{
  "best_global_step": 5000,
  "best_metric": 0.9333181977272034,
  "best_model_checkpoint": "/workspace/tanglish/checkpoints/qwen_lora_v1/lora/checkpoint-5000",
  "epoch": 1.0362694300518134,
  "eval_steps": 500,
  "global_step": 5000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "entropy": 1.4381825268268584,
      "epoch": 0.004145077720207254,
      "grad_norm": 2.1717000007629395,
      "learning_rate": 5.248618784530387e-06,
      "loss": 4.0762,
      "mean_token_accuracy": 0.4441531464457512,
      "num_tokens": 51480.0,
      "step": 20
    },
    {
      "entropy": 1.5037598952651023,
      "epoch": 0.008290155440414507,
      "grad_norm": 2.724255323410034,
      "learning_rate": 1.0773480662983426e-05,
      "loss": 3.8127,
      "mean_token_accuracy": 0.4525611069053411,
      "num_tokens": 103460.0,
      "step": 40
    },
    {
      "entropy": 1.901321442425251,
      "epoch": 0.012435233160621761,
      "grad_norm": 1.1939035654067993,
      "learning_rate": 1.6298342541436465e-05,
      "loss": 3.1292,
      "mean_token_accuracy": 0.48335910513997077,
      "num_tokens": 154920.0,
      "step": 60
    },
    {
      "entropy": 2.358680176734924,
      "epoch": 0.016580310880829015,
      "grad_norm": 0.7893019318580627,
      "learning_rate": 2.1823204419889505e-05,
      "loss": 2.3084,
      "mean_token_accuracy": 0.554056815057993,
      "num_tokens": 207786.0,
      "step": 80
    },
    {
      "entropy": 1.830135741829872,
      "epoch": 0.02072538860103627,
      "grad_norm": 0.5736128091812134,
      "learning_rate": 2.7348066298342545e-05,
      "loss": 1.708,
      "mean_token_accuracy": 0.6866068348288537,
      "num_tokens": 256847.0,
      "step": 100
    },
    {
      "entropy": 1.4562795095145702,
      "epoch": 0.024870466321243522,
      "grad_norm": 0.9956451058387756,
      "learning_rate": 3.287292817679558e-05,
      "loss": 1.4132,
      "mean_token_accuracy": 0.7327493987977505,
      "num_tokens": 309233.0,
      "step": 120
    },
    {
      "entropy": 1.4073016680777073,
      "epoch": 0.029015544041450778,
      "grad_norm": 0.8126519918441772,
      "learning_rate": 3.8397790055248625e-05,
      "loss": 1.3879,
      "mean_token_accuracy": 0.7353364005684853,
      "num_tokens": 361206.0,
      "step": 140
    },
    {
      "entropy": 1.28325657248497,
      "epoch": 0.03316062176165803,
      "grad_norm": 0.6408252120018005,
      "learning_rate": 4.392265193370166e-05,
      "loss": 1.2517,
      "mean_token_accuracy": 0.7543429806828499,
      "num_tokens": 412093.0,
      "step": 160
    },
    {
      "entropy": 1.219002117216587,
      "epoch": 0.03730569948186528,
      "grad_norm": 0.6377286911010742,
      "learning_rate": 4.94475138121547e-05,
      "loss": 1.2238,
      "mean_token_accuracy": 0.7544910028576851,
      "num_tokens": 463753.0,
      "step": 180
    },
    {
      "entropy": 1.2931843183934688,
      "epoch": 0.04145077720207254,
      "grad_norm": 0.7240749001502991,
      "learning_rate": 5.4972375690607735e-05,
      "loss": 1.3026,
      "mean_token_accuracy": 0.7400020457804203,
      "num_tokens": 515356.0,
      "step": 200
    },
    {
      "entropy": 1.2196356683969498,
      "epoch": 0.04559585492227979,
      "grad_norm": 0.7647127509117126,
      "learning_rate": 6.049723756906077e-05,
      "loss": 1.2411,
      "mean_token_accuracy": 0.7544264450669289,
      "num_tokens": 565236.0,
      "step": 220
    },
    {
      "entropy": 1.171409723162651,
      "epoch": 0.049740932642487044,
      "grad_norm": 0.7827390432357788,
      "learning_rate": 6.602209944751382e-05,
      "loss": 1.1776,
      "mean_token_accuracy": 0.7626889519393444,
      "num_tokens": 615515.0,
      "step": 240
    },
    {
      "entropy": 1.1406699948012828,
      "epoch": 0.0538860103626943,
      "grad_norm": 1.0029950141906738,
      "learning_rate": 7.154696132596685e-05,
      "loss": 1.1546,
      "mean_token_accuracy": 0.766391609609127,
      "num_tokens": 666062.0,
      "step": 260
    },
    {
      "entropy": 1.1219736263155937,
      "epoch": 0.058031088082901555,
      "grad_norm": 1.0503102540969849,
      "learning_rate": 7.70718232044199e-05,
      "loss": 1.1352,
      "mean_token_accuracy": 0.7641930885612964,
      "num_tokens": 719494.0,
      "step": 280
    },
    {
      "entropy": 1.1847783662378788,
      "epoch": 0.06217616580310881,
      "grad_norm": 0.8290141820907593,
      "learning_rate": 8.259668508287292e-05,
      "loss": 1.202,
      "mean_token_accuracy": 0.7585213594138622,
      "num_tokens": 772839.0,
      "step": 300
    },
    {
      "entropy": 1.1723797246813774,
      "epoch": 0.06632124352331606,
      "grad_norm": 0.877815842628479,
      "learning_rate": 8.812154696132597e-05,
      "loss": 1.1825,
      "mean_token_accuracy": 0.7599225588142872,
      "num_tokens": 825189.0,
      "step": 320
    },
    {
      "entropy": 1.1410918802022934,
      "epoch": 0.07046632124352331,
      "grad_norm": 0.7032957673072815,
      "learning_rate": 9.364640883977901e-05,
      "loss": 1.1333,
      "mean_token_accuracy": 0.7682646617293358,
      "num_tokens": 877294.0,
      "step": 340
    },
    {
      "entropy": 1.1239484950900078,
      "epoch": 0.07461139896373056,
      "grad_norm": 0.8480610847473145,
      "learning_rate": 9.917127071823204e-05,
      "loss": 1.1536,
      "mean_token_accuracy": 0.7669501408934594,
      "num_tokens": 927234.0,
      "step": 360
    },
    {
      "entropy": 1.110478015244007,
      "epoch": 0.07875647668393783,
      "grad_norm": 0.7844213843345642,
      "learning_rate": 0.0001046961325966851,
      "loss": 1.1255,
      "mean_token_accuracy": 0.7683746472001076,
      "num_tokens": 980398.0,
      "step": 380
    },
    {
      "entropy": 1.1618236370384694,
      "epoch": 0.08290155440414508,
      "grad_norm": 0.7246690988540649,
      "learning_rate": 0.00011022099447513811,
      "loss": 1.1729,
      "mean_token_accuracy": 0.7629252031445504,
      "num_tokens": 1030335.0,
      "step": 400
    },
    {
      "entropy": 1.1364129185676575,
      "epoch": 0.08704663212435233,
      "grad_norm": 0.7912014722824097,
      "learning_rate": 0.00011574585635359116,
      "loss": 1.152,
      "mean_token_accuracy": 0.7651066802442074,
      "num_tokens": 1083694.0,
      "step": 420
    },
    {
      "entropy": 1.0954272992908956,
      "epoch": 0.09119170984455958,
      "grad_norm": 0.7310007214546204,
      "learning_rate": 0.00012127071823204422,
      "loss": 1.1039,
      "mean_token_accuracy": 0.7700782015919685,
      "num_tokens": 1134346.0,
      "step": 440
    },
    {
      "entropy": 1.1179637335240842,
      "epoch": 0.09533678756476684,
      "grad_norm": 0.7415347099304199,
      "learning_rate": 0.00012679558011049723,
      "loss": 1.1267,
      "mean_token_accuracy": 0.7677439145743847,
      "num_tokens": 1188485.0,
      "step": 460
    },
    {
      "entropy": 1.0841590076684953,
      "epoch": 0.09948186528497409,
      "grad_norm": 0.8192847371101379,
      "learning_rate": 0.00013232044198895027,
      "loss": 1.095,
      "mean_token_accuracy": 0.7726523540914059,
      "num_tokens": 1239371.0,
      "step": 480
    },
    {
      "entropy": 1.0643972344696522,
      "epoch": 0.10362694300518134,
      "grad_norm": 0.8887308239936829,
      "learning_rate": 0.00013784530386740332,
      "loss": 1.0762,
      "mean_token_accuracy": 0.7769059598445892,
      "num_tokens": 1290491.0,
      "step": 500
    },
    {
      "epoch": 0.10362694300518134,
      "eval_entropy": 1.095322513439524,
      "eval_loss": 1.1184215545654297,
      "eval_mean_token_accuracy": 0.7680243507144958,
      "eval_num_tokens": 1290491.0,
      "eval_runtime": 198.0973,
      "eval_samples_per_second": 20.51,
      "eval_steps_per_second": 2.564,
      "step": 500
    },
    {
      "entropy": 1.0820845410227775,
      "epoch": 0.1077720207253886,
      "grad_norm": 0.7347564697265625,
      "learning_rate": 0.00014337016574585636,
      "loss": 1.0979,
      "mean_token_accuracy": 0.7743508994579316,
      "num_tokens": 1341205.0,
      "step": 520
    },
    {
      "entropy": 1.1001912213861942,
      "epoch": 0.11191709844559586,
      "grad_norm": 0.690479040145874,
      "learning_rate": 0.0001488950276243094,
      "loss": 1.1105,
      "mean_token_accuracy": 0.7694470949470997,
      "num_tokens": 1394461.0,
      "step": 540
    },
    {
      "entropy": 1.0999469496309757,
      "epoch": 0.11606217616580311,
      "grad_norm": 0.6917790174484253,
      "learning_rate": 0.00015441988950276245,
      "loss": 1.117,
      "mean_token_accuracy": 0.7682196438312531,
      "num_tokens": 1447828.0,
      "step": 560
    },
    {
      "entropy": 1.0735741600394249,
      "epoch": 0.12020725388601036,
      "grad_norm": 0.833886444568634,
      "learning_rate": 0.00015994475138121547,
      "loss": 1.086,
      "mean_token_accuracy": 0.7741858348250389,
      "num_tokens": 1498479.0,
      "step": 580
    },
    {
      "entropy": 1.0962664999067784,
      "epoch": 0.12435233160621761,
      "grad_norm": 0.6647179126739502,
      "learning_rate": 0.0001654696132596685,
      "loss": 1.0958,
      "mean_token_accuracy": 0.7741696208715438,
      "num_tokens": 1548137.0,
      "step": 600
    },
    {
      "entropy": 1.0694029107689857,
      "epoch": 0.12849740932642487,
      "grad_norm": 0.6861886382102966,
      "learning_rate": 0.00017099447513812155,
      "loss": 1.0829,
      "mean_token_accuracy": 0.7720654770731926,
      "num_tokens": 1599638.0,
      "step": 620
    },
    {
      "entropy": 1.075825183838606,
      "epoch": 0.13264248704663212,
      "grad_norm": 0.7000752687454224,
      "learning_rate": 0.0001765193370165746,
      "loss": 1.0825,
      "mean_token_accuracy": 0.7762658178806305,
      "num_tokens": 1650323.0,
      "step": 640
    },
    {
      "entropy": 1.083068310469389,
      "epoch": 0.13678756476683937,
      "grad_norm": 0.7568516731262207,
      "learning_rate": 0.00018204419889502764,
      "loss": 1.0918,
      "mean_token_accuracy": 0.7720131345093251,
      "num_tokens": 1701215.0,
      "step": 660
    },
    {
      "entropy": 1.0660202227532865,
      "epoch": 0.14093264248704662,
      "grad_norm": 0.8106701374053955,
      "learning_rate": 0.00018756906077348066,
      "loss": 1.073,
      "mean_token_accuracy": 0.7741113729774952,
      "num_tokens": 1751275.0,
      "step": 680
    },
    {
      "entropy": 1.00438876375556,
      "epoch": 0.14507772020725387,
      "grad_norm": 0.6197234988212585,
      "learning_rate": 0.0001930939226519337,
      "loss": 1.0234,
      "mean_token_accuracy": 0.7818360298871994,
      "num_tokens": 1803256.0,
      "step": 700
    },
    {
      "entropy": 1.1083124563097955,
      "epoch": 0.14922279792746113,
      "grad_norm": 0.6046066284179688,
      "learning_rate": 0.00019861878453038677,
      "loss": 1.122,
      "mean_token_accuracy": 0.7704763405025006,
      "num_tokens": 1855372.0,
      "step": 720
    },
    {
      "entropy": 1.097743584215641,
      "epoch": 0.1533678756476684,
      "grad_norm": 0.6518492102622986,
      "learning_rate": 0.00019999941280374287,
      "loss": 1.1116,
      "mean_token_accuracy": 0.7682732604444027,
      "num_tokens": 1906015.0,
      "step": 740
    },
    {
      "entropy": 1.1098480015993117,
      "epoch": 0.15751295336787566,
      "grad_norm": 0.6173190474510193,
      "learning_rate": 0.00019999680305650554,
      "loss": 1.1173,
      "mean_token_accuracy": 0.7627138867974281,
      "num_tokens": 1959638.0,
      "step": 760
    },
    {
      "entropy": 1.0928526185452938,
      "epoch": 0.1616580310880829,
      "grad_norm": 0.6099253296852112,
      "learning_rate": 0.00019999210556868917,
      "loss": 1.1156,
      "mean_token_accuracy": 0.7711292572319508,
      "num_tokens": 2010476.0,
      "step": 780
    },
    {
      "entropy": 1.0808037273585795,
      "epoch": 0.16580310880829016,
      "grad_norm": 0.6099389791488647,
      "learning_rate": 0.0001999853204383683,
      "loss": 1.087,
      "mean_token_accuracy": 0.7749647930264473,
      "num_tokens": 2060494.0,
      "step": 800
    },
    {
      "entropy": 1.048035841435194,
      "epoch": 0.1699481865284974,
      "grad_norm": 0.5170192122459412,
      "learning_rate": 0.0001999764478072033,
      "loss": 1.0642,
      "mean_token_accuracy": 0.7727734588086606,
      "num_tokens": 2113260.0,
      "step": 820
    },
    {
      "entropy": 1.0665065109729768,
      "epoch": 0.17409326424870467,
      "grad_norm": 0.6461181640625,
      "learning_rate": 0.00019996548786043768,
      "loss": 1.0795,
      "mean_token_accuracy": 0.777741952240467,
      "num_tokens": 2164157.0,
      "step": 840
    },
    {
      "entropy": 1.0095063984394073,
      "epoch": 0.17823834196891192,
      "grad_norm": 0.5376838445663452,
      "learning_rate": 0.00019995244082689396,
      "loss": 1.0238,
      "mean_token_accuracy": 0.7842500291764736,
      "num_tokens": 2215501.0,
      "step": 860
    },
    {
      "entropy": 1.0482122428715228,
      "epoch": 0.18238341968911917,
      "grad_norm": 0.5362586379051208,
      "learning_rate": 0.00019993730697896912,
      "loss": 1.0567,
      "mean_token_accuracy": 0.7757732585072518,
      "num_tokens": 2265908.0,
      "step": 880
    },
    {
      "entropy": 1.0375815689563752,
      "epoch": 0.18652849740932642,
      "grad_norm": 0.6826100945472717,
      "learning_rate": 0.0001999200866326287,
      "loss": 1.055,
      "mean_token_accuracy": 0.777643445879221,
      "num_tokens": 2316578.0,
      "step": 900
    },
    {
      "entropy": 1.0592326551675797,
      "epoch": 0.19067357512953367,
      "grad_norm": 0.4966115355491638,
      "learning_rate": 0.00019990078014740036,
      "loss": 1.0682,
      "mean_token_accuracy": 0.7757245123386383,
      "num_tokens": 2367491.0,
      "step": 920
    },
    {
      "entropy": 1.0522546865046025,
      "epoch": 0.19481865284974093,
      "grad_norm": 0.5403271317481995,
      "learning_rate": 0.00019987938792636632,
      "loss": 1.0597,
      "mean_token_accuracy": 0.7824762858450413,
      "num_tokens": 2418531.0,
      "step": 940
    },
    {
      "entropy": 1.0112147249281407,
      "epoch": 0.19896373056994818,
      "grad_norm": 0.591804027557373,
      "learning_rate": 0.00019985591041615496,
      "loss": 1.0179,
      "mean_token_accuracy": 0.7824166387319564,
      "num_tokens": 2468312.0,
      "step": 960
    },
    {
      "entropy": 1.0457121133804321,
      "epoch": 0.20310880829015543,
      "grad_norm": 0.4792396128177643,
      "learning_rate": 0.00019983034810693137,
      "loss": 1.0567,
      "mean_token_accuracy": 0.7786681324243545,
      "num_tokens": 2518124.0,
      "step": 980
    },
    {
      "entropy": 1.0062039978802204,
      "epoch": 0.20725388601036268,
      "grad_norm": 0.6134971976280212,
      "learning_rate": 0.00019980270153238742,
      "loss": 1.0284,
      "mean_token_accuracy": 0.7829495541751385,
      "num_tokens": 2569706.0,
      "step": 1000
    },
    {
      "epoch": 0.20725388601036268,
      "eval_entropy": 1.0445965941496722,
      "eval_loss": 1.059064507484436,
      "eval_mean_token_accuracy": 0.7763740899055962,
      "eval_num_tokens": 2569706.0,
      "eval_runtime": 197.7332,
      "eval_samples_per_second": 20.548,
      "eval_steps_per_second": 2.569,
      "step": 1000
    },
    {
      "entropy": 1.0440333500504493,
      "epoch": 0.21139896373056996,
      "grad_norm": 0.6377447843551636,
      "learning_rate": 0.0001997729712697303,
      "loss": 1.0571,
      "mean_token_accuracy": 0.7803341060876846,
      "num_tokens": 2618883.0,
      "step": 1020
    },
    {
      "entropy": 1.021172496676445,
      "epoch": 0.2155440414507772,
      "grad_norm": 0.5175086855888367,
      "learning_rate": 0.0001997411579396706,
      "loss": 1.0292,
      "mean_token_accuracy": 0.7807957075536252,
      "num_tokens": 2670732.0,
      "step": 1040
    },
    {
      "entropy": 1.0275346338748932,
      "epoch": 0.21968911917098446,
      "grad_norm": 0.5990492701530457,
      "learning_rate": 0.00019970726220640929,
      "loss": 1.0447,
      "mean_token_accuracy": 0.77860803976655,
      "num_tokens": 2722930.0,
      "step": 1060
    },
    {
      "entropy": 1.0109625704586507,
      "epoch": 0.22383419689119172,
      "grad_norm": 0.4519039988517761,
      "learning_rate": 0.00019967128477762404,
      "loss": 1.0286,
      "mean_token_accuracy": 0.7781259723007679,
      "num_tokens": 2775645.0,
      "step": 1080
    },
    {
      "entropy": 1.024503505975008,
      "epoch": 0.22797927461139897,
      "grad_norm": 0.5599638819694519,
      "learning_rate": 0.00019963322640445424,
      "loss": 1.0358,
      "mean_token_accuracy": 0.7802064262330533,
      "num_tokens": 2829031.0,
      "step": 1100
    },
    {
      "entropy": 1.000456291437149,
      "epoch": 0.23212435233160622,
      "grad_norm": 0.42756351828575134,
      "learning_rate": 0.0001995930878814854,
      "loss": 1.0071,
      "mean_token_accuracy": 0.7876958273351192,
      "num_tokens": 2882439.0,
      "step": 1120
    },
    {
      "entropy": 1.0316657554358244,
      "epoch": 0.23626943005181347,
      "grad_norm": 0.4877125322818756,
      "learning_rate": 0.0001995508700467325,
      "loss": 1.0422,
      "mean_token_accuracy": 0.7804410368204117,
      "num_tokens": 2935155.0,
      "step": 1140
    },
    {
      "entropy": 1.0400867745280267,
      "epoch": 0.24041450777202072,
      "grad_norm": 0.501750111579895,
      "learning_rate": 0.00019950657378162262,
      "loss": 1.06,
      "mean_token_accuracy": 0.7792659178376198,
      "num_tokens": 2986224.0,
      "step": 1160
    },
    {
      "entropy": 1.0030486956238747,
      "epoch": 0.24455958549222798,
      "grad_norm": 0.4440039396286011,
      "learning_rate": 0.00019946020001097644,
      "loss": 1.0081,
      "mean_token_accuracy": 0.7861276485025883,
      "num_tokens": 3037391.0,
      "step": 1180
    },
    {
      "entropy": 1.0251952454447746,
      "epoch": 0.24870466321243523,
      "grad_norm": 0.6198555827140808,
      "learning_rate": 0.00019941174970298888,
      "loss": 1.036,
      "mean_token_accuracy": 0.7791207812726497,
      "num_tokens": 3087069.0,
      "step": 1200
    },
    {
      "entropy": 1.0355947606265545,
      "epoch": 0.2528497409326425,
      "grad_norm": 0.43304187059402466,
      "learning_rate": 0.0001993612238692091,
      "loss": 1.0457,
      "mean_token_accuracy": 0.7767570614814758,
      "num_tokens": 3140164.0,
      "step": 1220
    },
    {
      "entropy": 1.0067585781216621,
      "epoch": 0.25699481865284973,
      "grad_norm": 0.5553335547447205,
      "learning_rate": 0.0001993086235645191,
      "loss": 1.0238,
      "mean_token_accuracy": 0.7811847984790802,
      "num_tokens": 3192879.0,
      "step": 1240
    },
    {
      "entropy": 1.0447653152048588,
      "epoch": 0.261139896373057,
      "grad_norm": 0.5924249291419983,
      "learning_rate": 0.0001992539498871119,
      "loss": 1.049,
      "mean_token_accuracy": 0.7775058671832085,
      "num_tokens": 3244981.0,
      "step": 1260
    },
    {
      "entropy": 1.0076527014374732,
      "epoch": 0.26528497409326424,
      "grad_norm": 0.5087696313858032,
      "learning_rate": 0.00019919720397846855,
      "loss": 1.018,
      "mean_token_accuracy": 0.784369270503521,
      "num_tokens": 3293733.0,
      "step": 1280
    },
    {
      "entropy": 1.0454526104032993,
      "epoch": 0.2694300518134715,
      "grad_norm": 0.5536513924598694,
      "learning_rate": 0.00019913838702333432,
      "loss": 1.0648,
      "mean_token_accuracy": 0.7793882772326469,
      "num_tokens": 3343268.0,
      "step": 1300
    },
    {
      "entropy": 1.0259532511234284,
      "epoch": 0.27357512953367874,
      "grad_norm": 0.5604416728019714,
      "learning_rate": 0.00019907750024969386,
      "loss": 1.0366,
      "mean_token_accuracy": 0.779184664785862,
      "num_tokens": 3395934.0,
      "step": 1320
    },
    {
      "entropy": 1.0157927133142948,
      "epoch": 0.277720207253886,
      "grad_norm": 0.45751333236694336,
      "learning_rate": 0.00019901454492874572,
      "loss": 1.0379,
      "mean_token_accuracy": 0.7799046277999878,
      "num_tokens": 3448448.0,
      "step": 1340
    },
    {
      "entropy": 1.013536487519741,
      "epoch": 0.28186528497409324,
      "grad_norm": 0.4128531813621521,
      "learning_rate": 0.00019894952237487566,
      "loss": 1.0201,
      "mean_token_accuracy": 0.7843268230557442,
      "num_tokens": 3499346.0,
      "step": 1360
    },
    {
      "entropy": 1.0443965464830398,
      "epoch": 0.2860103626943005,
      "grad_norm": 0.49815455079078674,
      "learning_rate": 0.0001988824339456294,
      "loss": 1.0593,
      "mean_token_accuracy": 0.7785378187894821,
      "num_tokens": 3551298.0,
      "step": 1380
    },
    {
      "entropy": 1.011622816324234,
      "epoch": 0.29015544041450775,
      "grad_norm": 0.46742376685142517,
      "learning_rate": 0.000198813281041684,
      "loss": 1.0193,
      "mean_token_accuracy": 0.7825383462011815,
      "num_tokens": 3602656.0,
      "step": 1400
    },
    {
      "entropy": 0.9844559028744697,
      "epoch": 0.29430051813471503,
      "grad_norm": 0.40691128373146057,
      "learning_rate": 0.0001987420651068189,
      "loss": 1.0042,
      "mean_token_accuracy": 0.7844808354973793,
      "num_tokens": 3654678.0,
      "step": 1420
    },
    {
      "entropy": 1.0059320218861103,
      "epoch": 0.29844559585492225,
      "grad_norm": 0.42339223623275757,
      "learning_rate": 0.00019866878762788562,
      "loss": 1.0142,
      "mean_token_accuracy": 0.7823195405304432,
      "num_tokens": 3705949.0,
      "step": 1440
    },
    {
      "entropy": 0.9958507776260376,
      "epoch": 0.30259067357512953,
      "grad_norm": 0.48249927163124084,
      "learning_rate": 0.00019859345013477665,
      "loss": 1.0121,
      "mean_token_accuracy": 0.7889884993433952,
      "num_tokens": 3756251.0,
      "step": 1460
    },
    {
      "entropy": 0.9966832913458348,
      "epoch": 0.3067357512953368,
      "grad_norm": 0.3815440237522125,
      "learning_rate": 0.00019851605420039377,
      "loss": 1.0226,
      "mean_token_accuracy": 0.7830308295786381,
      "num_tokens": 3808938.0,
      "step": 1480
    },
    {
      "entropy": 0.9885330185294151,
      "epoch": 0.31088082901554404,
      "grad_norm": 0.4722138047218323,
      "learning_rate": 0.00019843660144061497,
      "loss": 0.9955,
      "mean_token_accuracy": 0.7874660678207874,
      "num_tokens": 3858248.0,
      "step": 1500
    },
    {
      "epoch": 0.31088082901554404,
      "eval_entropy": 1.032557885595194,
      "eval_loss": 1.0242446660995483,
      "eval_mean_token_accuracy": 0.7811775283785317,
      "eval_num_tokens": 3858248.0,
      "eval_runtime": 198.2363,
      "eval_samples_per_second": 20.496,
      "eval_steps_per_second": 2.563,
      "step": 1500
    },
    {
      "entropy": 1.0101246651262046,
      "epoch": 0.3150259067357513,
      "grad_norm": 0.42561495304107666,
      "learning_rate": 0.00019835509351426073,
      "loss": 1.0177,
      "mean_token_accuracy": 0.7840446695685387,
      "num_tokens": 3909740.0,
      "step": 1520
    },
    {
      "entropy": 0.9913592956960201,
      "epoch": 0.31917098445595854,
      "grad_norm": 0.4773913025856018,
      "learning_rate": 0.00019827153212305954,
      "loss": 1.0043,
      "mean_token_accuracy": 0.7856849782168865,
      "num_tokens": 3958454.0,
      "step": 1540
    },
    {
      "entropy": 1.0170710057020187,
      "epoch": 0.3233160621761658,
      "grad_norm": 0.4388096034526825,
      "learning_rate": 0.00019818591901161225,
      "loss": 1.0284,
      "mean_token_accuracy": 0.7777703814208508,
      "num_tokens": 4010172.0,
      "step": 1560
    },
    {
      "entropy": 1.0317365549504758,
      "epoch": 0.32746113989637304,
      "grad_norm": 0.4321315884590149,
      "learning_rate": 0.00019809825596735567,
      "loss": 1.0507,
      "mean_token_accuracy": 0.782105403393507,
      "num_tokens": 4061342.0,
      "step": 1580
    },
    {
      "entropy": 1.0285033784806727,
      "epoch": 0.3316062176165803,
      "grad_norm": 0.4316249489784241,
      "learning_rate": 0.00019800854482052526,
      "loss": 1.053,
      "mean_token_accuracy": 0.7812598653137683,
      "num_tokens": 4112308.0,
      "step": 1600
    },
    {
      "entropy": 0.959014667570591,
      "epoch": 0.33575129533678755,
      "grad_norm": 0.4981176555156708,
      "learning_rate": 0.0001979167874441169,
      "loss": 0.9713,
      "mean_token_accuracy": 0.7900767177343369,
      "num_tokens": 4163721.0,
      "step": 1620
    },
    {
      "entropy": 0.9863104440271855,
      "epoch": 0.3398963730569948,
      "grad_norm": 0.5106677412986755,
      "learning_rate": 0.00019782298575384783,
      "loss": 0.9888,
      "mean_token_accuracy": 0.786780058592558,
      "num_tokens": 4214428.0,
      "step": 1640
    },
    {
      "entropy": 0.9868629761040211,
      "epoch": 0.34404145077720205,
      "grad_norm": 0.4558473527431488,
      "learning_rate": 0.00019772714170811662,
      "loss": 1.0034,
      "mean_token_accuracy": 0.7846312940120697,
      "num_tokens": 4267755.0,
      "step": 1660
    },
    {
      "entropy": 1.0307603180408478,
      "epoch": 0.34818652849740933,
      "grad_norm": 0.4738333225250244,
      "learning_rate": 0.00019762925730796224,
      "loss": 1.0414,
      "mean_token_accuracy": 0.7802501529455185,
      "num_tokens": 4317809.0,
      "step": 1680
    },
    {
      "entropy": 1.0127912051975727,
      "epoch": 0.35233160621761656,
      "grad_norm": 0.48111966252326965,
      "learning_rate": 0.0001975293345970224,
      "loss": 1.0147,
      "mean_token_accuracy": 0.7798892751336097,
      "num_tokens": 4370912.0,
      "step": 1700
    },
    {
      "entropy": 0.9599951460957528,
      "epoch": 0.35647668393782384,
      "grad_norm": 0.4724443256855011,
      "learning_rate": 0.00019742737566149078,
      "loss": 0.9585,
      "mean_token_accuracy": 0.7909531377255916,
      "num_tokens": 4423603.0,
      "step": 1720
    },
    {
      "entropy": 1.0278802808374166,
      "epoch": 0.3606217616580311,
      "grad_norm": 0.5354254245758057,
      "learning_rate": 0.0001973233826300735,
      "loss": 1.035,
      "mean_token_accuracy": 0.7811824880540371,
      "num_tokens": 4476943.0,
      "step": 1740
    },
    {
      "entropy": 0.9914717882871628,
      "epoch": 0.36476683937823834,
      "grad_norm": 0.47376251220703125,
      "learning_rate": 0.00019721735767394468,
      "loss": 1.0099,
      "mean_token_accuracy": 0.781830509006977,
      "num_tokens": 4527823.0,
      "step": 1760
    },
    {
      "entropy": 1.020705473423004,
      "epoch": 0.3689119170984456,
      "grad_norm": 0.4668932259082794,
      "learning_rate": 0.00019710930300670108,
      "loss": 1.0274,
      "mean_token_accuracy": 0.7800539046525955,
      "num_tokens": 4581082.0,
      "step": 1780
    },
    {
      "entropy": 1.0137963436543942,
      "epoch": 0.37305699481865284,
      "grad_norm": 0.4402451515197754,
      "learning_rate": 0.00019699922088431605,
      "loss": 1.0194,
      "mean_token_accuracy": 0.7841047510504723,
      "num_tokens": 4630834.0,
      "step": 1800
    },
    {
      "entropy": 0.9691961273550987,
      "epoch": 0.3772020725388601,
      "grad_norm": 0.4267026484012604,
      "learning_rate": 0.00019688711360509213,
      "loss": 0.9716,
      "mean_token_accuracy": 0.7890592306852341,
      "num_tokens": 4682795.0,
      "step": 1820
    },
    {
      "entropy": 0.9769839994609356,
      "epoch": 0.38134715025906735,
      "grad_norm": 0.3717665374279022,
      "learning_rate": 0.00019677298350961333,
      "loss": 0.9986,
      "mean_token_accuracy": 0.7845027431845665,
      "num_tokens": 4734983.0,
      "step": 1840
    },
    {
      "entropy": 1.0078977264463902,
      "epoch": 0.3854922279792746,
      "grad_norm": 0.5216906070709229,
      "learning_rate": 0.00019665683298069616,
      "loss": 1.0144,
      "mean_token_accuracy": 0.7834078282117843,
      "num_tokens": 4786331.0,
      "step": 1860
    },
    {
      "entropy": 0.9985935389995575,
      "epoch": 0.38963730569948185,
      "grad_norm": 0.408572793006897,
      "learning_rate": 0.00019653866444333985,
      "loss": 1.0096,
      "mean_token_accuracy": 0.7860454954206944,
      "num_tokens": 4836661.0,
      "step": 1880
    },
    {
      "entropy": 0.9866756744682789,
      "epoch": 0.39378238341968913,
      "grad_norm": 0.43202802538871765,
      "learning_rate": 0.00019641848036467576,
      "loss": 1.0044,
      "mean_token_accuracy": 0.7888299062848091,
      "num_tokens": 4887993.0,
      "step": 1900
    },
    {
      "entropy": 0.9607140734791756,
      "epoch": 0.39792746113989635,
      "grad_norm": 0.4784650206565857,
      "learning_rate": 0.00019629628325391595,
      "loss": 0.9777,
      "mean_token_accuracy": 0.7884210042655468,
      "num_tokens": 4937865.0,
      "step": 1920
    },
    {
      "entropy": 0.947428410500288,
      "epoch": 0.40207253886010363,
      "grad_norm": 0.5031050443649292,
      "learning_rate": 0.0001961720756623006,
      "loss": 0.9529,
      "mean_token_accuracy": 0.7923657827079296,
      "num_tokens": 4988659.0,
      "step": 1940
    },
    {
      "entropy": 0.9914572931826114,
      "epoch": 0.40621761658031086,
      "grad_norm": 0.46874579787254333,
      "learning_rate": 0.0001960458601830449,
      "loss": 1.0075,
      "mean_token_accuracy": 0.7866203740239144,
      "num_tokens": 5038807.0,
      "step": 1960
    },
    {
      "entropy": 1.0107652366161346,
      "epoch": 0.41036269430051814,
      "grad_norm": 0.5741726160049438,
      "learning_rate": 0.00019591763945128493,
      "loss": 1.0242,
      "mean_token_accuracy": 0.7811738647520542,
      "num_tokens": 5090621.0,
      "step": 1980
    },
    {
      "entropy": 1.022918675094843,
      "epoch": 0.41450777202072536,
      "grad_norm": 0.5132948756217957,
      "learning_rate": 0.00019578741614402245,
      "loss": 1.0264,
      "mean_token_accuracy": 0.7782419688999653,
      "num_tokens": 5143415.0,
      "step": 2000
    },
    {
      "epoch": 0.41450777202072536,
      "eval_entropy": 0.9987384857624535,
      "eval_loss": 1.0029442310333252,
      "eval_mean_token_accuracy": 0.7838117130628721,
      "eval_num_tokens": 5143415.0,
      "eval_runtime": 197.9171,
      "eval_samples_per_second": 20.529,
      "eval_steps_per_second": 2.567,
      "step": 2000
    },
    {
      "entropy": 0.961260611563921,
      "epoch": 0.41865284974093264,
      "grad_norm": 0.5051520466804504,
      "learning_rate": 0.00019565519298006924,
      "loss": 0.9695,
      "mean_token_accuracy": 0.7914115615189076,
      "num_tokens": 5194400.0,
      "step": 2020
    },
    {
      "entropy": 0.9982391525059938,
      "epoch": 0.4227979274611399,
      "grad_norm": 0.4538373351097107,
      "learning_rate": 0.00019552097271999022,
      "loss": 1.0146,
      "mean_token_accuracy": 0.7841442748904228,
      "num_tokens": 5245625.0,
      "step": 2040
    },
    {
      "entropy": 0.9815789975225926,
      "epoch": 0.42694300518134715,
      "grad_norm": 0.4690403342247009,
      "learning_rate": 0.00019538475816604578,
      "loss": 0.9959,
      "mean_token_accuracy": 0.7872560724616051,
      "num_tokens": 5295237.0,
      "step": 2060
    },
    {
      "entropy": 0.9912715546786786,
      "epoch": 0.4310880829015544,
      "grad_norm": 0.42687225341796875,
      "learning_rate": 0.00019524655216213334,
      "loss": 0.9866,
      "mean_token_accuracy": 0.7859402887523175,
      "num_tokens": 5348418.0,
      "step": 2080
    },
    {
      "entropy": 1.0102518431842327,
      "epoch": 0.43523316062176165,
      "grad_norm": 0.44384920597076416,
      "learning_rate": 0.00019510635759372805,
      "loss": 1.0238,
      "mean_token_accuracy": 0.7797439776360988,
      "num_tokens": 5399166.0,
      "step": 2100
    },
    {
      "entropy": 1.0160183981060982,
      "epoch": 0.43937823834196893,
      "grad_norm": 0.44171077013015747,
      "learning_rate": 0.00019496417738782228,
      "loss": 1.0228,
      "mean_token_accuracy": 0.7803056098520755,
      "num_tokens": 5451958.0,
      "step": 2120
    },
    {
      "entropy": 0.9693700931966305,
      "epoch": 0.44352331606217615,
      "grad_norm": 0.4557332694530487,
      "learning_rate": 0.00019482001451286488,
      "loss": 0.9848,
      "mean_token_accuracy": 0.7881055019795895,
      "num_tokens": 5504481.0,
      "step": 2140
    },
    {
      "entropy": 0.9666793867945671,
      "epoch": 0.44766839378238343,
      "grad_norm": 0.45267391204833984,
      "learning_rate": 0.00019467387197869883,
      "loss": 0.9834,
      "mean_token_accuracy": 0.7889953970909118,
      "num_tokens": 5551843.0,
      "step": 2160
    },
    {
      "entropy": 0.9949551515281201,
      "epoch": 0.45181347150259066,
      "grad_norm": 0.438851535320282,
      "learning_rate": 0.00019452575283649873,
      "loss": 1.0144,
      "mean_token_accuracy": 0.782796959578991,
      "num_tokens": 5604244.0,
      "step": 2180
    },
    {
      "entropy": 0.9963484168052673,
      "epoch": 0.45595854922279794,
      "grad_norm": 0.47502249479293823,
      "learning_rate": 0.00019437566017870678,
      "loss": 1.0137,
      "mean_token_accuracy": 0.7818049520254136,
      "num_tokens": 5655400.0,
      "step": 2200
    },
    {
      "entropy": 0.962771862745285,
      "epoch": 0.46010362694300516,
      "grad_norm": 0.42144647240638733,
      "learning_rate": 0.00019422359713896863,
      "loss": 0.9667,
      "mean_token_accuracy": 0.7884697504341602,
      "num_tokens": 5708666.0,
      "step": 2220
    },
    {
      "entropy": 0.9682818073779345,
      "epoch": 0.46424870466321244,
      "grad_norm": 0.3935494124889374,
      "learning_rate": 0.00019406956689206744,
      "loss": 0.9675,
      "mean_token_accuracy": 0.7886212930083275,
      "num_tokens": 5761629.0,
      "step": 2240
    },
    {
      "entropy": 1.037352129071951,
      "epoch": 0.46839378238341967,
      "grad_norm": 0.4130975902080536,
      "learning_rate": 0.00019391357265385807,
      "loss": 1.0403,
      "mean_token_accuracy": 0.7768575675785542,
      "num_tokens": 5812646.0,
      "step": 2260
    },
    {
      "entropy": 0.9832554623484612,
      "epoch": 0.47253886010362695,
      "grad_norm": 0.3829593360424042,
      "learning_rate": 0.00019375561768119958,
      "loss": 0.9976,
      "mean_token_accuracy": 0.7897411204874516,
      "num_tokens": 5862817.0,
      "step": 2280
    },
    {
      "entropy": 0.9487610541284084,
      "epoch": 0.47668393782383417,
      "grad_norm": 0.4701037108898163,
      "learning_rate": 0.00019359570527188754,
      "loss": 0.9638,
      "mean_token_accuracy": 0.792264959961176,
      "num_tokens": 5912716.0,
      "step": 2300
    },
    {
      "entropy": 0.954827756434679,
      "epoch": 0.48082901554404145,
      "grad_norm": 0.4769448935985565,
      "learning_rate": 0.00019343383876458485,
      "loss": 0.9595,
      "mean_token_accuracy": 0.7927994206547737,
      "num_tokens": 5963750.0,
      "step": 2320
    },
    {
      "entropy": 0.9979561999440193,
      "epoch": 0.48497409326424873,
      "grad_norm": 0.46122485399246216,
      "learning_rate": 0.0001932700215387524,
      "loss": 1.0035,
      "mean_token_accuracy": 0.7901160277426242,
      "num_tokens": 6013594.0,
      "step": 2340
    },
    {
      "entropy": 0.9591921351850032,
      "epoch": 0.48911917098445595,
      "grad_norm": 0.4695828855037689,
      "learning_rate": 0.00019310425701457814,
      "loss": 0.9747,
      "mean_token_accuracy": 0.7882664181292057,
      "num_tokens": 6066506.0,
      "step": 2360
    },
    {
      "entropy": 0.9904223918914795,
      "epoch": 0.49326424870466323,
      "grad_norm": 0.43157315254211426,
      "learning_rate": 0.00019293654865290604,
      "loss": 1.007,
      "mean_token_accuracy": 0.7831035263836383,
      "num_tokens": 6120258.0,
      "step": 2380
    },
    {
      "entropy": 0.9716513626277447,
      "epoch": 0.49740932642487046,
      "grad_norm": 0.4715091586112976,
      "learning_rate": 0.00019276689995516346,
      "loss": 0.9743,
      "mean_token_accuracy": 0.7874019861221313,
      "num_tokens": 6171154.0,
      "step": 2400
    },
    {
      "entropy": 0.9984053768217563,
      "epoch": 0.5015544041450777,
      "grad_norm": 0.4164447486400604,
      "learning_rate": 0.0001925953144632884,
      "loss": 1.0145,
      "mean_token_accuracy": 0.781635707616806,
      "num_tokens": 6222515.0,
      "step": 2420
    },
    {
      "entropy": 0.9804062567651272,
      "epoch": 0.505699481865285,
      "grad_norm": 0.45674899220466614,
      "learning_rate": 0.00019242179575965537,
      "loss": 0.9808,
      "mean_token_accuracy": 0.7865826293826104,
      "num_tokens": 6275939.0,
      "step": 2440
    },
    {
      "entropy": 0.9783611670136452,
      "epoch": 0.5098445595854922,
      "grad_norm": 0.4376664161682129,
      "learning_rate": 0.00019224634746700051,
      "loss": 0.992,
      "mean_token_accuracy": 0.7872013546526432,
      "num_tokens": 6325162.0,
      "step": 2460
    },
    {
      "entropy": 0.9830627031624317,
      "epoch": 0.5139896373056995,
      "grad_norm": 0.3772311508655548,
      "learning_rate": 0.00019206897324834614,
      "loss": 0.9948,
      "mean_token_accuracy": 0.7873731434345246,
      "num_tokens": 6376061.0,
      "step": 2480
    },
    {
      "entropy": 0.9824939467012882,
      "epoch": 0.5181347150259067,
      "grad_norm": 0.4799821376800537,
      "learning_rate": 0.0001918896768069242,
      "loss": 0.988,
      "mean_token_accuracy": 0.7887446887791156,
      "num_tokens": 6428028.0,
      "step": 2500
    },
    {
      "epoch": 0.5181347150259067,
      "eval_entropy": 0.9617302230023962,
      "eval_loss": 0.9882201552391052,
      "eval_mean_token_accuracy": 0.7864582590230806,
      "eval_num_tokens": 6428028.0,
      "eval_runtime": 197.0794,
      "eval_samples_per_second": 20.616,
      "eval_steps_per_second": 2.578,
      "step": 2500
    },
    {
      "entropy": 0.9796643897891044,
      "epoch": 0.522279792746114,
      "grad_norm": 0.45946526527404785,
      "learning_rate": 0.0001917084618860989,
      "loss": 0.9933,
      "mean_token_accuracy": 0.7855292394757271,
      "num_tokens": 6479048.0,
      "step": 2520
    },
    {
      "entropy": 0.9833916153758764,
      "epoch": 0.5264248704663212,
      "grad_norm": 0.3963780999183655,
      "learning_rate": 0.00019152533226928867,
      "loss": 0.9962,
      "mean_token_accuracy": 0.7817724473774433,
      "num_tokens": 6529826.0,
      "step": 2540
    },
    {
      "entropy": 0.9843226552009583,
      "epoch": 0.5305699481865285,
      "grad_norm": 0.4751962125301361,
      "learning_rate": 0.000191340291779887,
      "loss": 0.9994,
      "mean_token_accuracy": 0.7830713532865048,
      "num_tokens": 6580699.0,
      "step": 2560
    },
    {
      "entropy": 0.9555308140814305,
      "epoch": 0.5347150259067357,
      "grad_norm": 0.40511226654052734,
      "learning_rate": 0.00019115334428118286,
      "loss": 0.9761,
      "mean_token_accuracy": 0.7889509312808514,
      "num_tokens": 6633814.0,
      "step": 2580
    },
    {
      "entropy": 0.9578822962939739,
      "epoch": 0.538860103626943,
      "grad_norm": 0.4532525837421417,
      "learning_rate": 0.00019096449367627977,
      "loss": 0.9656,
      "mean_token_accuracy": 0.7934875927865506,
      "num_tokens": 6684053.0,
      "step": 2600
    },
    {
      "entropy": 0.9569033116102219,
      "epoch": 0.5430051813471503,
      "grad_norm": 0.5332207679748535,
      "learning_rate": 0.00019077374390801455,
      "loss": 0.9788,
      "mean_token_accuracy": 0.7917719252407551,
      "num_tokens": 6733968.0,
      "step": 2620
    },
    {
      "entropy": 0.9321629591286182,
      "epoch": 0.5471502590673575,
      "grad_norm": 0.4724487364292145,
      "learning_rate": 0.00019058109895887478,
      "loss": 0.9322,
      "mean_token_accuracy": 0.7971892558038235,
      "num_tokens": 6785674.0,
      "step": 2640
    },
    {
      "entropy": 0.9211927853524685,
      "epoch": 0.5512953367875648,
      "grad_norm": 0.4107973277568817,
      "learning_rate": 0.0001903865628509158,
      "loss": 0.9379,
      "mean_token_accuracy": 0.7973543092608452,
      "num_tokens": 6837186.0,
      "step": 2660
    },
    {
      "entropy": 0.9783548146486283,
      "epoch": 0.555440414507772,
      "grad_norm": 0.4260256588459015,
      "learning_rate": 0.0001901901396456768,
      "loss": 0.986,
      "mean_token_accuracy": 0.7868110507726669,
      "num_tokens": 6889132.0,
      "step": 2680
    },
    {
      "entropy": 0.9350843414664268,
      "epoch": 0.5595854922279793,
      "grad_norm": 0.43470704555511475,
      "learning_rate": 0.00018999183344409578,
      "loss": 0.9496,
      "mean_token_accuracy": 0.796025338023901,
      "num_tokens": 6940037.0,
      "step": 2700
    },
    {
      "entropy": 0.9442426189780235,
      "epoch": 0.5637305699481865,
      "grad_norm": 0.5161582827568054,
      "learning_rate": 0.00018979164838642417,
      "loss": 0.9485,
      "mean_token_accuracy": 0.7975613526999951,
      "num_tokens": 6990041.0,
      "step": 2720
    },
    {
      "entropy": 0.9514821983873845,
      "epoch": 0.5678756476683938,
      "grad_norm": 0.35760676860809326,
      "learning_rate": 0.00018958958865214026,
      "loss": 0.9622,
      "mean_token_accuracy": 0.7867028988897801,
      "num_tokens": 7044442.0,
      "step": 2740
    },
    {
      "entropy": 0.9858744382858277,
      "epoch": 0.572020725388601,
      "grad_norm": 0.4453827738761902,
      "learning_rate": 0.00018938565845986197,
      "loss": 1.0003,
      "mean_token_accuracy": 0.7861887782812118,
      "num_tokens": 7093249.0,
      "step": 2760
    },
    {
      "entropy": 0.9741125054657459,
      "epoch": 0.5761658031088083,
      "grad_norm": 0.433883935213089,
      "learning_rate": 0.00018917986206725887,
      "loss": 0.9804,
      "mean_token_accuracy": 0.7858184657990932,
      "num_tokens": 7144603.0,
      "step": 2780
    },
    {
      "entropy": 0.9409362945705653,
      "epoch": 0.5803108808290155,
      "grad_norm": 0.41870972514152527,
      "learning_rate": 0.000188972203770963,
      "loss": 0.9447,
      "mean_token_accuracy": 0.7908757150173187,
      "num_tokens": 7196254.0,
      "step": 2800
    },
    {
      "entropy": 0.9704828377813101,
      "epoch": 0.5844559585492228,
      "grad_norm": 0.37623921036720276,
      "learning_rate": 0.00018876268790647956,
      "loss": 0.9889,
      "mean_token_accuracy": 0.7871991626918315,
      "num_tokens": 7248841.0,
      "step": 2820
    },
    {
      "entropy": 0.9590335190296173,
      "epoch": 0.5886010362694301,
      "grad_norm": 0.42611730098724365,
      "learning_rate": 0.00018855131884809608,
      "loss": 0.9564,
      "mean_token_accuracy": 0.7912212461233139,
      "num_tokens": 7300448.0,
      "step": 2840
    },
    {
      "entropy": 0.9557400576770305,
      "epoch": 0.5927461139896373,
      "grad_norm": 0.41997987031936646,
      "learning_rate": 0.00018833810100879128,
      "loss": 0.9629,
      "mean_token_accuracy": 0.7843785919249058,
      "num_tokens": 7353316.0,
      "step": 2860
    },
    {
      "entropy": 0.962028032168746,
      "epoch": 0.5968911917098445,
      "grad_norm": 0.41813725233078003,
      "learning_rate": 0.00018812303884014283,
      "loss": 0.975,
      "mean_token_accuracy": 0.792021531611681,
      "num_tokens": 7402745.0,
      "step": 2880
    },
    {
      "entropy": 0.9566993176937103,
      "epoch": 0.6010362694300518,
      "grad_norm": 0.40095973014831543,
      "learning_rate": 0.00018790613683223437,
      "loss": 0.9677,
      "mean_token_accuracy": 0.7910853162407875,
      "num_tokens": 7453651.0,
      "step": 2900
    },
    {
      "entropy": 0.9973684482276439,
      "epoch": 0.6051813471502591,
      "grad_norm": 0.4190211296081543,
      "learning_rate": 0.00018768739951356198,
      "loss": 1.0137,
      "mean_token_accuracy": 0.7809802234172821,
      "num_tokens": 7507052.0,
      "step": 2920
    },
    {
      "entropy": 0.934837969392538,
      "epoch": 0.6093264248704663,
      "grad_norm": 0.4074060916900635,
      "learning_rate": 0.00018746683145093939,
      "loss": 0.9397,
      "mean_token_accuracy": 0.7906834915280342,
      "num_tokens": 7561383.0,
      "step": 2940
    },
    {
      "entropy": 0.950137410312891,
      "epoch": 0.6134715025906736,
      "grad_norm": 0.397320032119751,
      "learning_rate": 0.0001872444372494028,
      "loss": 0.9578,
      "mean_token_accuracy": 0.7921940423548222,
      "num_tokens": 7611992.0,
      "step": 2960
    },
    {
      "entropy": 0.9245664164423942,
      "epoch": 0.6176165803108808,
      "grad_norm": 0.4212367534637451,
      "learning_rate": 0.00018702022155211458,
      "loss": 0.9547,
      "mean_token_accuracy": 0.7904628373682498,
      "num_tokens": 7664532.0,
      "step": 2980
    },
    {
      "entropy": 0.9380590286105871,
      "epoch": 0.6217616580310881,
      "grad_norm": 0.3710251748561859,
      "learning_rate": 0.00018679418904026654,
      "loss": 0.9341,
      "mean_token_accuracy": 0.7926829554140568,
      "num_tokens": 7715917.0,
      "step": 3000
    },
    {
      "epoch": 0.6217616580310881,
      "eval_entropy": 0.9522759174033413,
      "eval_loss": 0.9706581830978394,
      "eval_mean_token_accuracy": 0.7886052683113128,
      "eval_num_tokens": 7715917.0,
      "eval_runtime": 197.1854,
      "eval_samples_per_second": 20.605,
      "eval_steps_per_second": 2.576,
      "step": 3000
    },
    {
      "entropy": 0.9760079063475132,
      "epoch": 0.6259067357512953,
      "grad_norm": 0.38734835386276245,
      "learning_rate": 0.00018656634443298202,
      "loss": 0.9954,
      "mean_token_accuracy": 0.7867716372013092,
      "num_tokens": 7769962.0,
      "step": 3020
    },
    {
      "entropy": 0.9858164794743061,
      "epoch": 0.6300518134715026,
      "grad_norm": 0.4148380160331726,
      "learning_rate": 0.00018633669248721736,
      "loss": 0.9893,
      "mean_token_accuracy": 0.785686731338501,
      "num_tokens": 7824173.0,
      "step": 3040
    },
    {
      "entropy": 0.9614122346043587,
      "epoch": 0.6341968911917099,
      "grad_norm": 0.4662570357322693,
      "learning_rate": 0.00018610523799766275,
      "loss": 0.9708,
      "mean_token_accuracy": 0.7884422443807125,
      "num_tokens": 7874061.0,
      "step": 3060
    },
    {
      "entropy": 0.9372418120503425,
      "epoch": 0.6383419689119171,
      "grad_norm": 0.43578842282295227,
      "learning_rate": 0.000185871985796642,
      "loss": 0.9501,
      "mean_token_accuracy": 0.7922198139131069,
      "num_tokens": 7925220.0,
      "step": 3080
    },
    {
      "entropy": 0.9477806560695171,
      "epoch": 0.6424870466321243,
      "grad_norm": 0.43452900648117065,
      "learning_rate": 0.0001856369407540116,
      "loss": 0.9672,
      "mean_token_accuracy": 0.7929503574967385,
      "num_tokens": 7977422.0,
      "step": 3100
    },
    {
      "entropy": 0.9531604364514351,
      "epoch": 0.6466321243523316,
      "grad_norm": 0.46345794200897217,
      "learning_rate": 0.0001854001077770592,
      "loss": 0.9618,
      "mean_token_accuracy": 0.7890139617025852,
      "num_tokens": 8030901.0,
      "step": 3120
    },
    {
      "entropy": 0.9154306463897228,
      "epoch": 0.6507772020725389,
      "grad_norm": 0.44431430101394653,
      "learning_rate": 0.000185161491810401,
      "loss": 0.9283,
      "mean_token_accuracy": 0.8002380356192589,
      "num_tokens": 8082737.0,
      "step": 3140
    },
    {
      "entropy": 0.9435916144400835,
      "epoch": 0.6549222797927461,
      "grad_norm": 0.3655269145965576,
      "learning_rate": 0.00018492109783587867,
      "loss": 0.9571,
      "mean_token_accuracy": 0.7891912691295146,
      "num_tokens": 8133737.0,
      "step": 3160
    },
    {
      "entropy": 0.9871173314750195,
      "epoch": 0.6590673575129534,
      "grad_norm": 0.45032334327697754,
      "learning_rate": 0.00018467893087245526,
      "loss": 0.9938,
      "mean_token_accuracy": 0.7833716459572315,
      "num_tokens": 8183945.0,
      "step": 3180
    },
    {
      "entropy": 0.9330227866768837,
      "epoch": 0.6632124352331606,
      "grad_norm": 0.4046294093132019,
      "learning_rate": 0.00018443499597611024,
      "loss": 0.9434,
      "mean_token_accuracy": 0.7933455288410187,
      "num_tokens": 8235327.0,
      "step": 3200
    },
    {
      "entropy": 0.9379861585795879,
      "epoch": 0.6673575129533679,
      "grad_norm": 0.4296433627605438,
      "learning_rate": 0.0001841892982397344,
      "loss": 0.9527,
      "mean_token_accuracy": 0.7907838642597198,
      "num_tokens": 8286392.0,
      "step": 3220
    },
    {
      "entropy": 0.9368982739746571,
      "epoch": 0.6715025906735751,
      "grad_norm": 0.5243311524391174,
      "learning_rate": 0.00018394184279302294,
      "loss": 0.9464,
      "mean_token_accuracy": 0.7926255285739898,
      "num_tokens": 8335527.0,
      "step": 3240
    },
    {
      "entropy": 0.9379900746047497,
      "epoch": 0.6756476683937824,
      "grad_norm": 0.397651344537735,
      "learning_rate": 0.0001836926348023689,
      "loss": 0.9504,
      "mean_token_accuracy": 0.791745250672102,
      "num_tokens": 8388161.0,
      "step": 3260
    },
    {
      "entropy": 0.9145612478256225,
      "epoch": 0.6797927461139897,
      "grad_norm": 0.43967580795288086,
      "learning_rate": 0.0001834416794707549,
      "loss": 0.9165,
      "mean_token_accuracy": 0.7961649537086487,
      "num_tokens": 8441889.0,
      "step": 3280
    },
    {
      "entropy": 0.926224023476243,
      "epoch": 0.6839378238341969,
      "grad_norm": 0.45205602049827576,
      "learning_rate": 0.0001831889820376448,
      "loss": 0.939,
      "mean_token_accuracy": 0.7933929577469826,
      "num_tokens": 8490987.0,
      "step": 3300
    },
    {
      "entropy": 0.9750065371394158,
      "epoch": 0.6880829015544041,
      "grad_norm": 0.43906891345977783,
      "learning_rate": 0.00018293454777887404,
      "loss": 0.9809,
      "mean_token_accuracy": 0.7898602753877639,
      "num_tokens": 8540947.0,
      "step": 3320
    },
    {
      "entropy": 0.9698948271572589,
      "epoch": 0.6922279792746114,
      "grad_norm": 0.4355306327342987,
      "learning_rate": 0.00018267838200653977,
      "loss": 0.9721,
      "mean_token_accuracy": 0.785855708271265,
      "num_tokens": 8592141.0,
      "step": 3340
    },
    {
      "entropy": 0.9159326329827309,
      "epoch": 0.6963730569948187,
      "grad_norm": 0.37019234895706177,
      "learning_rate": 0.00018242049006888974,
      "loss": 0.9234,
      "mean_token_accuracy": 0.8000622071325779,
      "num_tokens": 8642193.0,
      "step": 3360
    },
    {
      "entropy": 0.8978213213384152,
      "epoch": 0.7005181347150259,
      "grad_norm": 0.3895004391670227,
      "learning_rate": 0.00018216087735021072,
      "loss": 0.9101,
      "mean_token_accuracy": 0.7981852144002914,
      "num_tokens": 8693653.0,
      "step": 3380
    },
    {
      "entropy": 0.9253171287477017,
      "epoch": 0.7046632124352331,
      "grad_norm": 0.40894782543182373,
      "learning_rate": 0.00018189954927071603,
      "loss": 0.9329,
      "mean_token_accuracy": 0.7947318039834499,
      "num_tokens": 8744455.0,
      "step": 3400
    },
    {
      "entropy": 0.8997811481356621,
      "epoch": 0.7088082901554404,
      "grad_norm": 0.39331623911857605,
      "learning_rate": 0.00018163651128643248,
      "loss": 0.9036,
      "mean_token_accuracy": 0.8017535306513309,
      "num_tokens": 8796115.0,
      "step": 3420
    },
    {
      "entropy": 0.9271700248122215,
      "epoch": 0.7129533678756477,
      "grad_norm": 0.42100563645362854,
      "learning_rate": 0.0001813717688890864,
      "loss": 0.9429,
      "mean_token_accuracy": 0.7925452619791031,
      "num_tokens": 8846934.0,
      "step": 3440
    },
    {
      "entropy": 0.9542303301393986,
      "epoch": 0.7170984455958549,
      "grad_norm": 0.4123167395591736,
      "learning_rate": 0.00018110532760598895,
      "loss": 0.9613,
      "mean_token_accuracy": 0.7912637561559677,
      "num_tokens": 8895911.0,
      "step": 3460
    },
    {
      "entropy": 0.9155400857329369,
      "epoch": 0.7212435233160622,
      "grad_norm": 0.3995860517024994,
      "learning_rate": 0.00018083719299992067,
      "loss": 0.9283,
      "mean_token_accuracy": 0.796960674971342,
      "num_tokens": 8945784.0,
      "step": 3480
    },
    {
      "entropy": 0.934383400529623,
      "epoch": 0.7253886010362695,
      "grad_norm": 0.4636269509792328,
      "learning_rate": 0.00018056737066901563,
      "loss": 0.9472,
      "mean_token_accuracy": 0.7902762576937675,
      "num_tokens": 8996927.0,
      "step": 3500
    },
    {
      "epoch": 0.7253886010362695,
      "eval_entropy": 0.9803924484281089,
      "eval_loss": 0.9615499973297119,
      "eval_mean_token_accuracy": 0.7901124554121588,
      "eval_num_tokens": 8996927.0,
      "eval_runtime": 197.2805,
      "eval_samples_per_second": 20.595,
      "eval_steps_per_second": 2.575,
      "step": 3500
    },
    {
      "entropy": 0.9384146019816398,
      "epoch": 0.7295336787564767,
      "grad_norm": 0.39725178480148315,
      "learning_rate": 0.00018029586624664415,
      "loss": 0.9441,
      "mean_token_accuracy": 0.791342943161726,
      "num_tokens": 9047604.0,
      "step": 3520
    },
    {
      "entropy": 0.9334720097482204,
      "epoch": 0.7336787564766839,
      "grad_norm": 0.3954644501209259,
      "learning_rate": 0.00018002268540129543,
      "loss": 0.9417,
      "mean_token_accuracy": 0.795846226811409,
      "num_tokens": 9099017.0,
      "step": 3540
    },
    {
      "entropy": 0.907962579652667,
      "epoch": 0.7378238341968912,
      "grad_norm": 0.39940011501312256,
      "learning_rate": 0.0001797478338364592,
      "loss": 0.9206,
      "mean_token_accuracy": 0.7963109843432903,
      "num_tokens": 9149326.0,
      "step": 3560
    },
    {
      "entropy": 0.9557456441223622,
      "epoch": 0.7419689119170985,
      "grad_norm": 0.4359956383705139,
      "learning_rate": 0.0001794713172905066,
      "loss": 0.9658,
      "mean_token_accuracy": 0.7891202479600906,
      "num_tokens": 9200715.0,
      "step": 3580
    },
    {
      "entropy": 0.9499155536293984,
      "epoch": 0.7461139896373057,
      "grad_norm": 0.507350504398346,
      "learning_rate": 0.00017919314153657027,
      "loss": 0.9634,
      "mean_token_accuracy": 0.7931549407541751,
      "num_tokens": 9251534.0,
      "step": 3600
    },
    {
      "entropy": 0.9438272595405579,
      "epoch": 0.7502590673575129,
      "grad_norm": 0.4581449329853058,
      "learning_rate": 0.000178913312382424,
      "loss": 0.9584,
      "mean_token_accuracy": 0.7942813694477081,
      "num_tokens": 9302614.0,
      "step": 3620
    },
    {
      "entropy": 0.9297045323997736,
      "epoch": 0.7544041450777202,
      "grad_norm": 0.35071489214897156,
      "learning_rate": 0.00017863183567036137,
      "loss": 0.9414,
      "mean_token_accuracy": 0.7904184371232986,
      "num_tokens": 9353455.0,
      "step": 3640
    },
    {
      "entropy": 0.9570980660617352,
      "epoch": 0.7585492227979275,
      "grad_norm": 0.4017302989959717,
      "learning_rate": 0.0001783487172770738,
      "loss": 0.9741,
      "mean_token_accuracy": 0.7883284002542496,
      "num_tokens": 9405502.0,
      "step": 3660
    },
    {
      "entropy": 0.9436403628438711,
      "epoch": 0.7626943005181347,
      "grad_norm": 0.5486421585083008,
      "learning_rate": 0.00017806396311352782,
      "loss": 0.9596,
      "mean_token_accuracy": 0.7928663790225983,
      "num_tokens": 9455621.0,
      "step": 3680
    },
    {
      "entropy": 0.9594994522631168,
      "epoch": 0.7668393782383419,
      "grad_norm": 0.4096115231513977,
      "learning_rate": 0.00017777757912484175,
      "loss": 0.973,
      "mean_token_accuracy": 0.7888169556856155,
      "num_tokens": 9507088.0,
      "step": 3700
    },
    {
      "entropy": 0.940360489487648,
      "epoch": 0.7709844559585493,
      "grad_norm": 0.3976588845252991,
      "learning_rate": 0.00017748957129016147,
      "loss": 0.9386,
      "mean_token_accuracy": 0.7945226788520813,
      "num_tokens": 9558157.0,
      "step": 3720
    },
    {
      "entropy": 0.946375384926796,
      "epoch": 0.7751295336787565,
      "grad_norm": 0.4322788715362549,
      "learning_rate": 0.00017719994562253558,
      "loss": 0.951,
      "mean_token_accuracy": 0.7903898723423481,
      "num_tokens": 9611510.0,
      "step": 3740
    },
    {
      "entropy": 0.9477811448276043,
      "epoch": 0.7792746113989637,
      "grad_norm": 0.43729427456855774,
      "learning_rate": 0.00017690870816879014,
      "loss": 0.9661,
      "mean_token_accuracy": 0.7892401278018951,
      "num_tokens": 9662287.0,
      "step": 3760
    },
    {
      "entropy": 0.949357783794403,
      "epoch": 0.783419689119171,
      "grad_norm": 0.3893526792526245,
      "learning_rate": 0.0001766158650094019,
      "loss": 0.9518,
      "mean_token_accuracy": 0.793975007534027,
      "num_tokens": 9715516.0,
      "step": 3780
    },
    {
      "entropy": 0.9205279342830182,
      "epoch": 0.7875647668393783,
      "grad_norm": 0.46602073311805725,
      "learning_rate": 0.00017632142225837183,
      "loss": 0.9304,
      "mean_token_accuracy": 0.7955152101814746,
      "num_tokens": 9766133.0,
      "step": 3800
    },
    {
      "entropy": 0.9258571691811085,
      "epoch": 0.7917098445595855,
      "grad_norm": 0.39727622270584106,
      "learning_rate": 0.0001760253860630973,
      "loss": 0.9364,
      "mean_token_accuracy": 0.7936219781637192,
      "num_tokens": 9817516.0,
      "step": 3820
    },
    {
      "entropy": 0.9301609247922897,
      "epoch": 0.7958549222797927,
      "grad_norm": 0.43491441011428833,
      "learning_rate": 0.0001757277626042436,
      "loss": 0.9432,
      "mean_token_accuracy": 0.7936487175524235,
      "num_tokens": 9871371.0,
      "step": 3840
    },
    {
      "entropy": 0.9301216453313828,
      "epoch": 0.8,
      "grad_norm": 0.37254294753074646,
      "learning_rate": 0.0001754285580956152,
      "loss": 0.9328,
      "mean_token_accuracy": 0.7928548909723758,
      "num_tokens": 9923261.0,
      "step": 3860
    },
    {
      "entropy": 0.9340247794985771,
      "epoch": 0.8041450777202073,
      "grad_norm": 0.4496139585971832,
      "learning_rate": 0.00017512777878402568,
      "loss": 0.9544,
      "mean_token_accuracy": 0.7901494115591049,
      "num_tokens": 9973379.0,
      "step": 3880
    },
    {
      "entropy": 0.9295564122498036,
      "epoch": 0.8082901554404145,
      "grad_norm": 0.470420777797699,
      "learning_rate": 0.00017482543094916758,
      "loss": 0.9383,
      "mean_token_accuracy": 0.7970516562461853,
      "num_tokens": 10024029.0,
      "step": 3900
    },
    {
      "entropy": 0.936601135879755,
      "epoch": 0.8124352331606217,
      "grad_norm": 0.38742539286613464,
      "learning_rate": 0.0001745215209034811,
      "loss": 0.9611,
      "mean_token_accuracy": 0.7904755167663098,
      "num_tokens": 10077153.0,
      "step": 3920
    },
    {
      "entropy": 0.9289965465664863,
      "epoch": 0.816580310880829,
      "grad_norm": 0.4919315278530121,
      "learning_rate": 0.00017421605499202242,
      "loss": 0.9364,
      "mean_token_accuracy": 0.7947124816477299,
      "num_tokens": 10128837.0,
      "step": 3940
    },
    {
      "entropy": 0.9195003911852837,
      "epoch": 0.8207253886010363,
      "grad_norm": 0.43714791536331177,
      "learning_rate": 0.0001739090395923313,
      "loss": 0.9194,
      "mean_token_accuracy": 0.796821691840887,
      "num_tokens": 10180279.0,
      "step": 3960
    },
    {
      "entropy": 0.9256344065070152,
      "epoch": 0.8248704663212435,
      "grad_norm": 0.3864556550979614,
      "learning_rate": 0.00017360048111429767,
      "loss": 0.9483,
      "mean_token_accuracy": 0.7930616028606892,
      "num_tokens": 10230973.0,
      "step": 3980
    },
    {
      "entropy": 0.9640789791941643,
      "epoch": 0.8290155440414507,
      "grad_norm": 0.4234633147716522,
      "learning_rate": 0.00017329038600002807,
      "loss": 0.9681,
      "mean_token_accuracy": 0.7887310594320297,
      "num_tokens": 10283849.0,
      "step": 4000
    },
    {
      "epoch": 0.8290155440414507,
      "eval_entropy": 0.9277598650202038,
      "eval_loss": 0.9486145973205566,
      "eval_mean_token_accuracy": 0.7921127796642423,
      "eval_num_tokens": 10283849.0,
      "eval_runtime": 196.6351,
      "eval_samples_per_second": 20.663,
      "eval_steps_per_second": 2.583,
      "step": 4000
    },
    {
      "entropy": 0.939198125898838,
      "epoch": 0.8331606217616581,
      "grad_norm": 0.40732425451278687,
      "learning_rate": 0.00017297876072371103,
      "loss": 0.9512,
      "mean_token_accuracy": 0.7914685815572738,
      "num_tokens": 10336653.0,
      "step": 4020
    },
    {
      "entropy": 0.918938459455967,
      "epoch": 0.8373056994818653,
      "grad_norm": 0.5398784875869751,
      "learning_rate": 0.00017266561179148183,
      "loss": 0.9186,
      "mean_token_accuracy": 0.7927700787782669,
      "num_tokens": 10387330.0,
      "step": 4040
    },
    {
      "entropy": 0.9128594268113375,
      "epoch": 0.8414507772020725,
      "grad_norm": 0.43407782912254333,
      "learning_rate": 0.0001723509457412869,
      "loss": 0.931,
      "mean_token_accuracy": 0.7971027828752995,
      "num_tokens": 10438933.0,
      "step": 4060
    },
    {
      "entropy": 0.9039206847548484,
      "epoch": 0.8455958549222798,
      "grad_norm": 0.46065929532051086,
      "learning_rate": 0.000172034769142747,
      "loss": 0.9192,
      "mean_token_accuracy": 0.7961257986724377,
      "num_tokens": 10491722.0,
      "step": 4080
    },
    {
      "entropy": 0.9538728334009647,
      "epoch": 0.8497409326424871,
      "grad_norm": 0.45542430877685547,
      "learning_rate": 0.00017171708859702037,
      "loss": 0.9573,
      "mean_token_accuracy": 0.7893645465373993,
      "num_tokens": 10546885.0,
      "step": 4100
    },
    {
      "entropy": 0.9198624655604363,
      "epoch": 0.8538860103626943,
      "grad_norm": 0.3631950318813324,
      "learning_rate": 0.00017139791073666468,
      "loss": 0.9313,
      "mean_token_accuracy": 0.7931442521512508,
      "num_tokens": 10600746.0,
      "step": 4120
    },
    {
      "entropy": 0.9213518410921097,
      "epoch": 0.8580310880829015,
      "grad_norm": 0.4949886202812195,
      "learning_rate": 0.00017107724222549872,
      "loss": 0.9335,
      "mean_token_accuracy": 0.7955467194318772,
      "num_tokens": 10653477.0,
      "step": 4140
    },
    {
      "entropy": 0.9536297917366028,
      "epoch": 0.8621761658031089,
      "grad_norm": 0.4028345048427582,
      "learning_rate": 0.00017075508975846311,
      "loss": 0.9605,
      "mean_token_accuracy": 0.7890767879784107,
      "num_tokens": 10706433.0,
      "step": 4160
    },
    {
      "entropy": 0.9051979005336761,
      "epoch": 0.8663212435233161,
      "grad_norm": 0.4099532663822174,
      "learning_rate": 0.0001704314600614807,
      "loss": 0.9178,
      "mean_token_accuracy": 0.7976575642824173,
      "num_tokens": 10758950.0,
      "step": 4180
    },
    {
      "entropy": 0.9456002049148082,
      "epoch": 0.8704663212435233,
      "grad_norm": 0.44490471482276917,
      "learning_rate": 0.00017010635989131597,
      "loss": 0.9531,
      "mean_token_accuracy": 0.791679710894823,
      "num_tokens": 10810783.0,
      "step": 4200
    },
    {
      "entropy": 0.9459167554974556,
      "epoch": 0.8746113989637305,
      "grad_norm": 0.4577106535434723,
      "learning_rate": 0.0001697797960354341,
      "loss": 0.9609,
      "mean_token_accuracy": 0.7873589277267456,
      "num_tokens": 10863258.0,
      "step": 4220
    },
    {
      "entropy": 0.8723687015473842,
      "epoch": 0.8787564766839379,
      "grad_norm": 0.40351179242134094,
      "learning_rate": 0.00016945177531185922,
      "loss": 0.8735,
      "mean_token_accuracy": 0.8079392656683921,
      "num_tokens": 10914577.0,
      "step": 4240
    },
    {
      "entropy": 0.9348406352102756,
      "epoch": 0.8829015544041451,
      "grad_norm": 0.42083558440208435,
      "learning_rate": 0.00016912230456903198,
      "loss": 0.9457,
      "mean_token_accuracy": 0.7947359785437584,
      "num_tokens": 10966052.0,
      "step": 4260
    },
    {
      "entropy": 0.892266447097063,
      "epoch": 0.8870466321243523,
      "grad_norm": 0.413835346698761,
      "learning_rate": 0.0001687913906856667,
      "loss": 0.8917,
      "mean_token_accuracy": 0.7988453857600689,
      "num_tokens": 11017837.0,
      "step": 4280
    },
    {
      "entropy": 0.9510011091828346,
      "epoch": 0.8911917098445595,
      "grad_norm": 0.41305822134017944,
      "learning_rate": 0.00016845904057060767,
      "loss": 0.9565,
      "mean_token_accuracy": 0.7917753420770168,
      "num_tokens": 11068955.0,
      "step": 4300
    },
    {
      "entropy": 0.9034442327916622,
      "epoch": 0.8953367875647669,
      "grad_norm": 0.42375388741493225,
      "learning_rate": 0.000168125261162685,
      "loss": 0.9089,
      "mean_token_accuracy": 0.7968145549297333,
      "num_tokens": 11119250.0,
      "step": 4320
    },
    {
      "entropy": 0.9643872417509556,
      "epoch": 0.8994818652849741,
      "grad_norm": 0.4074117839336395,
      "learning_rate": 0.0001677900594305695,
      "loss": 0.9788,
      "mean_token_accuracy": 0.7875596329569816,
      "num_tokens": 11172260.0,
      "step": 4340
    },
    {
      "entropy": 0.92549347281456,
      "epoch": 0.9036269430051813,
      "grad_norm": 0.456617534160614,
      "learning_rate": 0.0001674534423726276,
      "loss": 0.9426,
      "mean_token_accuracy": 0.7943698972463608,
      "num_tokens": 11225236.0,
      "step": 4360
    },
    {
      "entropy": 0.9161198817193508,
      "epoch": 0.9077720207253887,
      "grad_norm": 0.5100305080413818,
      "learning_rate": 0.0001671154170167748,
      "loss": 0.9266,
      "mean_token_accuracy": 0.7931325800716877,
      "num_tokens": 11277166.0,
      "step": 4380
    },
    {
      "entropy": 0.8795958921313286,
      "epoch": 0.9119170984455959,
      "grad_norm": 0.4124648869037628,
      "learning_rate": 0.00016677599042032933,
      "loss": 0.8845,
      "mean_token_accuracy": 0.8016303740441799,
      "num_tokens": 11328093.0,
      "step": 4400
    },
    {
      "entropy": 0.9241185076534748,
      "epoch": 0.9160621761658031,
      "grad_norm": 0.47688350081443787,
      "learning_rate": 0.00016643516966986446,
      "loss": 0.9312,
      "mean_token_accuracy": 0.7907878324389458,
      "num_tokens": 11380215.0,
      "step": 4420
    },
    {
      "entropy": 0.8871541582047939,
      "epoch": 0.9202072538860103,
      "grad_norm": 0.44556674361228943,
      "learning_rate": 0.00016609296188106084,
      "loss": 0.9016,
      "mean_token_accuracy": 0.7992243818938732,
      "num_tokens": 11432104.0,
      "step": 4440
    },
    {
      "entropy": 0.9216676332056523,
      "epoch": 0.9243523316062177,
      "grad_norm": 0.43731555342674255,
      "learning_rate": 0.00016574937419855772,
      "loss": 0.9234,
      "mean_token_accuracy": 0.795772710442543,
      "num_tokens": 11482908.0,
      "step": 4460
    },
    {
      "entropy": 0.9152931831777096,
      "epoch": 0.9284974093264249,
      "grad_norm": 0.40907105803489685,
      "learning_rate": 0.00016540441379580387,
      "loss": 0.9271,
      "mean_token_accuracy": 0.7936010919511318,
      "num_tokens": 11535803.0,
      "step": 4480
    },
    {
      "entropy": 0.941212996840477,
      "epoch": 0.9326424870466321,
      "grad_norm": 0.469002902507782,
      "learning_rate": 0.00016505808787490792,
      "loss": 0.9491,
      "mean_token_accuracy": 0.793215012550354,
      "num_tokens": 11586430.0,
      "step": 4500
    },
    {
      "epoch": 0.9326424870466321,
      "eval_entropy": 0.8924233510972946,
      "eval_loss": 0.9386545419692993,
      "eval_mean_token_accuracy": 0.7933616472745505,
      "eval_num_tokens": 11586430.0,
      "eval_runtime": 197.7861,
      "eval_samples_per_second": 20.542,
      "eval_steps_per_second": 2.568,
      "step": 4500
    },
    {
      "entropy": 0.8890458445996046,
      "epoch": 0.9367875647668393,
      "grad_norm": 0.3855128884315491,
      "learning_rate": 0.00016471040366648776,
      "loss": 0.8964,
      "mean_token_accuracy": 0.8021313309669494,
      "num_tokens": 11639117.0,
      "step": 4520
    },
    {
      "entropy": 0.9057373393326997,
      "epoch": 0.9409326424870467,
      "grad_norm": 0.46859732270240784,
      "learning_rate": 0.00016436136842951977,
      "loss": 0.9213,
      "mean_token_accuracy": 0.7989621929824352,
      "num_tokens": 11689267.0,
      "step": 4540
    },
    {
      "entropy": 0.9198335148394108,
      "epoch": 0.9450777202072539,
      "grad_norm": 0.40903231501579285,
      "learning_rate": 0.0001640109894511872,
      "loss": 0.9284,
      "mean_token_accuracy": 0.7965403601527214,
      "num_tokens": 11740945.0,
      "step": 4560
    },
    {
      "entropy": 0.9119356460869312,
      "epoch": 0.9492227979274611,
      "grad_norm": 0.42043864727020264,
      "learning_rate": 0.00016365927404672807,
      "loss": 0.9249,
      "mean_token_accuracy": 0.796663673222065,
      "num_tokens": 11794600.0,
      "step": 4580
    },
    {
      "entropy": 0.9416319027543067,
      "epoch": 0.9533678756476683,
      "grad_norm": 0.41366055607795715,
      "learning_rate": 0.0001633062295592823,
      "loss": 0.9583,
      "mean_token_accuracy": 0.7909570150077343,
      "num_tokens": 11844693.0,
      "step": 4600
    },
    {
      "entropy": 0.9526241350919008,
      "epoch": 0.9575129533678757,
      "grad_norm": 0.41388946771621704,
      "learning_rate": 0.00016295186335973857,
      "loss": 0.9674,
      "mean_token_accuracy": 0.790430235862732,
      "num_tokens": 11893941.0,
      "step": 4620
    },
    {
      "entropy": 0.9134444534778595,
      "epoch": 0.9616580310880829,
      "grad_norm": 0.3884899914264679,
      "learning_rate": 0.0001625961828465804,
      "loss": 0.929,
      "mean_token_accuracy": 0.7953568585216999,
      "num_tokens": 11945935.0,
      "step": 4640
    },
    {
      "entropy": 0.9262224324047565,
      "epoch": 0.9658031088082901,
      "grad_norm": 0.4220770299434662,
      "learning_rate": 0.0001622391954457315,
      "loss": 0.9259,
      "mean_token_accuracy": 0.7925183959305286,
      "num_tokens": 11999075.0,
      "step": 4660
    },
    {
      "entropy": 0.931026178970933,
      "epoch": 0.9699481865284975,
      "grad_norm": 0.4672662615776062,
      "learning_rate": 0.00016188090861040107,
      "loss": 0.9433,
      "mean_token_accuracy": 0.7952560156583786,
      "num_tokens": 12047725.0,
      "step": 4680
    },
    {
      "entropy": 0.9058261185884475,
      "epoch": 0.9740932642487047,
      "grad_norm": 0.48223602771759033,
      "learning_rate": 0.00016152132982092784,
      "loss": 0.9122,
      "mean_token_accuracy": 0.7959005422890186,
      "num_tokens": 12099181.0,
      "step": 4700
    },
    {
      "entropy": 0.9087510220706463,
      "epoch": 0.9782383419689119,
      "grad_norm": 0.423324316740036,
      "learning_rate": 0.00016116046658462414,
      "loss": 0.9114,
      "mean_token_accuracy": 0.7989039994776249,
      "num_tokens": 12152610.0,
      "step": 4720
    },
    {
      "entropy": 0.9042955547571182,
      "epoch": 0.9823834196891191,
      "grad_norm": 0.46391963958740234,
      "learning_rate": 0.00016079832643561904,
      "loss": 0.9199,
      "mean_token_accuracy": 0.7983533076941967,
      "num_tokens": 12202583.0,
      "step": 4740
    },
    {
      "entropy": 0.9194691389799118,
      "epoch": 0.9865284974093265,
      "grad_norm": 0.4457080364227295,
      "learning_rate": 0.00016043491693470114,
      "loss": 0.9331,
      "mean_token_accuracy": 0.7974948555231094,
      "num_tokens": 12253776.0,
      "step": 4760
    },
    {
      "entropy": 0.8977299593389034,
      "epoch": 0.9906735751295337,
      "grad_norm": 0.47106996178627014,
      "learning_rate": 0.00016007024566916058,
      "loss": 0.9083,
      "mean_token_accuracy": 0.795095720142126,
      "num_tokens": 12305175.0,
      "step": 4780
    },
    {
      "entropy": 0.9298750199377537,
      "epoch": 0.9948186528497409,
      "grad_norm": 0.5093875527381897,
      "learning_rate": 0.00015970432025263074,
      "loss": 0.9329,
      "mean_token_accuracy": 0.793617632240057,
      "num_tokens": 12357750.0,
      "step": 4800
    },
    {
      "entropy": 0.9333759970963001,
      "epoch": 0.9989637305699481,
      "grad_norm": 0.41947904229164124,
      "learning_rate": 0.0001593371483249293,
      "loss": 0.9529,
      "mean_token_accuracy": 0.793743172287941,
      "num_tokens": 12409215.0,
      "step": 4820
    },
    {
      "entropy": 0.8986701011657715,
      "epoch": 1.0031088082901554,
      "grad_norm": 0.5347554683685303,
      "learning_rate": 0.00015896873755189868,
      "loss": 0.9002,
      "mean_token_accuracy": 0.8045198410749436,
      "num_tokens": 12460502.0,
      "step": 4840
    },
    {
      "entropy": 0.8791909657418728,
      "epoch": 1.0072538860103626,
      "grad_norm": 0.5265800952911377,
      "learning_rate": 0.00015859909562524594,
      "loss": 0.8929,
      "mean_token_accuracy": 0.8018290206789971,
      "num_tokens": 12510710.0,
      "step": 4860
    },
    {
      "entropy": 0.8866066768765449,
      "epoch": 1.01139896373057,
      "grad_norm": 0.5650938749313354,
      "learning_rate": 0.00015822823026238235,
      "loss": 0.8927,
      "mean_token_accuracy": 0.8024179935455322,
      "num_tokens": 12561257.0,
      "step": 4880
    },
    {
      "entropy": 0.8647308357059955,
      "epoch": 1.0155440414507773,
      "grad_norm": 0.4968569576740265,
      "learning_rate": 0.0001578561492062621,
      "loss": 0.8744,
      "mean_token_accuracy": 0.8047979861497879,
      "num_tokens": 12610438.0,
      "step": 4900
    },
    {
      "entropy": 0.8654443889856338,
      "epoch": 1.0196891191709845,
      "grad_norm": 0.48992252349853516,
      "learning_rate": 0.0001574828602252207,
      "loss": 0.87,
      "mean_token_accuracy": 0.8046348355710506,
      "num_tokens": 12661271.0,
      "step": 4920
    },
    {
      "entropy": 0.8522643312811852,
      "epoch": 1.0238341968911917,
      "grad_norm": 0.4607287049293518,
      "learning_rate": 0.00015710837111281286,
      "loss": 0.8611,
      "mean_token_accuracy": 0.8077139921486378,
      "num_tokens": 12709628.0,
      "step": 4940
    },
    {
      "entropy": 0.8478192627429962,
      "epoch": 1.027979274611399,
      "grad_norm": 0.537442684173584,
      "learning_rate": 0.0001567326896876497,
      "loss": 0.8664,
      "mean_token_accuracy": 0.8070318132638932,
      "num_tokens": 12760895.0,
      "step": 4960
    },
    {
      "entropy": 0.890300364792347,
      "epoch": 1.0321243523316062,
      "grad_norm": 0.4382570683956146,
      "learning_rate": 0.00015635582379323556,
      "loss": 0.8902,
      "mean_token_accuracy": 0.7996452160179615,
      "num_tokens": 12812211.0,
      "step": 4980
    },
    {
      "entropy": 0.8272694803774356,
      "epoch": 1.0362694300518134,
      "grad_norm": 0.41449499130249023,
      "learning_rate": 0.0001559777812978042,
      "loss": 0.8292,
      "mean_token_accuracy": 0.811527369171381,
      "num_tokens": 12861861.0,
      "step": 5000
    },
    {
      "epoch": 1.0362694300518134,
      "eval_entropy": 0.8690557431752287,
      "eval_loss": 0.9333181977272034,
      "eval_mean_token_accuracy": 0.7949006537518163,
      "eval_num_tokens": 12861861.0,
      "eval_runtime": 198.9365,
      "eval_samples_per_second": 20.424,
      "eval_steps_per_second": 2.554,
      "step": 5000
    }
  ],
  "logging_steps": 20,
  "max_steps": 14475,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 4,
        "early_stopping_threshold": 0.005
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 8.874798510427546e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}