{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 25,
  "global_step": 2926,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "entropy": 1.2600676015019416,
      "epoch": 0.003417927027257968,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00015000000000000001,
      "loss": 1.3553205490112306,
      "mean_token_accuracy": 0.7003426559269428,
      "num_tokens": 50495.0,
      "step": 10
    },
    {
      "entropy": 1.0893292173743248,
      "epoch": 0.006835854054515936,
      "grad_norm": 0.53125,
      "learning_rate": 2.9289321881345254e-05,
      "loss": 1.1145251274108887,
      "mean_token_accuracy": 0.7432689998298884,
      "num_tokens": 99591.0,
      "step": 20
    },
    {
      "epoch": 0.00854481756814492,
      "eval_entropy": 1.0693371649831533,
      "eval_loss": 1.040876865386963,
      "eval_mean_token_accuracy": 0.7473227139562368,
      "eval_num_tokens": 125580.0,
      "eval_runtime": 32.1389,
      "eval_samples_per_second": 7.965,
      "eval_steps_per_second": 0.996,
      "step": 25
    },
    {
      "entropy": 1.136149488389492,
      "epoch": 0.010253781081773904,
      "grad_norm": 0.6952905654907227,
      "learning_rate": 6.59090909090909e-05,
      "loss": 1.118192481994629,
      "mean_token_accuracy": 0.7425699919462204,
      "num_tokens": 23478.0,
      "step": 30
    },
    {
      "entropy": 1.143788156658411,
      "epoch": 0.013671708109031872,
      "grad_norm": 0.708786129951477,
      "learning_rate": 8.863636363636364e-05,
      "loss": 1.1115641593933105,
      "mean_token_accuracy": 0.7380502671003342,
      "num_tokens": 74079.0,
      "step": 40
    },
    {
      "entropy": 1.031112465262413,
      "epoch": 0.01708963513628984,
      "grad_norm": 0.551914393901825,
      "learning_rate": 0.00011136363636363636,
      "loss": 1.0043062210083007,
      "mean_token_accuracy": 0.7607270315289497,
      "num_tokens": 121523.0,
      "step": 50
    },
    {
      "epoch": 0.01708963513628984,
      "eval_entropy": 0.9792779758572578,
      "eval_loss": 0.9474607706069946,
      "eval_mean_token_accuracy": 0.7642499133944511,
      "eval_num_tokens": 121523.0,
      "eval_runtime": 32.42,
      "eval_samples_per_second": 7.896,
      "eval_steps_per_second": 0.987,
      "step": 50
    },
    {
      "entropy": 1.0256493411958219,
      "epoch": 0.020507562163547807,
      "grad_norm": 0.6798339486122131,
      "learning_rate": 0.0001340909090909091,
      "loss": 0.9909649848937988,
      "mean_token_accuracy": 0.7597807608544827,
      "num_tokens": 169298.0,
      "step": 60
    },
    {
      "entropy": 1.0200331829488278,
      "epoch": 0.023925489190805775,
      "grad_norm": 0.7172015309333801,
      "learning_rate": 0.00015681818181818182,
      "loss": 0.9925822257995606,
      "mean_token_accuracy": 0.760905759036541,
      "num_tokens": 215812.0,
      "step": 70
    },
    {
      "epoch": 0.02563445270443476,
      "eval_entropy": 0.9389796517789364,
      "eval_loss": 0.9137760996818542,
      "eval_mean_token_accuracy": 0.7696583028882742,
      "eval_num_tokens": 238841.0,
      "eval_runtime": 32.4362,
      "eval_samples_per_second": 7.892,
      "eval_steps_per_second": 0.987,
      "step": 75
    },
    {
      "entropy": 1.0091111920773983,
      "epoch": 0.027343416218063743,
      "grad_norm": 0.5625494122505188,
      "learning_rate": 0.00017954545454545456,
      "loss": 0.9961751937866211,
      "mean_token_accuracy": 0.7597843900322914,
      "num_tokens": 262446.0,
      "step": 80
    },
    {
      "entropy": 1.0306036248803139,
      "epoch": 0.03076134324532171,
      "grad_norm": 0.6916620135307312,
      "learning_rate": 0.00019999993873041723,
      "loss": 1.008632755279541,
      "mean_token_accuracy": 0.7560263566672802,
      "num_tokens": 307540.0,
      "step": 90
    },
    {
      "entropy": 1.0128997907042503,
      "epoch": 0.03417927027257968,
      "grad_norm": 0.5265458226203918,
      "learning_rate": 0.00019999258647132646,
      "loss": 0.9914002418518066,
      "mean_token_accuracy": 0.7570689752697944,
      "num_tokens": 354777.0,
      "step": 100
    },
    {
      "epoch": 0.03417927027257968,
      "eval_entropy": 0.9012311659753323,
      "eval_loss": 0.8966313600540161,
      "eval_mean_token_accuracy": 0.7727285120636225,
      "eval_num_tokens": 354777.0,
      "eval_runtime": 32.4154,
      "eval_samples_per_second": 7.897,
      "eval_steps_per_second": 0.987,
      "step": 100
    },
    {
      "entropy": 0.9638666190207005,
      "epoch": 0.03759719729983765,
      "grad_norm": 0.5038418769836426,
      "learning_rate": 0.00019997298132799407,
      "loss": 0.9364572525024414,
      "mean_token_accuracy": 0.7678967766463757,
      "num_tokens": 405355.0,
      "step": 110
    },
    {
      "entropy": 0.9465805731713772,
      "epoch": 0.041015124327095615,
      "grad_norm": 0.4934922456741333,
      "learning_rate": 0.00019994112570279366,
      "loss": 0.9496530532836914,
      "mean_token_accuracy": 0.7676978453993797,
      "num_tokens": 453525.0,
      "step": 120
    },
    {
      "epoch": 0.0427240878407246,
      "eval_entropy": 0.8910342156887054,
      "eval_loss": 0.8766279220581055,
      "eval_mean_token_accuracy": 0.7756589185446501,
      "eval_num_tokens": 478081.0,
      "eval_runtime": 32.4243,
      "eval_samples_per_second": 7.895,
      "eval_steps_per_second": 0.987,
      "step": 125
    },
    {
      "entropy": 1.0112694174051284,
      "epoch": 0.04443305135435358,
      "grad_norm": 0.5341691374778748,
      "learning_rate": 0.00019989702349924754,
      "loss": 0.9902518272399903,
      "mean_token_accuracy": 0.7600088074803353,
      "num_tokens": 500728.0,
      "step": 130
    },
    {
      "entropy": 1.0307338863611222,
      "epoch": 0.04785097838161155,
      "grad_norm": 0.4941849112510681,
      "learning_rate": 0.00019984068012154823,
      "loss": 0.9965475082397461,
      "mean_token_accuracy": 0.7593892373144626,
      "num_tokens": 546115.0,
      "step": 140
    },
    {
      "entropy": 0.9555066972970963,
      "epoch": 0.05126890540886952,
      "grad_norm": 0.46804171800613403,
      "learning_rate": 0.00019977210247389647,
      "loss": 0.938600730895996,
      "mean_token_accuracy": 0.7718380980193615,
      "num_tokens": 591332.0,
      "step": 150
    },
    {
      "epoch": 0.05126890540886952,
      "eval_entropy": 0.8860863335430622,
      "eval_loss": 0.8679022789001465,
      "eval_mean_token_accuracy": 0.7788128759711981,
      "eval_num_tokens": 591332.0,
      "eval_runtime": 32.463,
      "eval_samples_per_second": 7.886,
      "eval_steps_per_second": 0.986,
      "step": 150
    },
    {
      "entropy": 0.9176225580275059,
      "epoch": 0.054686832436127486,
      "grad_norm": 0.5647802352905273,
      "learning_rate": 0.00019969129895965502,
      "loss": 0.9094284057617188,
      "mean_token_accuracy": 0.7718568064272404,
      "num_tokens": 638922.0,
      "step": 160
    },
    {
      "entropy": 0.9221074335277081,
      "epoch": 0.058104759463385454,
      "grad_norm": 0.4803485572338104,
      "learning_rate": 0.00019959827948031898,
      "loss": 0.895450496673584,
      "mean_token_accuracy": 0.7760300859808922,
      "num_tokens": 687534.0,
      "step": 170
    },
    {
      "epoch": 0.05981372297701444,
      "eval_entropy": 0.8816897980868816,
      "eval_loss": 0.8514255881309509,
      "eval_mean_token_accuracy": 0.7821575775742531,
      "eval_num_tokens": 712836.0,
      "eval_runtime": 32.5266,
      "eval_samples_per_second": 7.87,
      "eval_steps_per_second": 0.984,
      "step": 175
    },
    {
      "entropy": 0.9514009088277817,
      "epoch": 0.06152268649064342,
      "grad_norm": 0.5159069299697876,
      "learning_rate": 0.00019949305543430254,
      "loss": 0.9298910140991211,
      "mean_token_accuracy": 0.7662424340844154,
      "num_tokens": 737751.0,
      "step": 180
    },
    {
      "entropy": 0.936088802292943,
      "epoch": 0.0649406135179014,
      "grad_norm": 0.5905720591545105,
      "learning_rate": 0.0001993756397155421,
      "loss": 0.9271375656127929,
      "mean_token_accuracy": 0.7713530980050564,
      "num_tokens": 783903.0,
      "step": 190
    },
    {
      "entropy": 0.8810944482684135,
      "epoch": 0.06835854054515936,
      "grad_norm": 0.5631661415100098,
      "learning_rate": 0.0001992460467119164,
      "loss": 0.8553449630737304,
      "mean_token_accuracy": 0.7837967626750469,
      "num_tokens": 835058.0,
      "step": 200
    },
    {
      "epoch": 0.06835854054515936,
      "eval_entropy": 0.8469126336276531,
      "eval_loss": 0.843980073928833,
      "eval_mean_token_accuracy": 0.7836501188576221,
      "eval_num_tokens": 835058.0,
      "eval_runtime": 32.5036,
      "eval_samples_per_second": 7.876,
      "eval_steps_per_second": 0.985,
      "step": 200
    },
    {
      "entropy": 0.9015118815004826,
      "epoch": 0.07177646757241733,
      "grad_norm": 0.5022907257080078,
      "learning_rate": 0.00019910429230348347,
      "loss": 0.8947334289550781,
      "mean_token_accuracy": 0.775901124626398,
      "num_tokens": 881548.0,
      "step": 210
    },
    {
      "entropy": 0.9544241480529309,
      "epoch": 0.0751943945996753,
      "grad_norm": 0.6783669590950012,
      "learning_rate": 0.00019895039386053464,
      "loss": 0.939936351776123,
      "mean_token_accuracy": 0.7679277427494526,
      "num_tokens": 930380.0,
      "step": 220
    },
    {
      "epoch": 0.07690335811330429,
      "eval_entropy": 0.86623908393085,
      "eval_loss": 0.8365424871444702,
      "eval_mean_token_accuracy": 0.7847181409597397,
      "eval_num_tokens": 953371.0,
      "eval_runtime": 32.6302,
      "eval_samples_per_second": 7.845,
      "eval_steps_per_second": 0.981,
      "step": 225
    },
    {
      "entropy": 0.9503094263374805,
      "epoch": 0.07861232162693327,
      "grad_norm": 0.5332171320915222,
      "learning_rate": 0.00019878437024146604,
      "loss": 0.9248787879943847,
      "mean_token_accuracy": 0.7707160159945488,
      "num_tokens": 977205.0,
      "step": 230
    },
    {
      "entropy": 0.9480014778673649,
      "epoch": 0.08203024865419123,
      "grad_norm": 0.636835515499115,
      "learning_rate": 0.0001986062417904677,
      "loss": 0.9326526641845703,
      "mean_token_accuracy": 0.7707653798162937,
      "num_tokens": 1023070.0,
      "step": 240
    },
    {
      "entropy": 0.9017960362136364,
      "epoch": 0.0854481756814492,
      "grad_norm": 0.6051347851753235,
      "learning_rate": 0.00019841603033503077,
      "loss": 0.890564250946045,
      "mean_token_accuracy": 0.7787797845900059,
      "num_tokens": 1068509.0,
      "step": 250
    },
    {
      "epoch": 0.0854481756814492,
      "eval_entropy": 0.8300079107284546,
      "eval_loss": 0.8322861194610596,
      "eval_mean_token_accuracy": 0.7865318432450294,
      "eval_num_tokens": 1068509.0,
      "eval_runtime": 32.5945,
      "eval_samples_per_second": 7.854,
      "eval_steps_per_second": 0.982,
      "step": 250
    },
    {
      "entropy": 0.9348328456282615,
      "epoch": 0.08886610270870716,
      "grad_norm": 0.5464855432510376,
      "learning_rate": 0.00019821375918327268,
      "loss": 0.9178380966186523,
      "mean_token_accuracy": 0.7756208308041096,
      "num_tokens": 1114975.0,
      "step": 260
    },
    {
      "entropy": 0.8812198378145695,
      "epoch": 0.09228402973596514,
      "grad_norm": 0.4713146686553955,
      "learning_rate": 0.0001979994531210809,
      "loss": 0.8591255187988281,
      "mean_token_accuracy": 0.7836810939013958,
      "num_tokens": 1162702.0,
      "step": 270
    },
    {
      "epoch": 0.09399299324959412,
      "eval_entropy": 0.8178850021213293,
      "eval_loss": 0.8219032287597656,
      "eval_mean_token_accuracy": 0.7888985127210617,
      "eval_num_tokens": 1187390.0,
      "eval_runtime": 32.6097,
      "eval_samples_per_second": 7.85,
      "eval_steps_per_second": 0.981,
      "step": 275
    },
    {
      "entropy": 0.7962687697261572,
      "epoch": 0.0957019567632231,
      "grad_norm": 0.45879796147346497,
      "learning_rate": 0.00019777313840907613,
      "loss": 0.778873348236084,
      "mean_token_accuracy": 0.7989643529057503,
      "num_tokens": 1212186.0,
      "step": 280
    },
    {
      "entropy": 0.9441293470561505,
      "epoch": 0.09911988379048108,
      "grad_norm": 0.5591995120048523,
      "learning_rate": 0.0001975348427793939,
      "loss": 0.931028175354004,
      "mean_token_accuracy": 0.7705901082605123,
      "num_tokens": 1261641.0,
      "step": 290
    },
    {
      "entropy": 0.9185084633529186,
      "epoch": 0.10253781081773904,
      "grad_norm": 0.5129427909851074,
      "learning_rate": 0.0001972845954322866,
      "loss": 0.9040238380432128,
      "mean_token_accuracy": 0.7761635169386863,
      "num_tokens": 1309392.0,
      "step": 300
    },
    {
      "epoch": 0.10253781081773904,
      "eval_entropy": 0.8404651805758476,
      "eval_loss": 0.8196567296981812,
      "eval_mean_token_accuracy": 0.7885849270969629,
      "eval_num_tokens": 1309392.0,
      "eval_runtime": 32.6325,
      "eval_samples_per_second": 7.845,
      "eval_steps_per_second": 0.981,
      "step": 300
    },
    {
      "entropy": 0.9515121657401323,
      "epoch": 0.10595573784499701,
      "grad_norm": 0.6107913851737976,
      "learning_rate": 0.00019702242703254546,
      "loss": 0.9385555267333985,
      "mean_token_accuracy": 0.7681696146726609,
      "num_tokens": 1359229.0,
      "step": 310
    },
    {
      "entropy": 0.914220716059208,
      "epoch": 0.10937366487225497,
      "grad_norm": 0.5568957328796387,
      "learning_rate": 0.00019674836970574254,
      "loss": 0.8987955093383789,
      "mean_token_accuracy": 0.7780232340097427,
      "num_tokens": 1404361.0,
      "step": 320
    },
    {
      "epoch": 0.11108262838588397,
      "eval_entropy": 0.8228043504059315,
      "eval_loss": 0.8159385919570923,
      "eval_mean_token_accuracy": 0.7894312981516123,
      "eval_num_tokens": 1424972.0,
      "eval_runtime": 32.6314,
      "eval_samples_per_second": 7.845,
      "eval_steps_per_second": 0.981,
      "step": 325
    },
    {
      "entropy": 1.0007071375846863,
      "epoch": 0.11279159189951295,
      "grad_norm": 0.6453739404678345,
      "learning_rate": 0.00019646245703429456,
      "loss": 0.9830230712890625,
      "mean_token_accuracy": 0.7651846624910832,
      "num_tokens": 1447390.0,
      "step": 330
    },
    {
      "entropy": 0.903276839107275,
      "epoch": 0.11620951892677091,
      "grad_norm": 0.584945023059845,
      "learning_rate": 0.00019616472405334747,
      "loss": 0.8930956840515136,
      "mean_token_accuracy": 0.7818102724850178,
      "num_tokens": 1495240.0,
      "step": 340
    },
    {
      "entropy": 0.8742344193160534,
      "epoch": 0.11962744595402888,
      "grad_norm": 0.530391275882721,
      "learning_rate": 0.00019585520724648353,
      "loss": 0.8604484558105469,
      "mean_token_accuracy": 0.7872159354388714,
      "num_tokens": 1542650.0,
      "step": 350
    },
    {
      "epoch": 0.11962744595402888,
      "eval_entropy": 0.8151113502681255,
      "eval_loss": 0.8080641031265259,
      "eval_mean_token_accuracy": 0.7918843850493431,
      "eval_num_tokens": 1542650.0,
      "eval_runtime": 32.6195,
      "eval_samples_per_second": 7.848,
      "eval_steps_per_second": 0.981,
      "step": 350
    },
    {
      "entropy": 0.9260273765772581,
      "epoch": 0.12304537298128684,
      "grad_norm": 0.5639971494674683,
      "learning_rate": 0.00019553394454125057,
      "loss": 0.9099954605102539,
      "mean_token_accuracy": 0.7739293292164803,
      "num_tokens": 1590267.0,
      "step": 360
    },
    {
      "entropy": 0.8605987407267094,
      "epoch": 0.1264633000085448,
      "grad_norm": 0.5054117441177368,
      "learning_rate": 0.00019520097530451445,
      "loss": 0.8408944129943847,
      "mean_token_accuracy": 0.7852211132645607,
      "num_tokens": 1643921.0,
      "step": 370
    },
    {
      "epoch": 0.1281722635221738,
      "eval_entropy": 0.7877141386270523,
      "eval_loss": 0.803894579410553,
      "eval_mean_token_accuracy": 0.7931277714669704,
      "eval_num_tokens": 1668079.0,
      "eval_runtime": 32.6553,
      "eval_samples_per_second": 7.839,
      "eval_steps_per_second": 0.98,
      "step": 375
    },
    {
      "entropy": 0.8402018003165722,
      "epoch": 0.1298812270358028,
      "grad_norm": 0.7080391049385071,
      "learning_rate": 0.00019485634033763506,
      "loss": 0.8387722969055176,
      "mean_token_accuracy": 0.7905740939080715,
      "num_tokens": 1693171.0,
      "step": 380
    },
    {
      "entropy": 0.8122920129448176,
      "epoch": 0.13329915406306075,
      "grad_norm": 0.4884185791015625,
      "learning_rate": 0.00019450008187146684,
      "loss": 0.7884160041809082,
      "mean_token_accuracy": 0.7995197549462318,
      "num_tokens": 1741842.0,
      "step": 390
    },
    {
      "entropy": 0.8610376358032227,
      "epoch": 0.13671708109031872,
      "grad_norm": 0.6019439697265625,
      "learning_rate": 0.0001941322435611836,
      "loss": 0.8544740676879883,
      "mean_token_accuracy": 0.7863288387656212,
      "num_tokens": 1789054.0,
      "step": 400
    },
    {
      "epoch": 0.13671708109031872,
      "eval_entropy": 0.8057211544364691,
      "eval_loss": 0.8004281520843506,
      "eval_mean_token_accuracy": 0.7926788628101349,
      "eval_num_tokens": 1789054.0,
      "eval_runtime": 32.6897,
      "eval_samples_per_second": 7.831,
      "eval_steps_per_second": 0.979,
      "step": 400
    },
    {
      "entropy": 0.9050097674131393,
      "epoch": 0.14013500811757668,
      "grad_norm": 0.701719343662262,
      "learning_rate": 0.00019375287048092928,
      "loss": 0.8894415855407715,
      "mean_token_accuracy": 0.7790660828351974,
      "num_tokens": 1836694.0,
      "step": 410
    },
    {
      "entropy": 0.8502700824290514,
      "epoch": 0.14355293514483466,
      "grad_norm": 0.5158707499504089,
      "learning_rate": 0.00019336200911829462,
      "loss": 0.822513484954834,
      "mean_token_accuracy": 0.7917489908635617,
      "num_tokens": 1884795.0,
      "step": 420
    },
    {
      "epoch": 0.14526189865846365,
      "eval_entropy": 0.7786619961261749,
      "eval_loss": 0.7941492795944214,
      "eval_mean_token_accuracy": 0.795093884691596,
      "eval_num_tokens": 1906817.0,
      "eval_runtime": 32.6621,
      "eval_samples_per_second": 7.838,
      "eval_steps_per_second": 0.98,
      "step": 425
    },
    {
      "entropy": 0.8885743632912636,
      "epoch": 0.14697086217209263,
      "grad_norm": 0.5763893723487854,
      "learning_rate": 0.00019295970736862064,
      "loss": 0.8765960693359375,
      "mean_token_accuracy": 0.7833322003483772,
      "num_tokens": 1931776.0,
      "step": 430
    },
    {
      "entropy": 0.8891543313860893,
      "epoch": 0.1503887891993506,
      "grad_norm": 0.5412898063659668,
      "learning_rate": 0.0001925460145291297,
      "loss": 0.8739751815795899,
      "mean_token_accuracy": 0.7810932233929634,
      "num_tokens": 1981794.0,
      "step": 440
    },
    {
      "entropy": 0.8810034863650799,
      "epoch": 0.15380671622660858,
      "grad_norm": 0.4799567759037018,
      "learning_rate": 0.00019212098129288473,
      "loss": 0.8641800880432129,
      "mean_token_accuracy": 0.785812397301197,
      "num_tokens": 2032831.0,
      "step": 450
    },
    {
      "epoch": 0.15380671622660858,
      "eval_entropy": 0.8059570249170065,
      "eval_loss": 0.7902603149414062,
      "eval_mean_token_accuracy": 0.7959560509771109,
      "eval_num_tokens": 2032831.0,
      "eval_runtime": 32.6415,
      "eval_samples_per_second": 7.843,
      "eval_steps_per_second": 0.98,
      "step": 450
    },
    {
      "entropy": 0.8998822886496782,
      "epoch": 0.15722464325386654,
      "grad_norm": 0.6191452145576477,
      "learning_rate": 0.0001916846597425773,
      "loss": 0.878938865661621,
      "mean_token_accuracy": 0.7810092575848102,
      "num_tokens": 2082699.0,
      "step": 460
    },
    {
      "entropy": 0.9274133644998074,
      "epoch": 0.1606425702811245,
      "grad_norm": 0.5060327649116516,
      "learning_rate": 0.00019123710334414552,
      "loss": 0.9187069892883301,
      "mean_token_accuracy": 0.7734548069536686,
      "num_tokens": 2129343.0,
      "step": 470
    },
    {
      "epoch": 0.16235153379475348,
      "eval_entropy": 0.7861206252127886,
      "eval_loss": 0.7843623161315918,
      "eval_mean_token_accuracy": 0.7965397275984287,
      "eval_num_tokens": 2154727.0,
      "eval_runtime": 32.6107,
      "eval_samples_per_second": 7.85,
      "eval_steps_per_second": 0.981,
      "step": 475
    },
    {
      "entropy": 0.8581852417439222,
      "epoch": 0.16406049730838246,
      "grad_norm": 0.5885692834854126,
      "learning_rate": 0.0001907783669402227,
      "loss": 0.8446619033813476,
      "mean_token_accuracy": 0.7875718802213669,
      "num_tokens": 2178667.0,
      "step": 480
    },
    {
      "entropy": 0.9119165766984224,
      "epoch": 0.16747842433564045,
      "grad_norm": 0.5744176506996155,
      "learning_rate": 0.00019030850674341666,
      "loss": 0.9125348091125488,
      "mean_token_accuracy": 0.7730302430689335,
      "num_tokens": 2228020.0,
      "step": 490
    },
    {
      "entropy": 0.946759483218193,
      "epoch": 0.1708963513628984,
      "grad_norm": 0.48804807662963867,
      "learning_rate": 0.00018982758032942186,
      "loss": 0.9263985633850098,
      "mean_token_accuracy": 0.77376104965806,
      "num_tokens": 2274822.0,
      "step": 500
    },
    {
      "epoch": 0.1708963513628984,
      "eval_entropy": 0.7850293070077896,
      "eval_loss": 0.7818621397018433,
      "eval_mean_token_accuracy": 0.7967414036393166,
      "eval_num_tokens": 2274822.0,
      "eval_runtime": 32.6612,
      "eval_samples_per_second": 7.838,
      "eval_steps_per_second": 0.98,
      "step": 500
    },
    {
      "entropy": 0.850209466740489,
      "epoch": 0.17431427839015637,
      "grad_norm": 0.5423233509063721,
      "learning_rate": 0.00018933564662996393,
      "loss": 0.8323702812194824,
      "mean_token_accuracy": 0.790254071354866,
      "num_tokens": 2324238.0,
      "step": 510
    },
    {
      "entropy": 0.8254084564745426,
      "epoch": 0.17773220541741433,
      "grad_norm": 0.6329611539840698,
      "learning_rate": 0.0001888327659255786,
      "loss": 0.8157858848571777,
      "mean_token_accuracy": 0.7952801741659641,
      "num_tokens": 2369938.0,
      "step": 520
    },
    {
      "epoch": 0.1794411689310433,
      "eval_entropy": 0.7873559519648552,
      "eval_loss": 0.7806037664413452,
      "eval_mean_token_accuracy": 0.7967518828809261,
      "eval_num_tokens": 2396938.0,
      "eval_runtime": 32.5735,
      "eval_samples_per_second": 7.859,
      "eval_steps_per_second": 0.982,
      "step": 525
    },
    {
      "entropy": 0.7927372690290213,
      "epoch": 0.18115013244467232,
      "grad_norm": 0.5249568819999695,
      "learning_rate": 0.00018831899983822475,
      "loss": 0.7778078556060791,
      "mean_token_accuracy": 0.8018042631447315,
      "num_tokens": 2418510.0,
      "step": 530
    },
    {
      "entropy": 0.8573413282632828,
      "epoch": 0.18456805947193028,
      "grad_norm": 0.597499430179596,
      "learning_rate": 0.00018779441132373362,
      "loss": 0.8551907539367676,
      "mean_token_accuracy": 0.7869668185710907,
      "num_tokens": 2464421.0,
      "step": 540
    },
    {
      "entropy": 0.9104309018701315,
      "epoch": 0.18798598649918824,
      "grad_norm": 0.5782085061073303,
      "learning_rate": 0.0001872590646640942,
      "loss": 0.8853252410888672,
      "mean_token_accuracy": 0.782588554173708,
      "num_tokens": 2506949.0,
      "step": 550
    },
    {
      "epoch": 0.18798598649918824,
      "eval_entropy": 0.7909549381583929,
      "eval_loss": 0.7767490148544312,
      "eval_mean_token_accuracy": 0.7989067882299423,
      "eval_num_tokens": 2506949.0,
      "eval_runtime": 32.5781,
      "eval_samples_per_second": 7.858,
      "eval_steps_per_second": 0.982,
      "step": 550
    },
    {
      "entropy": 0.9016286168247462,
      "epoch": 0.1914039135264462,
      "grad_norm": 0.6918966770172119,
      "learning_rate": 0.0001867130254595763,
      "loss": 0.8875547409057617,
      "mean_token_accuracy": 0.7801647335290909,
      "num_tokens": 2551893.0,
      "step": 560
    },
    {
      "entropy": 0.7990992326289416,
      "epoch": 0.1948218405537042,
      "grad_norm": 0.44646406173706055,
      "learning_rate": 0.00018615636062069193,
      "loss": 0.7881239891052246,
      "mean_token_accuracy": 0.8021551303565502,
      "num_tokens": 2601200.0,
      "step": 570
    },
    {
      "epoch": 0.19653080406733317,
      "eval_entropy": 0.7736214082688093,
      "eval_loss": 0.7739249467849731,
      "eval_mean_token_accuracy": 0.7984153404831886,
      "eval_num_tokens": 2625970.0,
      "eval_runtime": 32.6428,
      "eval_samples_per_second": 7.842,
      "eval_steps_per_second": 0.98,
      "step": 575
    },
    {
      "entropy": 0.8380344811826944,
      "epoch": 0.19823976758096215,
      "grad_norm": 0.5394240617752075,
      "learning_rate": 0.0001855891383599964,
      "loss": 0.8400195121765137,
      "mean_token_accuracy": 0.7881024792790413,
      "num_tokens": 2648956.0,
      "step": 580
    },
    {
      "entropy": 0.8434151515364647,
      "epoch": 0.2016576946082201,
      "grad_norm": 0.5002357959747314,
      "learning_rate": 0.00018501142818372964,
      "loss": 0.8170535087585449,
      "mean_token_accuracy": 0.794566561281681,
      "num_tokens": 2694536.0,
      "step": 590
    },
    {
      "entropy": 0.7922864239662886,
      "epoch": 0.20507562163547807,
      "grad_norm": 0.5080456137657166,
      "learning_rate": 0.00018442330088329882,
      "loss": 0.7768840312957763,
      "mean_token_accuracy": 0.7996326476335526,
      "num_tokens": 2745360.0,
      "step": 600
    },
    {
      "epoch": 0.20507562163547807,
      "eval_entropy": 0.7685494627803564,
      "eval_loss": 0.7737076282501221,
      "eval_mean_token_accuracy": 0.7998364306986332,
      "eval_num_tokens": 2745360.0,
      "eval_runtime": 32.6306,
      "eval_samples_per_second": 7.845,
      "eval_steps_per_second": 0.981,
      "step": 600
    },
    {
      "entropy": 0.8512289695441723,
      "epoch": 0.20849354866273606,
      "grad_norm": 0.8412303328514099,
      "learning_rate": 0.00018382482852660414,
      "loss": 0.8491168022155762,
      "mean_token_accuracy": 0.7873495660722256,
      "num_tokens": 2790839.0,
      "step": 610
    },
    {
      "entropy": 0.8139516353607178,
      "epoch": 0.21191147568999402,
      "grad_norm": 0.5720385313034058,
      "learning_rate": 0.00018321608444920738,
      "loss": 0.7931262969970703,
      "mean_token_accuracy": 0.7983397327363491,
      "num_tokens": 2843698.0,
      "step": 620
    },
    {
      "epoch": 0.213620439203623,
      "eval_entropy": 0.7624373491853476,
      "eval_loss": 0.7696622610092163,
      "eval_mean_token_accuracy": 0.8000017125159502,
      "eval_num_tokens": 2870945.0,
      "eval_runtime": 32.5975,
      "eval_samples_per_second": 7.853,
      "eval_steps_per_second": 0.982,
      "step": 625
    },
    {
      "entropy": 0.8417224410921336,
      "epoch": 0.21532940271725198,
      "grad_norm": 0.5085984468460083,
      "learning_rate": 0.0001825971432453457,
      "loss": 0.8314865112304688,
      "mean_token_accuracy": 0.7912811681628227,
      "num_tokens": 2894559.0,
      "step": 630
    },
    {
      "entropy": 0.8762644857168198,
      "epoch": 0.21874732974450994,
      "grad_norm": 0.5597890019416809,
      "learning_rate": 0.00018196808075879114,
      "loss": 0.8614648818969727,
      "mean_token_accuracy": 0.7858939148485661,
      "num_tokens": 2944113.0,
      "step": 640
    },
    {
      "entropy": 0.8845491908490658,
      "epoch": 0.22216525677176793,
      "grad_norm": 0.5694191455841064,
      "learning_rate": 0.00018132897407355657,
      "loss": 0.8676918029785157,
      "mean_token_accuracy": 0.7828861311078071,
      "num_tokens": 2992188.0,
      "step": 650
    },
    {
      "epoch": 0.22216525677176793,
      "eval_entropy": 0.7651609890162945,
      "eval_loss": 0.7629461884498596,
      "eval_mean_token_accuracy": 0.8016721252351999,
      "eval_num_tokens": 2992188.0,
      "eval_runtime": 32.6395,
      "eval_samples_per_second": 7.843,
      "eval_steps_per_second": 0.98,
      "step": 650
    },
    {
      "entropy": 0.9146428674459457,
      "epoch": 0.2255831837990259,
      "grad_norm": 0.5276857614517212,
      "learning_rate": 0.00018067990150445016,
      "loss": 0.8943714141845703,
      "mean_token_accuracy": 0.7790229134261608,
      "num_tokens": 3039298.0,
      "step": 660
    },
    {
      "entropy": 0.7951211579144001,
      "epoch": 0.22900111082628385,
      "grad_norm": 0.5947771668434143,
      "learning_rate": 0.0001800209425874789,
      "loss": 0.7826883792877197,
      "mean_token_accuracy": 0.8007356464862824,
      "num_tokens": 3088419.0,
      "step": 670
    },
    {
      "epoch": 0.23071007433991284,
      "eval_entropy": 0.7665124572813511,
      "eval_loss": 0.7581461668014526,
      "eval_mean_token_accuracy": 0.8026615120470524,
      "eval_num_tokens": 3113387.0,
      "eval_runtime": 32.6339,
      "eval_samples_per_second": 7.845,
      "eval_steps_per_second": 0.981,
      "step": 675
    },
    {
      "entropy": 0.8008962716907263,
      "epoch": 0.23241903785354182,
      "grad_norm": 0.5073366761207581,
      "learning_rate": 0.0001793521780701024,
      "loss": 0.7925901889801026,
      "mean_token_accuracy": 0.7972975365817547,
      "num_tokens": 3138173.0,
      "step": 680
    },
    {
      "entropy": 0.8295211162418127,
      "epoch": 0.2358369648807998,
      "grad_norm": 0.6248682737350464,
      "learning_rate": 0.000178673689901338,
      "loss": 0.8191262245178222,
      "mean_token_accuracy": 0.7940905131399632,
      "num_tokens": 3182437.0,
      "step": 690
    },
    {
      "entropy": 0.8073327772319316,
      "epoch": 0.23925489190805777,
      "grad_norm": 0.582673966884613,
      "learning_rate": 0.00017798556122171938,
      "loss": 0.7840201377868652,
      "mean_token_accuracy": 0.7995604500174522,
      "num_tokens": 3229286.0,
      "step": 700
    },
    {
      "epoch": 0.23925489190805777,
      "eval_entropy": 0.7552605923265219,
      "eval_loss": 0.7581686973571777,
      "eval_mean_token_accuracy": 0.8022663723677397,
      "eval_num_tokens": 3229286.0,
      "eval_runtime": 32.7242,
      "eval_samples_per_second": 7.823,
      "eval_steps_per_second": 0.978,
      "step": 700
    },
    {
      "entropy": 0.7811179537326097,
      "epoch": 0.24267281893531573,
      "grad_norm": 0.5043304562568665,
      "learning_rate": 0.00017728787635310826,
      "loss": 0.7765492916107177,
      "mean_token_accuracy": 0.8002728261053562,
      "num_tokens": 3278460.0,
      "step": 710
    },
    {
      "entropy": 0.8374413985759019,
      "epoch": 0.2460907459625737,
      "grad_norm": 0.5113959312438965,
      "learning_rate": 0.000176580720788362,
      "loss": 0.8117224693298339,
      "mean_token_accuracy": 0.7954827964305877,
      "num_tokens": 3323710.0,
      "step": 720
    },
    {
      "epoch": 0.2477997094762027,
      "eval_entropy": 0.7489179540425539,
      "eval_loss": 0.7608813643455505,
      "eval_mean_token_accuracy": 0.8018700126558542,
      "eval_num_tokens": 3348967.0,
      "eval_runtime": 32.6984,
      "eval_samples_per_second": 7.829,
      "eval_steps_per_second": 0.979,
      "step": 725
    },
    {
      "entropy": 0.8640857309103012,
      "epoch": 0.24950867298983168,
      "grad_norm": 0.5439287424087524,
      "learning_rate": 0.00017586418118085742,
      "loss": 0.861956787109375,
      "mean_token_accuracy": 0.782987778633833,
      "num_tokens": 3374483.0,
      "step": 730
    },
    {
      "entropy": 0.8198690854012967,
      "epoch": 0.2529266000170896,
      "grad_norm": 0.5498844981193542,
      "learning_rate": 0.00017513834533387256,
      "loss": 0.7932277679443359,
      "mean_token_accuracy": 0.7930847354233265,
      "num_tokens": 3425284.0,
      "step": 740
    },
    {
      "entropy": 0.8567254289984703,
      "epoch": 0.2563445270443476,
      "grad_norm": 0.575046718120575,
      "learning_rate": 0.00017440330218982717,
      "loss": 0.844538688659668,
      "mean_token_accuracy": 0.7879870742559433,
      "num_tokens": 3473198.0,
      "step": 750
    },
    {
      "epoch": 0.2563445270443476,
      "eval_entropy": 0.7522654607892036,
      "eval_loss": 0.7591322064399719,
      "eval_mean_token_accuracy": 0.8022094331681728,
      "eval_num_tokens": 3473198.0,
      "eval_runtime": 32.5916,
      "eval_samples_per_second": 7.855,
      "eval_steps_per_second": 0.982,
      "step": 750
    },
    {
      "entropy": 0.8989825204014779,
      "epoch": 0.2597624540716056,
      "grad_norm": 0.46154603362083435,
      "learning_rate": 0.0001736591418193844,
      "loss": 0.8899558067321778,
      "mean_token_accuracy": 0.7821005254983902,
      "num_tokens": 3522189.0,
      "step": 760
    },
    {
      "entropy": 0.8200666073709726,
      "epoch": 0.26318038109886355,
      "grad_norm": 0.6104792952537537,
      "learning_rate": 0.0001729059554104131,
      "loss": 0.7961187839508057,
      "mean_token_accuracy": 0.7963723540306091,
      "num_tokens": 3572845.0,
      "step": 770
    },
    {
      "epoch": 0.2648893446124925,
      "eval_entropy": 0.7473378330469131,
      "eval_loss": 0.7550503015518188,
      "eval_mean_token_accuracy": 0.8024975471198559,
      "eval_num_tokens": 3597729.0,
      "eval_runtime": 32.6181,
      "eval_samples_per_second": 7.848,
      "eval_steps_per_second": 0.981,
      "step": 775
    },
    {
      "entropy": 0.8045445740222931,
      "epoch": 0.2665983081261215,
      "grad_norm": 0.5377092361450195,
      "learning_rate": 0.0001721438352568144,
      "loss": 0.7989499568939209,
      "mean_token_accuracy": 0.796073941886425,
      "num_tokens": 3620934.0,
      "step": 780
    },
    {
      "entropy": 0.7875147189944982,
      "epoch": 0.27001623515337947,
      "grad_norm": 0.5906500220298767,
      "learning_rate": 0.0001713728747472119,
      "loss": 0.7800149917602539,
      "mean_token_accuracy": 0.8019427113234997,
      "num_tokens": 3666872.0,
      "step": 790
    },
    {
      "entropy": 0.8653329361230135,
      "epoch": 0.27343416218063743,
      "grad_norm": 0.6267643570899963,
      "learning_rate": 0.00017059316835350808,
      "loss": 0.8605043411254882,
      "mean_token_accuracy": 0.7878318265080452,
      "num_tokens": 3713318.0,
      "step": 800
    },
    {
      "epoch": 0.27343416218063743,
      "eval_entropy": 0.770980479195714,
      "eval_loss": 0.7552855014801025,
      "eval_mean_token_accuracy": 0.8025437407195568,
      "eval_num_tokens": 3713318.0,
      "eval_runtime": 32.6834,
      "eval_samples_per_second": 7.833,
      "eval_steps_per_second": 0.979,
      "step": 800
    },
    {
      "entropy": 0.795812277123332,
      "epoch": 0.2768520892078954,
      "grad_norm": 0.5157990455627441,
      "learning_rate": 0.00016980481161930794,
      "loss": 0.7769721508026123,
      "mean_token_accuracy": 0.7985086388885975,
      "num_tokens": 3761523.0,
      "step": 810
    },
    {
      "entropy": 0.7850888863205909,
      "epoch": 0.28027001623515335,
      "grad_norm": 0.5042206048965454,
      "learning_rate": 0.00016900790114821122,
      "loss": 0.7836944103240967,
      "mean_token_accuracy": 0.7981749847531319,
      "num_tokens": 3813836.0,
      "step": 820
    },
    {
      "epoch": 0.2819789797487824,
      "eval_entropy": 0.7642055042088032,
      "eval_loss": 0.7502511739730835,
      "eval_mean_token_accuracy": 0.8033925238996744,
      "eval_num_tokens": 3838428.0,
      "eval_runtime": 32.5371,
      "eval_samples_per_second": 7.868,
      "eval_steps_per_second": 0.983,
      "step": 825
    },
    {
      "entropy": 0.8371281690895558,
      "epoch": 0.28368794326241137,
      "grad_norm": 0.5662909150123596,
      "learning_rate": 0.00016820253459197494,
      "loss": 0.8017690658569336,
      "mean_token_accuracy": 0.7955864246934652,
      "num_tokens": 3861556.0,
      "step": 830
    },
    {
      "entropy": 0.8396706741303206,
      "epoch": 0.28710587028966933,
      "grad_norm": 0.6416739225387573,
      "learning_rate": 0.00016738881063854723,
      "loss": 0.8341230392456055,
      "mean_token_accuracy": 0.7900606267154217,
      "num_tokens": 3905123.0,
      "step": 840
    },
    {
      "entropy": 0.8452721174806357,
      "epoch": 0.2905237973169273,
      "grad_norm": 0.6046516299247742,
      "learning_rate": 0.0001665668289999743,
      "loss": 0.8393696784973145,
      "mean_token_accuracy": 0.7913000397384167,
      "num_tokens": 3951295.0,
      "step": 850
    },
    {
      "epoch": 0.2905237973169273,
      "eval_entropy": 0.7603745739907026,
      "eval_loss": 0.7504570484161377,
      "eval_mean_token_accuracy": 0.8041685316711664,
      "eval_num_tokens": 3951295.0,
      "eval_runtime": 32.5482,
      "eval_samples_per_second": 7.865,
      "eval_steps_per_second": 0.983,
      "step": 850
    },
    {
      "entropy": 0.8369365192949771,
      "epoch": 0.29394172434418525,
      "grad_norm": 0.4931451082229614,
      "learning_rate": 0.000165736690400182,
      "loss": 0.8118236541748047,
      "mean_token_accuracy": 0.7955066107213498,
      "num_tokens": 3996465.0,
      "step": 860
    },
    {
      "entropy": 0.8153265129774809,
      "epoch": 0.2973596513714432,
      "grad_norm": 0.48623228073120117,
      "learning_rate": 0.00016489849656263337,
      "loss": 0.8089279174804688,
      "mean_token_accuracy": 0.7952679857611656,
      "num_tokens": 4043163.0,
      "step": 870
    },
    {
      "epoch": 0.2990686148850722,
      "eval_entropy": 0.7621881738305092,
      "eval_loss": 0.7468876838684082,
      "eval_mean_token_accuracy": 0.8038825057446957,
      "eval_num_tokens": 4064909.0,
      "eval_runtime": 32.5277,
      "eval_samples_per_second": 7.87,
      "eval_steps_per_second": 0.984,
      "step": 875
    },
    {
      "entropy": 0.8572983916848897,
      "epoch": 0.3007775783987012,
      "grad_norm": 0.48486313223838806,
      "learning_rate": 0.00016405235019786342,
      "loss": 0.8545586585998535,
      "mean_token_accuracy": 0.788145849853754,
      "num_tokens": 4090312.0,
      "step": 880
    },
    {
      "entropy": 0.8321780387312174,
      "epoch": 0.30419550542595913,
      "grad_norm": 0.5222928524017334,
      "learning_rate": 0.00016319835499089357,
      "loss": 0.8050444602966309,
      "mean_token_accuracy": 0.7938913144171238,
      "num_tokens": 4136984.0,
      "step": 890
    },
    {
      "entropy": 0.8131867442280054,
      "epoch": 0.30761343245321715,
      "grad_norm": 0.5593435168266296,
      "learning_rate": 0.00016233661558852589,
      "loss": 0.8005794525146485,
      "mean_token_accuracy": 0.7981902278959752,
      "num_tokens": 4182930.0,
      "step": 900
    },
    {
      "epoch": 0.30761343245321715,
      "eval_entropy": 0.7397047821432352,
      "eval_loss": 0.7391353845596313,
      "eval_mean_token_accuracy": 0.807006636634469,
      "eval_num_tokens": 4182930.0,
      "eval_runtime": 32.4819,
      "eval_samples_per_second": 7.881,
      "eval_steps_per_second": 0.985,
      "step": 900
    },
    {
      "entropy": 0.818829319998622,
      "epoch": 0.3110313594804751,
      "grad_norm": 0.5209222435951233,
      "learning_rate": 0.00016146723758652022,
      "loss": 0.8059531211853027,
      "mean_token_accuracy": 0.7948882877826691,
      "num_tokens": 4231251.0,
      "step": 910
    },
    {
      "entropy": 0.8251776993274689,
      "epoch": 0.3144492865077331,
      "grad_norm": 0.5925418734550476,
      "learning_rate": 0.00016059032751665454,
      "loss": 0.8086865425109864,
      "mean_token_accuracy": 0.7940934114158154,
      "num_tokens": 4280166.0,
      "step": 920
    },
    {
      "epoch": 0.31615825002136205,
      "eval_entropy": 0.741329038515687,
      "eval_loss": 0.7370421290397644,
      "eval_mean_token_accuracy": 0.8069011252373457,
      "eval_num_tokens": 4307100.0,
      "eval_runtime": 32.4627,
      "eval_samples_per_second": 7.886,
      "eval_steps_per_second": 0.986,
      "step": 925
    },
    {
      "entropy": 0.765642897412181,
      "epoch": 0.31786721353499103,
      "grad_norm": 0.5677841305732727,
      "learning_rate": 0.00015970599283367078,
      "loss": 0.7514712333679199,
      "mean_token_accuracy": 0.8063071981072426,
      "num_tokens": 4336292.0,
      "step": 930
    },
    {
      "entropy": 0.7642712559551,
      "epoch": 0.321285140562249,
      "grad_norm": 0.5148168206214905,
      "learning_rate": 0.00015881434190210763,
      "loss": 0.7541075706481933,
      "mean_token_accuracy": 0.8060836717486382,
      "num_tokens": 4388266.0,
      "step": 940
    },
    {
      "entropy": 0.8410390850156546,
      "epoch": 0.32470306758950696,
      "grad_norm": 0.7363119125366211,
      "learning_rate": 0.00015791548398302164,
      "loss": 0.8303253173828125,
      "mean_token_accuracy": 0.7919544480741024,
      "num_tokens": 4433500.0,
      "step": 950
    },
    {
      "epoch": 0.32470306758950696,
      "eval_entropy": 0.7385244891047478,
      "eval_loss": 0.7364199161529541,
      "eval_mean_token_accuracy": 0.8064117543399334,
      "eval_num_tokens": 4433500.0,
      "eval_runtime": 32.4989,
      "eval_samples_per_second": 7.877,
      "eval_steps_per_second": 0.985,
      "step": 950
    },
    {
      "entropy": 0.8260718151926995,
      "epoch": 0.3281209946167649,
      "grad_norm": 0.5341859459877014,
      "learning_rate": 0.00015700952922059874,
      "loss": 0.8289309501647949,
      "mean_token_accuracy": 0.7920399971306324,
      "num_tokens": 4479489.0,
      "step": 960
    },
    {
      "entropy": 0.7881720349192619,
      "epoch": 0.3315389216440229,
      "grad_norm": 0.7280305027961731,
      "learning_rate": 0.00015609658862865725,
      "loss": 0.7682159423828125,
      "mean_token_accuracy": 0.8035141505300999,
      "num_tokens": 4530797.0,
      "step": 970
    },
    {
      "epoch": 0.33324788515765186,
      "eval_entropy": 0.7407499924302101,
      "eval_loss": 0.7331442832946777,
      "eval_mean_token_accuracy": 0.8074121940881014,
      "eval_num_tokens": 4559773.0,
      "eval_runtime": 32.5019,
      "eval_samples_per_second": 7.876,
      "eval_steps_per_second": 0.985,
      "step": 975
    },
    {
      "entropy": 0.7880524016916752,
      "epoch": 0.3349568486712809,
      "grad_norm": 0.6096550226211548,
      "learning_rate": 0.0001551767740770446,
      "loss": 0.7723751068115234,
      "mean_token_accuracy": 0.8043812476098537,
      "num_tokens": 4581669.0,
      "step": 980
    },
    {
      "entropy": 0.7872831225395203,
      "epoch": 0.33837477569853885,
      "grad_norm": 0.5057272911071777,
      "learning_rate": 0.00015425019827792902,
      "loss": 0.7812077045440674,
      "mean_token_accuracy": 0.8002652741968632,
      "num_tokens": 4630513.0,
      "step": 990
    },
    {
      "entropy": 0.8601962827146054,
      "epoch": 0.3417927027257968,
      "grad_norm": 0.6072553992271423,
      "learning_rate": 0.00015331697477198792,
      "loss": 0.8457524299621582,
      "mean_token_accuracy": 0.7890366986393929,
      "num_tokens": 4676077.0,
      "step": 1000
    },
    {
      "epoch": 0.3417927027257968,
      "eval_entropy": 0.7451446056365967,
      "eval_loss": 0.7285513281822205,
      "eval_mean_token_accuracy": 0.80840396694839,
      "eval_num_tokens": 4676077.0,
      "eval_runtime": 32.5072,
      "eval_samples_per_second": 7.875,
      "eval_steps_per_second": 0.984,
      "step": 1000
    },
    {
      "entropy": 0.8780841417610645,
      "epoch": 0.3452106297530548,
      "grad_norm": 0.6355507969856262,
      "learning_rate": 0.00015237721791449497,
      "loss": 0.8509461402893066,
      "mean_token_accuracy": 0.7890427649021149,
      "num_tokens": 4719356.0,
      "step": 1010
    },
    {
      "entropy": 0.8176627814769745,
      "epoch": 0.34862855678031274,
      "grad_norm": 0.5865179300308228,
      "learning_rate": 0.00015143104286130722,
      "loss": 0.8123026847839355,
      "mean_token_accuracy": 0.795345525443554,
      "num_tokens": 4768569.0,
      "step": 1020
    },
    {
      "epoch": 0.3503375202939417,
      "eval_entropy": 0.7301500011235476,
      "eval_loss": 0.7274665832519531,
      "eval_mean_token_accuracy": 0.8079865016043186,
      "eval_num_tokens": 4793965.0,
      "eval_runtime": 32.5203,
      "eval_samples_per_second": 7.872,
      "eval_steps_per_second": 0.984,
      "step": 1025
    },
    {
      "entropy": 0.8386632338166237,
      "epoch": 0.3520464838075707,
      "grad_norm": 0.5760496854782104,
      "learning_rate": 0.00015047856555475405,
      "loss": 0.8192385673522949,
      "mean_token_accuracy": 0.7943773828446865,
      "num_tokens": 4817157.0,
      "step": 1030
    },
    {
      "entropy": 0.8414398148655892,
      "epoch": 0.35546441083482866,
      "grad_norm": 0.6798858046531677,
      "learning_rate": 0.00014951990270942991,
      "loss": 0.8329007148742675,
      "mean_token_accuracy": 0.790632463246584,
      "num_tokens": 4866025.0,
      "step": 1040
    },
    {
      "entropy": 0.8049224857240915,
      "epoch": 0.3588823378620866,
      "grad_norm": 0.6678990721702576,
      "learning_rate": 0.00014855517179789238,
      "loss": 0.7853578567504883,
      "mean_token_accuracy": 0.7974557019770145,
      "num_tokens": 4915676.0,
      "step": 1050
    },
    {
      "epoch": 0.3588823378620866,
      "eval_entropy": 0.7328374776989222,
      "eval_loss": 0.7274249792098999,
      "eval_mean_token_accuracy": 0.8089383095502853,
      "eval_num_tokens": 4915676.0,
      "eval_runtime": 32.4738,
      "eval_samples_per_second": 7.883,
      "eval_steps_per_second": 0.985,
      "step": 1050
    },
    {
      "entropy": 0.827161880582571,
      "epoch": 0.36230026488934464,
      "grad_norm": 0.5211610794067383,
      "learning_rate": 0.0001475844910362673,
      "loss": 0.8180788040161133,
      "mean_token_accuracy": 0.7956268422305584,
      "num_tokens": 4963558.0,
      "step": 1060
    },
    {
      "entropy": 0.8426773726940155,
      "epoch": 0.3657181919166026,
      "grad_norm": 0.5229407548904419,
      "learning_rate": 0.0001466079793697628,
      "loss": 0.8286858558654785,
      "mean_token_accuracy": 0.791385481506586,
      "num_tokens": 5013165.0,
      "step": 1070
    },
    {
      "epoch": 0.3674271554302316,
      "eval_entropy": 0.7287974283099174,
      "eval_loss": 0.7251026034355164,
      "eval_mean_token_accuracy": 0.8087449353188276,
      "eval_num_tokens": 5037412.0,
      "eval_runtime": 32.4973,
      "eval_samples_per_second": 7.878,
      "eval_steps_per_second": 0.985,
      "step": 1075
    },
    {
      "entropy": 0.7869369875639677,
      "epoch": 0.36913611894386056,
      "grad_norm": 0.5398491024971008,
      "learning_rate": 0.00014562575645809392,
      "loss": 0.76619873046875,
      "mean_token_accuracy": 0.8015293650329113,
      "num_tokens": 5064171.0,
      "step": 1080
    },
    {
      "entropy": 0.7933939069509506,
      "epoch": 0.3725540459711185,
      "grad_norm": 0.5217536091804504,
      "learning_rate": 0.00014463794266081993,
      "loss": 0.7810101985931397,
      "mean_token_accuracy": 0.7977319084107876,
      "num_tokens": 5114271.0,
      "step": 1090
    },
    {
      "entropy": 0.8145204957574605,
      "epoch": 0.3759719729983765,
      "grad_norm": 0.6378165483474731,
      "learning_rate": 0.0001436446590225957,
      "loss": 0.7995360851287842,
      "mean_token_accuracy": 0.7963194355368615,
      "num_tokens": 5164849.0,
      "step": 1100
    },
    {
      "epoch": 0.3759719729983765,
      "eval_entropy": 0.7224072255194187,
      "eval_loss": 0.7258445024490356,
      "eval_mean_token_accuracy": 0.8094426840543747,
      "eval_num_tokens": 5164849.0,
      "eval_runtime": 32.4659,
      "eval_samples_per_second": 7.885,
      "eval_steps_per_second": 0.986,
      "step": 1100
    },
    {
      "entropy": 0.8088543228805065,
      "epoch": 0.37938990002563444,
      "grad_norm": 0.5665172338485718,
      "learning_rate": 0.00014264602725833889,
      "loss": 0.7946016311645507,
      "mean_token_accuracy": 0.802227259427309,
      "num_tokens": 5212036.0,
      "step": 1110
    },
    {
      "entropy": 0.8042231686413288,
      "epoch": 0.3828078270528924,
      "grad_norm": 0.5462493300437927,
      "learning_rate": 0.00014164216973831554,
      "loss": 0.7905406475067138,
      "mean_token_accuracy": 0.7975196823477745,
      "num_tokens": 5259601.0,
      "step": 1120
    },
    {
      "epoch": 0.3845167905665214,
      "eval_entropy": 0.7332466039806604,
      "eval_loss": 0.7217105031013489,
      "eval_mean_token_accuracy": 0.8097752016037703,
      "eval_num_tokens": 5285653.0,
      "eval_runtime": 32.5314,
      "eval_samples_per_second": 7.869,
      "eval_steps_per_second": 0.984,
      "step": 1125
    },
    {
      "entropy": 0.7697301335632801,
      "epoch": 0.38622575408015036,
      "grad_norm": 0.46639689803123474,
      "learning_rate": 0.000140633209473145,
      "loss": 0.749844741821289,
      "mean_token_accuracy": 0.8087578162550926,
      "num_tokens": 5308035.0,
      "step": 1130
    },
    {
      "entropy": 0.7382064629346132,
      "epoch": 0.3896436811074084,
      "grad_norm": 0.6214824318885803,
      "learning_rate": 0.00013961927009872635,
      "loss": 0.7318979740142822,
      "mean_token_accuracy": 0.8110274896025658,
      "num_tokens": 5358532.0,
      "step": 1140
    },
    {
      "entropy": 0.8200155161321163,
      "epoch": 0.39306160813466634,
      "grad_norm": 0.5116494297981262,
      "learning_rate": 0.00013860047586108847,
      "loss": 0.8156219482421875,
      "mean_token_accuracy": 0.7928235337138176,
      "num_tokens": 5404679.0,
      "step": 1150
    },
    {
      "epoch": 0.39306160813466634,
      "eval_entropy": 0.7200065050274134,
      "eval_loss": 0.7224220633506775,
      "eval_mean_token_accuracy": 0.8092462345957756,
      "eval_num_tokens": 5404679.0,
      "eval_runtime": 32.5201,
      "eval_samples_per_second": 7.872,
      "eval_steps_per_second": 0.984,
      "step": 1150
    },
    {
      "entropy": 0.7471101813018322,
      "epoch": 0.3964795351619243,
      "grad_norm": 0.6286695003509521,
      "learning_rate": 0.000137576951601165,
      "loss": 0.721110486984253,
      "mean_token_accuracy": 0.8117256015539169,
      "num_tokens": 5457566.0,
      "step": 1160
    },
    {
      "entropy": 0.7424865119159222,
      "epoch": 0.39989746218918226,
      "grad_norm": 0.5141443014144897,
      "learning_rate": 0.00013654882273949687,
      "loss": 0.7295090675354003,
      "mean_token_accuracy": 0.8102657824754715,
      "num_tokens": 5506965.0,
      "step": 1170
    },
    {
      "epoch": 0.40160642570281124,
      "eval_entropy": 0.7222198080271482,
      "eval_loss": 0.7179678678512573,
      "eval_mean_token_accuracy": 0.8107004519551992,
      "eval_num_tokens": 5531529.0,
      "eval_runtime": 32.5483,
      "eval_samples_per_second": 7.865,
      "eval_steps_per_second": 0.983,
      "step": 1175
    },
    {
      "entropy": 0.8053314633667469,
      "epoch": 0.4033153892164402,
      "grad_norm": 0.7132169008255005,
      "learning_rate": 0.00013551621526086327,
      "loss": 0.7890602588653565,
      "mean_token_accuracy": 0.7956084564328194,
      "num_tokens": 5554204.0,
      "step": 1180
    },
    {
      "entropy": 0.8563751470297575,
      "epoch": 0.4067333162436982,
      "grad_norm": 0.6268016695976257,
      "learning_rate": 0.00013447925569884376,
      "loss": 0.8365967750549317,
      "mean_token_accuracy": 0.7909242726862431,
      "num_tokens": 5597625.0,
      "step": 1190
    },
    {
      "entropy": 0.8078042831271887,
      "epoch": 0.41015124327095615,
      "grad_norm": 0.6837403178215027,
      "learning_rate": 0.00013343807112031327,
      "loss": 0.7975335121154785,
      "mean_token_accuracy": 0.7958420798182487,
      "num_tokens": 5645765.0,
      "step": 1200
    },
    {
      "epoch": 0.41015124327095615,
      "eval_entropy": 0.711488027125597,
      "eval_loss": 0.7188196778297424,
      "eval_mean_token_accuracy": 0.8109075669199228,
      "eval_num_tokens": 5645765.0,
      "eval_runtime": 32.5021,
      "eval_samples_per_second": 7.876,
      "eval_steps_per_second": 0.985,
      "step": 1200
    },
    {
      "entropy": 0.7705740612000227,
      "epoch": 0.4135691702982141,
      "grad_norm": 0.5480158925056458,
      "learning_rate": 0.00013239278910987152,
      "loss": 0.7573784828186035,
      "mean_token_accuracy": 0.8046229638159275,
      "num_tokens": 5694447.0,
      "step": 1210
    },
    {
      "entropy": 0.8250246964395046,
      "epoch": 0.4169870973254721,
      "grad_norm": 0.6693493127822876,
      "learning_rate": 0.00013134353775420894,
      "loss": 0.7927688598632813,
      "mean_token_accuracy": 0.7985307432711124,
      "num_tokens": 5742053.0,
      "step": 1220
    },
    {
      "epoch": 0.4186960608391011,
      "eval_entropy": 0.7028289344161749,
      "eval_loss": 0.7150291204452515,
      "eval_mean_token_accuracy": 0.8113893959671259,
      "eval_num_tokens": 5765131.0,
      "eval_runtime": 32.5146,
      "eval_samples_per_second": 7.873,
      "eval_steps_per_second": 0.984,
      "step": 1225
    },
    {
      "entropy": 0.7481278918683529,
      "epoch": 0.4204050243527301,
      "grad_norm": 0.5626993179321289,
      "learning_rate": 0.00013029044562641134,
      "loss": 0.7459975242614746,
      "mean_token_accuracy": 0.810116047412157,
      "num_tokens": 5788534.0,
      "step": 1230
    },
    {
      "entropy": 0.8283022619783879,
      "epoch": 0.42382295137998804,
      "grad_norm": 0.6033257842063904,
      "learning_rate": 0.00012923364177020482,
      "loss": 0.8174246788024903,
      "mean_token_accuracy": 0.7946907714009285,
      "num_tokens": 5834731.0,
      "step": 1240
    },
    {
      "entropy": 0.762397014349699,
      "epoch": 0.427240878407246,
      "grad_norm": 0.5459337830543518,
      "learning_rate": 0.00012817325568414297,
      "loss": 0.7549442291259766,
      "mean_token_accuracy": 0.8046312466263771,
      "num_tokens": 5888788.0,
      "step": 1250
    },
    {
      "epoch": 0.427240878407246,
      "eval_entropy": 0.7237848732620478,
      "eval_loss": 0.7120697498321533,
      "eval_mean_token_accuracy": 0.8127239029854536,
      "eval_num_tokens": 5888788.0,
      "eval_runtime": 32.5537,
      "eval_samples_per_second": 7.864,
      "eval_steps_per_second": 0.983,
      "step": 1250
    },
    {
      "entropy": 0.8118621993809938,
      "epoch": 0.43065880543450397,
      "grad_norm": 0.5670892000198364,
      "learning_rate": 0.00012710941730573836,
      "loss": 0.786963415145874,
      "mean_token_accuracy": 0.7991264708340168,
      "num_tokens": 5940945.0,
      "step": 1260
    },
    {
      "entropy": 0.8265308391302824,
      "epoch": 0.4340767324617619,
      "grad_norm": 0.58514404296875,
      "learning_rate": 0.0001260422569955403,
      "loss": 0.8180745124816895,
      "mean_token_accuracy": 0.7946577452123165,
      "num_tokens": 5987771.0,
      "step": 1270
    },
    {
      "epoch": 0.4357856959753909,
      "eval_entropy": 0.7171956170350313,
      "eval_loss": 0.7113481163978577,
      "eval_mean_token_accuracy": 0.8121560588479042,
      "eval_num_tokens": 6013715.0,
      "eval_runtime": 32.5293,
      "eval_samples_per_second": 7.87,
      "eval_steps_per_second": 0.984,
      "step": 1275
    },
    {
      "entropy": 0.8262432683259249,
      "epoch": 0.4374946594890199,
      "grad_norm": 0.652851939201355,
      "learning_rate": 0.00012497190552116083,
      "loss": 0.8104188919067383,
      "mean_token_accuracy": 0.7939992636442185,
      "num_tokens": 6039288.0,
      "step": 1280
    },
    {
      "entropy": 0.7597002040594816,
      "epoch": 0.4409125865162779,
      "grad_norm": 0.5283879637718201,
      "learning_rate": 0.0001238984940412505,
      "loss": 0.7422741889953614,
      "mean_token_accuracy": 0.8095363132655621,
      "num_tokens": 6087651.0,
      "step": 1290
    },
    {
      "entropy": 0.7729848586022854,
      "epoch": 0.44433051354353587,
      "grad_norm": 0.6668198704719543,
      "learning_rate": 0.00012282215408942674,
      "loss": 0.7518921852111816,
      "mean_token_accuracy": 0.8074666738510132,
      "num_tokens": 6133232.0,
      "step": 1300
    },
    {
      "epoch": 0.44433051354353587,
      "eval_entropy": 0.7048162054270506,
      "eval_loss": 0.7114033699035645,
      "eval_mean_token_accuracy": 0.8132472056895494,
      "eval_num_tokens": 6133232.0,
      "eval_runtime": 32.5592,
      "eval_samples_per_second": 7.863,
      "eval_steps_per_second": 0.983,
      "step": 1300
    },
    {
      "entropy": 0.7919327396899462,
      "epoch": 0.4477484405707938,
      "grad_norm": 0.7264841198921204,
      "learning_rate": 0.00012174301755815571,
      "loss": 0.784566068649292,
      "mean_token_accuracy": 0.801679265499115,
      "num_tokens": 6177149.0,
      "step": 1310
    },
    {
      "entropy": 0.7804473422467708,
      "epoch": 0.4511663675980518,
      "grad_norm": 0.7080020308494568,
      "learning_rate": 0.00012066121668259072,
      "loss": 0.7683864593505859,
      "mean_token_accuracy": 0.8041128098964692,
      "num_tokens": 6224301.0,
      "step": 1320
    },
    {
      "epoch": 0.45287533111168077,
      "eval_entropy": 0.7122832704335451,
      "eval_loss": 0.7075121402740479,
      "eval_mean_token_accuracy": 0.8129062969237566,
      "eval_num_tokens": 6249760.0,
      "eval_runtime": 32.5042,
      "eval_samples_per_second": 7.876,
      "eval_steps_per_second": 0.984,
      "step": 1325
    },
    {
      "entropy": 0.7673846136778593,
      "epoch": 0.45458429462530975,
      "grad_norm": 0.47276806831359863,
      "learning_rate": 0.00011957688402436826,
      "loss": 0.7500048637390136,
      "mean_token_accuracy": 0.8052079066634178,
      "num_tokens": 6273667.0,
      "step": 1330
    },
    {
      "entropy": 0.7663803111761809,
      "epoch": 0.4580022216525677,
      "grad_norm": 0.5211388468742371,
      "learning_rate": 0.00011849015245536424,
      "loss": 0.7578155040740967,
      "mean_token_accuracy": 0.8071829579770565,
      "num_tokens": 6323164.0,
      "step": 1340
    },
    {
      "entropy": 0.8483320087194443,
      "epoch": 0.46142014867982567,
      "grad_norm": 0.5593488216400146,
      "learning_rate": 0.0001174011551414121,
      "loss": 0.8137017250061035,
      "mean_token_accuracy": 0.7960939630866051,
      "num_tokens": 6367928.0,
      "step": 1350
    },
    {
      "epoch": 0.46142014867982567,
      "eval_entropy": 0.7004928775131702,
      "eval_loss": 0.7064136266708374,
      "eval_mean_token_accuracy": 0.8134569376707077,
      "eval_num_tokens": 6367928.0,
      "eval_runtime": 32.4753,
      "eval_samples_per_second": 7.883,
      "eval_steps_per_second": 0.985,
      "step": 1350
    },
    {
      "entropy": 0.7482695568352937,
      "epoch": 0.46483807570708363,
      "grad_norm": 0.6436233520507812,
      "learning_rate": 0.00011631002552598503,
      "loss": 0.7543563365936279,
      "mean_token_accuracy": 0.8053471855819225,
      "num_tokens": 6414719.0,
      "step": 1360
    },
    {
      "entropy": 0.7612337287515402,
      "epoch": 0.46825600273434165,
      "grad_norm": 0.6195892691612244,
      "learning_rate": 0.0001152168973138439,
      "loss": 0.7435910701751709,
      "mean_token_accuracy": 0.8076206877827644,
      "num_tokens": 6465494.0,
      "step": 1370
    },
    {
      "epoch": 0.46996496624797063,
      "eval_entropy": 0.7111266031861305,
      "eval_loss": 0.7041518688201904,
      "eval_mean_token_accuracy": 0.8130570650100708,
      "eval_num_tokens": 6489945.0,
      "eval_runtime": 32.5587,
      "eval_samples_per_second": 7.863,
      "eval_steps_per_second": 0.983,
      "step": 1375
    },
    {
      "entropy": 0.7827417716383934,
      "epoch": 0.4716739297615996,
      "grad_norm": 0.5481026768684387,
      "learning_rate": 0.0001141219044546536,
      "loss": 0.7692788124084473,
      "mean_token_accuracy": 0.8012573964893818,
      "num_tokens": 6514928.0,
      "step": 1380
    },
    {
      "entropy": 0.8570831570774317,
      "epoch": 0.47509185678885757,
      "grad_norm": 0.6454991102218628,
      "learning_rate": 0.00011302518112656897,
      "loss": 0.8441422462463379,
      "mean_token_accuracy": 0.786506038159132,
      "num_tokens": 6561291.0,
      "step": 1390
    },
    {
      "entropy": 0.8111637011170387,
      "epoch": 0.47850978381611553,
      "grad_norm": 0.5012980103492737,
      "learning_rate": 0.00011192686171979288,
      "loss": 0.8026445388793946,
      "mean_token_accuracy": 0.7964430071413517,
      "num_tokens": 6606995.0,
      "step": 1400
    },
    {
      "epoch": 0.47850978381611553,
      "eval_entropy": 0.6970334611833096,
      "eval_loss": 0.7001909017562866,
      "eval_mean_token_accuracy": 0.8144857585430145,
      "eval_num_tokens": 6606995.0,
      "eval_runtime": 32.5377,
      "eval_samples_per_second": 7.868,
      "eval_steps_per_second": 0.983,
      "step": 1400
    },
    {
      "entropy": 0.8142560321837664,
      "epoch": 0.4819277108433735,
      "grad_norm": 0.6923454999923706,
      "learning_rate": 0.00011082708082010839,
      "loss": 0.7896495342254639,
      "mean_token_accuracy": 0.7999372504651546,
      "num_tokens": 6651262.0,
      "step": 1410
    },
    {
      "entropy": 0.7541784971952439,
      "epoch": 0.48534563787063145,
      "grad_norm": 0.5750320553779602,
      "learning_rate": 0.0001097259731923869,
      "loss": 0.7421460151672363,
      "mean_token_accuracy": 0.8080541737377643,
      "num_tokens": 6699436.0,
      "step": 1420
    },
    {
      "epoch": 0.48705460138426043,
      "eval_entropy": 0.7030851934105158,
      "eval_loss": 0.6975977420806885,
      "eval_mean_token_accuracy": 0.8151122163981199,
      "eval_num_tokens": 6723464.0,
      "eval_runtime": 32.5721,
      "eval_samples_per_second": 7.859,
      "eval_steps_per_second": 0.982,
      "step": 1425
    },
    {
      "entropy": 0.7732356287539005,
      "epoch": 0.4887635648978894,
      "grad_norm": 0.6692112684249878,
      "learning_rate": 0.00010862367376407432,
      "loss": 0.7660270690917969,
      "mean_token_accuracy": 0.8035531021654606,
      "num_tokens": 6747857.0,
      "step": 1430
    },
    {
      "entropy": 0.7560388691723346,
      "epoch": 0.4921814919251474,
      "grad_norm": 0.5274794101715088,
      "learning_rate": 0.00010752031760865728,
      "loss": 0.7368973731994629,
      "mean_token_accuracy": 0.8090716004371643,
      "num_tokens": 6796429.0,
      "step": 1440
    },
    {
      "entropy": 0.7908974122256041,
      "epoch": 0.4955994189524054,
      "grad_norm": 0.4911586046218872,
      "learning_rate": 0.00010641603992911158,
      "loss": 0.7843417644500732,
      "mean_token_accuracy": 0.8022970989346504,
      "num_tokens": 6843753.0,
      "step": 1450
    },
    {
      "epoch": 0.4955994189524054,
      "eval_entropy": 0.7044373229146004,
      "eval_loss": 0.6949427127838135,
      "eval_mean_token_accuracy": 0.8151835780590773,
      "eval_num_tokens": 6843753.0,
      "eval_runtime": 32.5751,
      "eval_samples_per_second": 7.859,
      "eval_steps_per_second": 0.982,
      "step": 1450
    },
    {
      "entropy": 0.7853588089346886,
      "epoch": 0.49901734597966335,
      "grad_norm": 0.4415193796157837,
      "learning_rate": 0.00010531097604133473,
      "loss": 0.7712342739105225,
      "mean_token_accuracy": 0.8053767092525959,
      "num_tokens": 6889412.0,
      "step": 1460
    },
    {
      "entropy": 0.730915354192257,
      "epoch": 0.5024352730069213,
      "grad_norm": 0.47233015298843384,
      "learning_rate": 0.00010420526135756444,
      "loss": 0.7267525672912598,
      "mean_token_accuracy": 0.8125036112964154,
      "num_tokens": 6939916.0,
      "step": 1470
    },
    {
      "epoch": 0.5041442365205503,
      "eval_entropy": 0.7217494286596775,
      "eval_loss": 0.6922969818115234,
      "eval_mean_token_accuracy": 0.8169958014041185,
      "eval_num_tokens": 6963309.0,
      "eval_runtime": 32.513,
      "eval_samples_per_second": 7.874,
      "eval_steps_per_second": 0.984,
      "step": 1475
    },
    {
      "entropy": 0.7474553748965264,
      "epoch": 0.5058532000341792,
      "grad_norm": 0.5465685129165649,
      "learning_rate": 0.00010309903136978575,
      "loss": 0.7162308692932129,
      "mean_token_accuracy": 0.8151901520788669,
      "num_tokens": 6986979.0,
      "step": 1480
    },
    {
      "entropy": 0.8350271973758936,
      "epoch": 0.5092711270614373,
      "grad_norm": 0.8223411440849304,
      "learning_rate": 0.00010199242163312794,
      "loss": 0.8277231216430664,
      "mean_token_accuracy": 0.7921891517937183,
      "num_tokens": 7037591.0,
      "step": 1490
    },
    {
      "entropy": 0.8443024232983589,
      "epoch": 0.5126890540886953,
      "grad_norm": 0.6216680407524109,
      "learning_rate": 0.00010088556774925387,
      "loss": 0.826251220703125,
      "mean_token_accuracy": 0.7934649467468262,
      "num_tokens": 7082012.0,
      "step": 1500
    },
    {
      "epoch": 0.5126890540886953,
      "eval_entropy": 0.6949329674243927,
      "eval_loss": 0.6909105777740479,
      "eval_mean_token_accuracy": 0.8162583336234093,
      "eval_num_tokens": 7082012.0,
      "eval_runtime": 32.5406,
      "eval_samples_per_second": 7.867,
      "eval_steps_per_second": 0.983,
      "step": 1500
    },
    {
      "entropy": 0.7664934404194355,
      "epoch": 0.5161069811159532,
      "grad_norm": 0.4688818156719208,
      "learning_rate": 9.97786053497438e-05,
      "loss": 0.7553867816925048,
      "mean_token_accuracy": 0.8041857704520226,
      "num_tokens": 7129239.0,
      "step": 1510
    },
    {
      "entropy": 0.7467406760901213,
      "epoch": 0.5195249081432112,
      "grad_norm": 0.5349637269973755,
      "learning_rate": 9.86716700794751e-05,
      "loss": 0.7290421009063721,
      "mean_token_accuracy": 0.810805543512106,
      "num_tokens": 7176709.0,
      "step": 1520
    },
    {
      "epoch": 0.5212338716568401,
      "eval_entropy": 0.6912414748221636,
      "eval_loss": 0.6880679130554199,
      "eval_mean_token_accuracy": 0.8175152335315943,
      "eval_num_tokens": 7198599.0,
      "eval_runtime": 32.5234,
      "eval_samples_per_second": 7.871,
      "eval_steps_per_second": 0.984,
      "step": 1525
    },
    {
      "entropy": 0.7089745387434959,
      "epoch": 0.5229428351704691,
      "grad_norm": 0.4970061779022217,
      "learning_rate": 9.756489758000105e-05,
      "loss": 0.691909646987915,
      "mean_token_accuracy": 0.8186588749289513,
      "num_tokens": 7225416.0,
      "step": 1530
    },
    {
      "entropy": 0.7925937842577696,
      "epoch": 0.5263607621977271,
      "grad_norm": 0.6043300628662109,
      "learning_rate": 9.645842347292908e-05,
      "loss": 0.788155460357666,
      "mean_token_accuracy": 0.7994618818163872,
      "num_tokens": 7273046.0,
      "step": 1540
    },
    {
      "entropy": 0.6987096827477217,
      "epoch": 0.529778689224985,
      "grad_norm": 0.5128796696662903,
      "learning_rate": 9.535238334330234e-05,
      "loss": 0.6659093856811523,
      "mean_token_accuracy": 0.8243751518428326,
      "num_tokens": 7322822.0,
      "step": 1550
    },
    {
      "epoch": 0.529778689224985,
      "eval_entropy": 0.7001140043139458,
      "eval_loss": 0.6867726445198059,
      "eval_mean_token_accuracy": 0.8168847281485796,
      "eval_num_tokens": 7322822.0,
      "eval_runtime": 32.4995,
      "eval_samples_per_second": 7.877,
      "eval_steps_per_second": 0.985,
      "step": 1550
    },
    {
      "entropy": 0.7453165553510189,
      "epoch": 0.533196616252243,
      "grad_norm": 0.6003188490867615,
      "learning_rate": 9.424691272298522e-05,
      "loss": 0.7359031200408935,
      "mean_token_accuracy": 0.8109373219311238,
      "num_tokens": 7370793.0,
      "step": 1560
    },
    {
      "entropy": 0.7829654891043901,
      "epoch": 0.536614543279501,
      "grad_norm": 0.6426380276679993,
      "learning_rate": 9.314214707405578e-05,
      "loss": 0.7814960479736328,
      "mean_token_accuracy": 0.802507346868515,
      "num_tokens": 7415437.0,
      "step": 1570
    },
    {
      "epoch": 0.53832350679313,
      "eval_entropy": 0.6993504334241152,
      "eval_loss": 0.6848057508468628,
      "eval_mean_token_accuracy": 0.817769967019558,
      "eval_num_tokens": 7437929.0,
      "eval_runtime": 32.4676,
      "eval_samples_per_second": 7.885,
      "eval_steps_per_second": 0.986,
      "step": 1575
    },
    {
      "entropy": 0.7911600664258003,
      "epoch": 0.5400324703067589,
      "grad_norm": 0.46196243166923523,
      "learning_rate": 9.20382217722062e-05,
      "loss": 0.7808879852294922,
      "mean_token_accuracy": 0.8050209023058414,
      "num_tokens": 7461751.0,
      "step": 1580
    },
    {
      "entropy": 0.8003505248576402,
      "epoch": 0.5434503973340169,
      "grad_norm": 0.6887136697769165,
      "learning_rate": 9.093527209015428e-05,
      "loss": 0.7642087459564209,
      "mean_token_accuracy": 0.8055305682122708,
      "num_tokens": 7509191.0,
      "step": 1590
    },
    {
      "entropy": 0.8003017157316208,
      "epoch": 0.5468683243612749,
      "grad_norm": 0.7519078850746155,
      "learning_rate": 8.983343318106745e-05,
      "loss": 0.7946379661560059,
      "mean_token_accuracy": 0.8008874267339706,
      "num_tokens": 7557898.0,
      "step": 1600
    },
    {
      "epoch": 0.5468683243612749,
      "eval_entropy": 0.6895345561206341,
      "eval_loss": 0.682197630405426,
      "eval_mean_token_accuracy": 0.8186191953718662,
      "eval_num_tokens": 7557898.0,
      "eval_runtime": 32.5239,
      "eval_samples_per_second": 7.871,
      "eval_steps_per_second": 0.984,
      "step": 1600
    },
    {
      "entropy": 0.7015115257352591,
      "epoch": 0.5502862513885328,
      "grad_norm": 0.45591458678245544,
      "learning_rate": 8.873284006200129e-05,
      "loss": 0.6760759353637695,
      "mean_token_accuracy": 0.8240564435720443,
      "num_tokens": 7606551.0,
      "step": 1610
    },
    {
      "entropy": 0.6873151533305645,
      "epoch": 0.5537041784157908,
      "grad_norm": 0.7503350377082825,
      "learning_rate": 8.763362759735486e-05,
      "loss": 0.6830436706542968,
      "mean_token_accuracy": 0.8215327501296997,
      "num_tokens": 7655943.0,
      "step": 1620
    },
    {
      "epoch": 0.5554131419294198,
      "eval_entropy": 0.6954763662070036,
      "eval_loss": 0.6825394630432129,
      "eval_mean_token_accuracy": 0.8192094713449478,
      "eval_num_tokens": 7679462.0,
      "eval_runtime": 32.4902,
      "eval_samples_per_second": 7.879,
      "eval_steps_per_second": 0.985,
      "step": 1625
    },
    {
      "entropy": 0.7175914201885462,
      "epoch": 0.5571221054430487,
      "grad_norm": 0.506363570690155,
      "learning_rate": 8.65359304823447e-05,
      "loss": 0.7104014396667481,
      "mean_token_accuracy": 0.8144126996397972,
      "num_tokens": 7704856.0,
      "step": 1630
    },
    {
      "entropy": 0.684526550769806,
      "epoch": 0.5605400324703067,
      "grad_norm": 0.47179755568504333,
      "learning_rate": 8.543988322649954e-05,
      "loss": 0.6660084247589111,
      "mean_token_accuracy": 0.8225892163813114,
      "num_tokens": 7756433.0,
      "step": 1640
    },
    {
      "entropy": 0.8113386046141386,
      "epoch": 0.5639579594975648,
      "grad_norm": 0.5647494196891785,
      "learning_rate": 8.434562013717774e-05,
      "loss": 0.7892371654510498,
      "mean_token_accuracy": 0.8024112187325955,
      "num_tokens": 7800781.0,
      "step": 1650
    },
    {
      "epoch": 0.5639579594975648,
      "eval_entropy": 0.6803799960762262,
      "eval_loss": 0.6814059019088745,
      "eval_mean_token_accuracy": 0.8198210597038269,
      "eval_num_tokens": 7800781.0,
      "eval_runtime": 32.5489,
      "eval_samples_per_second": 7.865,
      "eval_steps_per_second": 0.983,
      "step": 1650
    },
    {
      "entropy": 0.7244523506611585,
      "epoch": 0.5673758865248227,
      "grad_norm": 0.48115718364715576,
      "learning_rate": 8.325327530310972e-05,
      "loss": 0.7086876392364502,
      "mean_token_accuracy": 0.8146171241998672,
      "num_tokens": 7850500.0,
      "step": 1660
    },
    {
      "entropy": 0.7694044571369887,
      "epoch": 0.5707938135520807,
      "grad_norm": 0.7885907888412476,
      "learning_rate": 8.216298257796678e-05,
      "loss": 0.7577213764190673,
      "mean_token_accuracy": 0.8088521212339401,
      "num_tokens": 7894460.0,
      "step": 1670
    },
    {
      "epoch": 0.5725027770657096,
      "eval_entropy": 0.689734099432826,
      "eval_loss": 0.6795113682746887,
      "eval_mean_token_accuracy": 0.8193087466061115,
      "eval_num_tokens": 7915198.0,
      "eval_runtime": 32.4654,
      "eval_samples_per_second": 7.885,
      "eval_steps_per_second": 0.986,
      "step": 1675
    },
    {
      "entropy": 0.7575687125325203,
      "epoch": 0.5742117405793387,
      "grad_norm": 0.5657883882522583,
      "learning_rate": 8.107487556395901e-05,
      "loss": 0.7516120433807373,
      "mean_token_accuracy": 0.8099164560437202,
      "num_tokens": 7938871.0,
      "step": 1680
    },
    {
      "entropy": 0.7479819674044847,
      "epoch": 0.5776296676065966,
      "grad_norm": 0.5981023907661438,
      "learning_rate": 7.998908759546405e-05,
      "loss": 0.7331737995147705,
      "mean_token_accuracy": 0.8109016142785549,
      "num_tokens": 7986967.0,
      "step": 1690
    },
    {
      "entropy": 0.7127784315496684,
      "epoch": 0.5810475946338546,
      "grad_norm": 0.5446374416351318,
      "learning_rate": 7.890575172268858e-05,
      "loss": 0.6960432529449463,
      "mean_token_accuracy": 0.816921915858984,
      "num_tokens": 8038667.0,
      "step": 1700
    },
    {
      "epoch": 0.5810475946338546,
      "eval_entropy": 0.6878421176224947,
      "eval_loss": 0.6786676645278931,
      "eval_mean_token_accuracy": 0.8191769048571587,
      "eval_num_tokens": 8038667.0,
      "eval_runtime": 32.6367,
      "eval_samples_per_second": 7.844,
      "eval_steps_per_second": 0.98,
      "step": 1700
    },
    {
      "entropy": 0.7568402998149395,
      "epoch": 0.5844655216611125,
      "grad_norm": 0.5637646913528442,
      "learning_rate": 7.782500069536443e-05,
      "loss": 0.7467115879058838,
      "mean_token_accuracy": 0.8103341706097126,
      "num_tokens": 8085866.0,
      "step": 1710
    },
    {
      "entropy": 0.7543296214193106,
      "epoch": 0.5878834486883705,
      "grad_norm": 0.5011135935783386,
      "learning_rate": 7.674696694648183e-05,
      "loss": 0.7485779762268067,
      "mean_token_accuracy": 0.8080232337117195,
      "num_tokens": 8134284.0,
      "step": 1720
    },
    {
      "epoch": 0.5895924122019995,
      "eval_entropy": 0.6893986221402884,
      "eval_loss": 0.6773992776870728,
      "eval_mean_token_accuracy": 0.8196295741945505,
      "eval_num_tokens": 8156806.0,
      "eval_runtime": 32.6057,
      "eval_samples_per_second": 7.851,
      "eval_steps_per_second": 0.981,
      "step": 1725
    },
    {
      "entropy": 0.8183664843440056,
      "epoch": 0.5913013757156285,
      "grad_norm": 0.5187571048736572,
      "learning_rate": 7.567178257606146e-05,
      "loss": 0.7994352817535401,
      "mean_token_accuracy": 0.7991348020732403,
      "num_tokens": 8180060.0,
      "step": 1730
    },
    {
      "entropy": 0.862914065271616,
      "epoch": 0.5947193027428864,
      "grad_norm": 0.6644055247306824,
      "learning_rate": 7.4599579334967e-05,
      "loss": 0.8406139373779297,
      "mean_token_accuracy": 0.7935954198241234,
      "num_tokens": 8225765.0,
      "step": 1740
    },
    {
      "entropy": 0.7744019731879235,
      "epoch": 0.5981372297701444,
      "grad_norm": 0.5454498529434204,
      "learning_rate": 7.353048860876064e-05,
      "loss": 0.7569465160369873,
      "mean_token_accuracy": 0.8064947456121445,
      "num_tokens": 8273313.0,
      "step": 1750
    },
    {
      "epoch": 0.5981372297701444,
      "eval_entropy": 0.6880827695131302,
      "eval_loss": 0.6778668761253357,
      "eval_mean_token_accuracy": 0.8190348986536264,
      "eval_num_tokens": 8273313.0,
      "eval_runtime": 32.4973,
      "eval_samples_per_second": 7.878,
      "eval_steps_per_second": 0.985,
      "step": 1750
    },
    {
      "entropy": 0.6878764893859625,
      "epoch": 0.6015551567974023,
      "grad_norm": 0.5708960294723511,
      "learning_rate": 7.246464140160365e-05,
      "loss": 0.6662783622741699,
      "mean_token_accuracy": 0.8213848076760769,
      "num_tokens": 8322356.0,
      "step": 1760
    },
    {
      "entropy": 0.7368987411260605,
      "epoch": 0.6049730838246603,
      "grad_norm": 0.4885517358779907,
      "learning_rate": 7.1402168320203e-05,
      "loss": 0.734394121170044,
      "mean_token_accuracy": 0.8115066438913345,
      "num_tokens": 8371767.0,
      "step": 1770
    },
    {
      "epoch": 0.6066820473382893,
      "eval_entropy": 0.6817968115210533,
      "eval_loss": 0.6757955551147461,
      "eval_mean_token_accuracy": 0.8204179033637047,
      "eval_num_tokens": 8395150.0,
      "eval_runtime": 32.4907,
      "eval_samples_per_second": 7.879,
      "eval_steps_per_second": 0.985,
      "step": 1775
    },
    {
      "entropy": 0.7217590250074863,
      "epoch": 0.6083910108519183,
      "grad_norm": 0.54530930519104,
      "learning_rate": 7.034319955780735e-05,
      "loss": 0.6975030899047852,
      "mean_token_accuracy": 0.8191726803779602,
      "num_tokens": 8418891.0,
      "step": 1780
    },
    {
      "entropy": 0.759762617573142,
      "epoch": 0.6118089378791762,
      "grad_norm": 0.5948047041893005,
      "learning_rate": 6.928786487825348e-05,
      "loss": 0.7440184116363525,
      "mean_token_accuracy": 0.807938826829195,
      "num_tokens": 8467595.0,
      "step": 1790
    },
    {
      "entropy": 0.7751124188303947,
      "epoch": 0.6152268649064343,
      "grad_norm": 0.7067475914955139,
      "learning_rate": 6.823629360006504e-05,
      "loss": 0.7460795879364014,
      "mean_token_accuracy": 0.8089013800024987,
      "num_tokens": 8516779.0,
      "step": 1800
    },
    {
      "epoch": 0.6152268649064343,
      "eval_entropy": 0.6731869708746672,
      "eval_loss": 0.673740029335022,
      "eval_mean_token_accuracy": 0.8215623293071985,
      "eval_num_tokens": 8516779.0,
      "eval_runtime": 32.5195,
      "eval_samples_per_second": 7.872,
      "eval_steps_per_second": 0.984,
      "step": 1800
    },
    {
      "entropy": 0.7652524448931217,
      "epoch": 0.6186447919336923,
      "grad_norm": 0.573112964630127,
      "learning_rate": 6.718861458060634e-05,
      "loss": 0.7655494689941407,
      "mean_token_accuracy": 0.8049928575754166,
      "num_tokens": 8563658.0,
      "step": 1810
    },
    {
      "entropy": 0.7533473178744317,
      "epoch": 0.6220627189609502,
      "grad_norm": 0.5001428723335266,
      "learning_rate": 6.614495620029238e-05,
      "loss": 0.731645154953003,
      "mean_token_accuracy": 0.815316341817379,
      "num_tokens": 8607280.0,
      "step": 1820
    },
    {
      "epoch": 0.6237716824745791,
      "eval_entropy": 0.6760462690144777,
      "eval_loss": 0.6731796264648438,
      "eval_mean_token_accuracy": 0.8210084475576878,
      "eval_num_tokens": 8631570.0,
      "eval_runtime": 32.5048,
      "eval_samples_per_second": 7.876,
      "eval_steps_per_second": 0.984,
      "step": 1825
    },
    {
      "entropy": 0.7785196773707866,
      "epoch": 0.6254806459882082,
      "grad_norm": 0.6337042450904846,
      "learning_rate": 6.51054463468574e-05,
      "loss": 0.7558555126190185,
      "mean_token_accuracy": 0.8058261357247829,
      "num_tokens": 8655345.0,
      "step": 1830
    },
    {
      "entropy": 0.7649154797196388,
      "epoch": 0.6288985730154661,
      "grad_norm": 0.5913355946540833,
      "learning_rate": 6.407021239968371e-05,
      "loss": 0.7588380336761474,
      "mean_token_accuracy": 0.8053297519683837,
      "num_tokens": 8700351.0,
      "step": 1840
    },
    {
      "entropy": 0.7844179723411798,
      "epoch": 0.6323165000427241,
      "grad_norm": 0.6086034178733826,
      "learning_rate": 6.303938121419295e-05,
      "loss": 0.7802872657775879,
      "mean_token_accuracy": 0.8035823844373227,
      "num_tokens": 8751552.0,
      "step": 1850
    },
    {
      "epoch": 0.6323165000427241,
      "eval_entropy": 0.6918724570423365,
      "eval_loss": 0.6699243187904358,
      "eval_mean_token_accuracy": 0.8219087645411491,
      "eval_num_tokens": 8751552.0,
      "eval_runtime": 32.5593,
      "eval_samples_per_second": 7.863,
      "eval_steps_per_second": 0.983,
      "step": 1850
    },
    {
      "entropy": 0.7972814869135618,
      "epoch": 0.6357344270699821,
      "grad_norm": 0.5157380700111389,
      "learning_rate": 6.201307910630146e-05,
      "loss": 0.7538751602172852,
      "mean_token_accuracy": 0.8092072702944278,
      "num_tokens": 8795469.0,
      "step": 1860
    },
    {
      "entropy": 0.7942239165306091,
      "epoch": 0.63915235409724,
      "grad_norm": 0.6042788028717041,
      "learning_rate": 6.099143183694185e-05,
      "loss": 0.7757480144500732,
      "mean_token_accuracy": 0.8043014384806156,
      "num_tokens": 8839975.0,
      "step": 1870
    },
    {
      "epoch": 0.6408613176108691,
      "eval_entropy": 0.6643337700515985,
      "eval_loss": 0.6681345701217651,
      "eval_mean_token_accuracy": 0.822357814759016,
      "eval_num_tokens": 8864636.0,
      "eval_runtime": 32.502,
      "eval_samples_per_second": 7.876,
      "eval_steps_per_second": 0.985,
      "step": 1875
    },
    {
      "entropy": 0.7517622873187065,
      "epoch": 0.642570281124498,
      "grad_norm": 0.6731847524642944,
      "learning_rate": 5.9974564596652375e-05,
      "loss": 0.7436028480529785,
      "mean_token_accuracy": 0.8090323090553284,
      "num_tokens": 8886750.0,
      "step": 1880
    },
    {
      "entropy": 0.734363779425621,
      "epoch": 0.645988208151756,
      "grad_norm": 0.5949045419692993,
      "learning_rate": 5.8962601990236545e-05,
      "loss": 0.7188807010650635,
      "mean_token_accuracy": 0.8148549936711789,
      "num_tokens": 8934893.0,
      "step": 1890
    },
    {
      "entropy": 0.7205371145159006,
      "epoch": 0.6494061351790139,
      "grad_norm": 0.6062500476837158,
      "learning_rate": 5.79556680214941e-05,
      "loss": 0.7039190292358398,
      "mean_token_accuracy": 0.8170817263424397,
      "num_tokens": 8985225.0,
      "step": 1900
    },
    {
      "epoch": 0.6494061351790139,
      "eval_entropy": 0.6739299017935991,
      "eval_loss": 0.6656136512756348,
      "eval_mean_token_accuracy": 0.8219083081930876,
      "eval_num_tokens": 8985225.0,
      "eval_runtime": 32.5695,
      "eval_samples_per_second": 7.86,
      "eval_steps_per_second": 0.983,
      "step": 1900
    },
    {
      "entropy": 0.7479930385947228,
      "epoch": 0.6528240622062719,
      "grad_norm": 0.6024976968765259,
      "learning_rate": 5.695388607802603e-05,
      "loss": 0.7273672103881836,
      "mean_token_accuracy": 0.8128669574856758,
      "num_tokens": 9034378.0,
      "step": 1910
    },
    {
      "entropy": 0.7050946086645127,
      "epoch": 0.6562419892335298,
      "grad_norm": 0.6390378475189209,
      "learning_rate": 5.5957378916114766e-05,
      "loss": 0.6898663520812989,
      "mean_token_accuracy": 0.8202776715159417,
      "num_tokens": 9083822.0,
      "step": 1920
    },
    {
      "epoch": 0.6579509527471589,
      "eval_entropy": 0.6668643578886986,
      "eval_loss": 0.6642057299613953,
      "eval_mean_token_accuracy": 0.8225639518350363,
      "eval_num_tokens": 9106403.0,
      "eval_runtime": 32.5658,
      "eval_samples_per_second": 7.861,
      "eval_steps_per_second": 0.983,
      "step": 1925
    },
    {
      "entropy": 0.7865349762141705,
      "epoch": 0.6596599162607878,
      "grad_norm": 0.5823968052864075,
      "learning_rate": 5.4966268645682037e-05,
      "loss": 0.773893404006958,
      "mean_token_accuracy": 0.8043547578155994,
      "num_tokens": 9131386.0,
      "step": 1930
    },
    {
      "entropy": 0.7573930438607931,
      "epoch": 0.6630778432880458,
      "grad_norm": 0.5634140372276306,
      "learning_rate": 5.398067671532554e-05,
      "loss": 0.7383981227874756,
      "mean_token_accuracy": 0.810357116907835,
      "num_tokens": 9181672.0,
      "step": 1940
    },
    {
      "entropy": 0.7444441251456737,
      "epoch": 0.6664957703153037,
      "grad_norm": 0.5634564161300659,
      "learning_rate": 5.30007238974371e-05,
      "loss": 0.715367317199707,
      "mean_token_accuracy": 0.8161345295608043,
      "num_tokens": 9229065.0,
      "step": 1950
    },
    {
      "epoch": 0.6664957703153037,
      "eval_entropy": 0.667938893660903,
      "eval_loss": 0.6625745296478271,
      "eval_mean_token_accuracy": 0.8220944050699472,
      "eval_num_tokens": 9229065.0,
      "eval_runtime": 32.5889,
      "eval_samples_per_second": 7.855,
      "eval_steps_per_second": 0.982,
      "step": 1950
    },
    {
      "entropy": 0.7767221700400114,
      "epoch": 0.6699136973425618,
      "grad_norm": 0.5166037678718567,
      "learning_rate": 5.202653027340341e-05,
      "loss": 0.765186071395874,
      "mean_token_accuracy": 0.8100834608078002,
      "num_tokens": 9274928.0,
      "step": 1960
    },
    {
      "entropy": 0.7660458698868752,
      "epoch": 0.6733316243698197,
      "grad_norm": 0.5478764176368713,
      "learning_rate": 5.105821521889147e-05,
      "loss": 0.7520006656646728,
      "mean_token_accuracy": 0.8094162195920944,
      "num_tokens": 9325169.0,
      "step": 1970
    },
    {
      "epoch": 0.6750405878834487,
      "eval_entropy": 0.6612209193408489,
      "eval_loss": 0.6612023115158081,
      "eval_mean_token_accuracy": 0.8231339156627655,
      "eval_num_tokens": 9349091.0,
      "eval_runtime": 32.6144,
      "eval_samples_per_second": 7.849,
      "eval_steps_per_second": 0.981,
      "step": 1975
    },
    {
      "entropy": 0.785111240670085,
      "epoch": 0.6767495513970777,
      "grad_norm": 0.721108615398407,
      "learning_rate": 5.009589738922052e-05,
      "loss": 0.7930411338806153,
      "mean_token_accuracy": 0.8013198770582676,
      "num_tokens": 9369995.0,
      "step": 1980
    },
    {
      "entropy": 0.7857901852577924,
      "epoch": 0.6801674784243357,
      "grad_norm": 0.6027366518974304,
      "learning_rate": 4.9139694704822326e-05,
      "loss": 0.7535711765289307,
      "mean_token_accuracy": 0.8079273730516434,
      "num_tokens": 9417148.0,
      "step": 1990
    },
    {
      "entropy": 0.7289311088621616,
      "epoch": 0.6835854054515936,
      "grad_norm": 0.552121102809906,
      "learning_rate": 4.8189724336791456e-05,
      "loss": 0.6994921207427979,
      "mean_token_accuracy": 0.8187321200966835,
      "num_tokens": 9468520.0,
      "step": 2000
    },
    {
      "epoch": 0.6835854054515936,
      "eval_entropy": 0.6680661998689175,
      "eval_loss": 0.6595595479011536,
      "eval_mean_token_accuracy": 0.8238746467977762,
      "eval_num_tokens": 9468520.0,
      "eval_runtime": 32.5862,
      "eval_samples_per_second": 7.856,
      "eval_steps_per_second": 0.982,
      "step": 2000
    },
    {
      "entropy": 0.7077226992696524,
      "epoch": 0.6870033324788516,
      "grad_norm": 0.5503117442131042,
      "learning_rate": 4.7246102692527314e-05,
      "loss": 0.6956156730651856,
      "mean_token_accuracy": 0.8213571734726429,
      "num_tokens": 9520166.0,
      "step": 2010
    },
    {
      "entropy": 0.7663476031273604,
      "epoch": 0.6904212595061096,
      "grad_norm": 0.6200604438781738,
      "learning_rate": 4.630894540146994e-05,
      "loss": 0.7606002807617187,
      "mean_token_accuracy": 0.8068253338336945,
      "num_tokens": 9570547.0,
      "step": 2020
    },
    {
      "epoch": 0.6921302230197385,
      "eval_entropy": 0.6676463298499584,
      "eval_loss": 0.6581661701202393,
      "eval_mean_token_accuracy": 0.8241461012512445,
      "eval_num_tokens": 9593672.0,
      "eval_runtime": 32.5143,
      "eval_samples_per_second": 7.873,
      "eval_steps_per_second": 0.984,
      "step": 2025
    },
    {
      "entropy": 0.7571568485349417,
      "epoch": 0.6938391865333675,
      "grad_norm": 0.6294745802879333,
      "learning_rate": 4.537836730093077e-05,
      "loss": 0.7340333461761475,
      "mean_token_accuracy": 0.8121012486517429,
      "num_tokens": 9618466.0,
      "step": 2030
    },
    {
      "entropy": 0.7618828386068344,
      "epoch": 0.6972571135606255,
      "grad_norm": 0.5088632106781006,
      "learning_rate": 4.4454482422021e-05,
      "loss": 0.744665002822876,
      "mean_token_accuracy": 0.8096645012497902,
      "num_tokens": 9667228.0,
      "step": 2040
    },
    {
      "entropy": 0.7383288767188787,
      "epoch": 0.7006750405878834,
      "grad_norm": 0.5766125321388245,
      "learning_rate": 4.3537403975678315e-05,
      "loss": 0.7258750915527343,
      "mean_token_accuracy": 0.8161473520100117,
      "num_tokens": 9713714.0,
      "step": 2050
    },
    {
      "epoch": 0.7006750405878834,
      "eval_entropy": 0.6651978325098753,
      "eval_loss": 0.6558593511581421,
      "eval_mean_token_accuracy": 0.8251358140259981,
      "eval_num_tokens": 9713714.0,
      "eval_runtime": 32.5025,
      "eval_samples_per_second": 7.876,
      "eval_steps_per_second": 0.985,
      "step": 2050
    },
    {
      "entropy": 0.7635805677622557,
      "epoch": 0.7040929676151414,
      "grad_norm": 0.6300836801528931,
      "learning_rate": 4.262724433879427e-05,
      "loss": 0.7414310455322266,
      "mean_token_accuracy": 0.8105963818728924,
      "num_tokens": 9760727.0,
      "step": 2060
    },
    {
      "entropy": 0.7665166191756725,
      "epoch": 0.7075108946423994,
      "grad_norm": 0.5252206921577454,
      "learning_rate": 4.172411504044375e-05,
      "loss": 0.7504849433898926,
      "mean_token_accuracy": 0.8100805446505547,
      "num_tokens": 9808108.0,
      "step": 2070
    },
    {
      "epoch": 0.7092198581560284,
      "eval_entropy": 0.6657087150961161,
      "eval_loss": 0.6561708450317383,
      "eval_mean_token_accuracy": 0.8234847020357847,
      "eval_num_tokens": 9833878.0,
      "eval_runtime": 32.4853,
      "eval_samples_per_second": 7.88,
      "eval_steps_per_second": 0.985,
      "step": 2075
    },
    {
      "entropy": 0.7739077024161816,
      "epoch": 0.7109288216696573,
      "grad_norm": 0.7093223333358765,
      "learning_rate": 4.0828126748218654e-05,
      "loss": 0.753156042098999,
      "mean_token_accuracy": 0.8082724556326866,
      "num_tokens": 9855760.0,
      "step": 2080
    },
    {
      "entropy": 0.7730333257466555,
      "epoch": 0.7143467486969153,
      "grad_norm": 0.6004301309585571,
      "learning_rate": 3.993938925466674e-05,
      "loss": 0.7471571445465088,
      "mean_token_accuracy": 0.8072588540613651,
      "num_tokens": 9904182.0,
      "step": 2090
    },
    {
      "entropy": 0.7263076733797789,
      "epoch": 0.7177646757241732,
      "grad_norm": 0.5550140738487244,
      "learning_rate": 3.9058011463837974e-05,
      "loss": 0.7263860702514648,
      "mean_token_accuracy": 0.8129746846854686,
      "num_tokens": 9954352.0,
      "step": 2100
    },
    {
      "epoch": 0.7177646757241732,
      "eval_entropy": 0.6623709443956614,
      "eval_loss": 0.6543427109718323,
      "eval_mean_token_accuracy": 0.8249485101550817,
      "eval_num_tokens": 9954352.0,
      "eval_runtime": 32.6266,
      "eval_samples_per_second": 7.846,
      "eval_steps_per_second": 0.981,
      "step": 2100
    },
    {
      "entropy": 0.7489235173910856,
      "epoch": 0.7211826027514313,
      "grad_norm": 0.720840573310852,
      "learning_rate": 3.8184101377939476e-05,
      "loss": 0.7226673603057862,
      "mean_token_accuracy": 0.8142540246248245,
      "num_tokens": 10001696.0,
      "step": 2110
    },
    {
      "entropy": 0.689397944882512,
      "epoch": 0.7246005297786893,
      "grad_norm": 0.5519832968711853,
      "learning_rate": 3.73177660841015e-05,
      "loss": 0.6655806064605713,
      "mean_token_accuracy": 0.8236007198691369,
      "num_tokens": 10051435.0,
      "step": 2120
    },
    {
      "epoch": 0.7263094932923182,
      "eval_entropy": 0.6580320056527853,
      "eval_loss": 0.6523653864860535,
      "eval_mean_token_accuracy": 0.825190607458353,
      "eval_num_tokens": 10076452.0,
      "eval_runtime": 32.5452,
      "eval_samples_per_second": 7.866,
      "eval_steps_per_second": 0.983,
      "step": 2125
    },
    {
      "entropy": 0.7550150021910668,
      "epoch": 0.7280184568059472,
      "grad_norm": 0.609462320804596,
      "learning_rate": 3.645911174125479e-05,
      "loss": 0.7256326675415039,
      "mean_token_accuracy": 0.8133809432387352,
      "num_tokens": 10097596.0,
      "step": 2130
    },
    {
      "entropy": 0.7556207679212094,
      "epoch": 0.7314363838332052,
      "grad_norm": 0.509903609752655,
      "learning_rate": 3.560824356712238e-05,
      "loss": 0.7371315002441406,
      "mean_token_accuracy": 0.8115937277674675,
      "num_tokens": 10143130.0,
      "step": 2140
    },
    {
      "entropy": 0.7149132907390594,
      "epoch": 0.7348543108604632,
      "grad_norm": 0.6229613423347473,
      "learning_rate": 3.47652658253263e-05,
      "loss": 0.7009588241577148,
      "mean_token_accuracy": 0.8162322334945202,
      "num_tokens": 10192913.0,
      "step": 2150
    },
    {
      "epoch": 0.7348543108604632,
      "eval_entropy": 0.6570977903902531,
      "eval_loss": 0.6511036157608032,
      "eval_mean_token_accuracy": 0.8255328629165888,
      "eval_num_tokens": 10192913.0,
      "eval_runtime": 32.5048,
      "eval_samples_per_second": 7.876,
      "eval_steps_per_second": 0.984,
      "step": 2150
    },
    {
      "entropy": 0.7812049750238657,
      "epoch": 0.7382722378877211,
      "grad_norm": 0.682597279548645,
      "learning_rate": 3.393028181261142e-05,
      "loss": 0.7646748065948487,
      "mean_token_accuracy": 0.8081333816051484,
      "num_tokens": 10240506.0,
      "step": 2160
    },
    {
      "entropy": 0.7046744357794523,
      "epoch": 0.7416901649149791,
      "grad_norm": 0.6965001225471497,
      "learning_rate": 3.3103393846187445e-05,
      "loss": 0.6830405712127685,
      "mean_token_accuracy": 0.8225828550755978,
      "num_tokens": 10289324.0,
      "step": 2170
    },
    {
      "epoch": 0.743399128428608,
      "eval_entropy": 0.6515689790248871,
      "eval_loss": 0.6498388051986694,
      "eval_mean_token_accuracy": 0.8258727006614208,
      "eval_num_tokens": 10313829.0,
      "eval_runtime": 32.5295,
      "eval_samples_per_second": 7.87,
      "eval_steps_per_second": 0.984,
      "step": 2175
    },
    {
      "entropy": 0.7740226969122886,
      "epoch": 0.745108091942237,
      "grad_norm": 0.7107183933258057,
      "learning_rate": 3.228470325119164e-05,
      "loss": 0.7719791412353516,
      "mean_token_accuracy": 0.8062848366796971,
      "num_tokens": 10337899.0,
      "step": 2180
    },
    {
      "entropy": 0.6804906763136387,
      "epoch": 0.748526018969495,
      "grad_norm": 0.8012505173683167,
      "learning_rate": 3.147431034827208e-05,
      "loss": 0.6739416599273682,
      "mean_token_accuracy": 0.8224732950329781,
      "num_tokens": 10383023.0,
      "step": 2190
    },
    {
      "entropy": 0.7387298934161663,
      "epoch": 0.751943945996753,
      "grad_norm": 0.5948705077171326,
      "learning_rate": 3.067231444129507e-05,
      "loss": 0.7141981601715088,
      "mean_token_accuracy": 0.8123697496950626,
      "num_tokens": 10436562.0,
      "step": 2200
    },
    {
      "epoch": 0.751943945996753,
      "eval_entropy": 0.6572407782077789,
      "eval_loss": 0.6485446095466614,
      "eval_mean_token_accuracy": 0.8260926008224487,
      "eval_num_tokens": 10436562.0,
      "eval_runtime": 32.5179,
      "eval_samples_per_second": 7.873,
      "eval_steps_per_second": 0.984,
      "step": 2200
    },
    {
      "entropy": 0.7102626815438271,
      "epoch": 0.7553618730240109,
      "grad_norm": 0.45274314284324646,
      "learning_rate": 2.9878813805176253e-05,
      "loss": 0.6939650535583496,
      "mean_token_accuracy": 0.821110887080431,
      "num_tokens": 10484097.0,
      "step": 2210
    },
    {
      "entropy": 0.6937402199953795,
      "epoch": 0.7587798000512689,
      "grad_norm": 0.5389388203620911,
      "learning_rate": 2.9093905673838606e-05,
      "loss": 0.6805085659027099,
      "mean_token_accuracy": 0.8231124863028526,
      "num_tokens": 10531946.0,
      "step": 2220
    },
    {
      "epoch": 0.7604887635648979,
      "eval_entropy": 0.6562957186251879,
      "eval_loss": 0.6482586860656738,
      "eval_mean_token_accuracy": 0.8258698955178261,
      "eval_num_tokens": 10556164.0,
      "eval_runtime": 32.5336,
      "eval_samples_per_second": 7.869,
      "eval_steps_per_second": 0.984,
      "step": 2225
    },
    {
      "entropy": 0.7634148493409156,
      "epoch": 0.7621977270785268,
      "grad_norm": 0.5631259083747864,
      "learning_rate": 2.8317686228297192e-05,
      "loss": 0.7458550453186035,
      "mean_token_accuracy": 0.8041129514575005,
      "num_tokens": 10581095.0,
      "step": 2230
    },
    {
      "entropy": 0.7783615037798881,
      "epoch": 0.7656156541057848,
      "grad_norm": 0.4985562562942505,
      "learning_rate": 2.7550250584873637e-05,
      "loss": 0.7557303428649902,
      "mean_token_accuracy": 0.8075458481907845,
      "num_tokens": 10629253.0,
      "step": 2240
    },
    {
      "entropy": 0.7126111596822738,
      "epoch": 0.7690335811330428,
      "grad_norm": 0.5650884509086609,
      "learning_rate": 2.6791692783540633e-05,
      "loss": 0.6887269973754883,
      "mean_token_accuracy": 0.821088171005249,
      "num_tokens": 10678619.0,
      "step": 2250
    },
    {
      "epoch": 0.7690335811330428,
      "eval_entropy": 0.653578756377101,
      "eval_loss": 0.6471341252326965,
      "eval_mean_token_accuracy": 0.8266212977468967,
      "eval_num_tokens": 10678619.0,
      "eval_runtime": 32.5421,
      "eval_samples_per_second": 7.867,
      "eval_steps_per_second": 0.983,
      "step": 2250
    },
    {
      "entropy": 0.6648651383817196,
      "epoch": 0.7724515081603007,
      "grad_norm": 0.48509320616722107,
      "learning_rate": 2.604210577639855e-05,
      "loss": 0.6434546947479248,
      "mean_token_accuracy": 0.827840019762516,
      "num_tokens": 10728226.0,
      "step": 2260
    },
    {
      "entropy": 0.7198113534599543,
      "epoch": 0.7758694351875588,
      "grad_norm": 0.8218326568603516,
      "learning_rate": 2.5301581416285147e-05,
      "loss": 0.7089159965515137,
      "mean_token_accuracy": 0.8145917497575283,
      "num_tokens": 10778245.0,
      "step": 2270
    },
    {
      "epoch": 0.7775783987011877,
      "eval_entropy": 0.652136854827404,
      "eval_loss": 0.6467763185501099,
      "eval_mean_token_accuracy": 0.8270386923104525,
      "eval_num_tokens": 10801879.0,
      "eval_runtime": 32.5386,
      "eval_samples_per_second": 7.868,
      "eval_steps_per_second": 0.983,
      "step": 2275
    },
    {
      "entropy": 0.7346194788813591,
      "epoch": 0.7792873622148168,
      "grad_norm": 0.5933800935745239,
      "learning_rate": 2.457021044552028e-05,
      "loss": 0.7139677047729492,
      "mean_token_accuracy": 0.8131411246955395,
      "num_tokens": 10827578.0,
      "step": 2280
    },
    {
      "entropy": 0.756122961640358,
      "epoch": 0.7827052892420747,
      "grad_norm": 0.6672066450119019,
      "learning_rate": 2.3848082484786426e-05,
      "loss": 0.7328310489654541,
      "mean_token_accuracy": 0.8131688497960567,
      "num_tokens": 10873441.0,
      "step": 2290
    },
    {
      "entropy": 0.6612967252731323,
      "epoch": 0.7861232162693327,
      "grad_norm": 0.5208282470703125,
      "learning_rate": 2.3135286022146785e-05,
      "loss": 0.6452344417572021,
      "mean_token_accuracy": 0.8288177840411663,
      "num_tokens": 10919727.0,
      "step": 2300
    },
    {
      "epoch": 0.7861232162693327,
      "eval_entropy": 0.6515894420444965,
      "eval_loss": 0.6458343267440796,
      "eval_mean_token_accuracy": 0.8261187877506018,
      "eval_num_tokens": 10919727.0,
      "eval_runtime": 32.562,
      "eval_samples_per_second": 7.862,
      "eval_steps_per_second": 0.983,
      "step": 2300
    },
    {
      "entropy": 0.7304637894034386,
      "epoch": 0.7895411432965906,
      "grad_norm": 0.738002359867096,
      "learning_rate": 2.243190840220214e-05,
      "loss": 0.7215014457702636,
      "mean_token_accuracy": 0.8141652099788189,
      "num_tokens": 10969951.0,
      "step": 2310
    },
    {
      "entropy": 0.7662419803440571,
      "epoch": 0.7929590703238486,
      "grad_norm": 0.5868961811065674,
      "learning_rate": 2.173803581538775e-05,
      "loss": 0.7481863021850585,
      "mean_token_accuracy": 0.8079202443361282,
      "num_tokens": 11019244.0,
      "step": 2320
    },
    {
      "epoch": 0.7946680338374775,
      "eval_entropy": 0.650886744260788,
      "eval_loss": 0.6445406079292297,
      "eval_mean_token_accuracy": 0.8274833615869284,
      "eval_num_tokens": 11040646.0,
      "eval_runtime": 32.5078,
      "eval_samples_per_second": 7.875,
      "eval_steps_per_second": 0.984,
      "step": 2325
    },
    {
      "entropy": 0.702366477996111,
      "epoch": 0.7963769973511066,
      "grad_norm": 0.5595235824584961,
      "learning_rate": 2.1053753287411892e-05,
      "loss": 0.6811541080474853,
      "mean_token_accuracy": 0.822179126739502,
      "num_tokens": 11066388.0,
      "step": 2330
    },
    {
      "entropy": 0.75776182487607,
      "epoch": 0.7997949243783645,
      "grad_norm": 0.669806718826294,
      "learning_rate": 2.037914466883688e-05,
      "loss": 0.7357389450073242,
      "mean_token_accuracy": 0.811240018159151,
      "num_tokens": 11111032.0,
      "step": 2340
    },
    {
      "entropy": 0.6907448621466756,
      "epoch": 0.8032128514056225,
      "grad_norm": 0.5382040739059448,
      "learning_rate": 1.9714292624804275e-05,
      "loss": 0.6850921630859375,
      "mean_token_accuracy": 0.8219761483371257,
      "num_tokens": 11162497.0,
      "step": 2350
    },
    {
      "epoch": 0.8032128514056225,
      "eval_entropy": 0.6502471938729286,
      "eval_loss": 0.6438165903091431,
      "eval_mean_token_accuracy": 0.8278086725622416,
      "eval_num_tokens": 11162497.0,
      "eval_runtime": 32.5785,
      "eval_samples_per_second": 7.858,
      "eval_steps_per_second": 0.982,
      "step": 2350
    },
    {
      "entropy": 0.6725162204355002,
      "epoch": 0.8066307784328804,
      "grad_norm": 0.47493892908096313,
      "learning_rate": 1.905927862490512e-05,
      "loss": 0.6554149627685547,
      "mean_token_accuracy": 0.8262991026043892,
      "num_tokens": 11211623.0,
      "step": 2360
    },
    {
      "entropy": 0.7101132038980722,
      "epoch": 0.8100487054601384,
      "grad_norm": 0.5834583044052124,
      "learning_rate": 1.8414182933197023e-05,
      "loss": 0.6969385147094727,
      "mean_token_accuracy": 0.8214772522449494,
      "num_tokens": 11259236.0,
      "step": 2370
    },
    {
      "epoch": 0.8117576689737674,
      "eval_entropy": 0.6499252859503031,
      "eval_loss": 0.6433589458465576,
      "eval_mean_token_accuracy": 0.8272677790373564,
      "eval_num_tokens": 11281620.0,
      "eval_runtime": 32.6124,
      "eval_samples_per_second": 7.85,
      "eval_steps_per_second": 0.981,
      "step": 2375
    },
    {
      "entropy": 0.7414477836340666,
      "epoch": 0.8134666324873964,
      "grad_norm": 0.5209059119224548,
      "learning_rate": 1.7779084598368635e-05,
      "loss": 0.7268849372863769,
      "mean_token_accuracy": 0.8158441334962845,
      "num_tokens": 11306119.0,
      "step": 2380
    },
    {
      "entropy": 0.7430920496582984,
      "epoch": 0.8168845595146543,
      "grad_norm": 0.6851968169212341,
      "learning_rate": 1.7154061444053237e-05,
      "loss": 0.7404438972473144,
      "mean_token_accuracy": 0.8101868823170661,
      "num_tokens": 11355199.0,
      "step": 2390
    },
    {
      "entropy": 0.7235891550779343,
      "epoch": 0.8203024865419123,
      "grad_norm": 0.5767418742179871,
      "learning_rate": 1.6539190059292288e-05,
      "loss": 0.7079296112060547,
      "mean_token_accuracy": 0.8172104585915804,
      "num_tokens": 11404488.0,
      "step": 2400
    },
    {
      "epoch": 0.8203024865419123,
      "eval_entropy": 0.6570469662547112,
      "eval_loss": 0.6428108215332031,
      "eval_mean_token_accuracy": 0.8273015934973955,
      "eval_num_tokens": 11404488.0,
      "eval_runtime": 32.6527,
      "eval_samples_per_second": 7.84,
      "eval_steps_per_second": 0.98,
      "step": 2400
    },
    {
      "entropy": 0.7659241635352373,
      "epoch": 0.8237204135691703,
      "grad_norm": 0.5965316295623779,
      "learning_rate": 1.5934545789150623e-05,
      "loss": 0.7549265861511231,
      "mean_token_accuracy": 0.8072838112711906,
      "num_tokens": 11453374.0,
      "step": 2410
    },
    {
      "entropy": 0.7300883550196886,
      "epoch": 0.8271383405964282,
      "grad_norm": 0.5207374095916748,
      "learning_rate": 1.5340202725483487e-05,
      "loss": 0.7029576778411866,
      "mean_token_accuracy": 0.8164526544511318,
      "num_tokens": 11501777.0,
      "step": 2420
    },
    {
      "epoch": 0.8288473041100572,
      "eval_entropy": 0.6553569138050079,
      "eval_loss": 0.641933798789978,
      "eval_mean_token_accuracy": 0.8275820016860962,
      "eval_num_tokens": 11527261.0,
      "eval_runtime": 32.6148,
      "eval_samples_per_second": 7.849,
      "eval_steps_per_second": 0.981,
      "step": 2425
    },
    {
      "entropy": 0.7407207667827607,
      "epoch": 0.8305562676236863,
      "grad_norm": 0.49078667163848877,
      "learning_rate": 1.4756233697857746e-05,
      "loss": 0.71334810256958,
      "mean_token_accuracy": 0.8155447125434876,
      "num_tokens": 11551134.0,
      "step": 2430
    },
    {
      "entropy": 0.687175927311182,
      "epoch": 0.8339741946509442,
      "grad_norm": 0.6775448322296143,
      "learning_rate": 1.4182710264627341e-05,
      "loss": 0.6605133533477783,
      "mean_token_accuracy": 0.8263660438358784,
      "num_tokens": 11598933.0,
      "step": 2440
    },
    {
      "entropy": 0.7657349318265915,
      "epoch": 0.8373921216782022,
      "grad_norm": 0.7116291522979736,
      "learning_rate": 1.3619702704164783e-05,
      "loss": 0.7508980751037597,
      "mean_token_accuracy": 0.8105187758803367,
      "num_tokens": 11649337.0,
      "step": 2450
    },
    {
      "epoch": 0.8373921216782022,
      "eval_entropy": 0.6494570691138506,
      "eval_loss": 0.6415493488311768,
      "eval_mean_token_accuracy": 0.827751325443387,
      "eval_num_tokens": 11649337.0,
      "eval_runtime": 32.5345,
      "eval_samples_per_second": 7.869,
      "eval_steps_per_second": 0.984,
      "step": 2450
    },
    {
      "entropy": 0.7346297159790993,
      "epoch": 0.8408100487054602,
      "grad_norm": 0.7070845365524292,
      "learning_rate": 1.3067280006249261e-05,
      "loss": 0.72440505027771,
      "mean_token_accuracy": 0.812112944573164,
      "num_tokens": 11697361.0,
      "step": 2460
    },
    {
      "entropy": 0.6898879475891591,
      "epoch": 0.8442279757327181,
      "grad_norm": 0.5906277298927307,
      "learning_rate": 1.2525509863612916e-05,
      "loss": 0.6653994560241699,
      "mean_token_accuracy": 0.824587044864893,
      "num_tokens": 11743307.0,
      "step": 2470
    },
    {
      "epoch": 0.845936939246347,
      "eval_entropy": 0.6489466484636068,
      "eval_loss": 0.640968918800354,
      "eval_mean_token_accuracy": 0.8277594074606895,
      "eval_num_tokens": 11767237.0,
      "eval_runtime": 32.5474,
      "eval_samples_per_second": 7.865,
      "eval_steps_per_second": 0.983,
      "step": 2475
    },
    {
      "entropy": 0.7420150343328714,
      "epoch": 0.8476459027599761,
      "grad_norm": 0.6130893230438232,
      "learning_rate": 1.1994458663645835e-05,
      "loss": 0.7171807289123535,
      "mean_token_accuracy": 0.8172306433320046,
      "num_tokens": 11788968.0,
      "step": 2480
    },
    {
      "entropy": 0.6987740892916918,
      "epoch": 0.851063829787234,
      "grad_norm": 0.5828075408935547,
      "learning_rate": 1.1474191480261098e-05,
      "loss": 0.6787797451019287,
      "mean_token_accuracy": 0.8233332350850106,
      "num_tokens": 11839109.0,
      "step": 2490
    },
    {
      "entropy": 0.7257452422752977,
      "epoch": 0.854481756814492,
      "grad_norm": 0.5111470818519592,
      "learning_rate": 1.096477206592067e-05,
      "loss": 0.7015585422515869,
      "mean_token_accuracy": 0.8180849917232991,
      "num_tokens": 11889064.0,
      "step": 2500
    },
    {
      "epoch": 0.854481756814492,
      "eval_entropy": 0.6452019456773996,
      "eval_loss": 0.6404462456703186,
      "eval_mean_token_accuracy": 0.827375078573823,
      "eval_num_tokens": 11889064.0,
      "eval_runtime": 32.7054,
      "eval_samples_per_second": 7.827,
      "eval_steps_per_second": 0.978,
      "step": 2500
    },
    {
      "entropy": 0.7539002969861031,
      "epoch": 0.85789968384175,
      "grad_norm": 0.63209068775177,
      "learning_rate": 1.046626284382356e-05,
      "loss": 0.7466847419738769,
      "mean_token_accuracy": 0.8092853993177413,
      "num_tokens": 11935729.0,
      "step": 2510
    },
    {
      "entropy": 0.6735727831721305,
      "epoch": 0.8613176108690079,
      "grad_norm": 0.48319295048713684,
      "learning_rate": 9.978724900256265e-06,
      "loss": 0.6552696228027344,
      "mean_token_accuracy": 0.8258471079170704,
      "num_tokens": 11986975.0,
      "step": 2520
    },
    {
      "epoch": 0.863026574382637,
      "eval_entropy": 0.6486122142523527,
      "eval_loss": 0.6401728987693787,
      "eval_mean_token_accuracy": 0.8278232961893082,
      "eval_num_tokens": 12010780.0,
      "eval_runtime": 32.6273,
      "eval_samples_per_second": 7.846,
      "eval_steps_per_second": 0.981,
      "step": 2525
    },
    {
      "entropy": 0.7377024326473475,
      "epoch": 0.8647355378962659,
      "grad_norm": 0.6024383306503296,
      "learning_rate": 9.502217977107642e-06,
      "loss": 0.725794792175293,
      "mean_token_accuracy": 0.8168808005750179,
      "num_tokens": 12033819.0,
      "step": 2530
    },
    {
      "entropy": 0.6924330729991197,
      "epoch": 0.8681534649235239,
      "grad_norm": 0.5526846051216125,
      "learning_rate": 9.036800464548157e-06,
      "loss": 0.6714127540588379,
      "mean_token_accuracy": 0.8240827329456806,
      "num_tokens": 12082307.0,
      "step": 2540
    },
    {
      "entropy": 0.7305394794791937,
      "epoch": 0.8715713919507818,
      "grad_norm": 0.4549733102321625,
      "learning_rate": 8.582529393874927e-06,
      "loss": 0.7102427005767822,
      "mean_token_accuracy": 0.8165039896965027,
      "num_tokens": 12130018.0,
      "step": 2550
    },
    {
      "epoch": 0.8715713919507818,
      "eval_entropy": 0.6483217645436525,
      "eval_loss": 0.6397960782051086,
      "eval_mean_token_accuracy": 0.828344039618969,
      "eval_num_tokens": 12130018.0,
      "eval_runtime": 32.5906,
      "eval_samples_per_second": 7.855,
      "eval_steps_per_second": 0.982,
      "step": 2550
    },
    {
      "entropy": 0.6853050304576754,
      "epoch": 0.8749893189780398,
      "grad_norm": 0.5434498190879822,
      "learning_rate": 8.139460430523104e-06,
      "loss": 0.6624485492706299,
      "mean_token_accuracy": 0.8262315645813942,
      "num_tokens": 12178126.0,
      "step": 2560
    },
    {
      "entropy": 0.6818725261837244,
      "epoch": 0.8784072460052977,
      "grad_norm": 0.642160952091217,
      "learning_rate": 7.707647867244928e-06,
      "loss": 0.6672566890716553,
      "mean_token_accuracy": 0.8246341213583946,
      "num_tokens": 12227181.0,
      "step": 2570
    },
    {
      "epoch": 0.8801162095189268,
      "eval_entropy": 0.6478469502180815,
      "eval_loss": 0.6394245624542236,
      "eval_mean_token_accuracy": 0.8280942086130381,
      "eval_num_tokens": 12251947.0,
      "eval_runtime": 32.5909,
      "eval_samples_per_second": 7.855,
      "eval_steps_per_second": 0.982,
      "step": 2575
    },
    {
      "entropy": 0.6886890552937984,
      "epoch": 0.8818251730325558,
      "grad_norm": 0.472001850605011,
      "learning_rate": 7.287144617456654e-06,
      "loss": 0.6606461524963378,
      "mean_token_accuracy": 0.8270013853907585,
      "num_tokens": 12278219.0,
      "step": 2580
    },
    {
      "entropy": 0.7449534449726343,
      "epoch": 0.8852431000598138,
      "grad_norm": 0.4806518256664276,
      "learning_rate": 6.8780022087547305e-06,
      "loss": 0.7350006580352784,
      "mean_token_accuracy": 0.8134893275797367,
      "num_tokens": 12324590.0,
      "step": 2590
    },
    {
      "entropy": 0.6892618909478188,
      "epoch": 0.8886610270870717,
      "grad_norm": 0.8003678917884827,
      "learning_rate": 6.4802707766016825e-06,
      "loss": 0.6735070228576661,
      "mean_token_accuracy": 0.8235712304711342,
      "num_tokens": 12370465.0,
      "step": 2600
    },
    {
      "epoch": 0.8886610270870717,
      "eval_entropy": 0.6483608782291412,
      "eval_loss": 0.6388651132583618,
      "eval_mean_token_accuracy": 0.8282469250261784,
      "eval_num_tokens": 12370465.0,
      "eval_runtime": 32.6321,
      "eval_samples_per_second": 7.845,
      "eval_steps_per_second": 0.981,
      "step": 2600
    },
    {
      "entropy": 0.7182144347578288,
      "epoch": 0.8920789541143297,
      "grad_norm": 0.5940407514572144,
      "learning_rate": 6.09399905818262e-06,
      "loss": 0.6957920551300049,
      "mean_token_accuracy": 0.8216035731136799,
      "num_tokens": 12415923.0,
      "step": 2610
    },
    {
      "entropy": 0.7306089602410794,
      "epoch": 0.8954968811415877,
      "grad_norm": 0.5374943614006042,
      "learning_rate": 5.7192343864331255e-06,
      "loss": 0.7158644199371338,
      "mean_token_accuracy": 0.8140112727880477,
      "num_tokens": 12464233.0,
      "step": 2620
    },
    {
      "epoch": 0.8972058446552166,
      "eval_entropy": 0.6480556409806013,
      "eval_loss": 0.6388079524040222,
      "eval_mean_token_accuracy": 0.8276491928845644,
      "eval_num_tokens": 12490627.0,
      "eval_runtime": 32.6018,
      "eval_samples_per_second": 7.852,
      "eval_steps_per_second": 0.982,
      "step": 2625
    },
    {
      "entropy": 0.7090828090906143,
      "epoch": 0.8989148081688456,
      "grad_norm": 0.6297702193260193,
      "learning_rate": 5.3560226842390596e-06,
      "loss": 0.6868664264678955,
      "mean_token_accuracy": 0.8201880976557732,
      "num_tokens": 12518591.0,
      "step": 2630
    },
    {
      "entropy": 0.7153299488127232,
      "epoch": 0.9023327351961036,
      "grad_norm": 0.552191436290741,
      "learning_rate": 5.004408458809473e-06,
      "loss": 0.6930214405059815,
      "mean_token_accuracy": 0.8172111846506596,
      "num_tokens": 12569948.0,
      "step": 2640
    },
    {
      "entropy": 0.7137423895299435,
      "epoch": 0.9057506622233615,
      "grad_norm": 0.5232876539230347,
      "learning_rate": 4.664434796222528e-06,
      "loss": 0.7068160533905029,
      "mean_token_accuracy": 0.8190832555294036,
      "num_tokens": 12620238.0,
      "step": 2650
    },
    {
      "epoch": 0.9057506622233615,
      "eval_entropy": 0.6479176562279463,
      "eval_loss": 0.6384446024894714,
      "eval_mean_token_accuracy": 0.827843364328146,
      "eval_num_tokens": 12620238.0,
      "eval_runtime": 32.5681,
      "eval_samples_per_second": 7.86,
      "eval_steps_per_second": 0.983,
      "step": 2650
    },
    {
      "entropy": 0.6815611261874437,
      "epoch": 0.9091685892506195,
      "grad_norm": 0.7127875089645386,
      "learning_rate": 4.336143356146028e-06,
      "loss": 0.6582761764526367,
      "mean_token_accuracy": 0.8260635226964951,
      "num_tokens": 12669309.0,
      "step": 2660
    },
    {
      "entropy": 0.7074244495481252,
      "epoch": 0.9125865162778775,
      "grad_norm": 0.5414922833442688,
      "learning_rate": 4.019574366732426e-06,
      "loss": 0.6834642887115479,
      "mean_token_accuracy": 0.8240615390241146,
      "num_tokens": 12716960.0,
      "step": 2670
    },
    {
      "epoch": 0.9142954797915065,
      "eval_entropy": 0.6465785261243582,
      "eval_loss": 0.6383459568023682,
      "eval_mean_token_accuracy": 0.8281074147671461,
      "eval_num_tokens": 12737104.0,
      "eval_runtime": 32.6379,
      "eval_samples_per_second": 7.844,
      "eval_steps_per_second": 0.98,
      "step": 2675
    },
    {
      "entropy": 0.737743879109621,
      "epoch": 0.9160044433051354,
      "grad_norm": 0.7617110013961792,
      "learning_rate": 3.7147666196893693e-06,
      "loss": 0.72126145362854,
      "mean_token_accuracy": 0.8160571806132794,
      "num_tokens": 12760793.0,
      "step": 2680
    },
    {
      "entropy": 0.7016262900084257,
      "epoch": 0.9194223703323934,
      "grad_norm": 0.508277177810669,
      "learning_rate": 3.421757465526243e-06,
      "loss": 0.6778374671936035,
      "mean_token_accuracy": 0.8244507744908333,
      "num_tokens": 12809918.0,
      "step": 2690
    },
    {
      "entropy": 0.7006717912852765,
      "epoch": 0.9228402973596513,
      "grad_norm": 0.5380056500434875,
      "learning_rate": 3.140582808977355e-06,
      "loss": 0.6844939708709716,
      "mean_token_accuracy": 0.8254795782268047,
      "num_tokens": 12858956.0,
      "step": 2700
    },
    {
      "epoch": 0.9228402973596513,
      "eval_entropy": 0.6461731214076281,
      "eval_loss": 0.6381875872612,
      "eval_mean_token_accuracy": 0.8280313592404127,
      "eval_num_tokens": 12858956.0,
      "eval_runtime": 32.6956,
      "eval_samples_per_second": 7.83,
      "eval_steps_per_second": 0.979,
      "step": 2700
    },
    {
      "entropy": 0.6936084788292647,
      "epoch": 0.9262582243869093,
      "grad_norm": 0.7185409665107727,
      "learning_rate": 2.8712771046021768e-06,
      "loss": 0.6746735572814941,
      "mean_token_accuracy": 0.8217683151364327,
      "num_tokens": 12908472.0,
      "step": 2710
    },
    {
      "entropy": 0.7185607634484767,
      "epoch": 0.9296761514141673,
      "grad_norm": 0.6331292390823364,
      "learning_rate": 2.6138733525633895e-06,
      "loss": 0.700202465057373,
      "mean_token_accuracy": 0.8168184526264668,
      "num_tokens": 12957487.0,
      "step": 2720
    },
    {
      "epoch": 0.9313851149277963,
      "eval_entropy": 0.6467475481331348,
      "eval_loss": 0.6379902958869934,
      "eval_mean_token_accuracy": 0.8283456135541201,
      "eval_num_tokens": 12978759.0,
      "eval_runtime": 32.6564,
      "eval_samples_per_second": 7.839,
      "eval_steps_per_second": 0.98,
      "step": 2725
    },
    {
      "entropy": 0.7706727508455515,
      "epoch": 0.9330940784414252,
      "grad_norm": 0.5760138034820557,
      "learning_rate": 2.3684030945831158e-06,
      "loss": 0.7569189071655273,
      "mean_token_accuracy": 0.8080990195274353,
      "num_tokens": 13002599.0,
      "step": 2730
    },
    {
      "entropy": 0.7010563030838967,
      "epoch": 0.9365120054686833,
      "grad_norm": 0.5545640587806702,
      "learning_rate": 2.134896410077891e-06,
      "loss": 0.6833337783813477,
      "mean_token_accuracy": 0.8220656841993332,
      "num_tokens": 13051935.0,
      "step": 2740
    },
    {
      "entropy": 0.7135491766035557,
      "epoch": 0.9399299324959413,
      "grad_norm": 0.47150468826293945,
      "learning_rate": 1.9133819124727005e-06,
      "loss": 0.6957959651947021,
      "mean_token_accuracy": 0.8195874467492104,
      "num_tokens": 13102587.0,
      "step": 2750
    },
    {
      "epoch": 0.9399299324959413,
      "eval_entropy": 0.6465013269335032,
      "eval_loss": 0.6378064155578613,
      "eval_mean_token_accuracy": 0.8284552879631519,
      "eval_num_tokens": 13102587.0,
      "eval_runtime": 32.5024,
      "eval_samples_per_second": 7.876,
      "eval_steps_per_second": 0.985,
      "step": 2750
    },
    {
      "entropy": 0.718373591452837,
      "epoch": 0.9433478595231992,
      "grad_norm": 0.7023600339889526,
      "learning_rate": 1.7038867456948383e-06,
      "loss": 0.7024884223937988,
      "mean_token_accuracy": 0.8127410352230072,
      "num_tokens": 13151230.0,
      "step": 2760
    },
    {
      "entropy": 0.6812984146177768,
      "epoch": 0.9467657865504572,
      "grad_norm": 0.5730180144309998,
      "learning_rate": 1.5064365808477254e-06,
      "loss": 0.6751970291137696,
      "mean_token_accuracy": 0.825221062451601,
      "num_tokens": 13199667.0,
      "step": 2770
    },
    {
      "epoch": 0.9484747500640861,
      "eval_entropy": 0.6464472748339176,
      "eval_loss": 0.6376645565032959,
      "eval_mean_token_accuracy": 0.8284712266176939,
      "eval_num_tokens": 13222849.0,
      "eval_runtime": 32.5065,
      "eval_samples_per_second": 7.875,
      "eval_steps_per_second": 0.984,
      "step": 2775
    },
    {
      "entropy": 0.6846175715327263,
      "epoch": 0.9501837135777151,
      "grad_norm": 0.4973992705345154,
      "learning_rate": 1.321055613065203e-06,
      "loss": 0.6522854328155517,
      "mean_token_accuracy": 0.8291602872312069,
      "num_tokens": 13246877.0,
      "step": 2780
    },
    {
      "entropy": 0.7449381787329912,
      "epoch": 0.9536016406049731,
      "grad_norm": 0.5896213054656982,
      "learning_rate": 1.1477665585467257e-06,
      "loss": 0.7295150756835938,
      "mean_token_accuracy": 0.8137959085404873,
      "num_tokens": 13291131.0,
      "step": 2790
    },
    {
      "entropy": 0.7088073439896106,
      "epoch": 0.9570195676322311,
      "grad_norm": 0.48523518443107605,
      "learning_rate": 9.865906517738e-07,
      "loss": 0.6960892677307129,
      "mean_token_accuracy": 0.8195622898638248,
      "num_tokens": 13341632.0,
      "step": 2800
    },
    {
      "epoch": 0.9570195676322311,
      "eval_entropy": 0.6466101501137018,
      "eval_loss": 0.6376082897186279,
      "eval_mean_token_accuracy": 0.8284325748682022,
      "eval_num_tokens": 13341632.0,
      "eval_runtime": 32.5248,
      "eval_samples_per_second": 7.871,
      "eval_steps_per_second": 0.984,
      "step": 2800
    },
    {
      "entropy": 0.6897268868982792,
      "epoch": 0.960437494659489,
      "grad_norm": 0.5249438285827637,
      "learning_rate": 8.375476429078544e-07,
      "loss": 0.6755472183227539,
      "mean_token_accuracy": 0.8227805390954017,
      "num_tokens": 13395014.0,
      "step": 2810
    },
    {
      "entropy": 0.7521859399974347,
      "epoch": 0.963855421686747,
      "grad_norm": 0.5275380611419678,
      "learning_rate": 7.006557953702198e-07,
      "loss": 0.7428174018859863,
      "mean_token_accuracy": 0.8101296730339527,
      "num_tokens": 13447963.0,
      "step": 2820
    },
    {
      "epoch": 0.965564385200376,
      "eval_entropy": 0.6466463468968868,
      "eval_loss": 0.6376487612724304,
      "eval_mean_token_accuracy": 0.8285549189895391,
      "eval_num_tokens": 13471931.0,
      "eval_runtime": 32.4665,
      "eval_samples_per_second": 7.885,
      "eval_steps_per_second": 0.986,
      "step": 2825
    },
    {
      "entropy": 0.7089713286608458,
      "epoch": 0.9672733487140049,
      "grad_norm": 0.5778820514678955,
      "learning_rate": 5.759318836040861e-07,
      "loss": 0.6880727291107178,
      "mean_token_accuracy": 0.8235937155783176,
      "num_tokens": 13494899.0,
      "step": 2830
    },
    {
      "entropy": 0.7000694863498211,
      "epoch": 0.9706912757412629,
      "grad_norm": 0.7135331034660339,
      "learning_rate": 4.6339119101902473e-07,
      "loss": 0.6771084308624268,
      "mean_token_accuracy": 0.8221731469035148,
      "num_tokens": 13539972.0,
      "step": 2840
    },
    {
      "entropy": 0.6829410705715417,
      "epoch": 0.9741092027685209,
      "grad_norm": 0.6099685430526733,
      "learning_rate": 3.630475081181861e-07,
      "loss": 0.655033540725708,
      "mean_token_accuracy": 0.8259231664240361,
      "num_tokens": 13587391.0,
      "step": 2850
    },
    {
      "epoch": 0.9741092027685209,
      "eval_entropy": 0.6465384587645531,
      "eval_loss": 0.6375482082366943,
      "eval_mean_token_accuracy": 0.828503267839551,
      "eval_num_tokens": 13587391.0,
      "eval_runtime": 32.5191,
      "eval_samples_per_second": 7.872,
      "eval_steps_per_second": 0.984,
      "step": 2850
    },
    {
      "entropy": 0.7193181496113539,
      "epoch": 0.9775271297957788,
      "grad_norm": 0.6574059724807739,
      "learning_rate": 2.7491313080844074e-07,
      "loss": 0.7081467151641846,
      "mean_token_accuracy": 0.8188973918557167,
      "num_tokens": 13636095.0,
      "step": 2860
    },
    {
      "entropy": 0.7227353582158684,
      "epoch": 0.9809450568230368,
      "grad_norm": 0.6026723980903625,
      "learning_rate": 1.989988588936509e-07,
      "loss": 0.7178384304046631,
      "mean_token_accuracy": 0.8183843664824962,
      "num_tokens": 13685656.0,
      "step": 2870
    },
    {
      "epoch": 0.9826540203366658,
      "eval_entropy": 0.646482452750206,
      "eval_loss": 0.6376971006393433,
      "eval_mean_token_accuracy": 0.8282717633992434,
      "eval_num_tokens": 13710861.0,
      "eval_runtime": 32.5648,
      "eval_samples_per_second": 7.861,
      "eval_steps_per_second": 0.983,
      "step": 2875
    },
    {
      "entropy": 0.6969666961580515,
      "epoch": 0.9843629838502947,
      "grad_norm": 0.6726225018501282,
      "learning_rate": 1.3531399475132934e-07,
      "loss": 0.6770923137664795,
      "mean_token_accuracy": 0.8201491139829159,
      "num_tokens": 13736037.0,
      "step": 2880
    },
    {
      "entropy": 0.7437307950109243,
      "epoch": 0.9877809108775527,
      "grad_norm": 0.6295607089996338,
      "learning_rate": 8.386634219269551e-08,
      "loss": 0.7483133792877197,
      "mean_token_accuracy": 0.8097194038331509,
      "num_tokens": 13784654.0,
      "step": 2890
    },
    {
      "entropy": 0.7080582052469253,
      "epoch": 0.9911988379048108,
      "grad_norm": 0.5197203755378723,
      "learning_rate": 4.466220550641831e-08,
      "loss": 0.6913296222686768,
      "mean_token_accuracy": 0.8222644709050655,
      "num_tokens": 13832230.0,
      "step": 2900
    },
    {
      "epoch": 0.9911988379048108,
      "eval_entropy": 0.6466146521270275,
      "eval_loss": 0.6376142501831055,
      "eval_mean_token_accuracy": 0.8284369762986898,
      "eval_num_tokens": 13832230.0,
      "eval_runtime": 32.5153,
      "eval_samples_per_second": 7.873,
      "eval_steps_per_second": 0.984,
      "step": 2900
    },
    {
      "entropy": 0.7204008292406797,
      "epoch": 0.9946167649320687,
      "grad_norm": 0.6330618858337402,
      "learning_rate": 1.770638868614505e-08,
      "loss": 0.7113766670227051,
      "mean_token_accuracy": 0.8157438091933727,
      "num_tokens": 13881699.0,
      "step": 2910
    },
    {
      "entropy": 0.7834692295640707,
      "epoch": 0.9980346919593267,
      "grad_norm": 0.5942244529724121,
      "learning_rate": 3.0021948417724433e-09,
      "loss": 0.7758901119232178,
      "mean_token_accuracy": 0.8049186132848263,
      "num_tokens": 13926410.0,
      "step": 2920
    },
    {
      "epoch": 0.9997436554729556,
      "eval_entropy": 0.646532105281949,
      "eval_loss": 0.6376615762710571,
      "eval_mean_token_accuracy": 0.8283750601112843,
      "eval_num_tokens": 13951499.0,
      "eval_runtime": 32.5193,
      "eval_samples_per_second": 7.872,
      "eval_steps_per_second": 0.984,
      "step": 2925
    },
    {
      "epoch": 1.0,
      "eval_entropy": 0.646526113152504,
      "eval_loss": 0.6376044750213623,
      "eval_mean_token_accuracy": 0.828487666323781,
      "eval_num_tokens": 13954682.0,
      "eval_runtime": 32.5825,
      "eval_samples_per_second": 7.857,
      "eval_steps_per_second": 0.982,
      "step": 2926
    }
  ],
  "logging_steps": 10,
  "max_steps": 2926,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 25,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 9.507902841041303e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}