0,
      "eval_runtime": 16.2486,
      "eval_samples_per_second": 299.41,
      "eval_steps_per_second": 18.771,
      "step": 2400
    },
    {
      "entropy": 0.811096111536026,
      "epoch": 0.8395360913969189,
      "grad_norm": 2.4375,
      "learning_rate": 1.3045350354362495e-05,
      "loss": 0.7822922515869141,
      "mean_token_accuracy": 0.7949872577190399,
      "num_tokens": 15015328.0,
      "step": 2425
    },
    {
      "entropy": 0.8189793074131012,
      "epoch": 0.8481911026484334,
      "grad_norm": 2.3125,
      "learning_rate": 1.2911563027024392e-05,
      "loss": 0.8003229522705078,
      "mean_token_accuracy": 0.7943521749973297,
      "num_tokens": 15170350.0,
      "step": 2450
    },
    {
      "entropy": 0.8213041162490845,
      "epoch": 0.856846113899948,
      "grad_norm": 2.21875,
      "learning_rate": 1.2777203822730557e-05,
      "loss": 0.7970082092285157,
      "mean_token_accuracy": 0.7927951776981353,
      "num_tokens": 15324153.0,
      "step": 2475
    },
    {
      "entropy": 0.824720960855484,
      "epoch": 0.8655011251514627,
      "grad_norm": 2.125,
      "learning_rate": 1.2642299131750417e-05,
      "loss": 0.7908332824707032,
      "mean_token_accuracy": 0.7957222282886505,
      "num_tokens": 15473303.0,
      "step": 2500
    },
    {
      "entropy": 0.838714919090271,
      "epoch": 0.8741561364029773,
      "grad_norm": 2.09375,
      "learning_rate": 1.250687545149559e-05,
      "loss": 0.8137520599365234,
      "mean_token_accuracy": 0.7890112113952636,
      "num_tokens": 15629017.0,
      "step": 2525
    },
    {
      "entropy": 0.8311066961288452,
      "epoch": 0.882811147654492,
      "grad_norm": 2.1875,
      "learning_rate": 1.2370959381315393e-05,
      "loss": 0.8034115600585937,
      "mean_token_accuracy": 0.7926496362686157,
      "num_tokens": 15777876.0,
      "step": 2550
    },
    {
      "entropy": 0.8097344720363617,
      "epoch": 0.8914661589060066,
      "grad_norm": 2.203125,
      "learning_rate": 1.2234577617272282e-05,
      "loss": 0.786445083618164,
      "mean_token_accuracy": 0.7966820681095124,
      "num_tokens": 15929184.0,
      "step": 2575
    },
    {
      "entropy": 0.8254410099983215,
      "epoch": 0.9001211701575212,
      "grad_norm": 2.296875,
      "learning_rate": 1.209775694689837e-05,
      "loss": 0.8023792266845703,
      "mean_token_accuracy": 0.7923547208309174,
      "num_tokens": 16087411.0,
      "step": 2600
    },
    {
      "epoch": 0.9001211701575212,
      "eval_entropy": 0.8213896471945966,
      "eval_loss": 0.8008216619491577,
      "eval_mean_token_accuracy": 0.7924871415388389,
      "eval_num_tokens": 16087411.0,
      "eval_runtime": 16.2722,
      "eval_samples_per_second": 298.976,
      "eval_steps_per_second": 18.744,
      "step": 2600
    },
    {
      "entropy": 0.8255444514751434,
      "epoch": 0.9087761814090358,
      "grad_norm": 1.9140625,
      "learning_rate": 1.1960524243933875e-05,
      "loss": 0.8017765808105469,
      "mean_token_accuracy": 0.7921687912940979,
      "num_tokens": 16244042.0,
      "step": 2625
    },
    {
      "entropy": 0.8252915942668915,
      "epoch": 0.9174311926605505,
      "grad_norm": 2.21875,
      "learning_rate": 1.182290646304874e-05,
      "loss": 0.8029595947265625,
      "mean_token_accuracy": 0.7924495422840119,
      "num_tokens": 16399768.0,
      "step": 2650
    },
    {
      "entropy": 0.8625216126441956,
      "epoch": 0.9260862039120651,
      "grad_norm": 2.171875,
      "learning_rate": 1.1684930634548287e-05,
      "loss": 0.838686752319336,
      "mean_token_accuracy": 0.7851159811019898,
      "num_tokens": 16557371.0,
      "step": 2675
    },
    {
      "entropy": 0.8215836095809936,
      "epoch": 0.9347412151635797,
      "grad_norm": 2.265625,
      "learning_rate": 1.1546623859064053e-05,
      "loss": 0.7977464294433594,
      "mean_token_accuracy": 0.7929444241523743,
      "num_tokens": 16713045.0,
      "step": 2700
    },
    {
      "entropy": 0.8286124646663666,
      "epoch": 0.9433962264150944,
      "grad_norm": 2.078125,
      "learning_rate": 1.1408013302230765e-05,
      "loss": 0.802153091430664,
      "mean_token_accuracy": 0.7924544358253479,
      "num_tokens": 16867145.0,
      "step": 2725
    },
    {
      "entropy": 0.8237406086921691,
      "epoch": 0.9520512376666089,
      "grad_norm": 2.15625,
      "learning_rate": 1.126912618935063e-05,
      "loss": 0.8021310424804687,
      "mean_token_accuracy": 0.7944400310516357,
      "num_tokens": 17014455.0,
      "step": 2750
    },
    {
      "entropy": 0.826482676267624,
      "epoch": 0.9607062489181236,
      "grad_norm": 2.140625,
      "learning_rate": 1.1129989800045817e-05,
      "loss": 0.8052169036865234,
      "mean_token_accuracy": 0.7911915636062622,
      "num_tokens": 17175484.0,
      "step": 2775
    },
    {
      "entropy": 0.8201354122161866,
      "epoch": 0.9693612601696382,
      "grad_norm": 2.421875,
      "learning_rate": 1.099063146290032e-05,
      "loss": 0.7985061645507813,
      "mean_token_accuracy": 0.7938509380817413,
      "num_tokens": 17334068.0,
      "step": 2800
    },
    {
      "epoch": 0.9693612601696382,
      "eval_entropy": 0.8281321289109402,
      "eval_loss": 0.7994962334632874,
      "eval_mean_token_accuracy": 0.792740264681519,
      "eval_num_tokens": 17334068.0,
      "eval_runtime": 16.2806,
      "eval_samples_per_second": 298.822,
      "eval_steps_per_second": 18.734,
      "step": 2800
    },
    {
      "entropy": 0.8046394038200378,
      "epoch": 0.9780162714211529,
      "grad_norm": 2.46875,
      "learning_rate": 1.08510785500922e-05,
      "loss": 0.7775811004638672,
      "mean_token_accuracy": 0.7966744971275329,
      "num_tokens": 17483883.0,
      "step": 2825
    },
    {
      "entropy": 0.8306651365756988,
      "epoch": 0.9866712826726675,
      "grad_norm": 2.359375,
      "learning_rate": 1.0711358472017244e-05,
      "loss": 0.7954608154296875,
      "mean_token_accuracy": 0.7919278919696808,
      "num_tokens": 17639972.0,
      "step": 2850
    },
    {
      "entropy": 0.8295222818851471,
      "epoch": 0.9953262939241821,
      "grad_norm": 2.125,
      "learning_rate": 1.0571498671905146e-05,
      "loss": 0.8072916412353516,
      "mean_token_accuracy": 0.7907999694347382,
      "num_tokens": 17794430.0,
      "step": 2875
    },
    {
      "entropy": 0.8070371369926297,
      "epoch": 1.0038082049506665,
      "grad_norm": 2.09375,
      "learning_rate": 1.0431526620429207e-05,
      "loss": 0.760282211303711,
      "mean_token_accuracy": 0.7995899867038337,
      "num_tokens": 17946140.0,
      "step": 2900
    },
    {
      "entropy": 0.7756562912464142,
      "epoch": 1.0124632162021812,
      "grad_norm": 2.140625,
      "learning_rate": 1.0291469810310669e-05,
      "loss": 0.741202621459961,
      "mean_token_accuracy": 0.803013744354248,
      "num_tokens": 18100839.0,
      "step": 2925
    },
    {
      "entropy": 0.7452013885974884,
      "epoch": 1.0211182274536956,
      "grad_norm": 2.15625,
      "learning_rate": 1.0151355750918712e-05,
      "loss": 0.7064004516601563,
      "mean_token_accuracy": 0.8126378917694091,
      "num_tokens": 18256691.0,
      "step": 2950
    },
    {
      "entropy": 0.7718643927574158,
      "epoch": 1.0297732387052103,
      "grad_norm": 1.8671875,
      "learning_rate": 1.001121196286718e-05,
      "loss": 0.7311404418945312,
      "mean_token_accuracy": 0.8063716220855713,
      "num_tokens": 18418483.0,
      "step": 2975
    },
    {
      "entropy": 0.765803085565567,
      "epoch": 1.038428249956725,
      "grad_norm": 2.0,
      "learning_rate": 9.871065972609089e-06,
      "loss": 0.7332243347167968,
      "mean_token_accuracy": 0.8049838674068451,
      "num_tokens": 18576253.0,
      "step": 3000
    },
    {
      "epoch": 1.038428249956725,
      "eval_entropy": 0.7786700289757525,
      "eval_loss": 0.8012996912002563,
      "eval_mean_token_accuracy": 0.792937566022404,
      "eval_num_tokens": 18576253.0,
      "eval_runtime": 16.277,
      "eval_samples_per_second": 298.888,
      "eval_steps_per_second": 18.738,
      "step": 3000
    },
    {
      "entropy": 0.7836507785320282,
      "epoch": 1.0470832612082395,
      "grad_norm": 2.0625,
      "learning_rate": 9.730945307030009e-06,
      "loss": 0.7418560028076172,
      "mean_token_accuracy": 0.8028053915500641,
      "num_tokens": 18729564.0,
      "step": 3025
    },
    {
      "entropy": 0.7653721117973328,
      "epoch": 1.0557382724597542,
      "grad_norm": 2.03125,
      "learning_rate": 9.590877488041336e-06,
      "loss": 0.7263675689697265,
      "mean_token_accuracy": 0.807190809249878,
      "num_tokens": 18891538.0,
      "step": 3050
    },
    {
      "entropy": 0.7604654181003571,
      "epoch": 1.0643932837112688,
      "grad_norm": 2.296875,
      "learning_rate": 9.45089002717457e-06,
      "loss": 0.72638916015625,
      "mean_token_accuracy": 0.8084564554691315,
      "num_tokens": 19050606.0,
      "step": 3075
    },
    {
      "entropy": 0.7386153995990753,
      "epoch": 1.0730482949627835,
      "grad_norm": 1.9765625,
      "learning_rate": 9.311010420177597e-06,
      "loss": 0.706187973022461,
      "mean_token_accuracy": 0.8120154666900635,
      "num_tokens": 19208707.0,
      "step": 3100
    },
    {
      "entropy": 0.7525534296035766,
      "epoch": 1.081703306214298,
      "grad_norm": 2.171875,
      "learning_rate": 9.171266141614112e-06,
      "loss": 0.7135391235351562,
      "mean_token_accuracy": 0.8123032712936401,
      "num_tokens": 19362737.0,
      "step": 3125
    },
    {
      "entropy": 0.7687537455558777,
      "epoch": 1.0903583174658127,
      "grad_norm": 2.109375,
      "learning_rate": 9.031684639467159e-06,
      "loss": 0.734495849609375,
      "mean_token_accuracy": 0.8068291318416595,
      "num_tokens": 19521058.0,
      "step": 3150
    },
    {
      "entropy": 0.7806685280799865,
      "epoch": 1.0990133287173274,
      "grad_norm": 2.046875,
      "learning_rate": 8.892293329747923e-06,
      "loss": 0.7469673156738281,
      "mean_token_accuracy": 0.8041970551013946,
      "num_tokens": 19675603.0,
      "step": 3175
    },
    {
      "entropy": 0.7709089267253876,
      "epoch": 1.1076683399688418,
      "grad_norm": 2.15625,
      "learning_rate": 8.753119591110802e-06,
      "loss": 0.7363986206054688,
      "mean_token_accuracy": 0.8061197555065155,
      "num_tokens": 19828898.0,
      "step": 3200
    },
    {
      "epoch": 1.1076683399688418,
      "eval_entropy": 0.7721671974072691,
      "eval_loss": 0.8010231852531433,
      "eval_mean_token_accuracy": 0.7929807965872717,
      "eval_num_tokens": 19828898.0,
      "eval_runtime": 16.2529,
      "eval_samples_per_second": 299.331,
      "eval_steps_per_second": 18.766,
      "step": 3200
    },
    {
      "entropy": 0.7211595928668976,
      "epoch": 1.1163233512203565,
      "grad_norm": 2.046875,
      "learning_rate": 8.614190759475789e-06,
      "loss": 0.6826175689697266,
      "mean_token_accuracy": 0.8186909162998199,
      "num_tokens": 19982072.0,
      "step": 3225
    },
    {
      "entropy": 0.7708668720722198,
      "epoch": 1.1249783624718712,
      "grad_norm": 2.15625,
      "learning_rate": 8.475534122659275e-06,
      "loss": 0.7388634490966797,
      "mean_token_accuracy": 0.8042692136764527,
      "num_tokens": 20137744.0,
      "step": 3250
    },
    {
      "entropy": 0.7538685095310211,
      "epoch": 1.133633373723386,
      "grad_norm": 2.171875,
      "learning_rate": 8.337176915014293e-06,
      "loss": 0.7132722473144532,
      "mean_token_accuracy": 0.8105318355560303,
      "num_tokens": 20290524.0,
      "step": 3275
    },
    {
      "entropy": 0.7487189078330994,
      "epoch": 1.1422883849749004,
      "grad_norm": 2.046875,
      "learning_rate": 8.199146312081265e-06,
      "loss": 0.7089384460449218,
      "mean_token_accuracy": 0.8108018398284912,
      "num_tokens": 20443485.0,
      "step": 3300
    },
    {
      "entropy": 0.7543841946125031,
      "epoch": 1.150943396226415,
      "grad_norm": 2.0625,
      "learning_rate": 8.061469425250273e-06,
      "loss": 0.7227223205566407,
      "mean_token_accuracy": 0.8090525925159454,
      "num_tokens": 20595797.0,
      "step": 3325
    },
    {
      "entropy": 0.7593919610977173,
      "epoch": 1.1595984074779297,
      "grad_norm": 2.015625,
      "learning_rate": 7.924173296435987e-06,
      "loss": 0.7215187072753906,
      "mean_token_accuracy": 0.8093890881538391,
      "num_tokens": 20749554.0,
      "step": 3350
    },
    {
      "entropy": 0.7829481947422028,
      "epoch": 1.1682534187294444,
      "grad_norm": 2.171875,
      "learning_rate": 7.78728489276618e-06,
      "loss": 0.7491289520263672,
      "mean_token_accuracy": 0.8015105819702149,
      "num_tokens": 20909534.0,
      "step": 3375
    },
    {
      "entropy": 0.7392001152038574,
      "epoch": 1.176908429980959,
      "grad_norm": 2.1875,
      "learning_rate": 7.650831101284977e-06,
      "loss": 0.7020195007324219,
      "mean_token_accuracy": 0.8150424540042878,
      "num_tokens": 21059158.0,
      "step": 3400
    },
    {
      "epoch": 1.176908429980959,
      "eval_entropy": 0.7698653085309951,
      "eval_loss": 0.8008679747581482,
      "eval_mean_token_accuracy": 0.793300522155449,
      "eval_num_tokens": 21059158.0,
      "eval_runtime": 16.2799,
      "eval_samples_per_second": 298.835,
      "eval_steps_per_second": 18.735,
      "step": 3400
    },
    {
      "entropy": 0.768788868188858,
      "epoch": 1.1855634412324736,
      "grad_norm": 2.046875,
      "learning_rate": 7.514838723671799e-06,
      "loss": 0.7372692108154297,
      "mean_token_accuracy": 0.8063059449195862,
      "num_tokens": 21208688.0,
      "step": 3425
    },
    {
      "entropy": 0.7443365502357483,
      "epoch": 1.1942184524839883,
      "grad_norm": 1.875,
      "learning_rate": 7.379334470977093e-06,
      "loss": 0.710152587890625,
      "mean_token_accuracy": 0.811426283121109,
      "num_tokens": 21365731.0,
      "step": 3450
    },
    {
      "entropy": 0.7438737845420837,
      "epoch": 1.2028734637355027,
      "grad_norm": 1.96875,
      "learning_rate": 7.244344958375881e-06,
      "loss": 0.710167465209961,
      "mean_token_accuracy": 0.8120112872123718,
      "num_tokens": 21519688.0,
      "step": 3475
    },
    {
      "entropy": 0.7728407967090607,
      "epoch": 1.2115284749870174,
      "grad_norm": 2.046875,
      "learning_rate": 7.109896699940084e-06,
      "loss": 0.7461888885498047,
      "mean_token_accuracy": 0.8037091171741486,
      "num_tokens": 21676493.0,
      "step": 3500
    },
    {
      "entropy": 0.7640272963047028,
      "epoch": 1.2201834862385321,
      "grad_norm": 2.203125,
      "learning_rate": 6.976016103430765e-06,
      "loss": 0.7235647583007813,
      "mean_token_accuracy": 0.8083086097240448,
      "num_tokens": 21825664.0,
      "step": 3525
    },
    {
      "entropy": 0.7460694527626037,
      "epoch": 1.2288384974900468,
      "grad_norm": 2.03125,
      "learning_rate": 6.842729465111224e-06,
      "loss": 0.7100708770751953,
      "mean_token_accuracy": 0.8116568970680237,
      "num_tokens": 21976504.0,
      "step": 3550
    },
    {
      "entropy": 0.759437519311905,
      "epoch": 1.2374935087415613,
      "grad_norm": 2.609375,
      "learning_rate": 6.710062964581988e-06,
      "loss": 0.722847900390625,
      "mean_token_accuracy": 0.807286411523819,
      "num_tokens": 22133165.0,
      "step": 3575
    },
    {
      "entropy": 0.7708945345878601,
      "epoch": 1.246148519993076,
      "grad_norm": 2.015625,
      "learning_rate": 6.578042659638726e-06,
      "loss": 0.7426469421386719,
      "mean_token_accuracy": 0.8035933816432953,
      "num_tokens": 22290456.0,
      "step": 3600
    },
    {
      "epoch": 1.246148519993076,
      "eval_entropy": 0.7726592179204597,
      "eval_loss": 0.8001708984375,
      "eval_mean_token_accuracy": 0.7932597899046101,
      "eval_num_tokens": 22290456.0,
      "eval_runtime": 16.2656,
      "eval_samples_per_second": 299.097,
      "eval_steps_per_second": 18.751,
      "step": 3600
    },
    {
      "entropy": 0.7578576147556305,
      "epoch": 1.2548035312445907,
      "grad_norm": 2.078125,
      "learning_rate": 6.446694481154104e-06,
      "loss": 0.7225240325927734,
      "mean_token_accuracy": 0.8097064507007599,
      "num_tokens": 22446824.0,
      "step": 3625
    },
    {
      "entropy": 0.7762425947189331,
      "epoch": 1.2634585424961053,
      "grad_norm": 2.265625,
      "learning_rate": 6.3160442279845206e-06,
      "loss": 0.7412718200683593,
      "mean_token_accuracy": 0.805392199754715,
      "num_tokens": 22602588.0,
      "step": 3650
    },
    {
      "entropy": 0.7637201595306397,
      "epoch": 1.2721135537476198,
      "grad_norm": 2.25,
      "learning_rate": 6.186117561902836e-06,
      "loss": 0.7297930908203125,
      "mean_token_accuracy": 0.8078333747386932,
      "num_tokens": 22752707.0,
      "step": 3675
    },
    {
      "entropy": 0.7704725599288941,
      "epoch": 1.2807685649991345,
      "grad_norm": 2.125,
      "learning_rate": 6.056940002557969e-06,
      "loss": 0.7372798919677734,
      "mean_token_accuracy": 0.8056825184822083,
      "num_tokens": 22907803.0,
      "step": 3700
    },
    {
      "entropy": 0.7552802407741547,
      "epoch": 1.2894235762506492,
      "grad_norm": 2.15625,
      "learning_rate": 5.928536922462449e-06,
      "loss": 0.7166344451904297,
      "mean_token_accuracy": 0.8109099209308624,
      "num_tokens": 23059819.0,
      "step": 3725
    },
    {
      "entropy": 0.7541340529918671,
      "epoch": 1.2980785875021636,
      "grad_norm": 2.3125,
      "learning_rate": 5.80093354200886e-06,
      "loss": 0.7238423919677734,
      "mean_token_accuracy": 0.809191288948059,
      "num_tokens": 23217392.0,
      "step": 3750
    },
    {
      "entropy": 0.7676697385311126,
      "epoch": 1.3067335987536783,
      "grad_norm": 2.21875,
      "learning_rate": 5.674154924516129e-06,
      "loss": 0.7366065979003906,
      "mean_token_accuracy": 0.8054041469097137,
      "num_tokens": 23371073.0,
      "step": 3775
    },
    {
      "entropy": 0.7785192513465882,
      "epoch": 1.315388610005193,
      "grad_norm": 1.9140625,
      "learning_rate": 5.548225971306731e-06,
      "loss": 0.7439189147949219,
      "mean_token_accuracy": 0.8034746491909027,
      "num_tokens": 23524592.0,
      "step": 3800
    },
    {
      "epoch": 1.315388610005193,
      "eval_entropy": 0.7713183484116538,
      "eval_loss": 0.7998276352882385,
      "eval_mean_token_accuracy": 0.7933589818047695,
      "eval_num_tokens": 23524592.0,
      "eval_runtime": 16.2844,
      "eval_samples_per_second": 298.753,
      "eval_steps_per_second": 18.73,
      "step": 3800
    },
    {
      "entropy": 0.7570435631275177,
      "epoch": 1.3240436212567077,
      "grad_norm": 2.078125,
      "learning_rate": 5.423171416815651e-06,
      "loss": 0.7266172790527343,
      "mean_token_accuracy": 0.8094574224948883,
      "num_tokens": 23682141.0,
      "step": 3825
    },
    {
      "entropy": 0.76507066488266,
      "epoch": 1.3326986325082222,
      "grad_norm": 2.15625,
      "learning_rate": 5.299015823732163e-06,
      "loss": 0.7294864654541016,
      "mean_token_accuracy": 0.8064786922931672,
      "num_tokens": 23839050.0,
      "step": 3850
    },
    {
      "entropy": 0.7562107217311859,
      "epoch": 1.3413536437597369,
      "grad_norm": 2.0625,
      "learning_rate": 5.175783578175309e-06,
      "loss": 0.719031982421875,
      "mean_token_accuracy": 0.8091187167167664,
      "num_tokens": 23992205.0,
      "step": 3875
    },
    {
      "entropy": 0.7576862835884094,
      "epoch": 1.3500086550112516,
      "grad_norm": 2.109375,
      "learning_rate": 5.053498884904122e-06,
      "loss": 0.7270169830322266,
      "mean_token_accuracy": 0.8094841420650483,
      "num_tokens": 24139859.0,
      "step": 3900
    },
    {
      "entropy": 0.7541490840911865,
      "epoch": 1.3586636662627662,
      "grad_norm": 2.109375,
      "learning_rate": 4.932185762563381e-06,
      "loss": 0.7113311004638672,
      "mean_token_accuracy": 0.8112999868392944,
      "num_tokens": 24289196.0,
      "step": 3925
    },
    {
      "entropy": 0.7688007843494415,
      "epoch": 1.3673186775142807,
      "grad_norm": 2.1875,
      "learning_rate": 4.811868038965999e-06,
      "loss": 0.745551986694336,
      "mean_token_accuracy": 0.8050249302387238,
      "num_tokens": 24447938.0,
      "step": 3950
    },
    {
      "entropy": 0.7551292729377747,
      "epoch": 1.3759736887657954,
      "grad_norm": 2.1875,
      "learning_rate": 4.692569346412864e-06,
      "loss": 0.7206728363037109,
      "mean_token_accuracy": 0.8098892831802368,
      "num_tokens": 24598742.0,
      "step": 3975
    },
    {
      "entropy": 0.7587357807159424,
      "epoch": 1.38462870001731,
      "grad_norm": 2.125,
      "learning_rate": 4.574313117051059e-06,
      "loss": 0.7276847076416015,
      "mean_token_accuracy": 0.8077455127239227,
      "num_tokens": 24750406.0,
      "step": 4000
    },
    {
      "epoch": 1.38462870001731,
      "eval_entropy": 0.7717335020909544,
      "eval_loss": 0.7996781468391418,
      "eval_mean_token_accuracy": 0.7934077425081222,
      "eval_num_tokens": 24750406.0,
      "eval_runtime": 16.2676,
      "eval_samples_per_second": 299.061,
      "eval_steps_per_second": 18.749,
      "step": 4000
    },
    {
      "entropy": 0.7805033266544342,
      "epoch": 1.3932837112688246,
      "grad_norm": 2.078125,
      "learning_rate": 4.457122578271457e-06,
      "loss": 0.7509593200683594,
      "mean_token_accuracy": 0.8028152561187745,
      "num_tokens": 24906659.0,
      "step": 4025
    },
    {
      "entropy": 0.7705829620361329,
      "epoch": 1.4019387225203392,
      "grad_norm": 2.125,
      "learning_rate": 4.341020748146449e-06,
      "loss": 0.7392497253417969,
      "mean_token_accuracy": 0.8049503290653228,
      "num_tokens": 25066654.0,
      "step": 4050
    },
    {
      "entropy": 0.7581047600507737,
      "epoch": 1.410593733771854,
      "grad_norm": 2.0,
      "learning_rate": 4.226030430908869e-06,
      "loss": 0.7285484313964844,
      "mean_token_accuracy": 0.8085175144672394,
      "num_tokens": 25220415.0,
      "step": 4075
    },
    {
      "entropy": 0.7656786620616913,
      "epoch": 1.4192487450233684,
      "grad_norm": 2.21875,
      "learning_rate": 4.112174212472864e-06,
      "loss": 0.7323892211914063,
      "mean_token_accuracy": 0.8051603281497955,
      "num_tokens": 25374458.0,
      "step": 4100
    },
    {
      "entropy": 0.7522096228599549,
      "epoch": 1.427903756274883,
      "grad_norm": 2.171875,
      "learning_rate": 3.99947445599767e-06,
      "loss": 0.7212752532958985,
      "mean_token_accuracy": 0.8096666383743286,
      "num_tokens": 25525969.0,
      "step": 4125
    },
    {
      "entropy": 0.7608710610866547,
      "epoch": 1.4365587675263978,
      "grad_norm": 2.40625,
      "learning_rate": 3.887953297495121e-06,
      "loss": 0.7316089630126953,
      "mean_token_accuracy": 0.8074719154834747,
      "num_tokens": 25681452.0,
      "step": 4150
    },
    {
      "entropy": 0.7639850771427155,
      "epoch": 1.4452137787779125,
      "grad_norm": 2.140625,
      "learning_rate": 3.77763264148181e-06,
      "loss": 0.7320518493652344,
      "mean_token_accuracy": 0.809201009273529,
      "num_tokens": 25838398.0,
      "step": 4175
    },
    {
      "entropy": 0.7513358175754548,
      "epoch": 1.4538687900294271,
      "grad_norm": 2.21875,
      "learning_rate": 3.6685341566766785e-06,
      "loss": 0.7206896209716797,
      "mean_token_accuracy": 0.8095373523235321,
      "num_tokens": 25992640.0,
      "step": 4200
    },
    {
      "epoch": 1.4538687900294271,
      "eval_entropy": 0.7707103953986871,
      "eval_loss": 0.7995045185089111,
      "eval_mean_token_accuracy": 0.7933821330305005,
      "eval_num_tokens": 25992640.0,
      "eval_runtime": 16.2677,
      "eval_samples_per_second": 299.058,
      "eval_steps_per_second": 18.749,
      "step": 4200
    },
    {
      "entropy": 0.7595930647850037,
      "epoch": 1.4625238012809416,
      "grad_norm": 2.171875,
      "learning_rate": 3.5606792717449455e-06,
      "loss": 0.7264796447753906,
      "mean_token_accuracy": 0.8094517874717713,
      "num_tokens": 26141304.0,
      "step": 4225
    },
    {
      "entropy": 0.7630014491081237,
      "epoch": 1.4711788125324563,
      "grad_norm": 2.15625,
      "learning_rate": 3.4540891710891768e-06,
      "loss": 0.7287496185302734,
      "mean_token_accuracy": 0.808629115819931,
      "num_tokens": 26295315.0,
      "step": 4250
    },
    {
      "entropy": 0.7566706609725952,
      "epoch": 1.479833823783971,
      "grad_norm": 2.34375,
      "learning_rate": 3.3487847906883232e-06,
      "loss": 0.7267755126953125,
      "mean_token_accuracy": 0.8091002118587494,
      "num_tokens": 26450821.0,
      "step": 4275
    },
    {
      "entropy": 0.7702415359020233,
      "epoch": 1.4884888350354855,
      "grad_norm": 2.09375,
      "learning_rate": 3.2447868139855954e-06,
      "loss": 0.74352294921875,
      "mean_token_accuracy": 0.8064157390594482,
      "num_tokens": 26610364.0,
      "step": 4300
    },
    {
      "entropy": 0.78652991771698,
      "epoch": 1.4971438462870001,
      "grad_norm": 2.359375,
      "learning_rate": 3.142115667825868e-06,
      "loss": 0.7553633117675781,
      "mean_token_accuracy": 0.8014207696914672,
      "num_tokens": 26766393.0,
      "step": 4325
    },
    {
      "entropy": 0.7758943355083465,
      "epoch": 1.5057988575385148,
      "grad_norm": 2.0,
      "learning_rate": 3.0407915184435544e-06,
      "loss": 0.7414107513427735,
      "mean_token_accuracy": 0.8034427273273468,
      "num_tokens": 26928821.0,
      "step": 4350
    },
    {
      "entropy": 0.7746455824375152,
      "epoch": 1.5144538687900293,
      "grad_norm": 2.15625,
      "learning_rate": 2.9408342675016288e-06,
      "loss": 0.7338740539550781,
      "mean_token_accuracy": 0.8053066456317901,
      "num_tokens": 27081955.0,
      "step": 4375
    },
    {
      "entropy": 0.7544065153598786,
      "epoch": 1.5231088800415442,
      "grad_norm": 2.125,
      "learning_rate": 2.84226354818263e-06,
      "loss": 0.7168594360351562,
      "mean_token_accuracy": 0.8104126238822937,
      "num_tokens": 27233223.0,
      "step": 4400
    },
    {
      "epoch": 1.5231088800415442,
      "eval_entropy": 0.770115079156688,
      "eval_loss": 0.7993313074111938,
      "eval_mean_token_accuracy": 0.7934993093131019,
      "eval_num_tokens": 27233223.0,
      "eval_runtime": 16.2903,
      "eval_samples_per_second": 298.644,
      "eval_steps_per_second": 18.723,
      "step": 4400
    },
    {
      "entropy": 0.7756598269939423,
      "epoch": 1.5317638912930587,
      "grad_norm": 2.15625,
      "learning_rate": 2.74509872133238e-06,
      "loss": 0.750801010131836,
      "mean_token_accuracy": 0.8043915629386902,
      "num_tokens": 27385671.0,
      "step": 4425
    },
    {
      "entropy": 0.7870103931427002,
      "epoch": 1.5404189025445731,
      "grad_norm": 2.46875,
      "learning_rate": 2.6493588716572306e-06,
      "loss": 0.7561086273193359,
      "mean_token_accuracy": 0.8031144332885742,
      "num_tokens": 27542323.0,
      "step": 4450
    },
    {
      "entropy": 0.7758941638469696,
      "epoch": 1.549073913796088,
      "grad_norm": 2.359375,
      "learning_rate": 2.555062803975512e-06,
      "loss": 0.7479061889648437,
      "mean_token_accuracy": 0.8041390895843505,
      "num_tokens": 27695755.0,
      "step": 4475
    },
    {
      "entropy": 0.754768637418747,
      "epoch": 1.5577289250476025,
      "grad_norm": 2.4375,
      "learning_rate": 2.4622290395239613e-06,
      "loss": 0.7247459411621093,
      "mean_token_accuracy": 0.8090137577056885,
      "num_tokens": 27847694.0,
      "step": 4500
    },
    {
      "entropy": 0.7636888045072555,
      "epoch": 1.5663839362991172,
      "grad_norm": 2.140625,
      "learning_rate": 2.3708758123198904e-06,
      "loss": 0.7294938659667969,
      "mean_token_accuracy": 0.8078456258773804,
      "num_tokens": 28001389.0,
      "step": 4525
    },
    {
      "entropy": 0.7723946702480317,
      "epoch": 1.575038947550632,
      "grad_norm": 1.9453125,
      "learning_rate": 2.2810210655796915e-06,
      "loss": 0.7421239471435547,
      "mean_token_accuracy": 0.8051141202449799,
      "num_tokens": 28155633.0,
      "step": 4550
    },
    {
      "entropy": 0.7729098165035247,
      "epoch": 1.5836939588021464,
      "grad_norm": 1.9375,
      "learning_rate": 2.192682448194555e-06,
      "loss": 0.7385528564453125,
      "mean_token_accuracy": 0.8061604249477387,
      "num_tokens": 28316815.0,
      "step": 4575
    },
    {
      "entropy": 0.7528582894802094,
      "epoch": 1.592348970053661,
      "grad_norm": 2.234375,
      "learning_rate": 2.1058773112639054e-06,
      "loss": 0.719280014038086,
      "mean_token_accuracy": 0.8096978998184204,
      "num_tokens": 28473438.0,
      "step": 4600
    },
    {
      "epoch": 1.592348970053661,
      "eval_entropy": 0.7718443285246365,
      "eval_loss": 0.7991446852684021,
      "eval_mean_token_accuracy": 0.7934910416603088,
      "eval_num_tokens": 28473438.0,
      "eval_runtime": 16.2845,
      "eval_samples_per_second": 298.751,
      "eval_steps_per_second": 18.729,
      "step": 4600
    },
    {
      "entropy": 0.7752782785892487,
      "epoch": 1.6010039813051757,
      "grad_norm": 2.203125,
      "learning_rate": 2.020622704687398e-06,
      "loss": 0.7455803680419922,
      "mean_token_accuracy": 0.8029440557956695,
      "num_tokens": 28629933.0,
      "step": 4625
    },
    {
      "entropy": 0.7682564723491668,
      "epoch": 1.6096589925566902,
      "grad_norm": 2.28125,
      "learning_rate": 1.936935373816046e-06,
      "loss": 0.7349263000488281,
      "mean_token_accuracy": 0.8075251269340515,
      "num_tokens": 28779447.0,
      "step": 4650
    },
    {
      "entropy": 0.7602540409564972,
      "epoch": 1.6183140038082051,
      "grad_norm": 2.15625,
      "learning_rate": 1.8548317561631534e-06,
      "loss": 0.722344741821289,
      "mean_token_accuracy": 0.809203851222992,
      "num_tokens": 28935999.0,
      "step": 4675
    },
    {
      "entropy": 0.7502141880989075,
      "epoch": 1.6269690150597196,
      "grad_norm": 1.984375,
      "learning_rate": 1.7743279781757493e-06,
      "loss": 0.718406982421875,
      "mean_token_accuracy": 0.8091218245029449,
      "num_tokens": 29092219.0,
      "step": 4700
    },
    {
      "entropy": 0.7485624659061432,
      "epoch": 1.635624026311234,
      "grad_norm": 1.9375,
      "learning_rate": 1.6954398520670867e-06,
      "loss": 0.7138198852539063,
      "mean_token_accuracy": 0.8104638087749482,
      "num_tokens": 29249386.0,
      "step": 4725
    },
    {
      "entropy": 0.7689664399623871,
      "epoch": 1.644279037562749,
      "grad_norm": 2.0625,
      "learning_rate": 1.6181828727108772e-06,
      "loss": 0.7358683776855469,
      "mean_token_accuracy": 0.8066857266426086,
      "num_tokens": 29406880.0,
      "step": 4750
    },
    {
      "entropy": 0.7672972738742828,
      "epoch": 1.6529340488142634,
      "grad_norm": 2.40625,
      "learning_rate": 1.542572214597844e-06,
      "loss": 0.7319490051269532,
      "mean_token_accuracy": 0.8055110704898835,
      "num_tokens": 29558251.0,
      "step": 4775
    },
    {
      "entropy": 0.7510581874847412,
      "epoch": 1.6615890600657781,
      "grad_norm": 2.078125,
      "learning_rate": 1.4686227288552357e-06,
      "loss": 0.7089730072021484,
      "mean_token_accuracy": 0.811128660440445,
      "num_tokens": 29713889.0,
      "step": 4800
    },
    {
      "epoch": 1.6615890600657781,
      "eval_entropy": 0.7707234706057877,
      "eval_loss": 0.7992180585861206,
      "eval_mean_token_accuracy": 0.7935016180648178,
      "eval_num_tokens": 29713889.0,
      "eval_runtime": 16.3186,
      "eval_samples_per_second": 298.126,
      "eval_steps_per_second": 18.69,
      "step": 4800
    },
    {
      "entropy": 0.7817390990257264,
      "epoch": 1.6702440713172928,
      "grad_norm": 2.25,
      "learning_rate": 1.3963489403297992e-06,
      "loss": 0.7543077087402343,
      "mean_token_accuracy": 0.8011208462715149,
      "num_tokens": 29870315.0,
      "step": 4825
    },
    {
      "entropy": 0.7862332105636597,
      "epoch": 1.6788990825688073,
      "grad_norm": 2.125,
      "learning_rate": 1.3257650447349035e-06,
      "loss": 0.7527214050292969,
      "mean_token_accuracy": 0.8031284868717193,
      "num_tokens": 30023428.0,
      "step": 4850
    },
    {
      "entropy": 0.7391238558292389,
      "epoch": 1.687554093820322,
      "grad_norm": 2.046875,
      "learning_rate": 1.2568849058622435e-06,
      "loss": 0.7039041137695312,
      "mean_token_accuracy": 0.811707170009613,
      "num_tokens": 30181568.0,
      "step": 4875
    },
    {
      "entropy": 0.7737973690032959,
      "epoch": 1.6962091050718366,
      "grad_norm": 2.125,
      "learning_rate": 1.1897220528588005e-06,
      "loss": 0.7400728607177735,
      "mean_token_accuracy": 0.8040676033496856,
      "num_tokens": 30334041.0,
      "step": 4900
    },
    {
      "entropy": 0.7637208473682403,
      "epoch": 1.704864116323351,
      "grad_norm": 2.265625,
      "learning_rate": 1.1242896775694923e-06,
      "loss": 0.7268269348144532,
      "mean_token_accuracy": 0.8077472651004791,
      "num_tokens": 30488073.0,
      "step": 4925
    },
    {
      "entropy": 0.7567976391315461,
      "epoch": 1.7135191275748658,
      "grad_norm": 1.984375,
      "learning_rate": 1.0606006319460794e-06,
      "loss": 0.7226065826416016,
      "mean_token_accuracy": 0.8086667931079865,
      "num_tokens": 30643659.0,
      "step": 4950
    },
    {
      "entropy": 0.7528193962574005,
      "epoch": 1.7221741388263805,
      "grad_norm": 2.09375,
      "learning_rate": 9.986674255228546e-07,
      "loss": 0.7181925964355469,
      "mean_token_accuracy": 0.8110329484939576,
      "num_tokens": 30794462.0,
      "step": 4975
    },
    {
      "entropy": 0.7771501529216767,
      "epoch": 1.730829150077895,
      "grad_norm": 2.203125,
      "learning_rate": 9.385022229595631e-07,
      "loss": 0.7418533325195312,
      "mean_token_accuracy": 0.8045794427394867,
      "num_tokens": 30945283.0,
      "step": 5000
    },
    {
      "epoch": 1.730829150077895,
      "eval_entropy": 0.7709188702653667,
      "eval_loss": 0.7991884350776672,
      "eval_mean_token_accuracy": 0.7934240538565839,
      "eval_num_tokens": 30945283.0,
      "eval_runtime": 16.2662,
      "eval_samples_per_second": 299.086,
      "eval_steps_per_second": 18.751,
      "step": 5000
    },
    {
      "entropy": 0.7722408485412597,
      "epoch": 1.7394841613294099,
      "grad_norm": 2.0625,
      "learning_rate": 8.801168416520733e-07,
      "loss": 0.7394268035888671,
      "mean_token_accuracy": 0.806780993938446,
      "num_tokens": 31105352.0,
      "step": 5025
    },
    {
      "entropy": 0.7758919060230255,
      "epoch": 1.7481391725809243,
      "grad_norm": 2.0625,
      "learning_rate": 8.235227494112463e-07,
      "loss": 0.7348842620849609,
      "mean_token_accuracy": 0.8069607448577881,
      "num_tokens": 31253216.0,
      "step": 5050
    },
    {
      "entropy": 0.7573688685894012,
      "epoch": 1.756794183832439,
      "grad_norm": 2.0,
      "learning_rate": 7.687310622104982e-07,
      "loss": 0.7183684539794922,
      "mean_token_accuracy": 0.8097042238712311,
      "num_tokens": 31404175.0,
      "step": 5075
    },
    {
      "entropy": 0.7699829995632171,
      "epoch": 1.7654491950839537,
      "grad_norm": 2.21875,
      "learning_rate": 7.157525420024181e-07,
      "loss": 0.7363372802734375,
      "mean_token_accuracy": 0.8065412914752961,
      "num_tokens": 31553043.0,
      "step": 5100
    },
    {
      "entropy": 0.7569133055210113,
      "epoch": 1.7741042063354682,
      "grad_norm": 1.9609375,
      "learning_rate": 6.645975946049732e-07,
      "loss": 0.7242645263671875,
      "mean_token_accuracy": 0.8096802568435669,
      "num_tokens": 31708719.0,
      "step": 5125
    },
    {
      "entropy": 0.763712112903595,
      "epoch": 1.7827592175869829,
      "grad_norm": 2.125,
      "learning_rate": 6.152762676576352e-07,
      "loss": 0.7330033111572266,
      "mean_token_accuracy": 0.8064149105548859,
      "num_tokens": 31869942.0,
      "step": 5150
    },
    {
      "entropy": 0.7759097361564636,
      "epoch": 1.7914142288384975,
      "grad_norm": 2.203125,
      "learning_rate": 5.677982486478595e-07,
      "loss": 0.7459344482421875,
      "mean_token_accuracy": 0.8050018990039826,
      "num_tokens": 32024066.0,
      "step": 5175
    },
    {
      "entropy": 0.7608749055862427,
      "epoch": 1.800069240090012,
      "grad_norm": 2.09375,
      "learning_rate": 5.221728630083201e-07,
      "loss": 0.7280139923095703,
      "mean_token_accuracy": 0.8076766312122345,
      "num_tokens": 32180495.0,
      "step": 5200
    },
    {
      "epoch": 1.800069240090012,
      "eval_entropy": 0.7705254261610938,
      "eval_loss": 0.7991676330566406,
      "eval_mean_token_accuracy": 0.7934689746528375,
      "eval_num_tokens": 32180495.0,
      "eval_runtime": 16.2695,
      "eval_samples_per_second": 299.026,
      "eval_steps_per_second": 18.747,
      "step": 5200
    },
    {
      "entropy": 0.7587528014183045,
      "epoch": 1.8087242513415267,
      "grad_norm": 2.3125,
      "learning_rate": 4.784090722852353e-07,
      "loss": 0.7158480834960937,
      "mean_token_accuracy": 0.809622370004654,
      "num_tokens": 32338906.0,
      "step": 5225
    },
    {
      "entropy": 0.7616716659069062,
      "epoch": 1.8173792625930414,
      "grad_norm": 2.109375,
      "learning_rate": 4.365154723781895e-07,
      "loss": 0.7293673706054687,
      "mean_token_accuracy": 0.8077958106994629,
      "num_tokens": 32495631.0,
      "step": 5250
    },
    {
      "entropy": 0.737936383485794,
      "epoch": 1.8260342738445559,
      "grad_norm": 2.046875,
      "learning_rate": 3.965002918517569e-07,
      "loss": 0.7038777160644532,
      "mean_token_accuracy": 0.8131058704853058,
      "num_tokens": 32651647.0,
      "step": 5275
    },
    {
      "entropy": 0.7654093158245087,
      "epoch": 1.8346892850960708,
      "grad_norm": 2.21875,
      "learning_rate": 3.5837139031928315e-07,
      "loss": 0.7362342071533203,
      "mean_token_accuracy": 0.8066077816486359,
      "num_tokens": 32809688.0,
      "step": 5300
    },
    {
      "entropy": 0.7611633980274201,
      "epoch": 1.8433442963475852,
      "grad_norm": 2.25,
      "learning_rate": 3.221362568991293e-07,
      "loss": 0.7225106811523437,
      "mean_token_accuracy": 0.809012303352356,
      "num_tokens": 32962531.0,
      "step": 5325
    },
    {
      "entropy": 0.7749582767486572,
      "epoch": 1.8519993075991,
      "grad_norm": 2.28125,
      "learning_rate": 2.8780200874370165e-07,
      "loss": 0.7392816925048828,
      "mean_token_accuracy": 0.803524352312088,
      "num_tokens": 33122116.0,
      "step": 5350
    },
    {
      "entropy": 0.7610144579410553,
      "epoch": 1.8606543188506146,
      "grad_norm": 2.28125,
      "learning_rate": 2.5537538964151676e-07,
      "loss": 0.7229084014892578,
      "mean_token_accuracy": 0.8083787798881531,
      "num_tokens": 33273413.0,
      "step": 5375
    },
    {
      "entropy": 0.791879060268402,
      "epoch": 1.869309330102129,
      "grad_norm": 2.234375,
      "learning_rate": 2.248627686926219e-07,
      "loss": 0.7606393432617188,
      "mean_token_accuracy": 0.8025156807899475,
      "num_tokens": 33421728.0,
      "step": 5400
    },
    {
      "epoch": 1.869309330102129,
      "eval_entropy": 0.770491715630547,
      "eval_loss": 0.7991881370544434,
      "eval_mean_token_accuracy": 0.7935233919346919,
      "eval_num_tokens": 33421728.0,
      "eval_runtime": 16.2639,
      "eval_samples_per_second": 299.128,
      "eval_steps_per_second": 18.753,
      "step": 5400
    },
    {
      "entropy": 0.7707629156112671,
      "epoch": 1.8779643413536438,
      "grad_norm": 2.265625,
      "learning_rate": 1.9627013905760605e-07,
      "loss": 0.7430142974853515,
      "mean_token_accuracy": 0.8047948408126832,
      "num_tokens": 33576359.0,
      "step": 5425
    },
    {
      "entropy": 0.7577464747428894,
      "epoch": 1.8866193526051585,
      "grad_norm": 2.171875,
      "learning_rate": 1.6960311678044018e-07,
      "loss": 0.7168014526367188,
      "mean_token_accuracy": 0.8094685411453247,
      "num_tokens": 33723770.0,
      "step": 5450
    },
    {
      "entropy": 0.7747404301166534,
      "epoch": 1.895274363856673,
      "grad_norm": 2.0625,
      "learning_rate": 1.4486693968540743e-07,
      "loss": 0.7389591979980469,
      "mean_token_accuracy": 0.8045553112030029,
      "num_tokens": 33879362.0,
      "step": 5475
    },
    {
      "entropy": 0.7680295896530152,
      "epoch": 1.9039293751081876,
      "grad_norm": 2.046875,
      "learning_rate": 1.22066466348304e-07,
      "loss": 0.7351342010498046,
      "mean_token_accuracy": 0.8047092247009278,
      "num_tokens": 34035862.0,
      "step": 5500
    },
    {
      "entropy": 0.7560386550426483,
      "epoch": 1.9125843863597023,
      "grad_norm": 2.15625,
      "learning_rate": 1.0120617514214115e-07,
      "loss": 0.7233296966552735,
      "mean_token_accuracy": 0.8088453054428101,
      "num_tokens": 34192367.0,
      "step": 5525
    },
    {
      "entropy": 0.749894061088562,
      "epoch": 1.9212393976112168,
      "grad_norm": 2.234375,
      "learning_rate": 8.229016335752238e-08,
      "loss": 0.721561279296875,
      "mean_token_accuracy": 0.8098326814174652,
      "num_tokens": 34346639.0,
      "step": 5550
    },
    {
      "entropy": 0.7512237560749054,
      "epoch": 1.9298944088627317,
      "grad_norm": 2.203125,
      "learning_rate": 6.532214639786926e-08,
      "loss": 0.7133358001708985,
      "mean_token_accuracy": 0.8117510068416596,
      "num_tokens": 34501969.0,
      "step": 5575
    },
    {
      "entropy": 0.7577315938472747,
      "epoch": 1.9385494201142461,
      "grad_norm": 1.984375,
      "learning_rate": 5.03054570496575e-08,
      "loss": 0.7302328491210938,
      "mean_token_accuracy": 0.8082369923591614,
      "num_tokens": 34658561.0,
      "step": 5600
    },
    {
      "epoch": 1.9385494201142461,
      "eval_entropy": 0.7700900946484237,
      "eval_loss": 0.7992120981216431,
      "eval_mean_token_accuracy": 0.7935370582049011,
      "eval_num_tokens": 34658561.0,
      "eval_runtime": 16.284,
      "eval_samples_per_second": 298.76,
      "eval_steps_per_second": 18.73,
      "step": 5600
    },
    {
      "entropy": 0.7591597294807434,
      "epoch": 1.9472044313657608,
      "grad_norm": 2.203125,
      "learning_rate": 3.724304482781049e-08,
      "loss": 0.7255685424804688,
      "mean_token_accuracy": 0.808221675157547,
      "num_tokens": 34809418.0,
      "step": 5625
    },
    {
      "entropy": 0.7517114734649658,
      "epoch": 1.9558594426172755,
      "grad_norm": 2.078125,
      "learning_rate": 2.613747539635725e-08,
      "loss": 0.7198385620117187,
      "mean_token_accuracy": 0.8097265243530274,
      "num_tokens": 34958876.0,
      "step": 5650
    },
    {
      "entropy": 0.7789162063598633,
      "epoch": 1.96451445386879,
      "grad_norm": 2.15625,
      "learning_rate": 1.6990930064507694e-08,
      "loss": 0.7432637786865235,
      "mean_token_accuracy": 0.8060086870193481,
      "num_tokens": 35115995.0,
      "step": 5675
    },
    {
      "entropy": 0.7361849272251129,
      "epoch": 1.9731694651203047,
      "grad_norm": 2.015625,
      "learning_rate": 9.805205358197622e-09,
      "loss": 0.7050675201416016,
      "mean_token_accuracy": 0.8122973215579986,
      "num_tokens": 35274445.0,
      "step": 5700
    },
    {
      "entropy": 0.7350089454650879,
      "epoch": 1.9818244763718194,
      "grad_norm": 2.15625,
      "learning_rate": 4.5817126672331735e-09,
      "loss": 0.697210922241211,
      "mean_token_accuracy": 0.816340823173523,
      "num_tokens": 35428606.0,
      "step": 5725
    },
    {
      "entropy": 0.7550158965587616,
      "epoch": 1.9904794876233338,
      "grad_norm": 2.171875,
      "learning_rate": 1.3214779680637002e-09,
      "loss": 0.71933349609375,
      "mean_token_accuracy": 0.8101057040691376,
      "num_tokens": 35582057.0,
      "step": 5750
    },
    {
      "entropy": 0.7830293524265289,
      "epoch": 1.9991344988748485,
      "grad_norm": 2.171875,
      "learning_rate": 2.514162226741057e-11,
      "loss": 0.7570413970947265,
      "mean_token_accuracy": 0.8038502883911133,
      "num_tokens": 35740061.0,
      "step": 5775
    },
    {
      "epoch": 2.0,
      "eval_entropy": 0.7710395187628074,
      "eval_loss": 0.7991825342178345,
      "eval_mean_token_accuracy": 0.7934717604371367,
      "eval_num_tokens": 35756220.0,
      "eval_runtime": 16.2915,
      "eval_samples_per_second": 298.623,
      "eval_steps_per_second": 18.721,
      "step": 5778
    }
  ],
  "logging_steps": 25,
  "max_steps": 5778,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.4611218403146547e+17,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}