{
  "best_metric": 0.6748951077461243,
  "best_model_checkpoint": "miner_id_24/checkpoint-500",
  "epoch": 0.09961647656522389,
  "eval_steps": 100,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00019923295313044778,
      "grad_norm": 0.42145729064941406,
      "learning_rate": 5e-06,
      "loss": 0.6572,
      "step": 1
    },
    {
      "epoch": 0.00019923295313044778,
      "eval_loss": 1.0046130418777466,
      "eval_runtime": 262.6844,
      "eval_samples_per_second": 32.183,
      "eval_steps_per_second": 8.048,
      "step": 1
    },
    {
      "epoch": 0.00039846590626089557,
      "grad_norm": 0.8600642681121826,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 2
    },
    {
      "epoch": 0.0005976988593913434,
      "grad_norm": 0.6406773924827576,
      "learning_rate": 1.5e-05,
      "loss": 0.8492,
      "step": 3
    },
    {
      "epoch": 0.0007969318125217911,
      "grad_norm": 0.6716526746749878,
      "learning_rate": 2e-05,
      "loss": 0.9895,
      "step": 4
    },
    {
      "epoch": 0.0009961647656522388,
      "grad_norm": 0.49085474014282227,
      "learning_rate": 2.5e-05,
      "loss": 0.8941,
      "step": 5
    },
    {
      "epoch": 0.0011953977187826868,
      "grad_norm": 0.4583233594894409,
      "learning_rate": 3e-05,
      "loss": 0.8405,
      "step": 6
    },
    {
      "epoch": 0.0013946306719131345,
      "grad_norm": 0.3206585645675659,
      "learning_rate": 3.5e-05,
      "loss": 0.8055,
      "step": 7
    },
    {
      "epoch": 0.0015938636250435823,
      "grad_norm": 0.36268213391304016,
      "learning_rate": 4e-05,
      "loss": 0.8796,
      "step": 8
    },
    {
      "epoch": 0.00179309657817403,
      "grad_norm": 0.3162664771080017,
      "learning_rate": 4.5e-05,
      "loss": 0.8962,
      "step": 9
    },
    {
      "epoch": 0.0019923295313044776,
      "grad_norm": 0.3157712519168854,
      "learning_rate": 5e-05,
      "loss": 0.9107,
      "step": 10
    },
    {
      "epoch": 0.0021915624844349255,
      "grad_norm": 0.2892014980316162,
      "learning_rate": 5.500000000000001e-05,
      "loss": 0.931,
      "step": 11
    },
    {
      "epoch": 0.0023907954375653735,
      "grad_norm": 0.28385069966316223,
      "learning_rate": 6e-05,
      "loss": 0.8569,
      "step": 12
    },
    {
      "epoch": 0.002590028390695821,
      "grad_norm": 0.3141295909881592,
      "learning_rate": 6.500000000000001e-05,
      "loss": 0.9154,
      "step": 13
    },
    {
      "epoch": 0.002789261343826269,
      "grad_norm": 0.2827235162258148,
      "learning_rate": 7e-05,
      "loss": 0.7631,
      "step": 14
    },
    {
      "epoch": 0.0029884942969567166,
      "grad_norm": 0.31552836298942566,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.8641,
      "step": 15
    },
    {
      "epoch": 0.0031877272500871645,
      "grad_norm": 0.2979770600795746,
      "learning_rate": 8e-05,
      "loss": 0.8825,
      "step": 16
    },
    {
      "epoch": 0.003386960203217612,
      "grad_norm": 0.30036237835884094,
      "learning_rate": 8.5e-05,
      "loss": 0.8403,
      "step": 17
    },
    {
      "epoch": 0.00358619315634806,
      "grad_norm": 0.30925682187080383,
      "learning_rate": 9e-05,
      "loss": 0.8187,
      "step": 18
    },
    {
      "epoch": 0.0037854261094785076,
      "grad_norm": 0.3207469582557678,
      "learning_rate": 9.5e-05,
      "loss": 0.9215,
      "step": 19
    },
    {
      "epoch": 0.003984659062608955,
      "grad_norm": 0.25850704312324524,
      "learning_rate": 0.0001,
      "loss": 0.7215,
      "step": 20
    },
    {
      "epoch": 0.0041838920157394035,
      "grad_norm": 0.2752808630466461,
      "learning_rate": 9.999892908320647e-05,
      "loss": 0.7553,
      "step": 21
    },
    {
      "epoch": 0.004383124968869851,
      "grad_norm": 0.30526119470596313,
      "learning_rate": 9.999571637870036e-05,
      "loss": 0.7467,
      "step": 22
    },
    {
      "epoch": 0.004582357922000299,
      "grad_norm": 0.35731807351112366,
      "learning_rate": 9.999036202410325e-05,
      "loss": 0.832,
      "step": 23
    },
    {
      "epoch": 0.004781590875130747,
      "grad_norm": 0.34211745858192444,
      "learning_rate": 9.998286624877786e-05,
      "loss": 0.8119,
      "step": 24
    },
    {
      "epoch": 0.0049808238282611945,
      "grad_norm": 0.294607549905777,
      "learning_rate": 9.997322937381829e-05,
      "loss": 0.7457,
      "step": 25
    },
    {
      "epoch": 0.005180056781391642,
      "grad_norm": 0.3165847957134247,
      "learning_rate": 9.996145181203615e-05,
      "loss": 0.8323,
      "step": 26
    },
    {
      "epoch": 0.00537928973452209,
      "grad_norm": 0.316105455160141,
      "learning_rate": 9.994753406794301e-05,
      "loss": 0.7302,
      "step": 27
    },
    {
      "epoch": 0.005578522687652538,
      "grad_norm": 0.3156639635562897,
      "learning_rate": 9.99314767377287e-05,
      "loss": 0.8405,
      "step": 28
    },
    {
      "epoch": 0.005777755640782986,
      "grad_norm": 0.28763553500175476,
      "learning_rate": 9.991328050923581e-05,
      "loss": 0.7269,
      "step": 29
    },
    {
      "epoch": 0.005976988593913433,
      "grad_norm": 0.3319118916988373,
      "learning_rate": 9.989294616193017e-05,
      "loss": 0.7607,
      "step": 30
    },
    {
      "epoch": 0.006176221547043881,
      "grad_norm": 0.32836800813674927,
      "learning_rate": 9.98704745668676e-05,
      "loss": 0.867,
      "step": 31
    },
    {
      "epoch": 0.006375454500174329,
      "grad_norm": 0.3175886273384094,
      "learning_rate": 9.98458666866564e-05,
      "loss": 0.7733,
      "step": 32
    },
    {
      "epoch": 0.006574687453304777,
      "grad_norm": 0.31788307428359985,
      "learning_rate": 9.981912357541627e-05,
      "loss": 0.7723,
      "step": 33
    },
    {
      "epoch": 0.006773920406435224,
      "grad_norm": 0.28625956177711487,
      "learning_rate": 9.97902463787331e-05,
      "loss": 0.7911,
      "step": 34
    },
    {
      "epoch": 0.0069731533595656725,
      "grad_norm": 0.28825896978378296,
      "learning_rate": 9.975923633360985e-05,
      "loss": 0.7694,
      "step": 35
    },
    {
      "epoch": 0.00717238631269612,
      "grad_norm": 0.29436683654785156,
      "learning_rate": 9.972609476841367e-05,
      "loss": 0.7453,
      "step": 36
    },
    {
      "epoch": 0.007371619265826568,
      "grad_norm": 0.29670408368110657,
      "learning_rate": 9.969082310281891e-05,
      "loss": 0.7294,
      "step": 37
    },
    {
      "epoch": 0.007570852218957015,
      "grad_norm": 0.27974042296409607,
      "learning_rate": 9.965342284774632e-05,
      "loss": 0.7269,
      "step": 38
    },
    {
      "epoch": 0.0077700851720874636,
      "grad_norm": 0.32981932163238525,
      "learning_rate": 9.961389560529836e-05,
      "loss": 0.8839,
      "step": 39
    },
    {
      "epoch": 0.00796931812521791,
      "grad_norm": 0.2603093683719635,
      "learning_rate": 9.957224306869053e-05,
      "loss": 0.7584,
      "step": 40
    },
    {
      "epoch": 0.008168551078348359,
      "grad_norm": 0.2658812403678894,
      "learning_rate": 9.952846702217886e-05,
      "loss": 0.6587,
      "step": 41
    },
    {
      "epoch": 0.008367784031478807,
      "grad_norm": 0.2873687446117401,
      "learning_rate": 9.948256934098352e-05,
      "loss": 0.7437,
      "step": 42
    },
    {
      "epoch": 0.008567016984609254,
      "grad_norm": 0.29043227434158325,
      "learning_rate": 9.943455199120837e-05,
      "loss": 0.7514,
      "step": 43
    },
    {
      "epoch": 0.008766249937739702,
      "grad_norm": 0.2623406648635864,
      "learning_rate": 9.938441702975689e-05,
      "loss": 0.6613,
      "step": 44
    },
    {
      "epoch": 0.00896548289087015,
      "grad_norm": 0.33088845014572144,
      "learning_rate": 9.933216660424395e-05,
      "loss": 0.7904,
      "step": 45
    },
    {
      "epoch": 0.009164715844000597,
      "grad_norm": 0.3347594439983368,
      "learning_rate": 9.927780295290389e-05,
      "loss": 0.8104,
      "step": 46
    },
    {
      "epoch": 0.009363948797131046,
      "grad_norm": 0.29131022095680237,
      "learning_rate": 9.922132840449459e-05,
      "loss": 0.7463,
      "step": 47
    },
    {
      "epoch": 0.009563181750261494,
      "grad_norm": 0.30381011962890625,
      "learning_rate": 9.916274537819775e-05,
      "loss": 0.8213,
      "step": 48
    },
    {
      "epoch": 0.00976241470339194,
      "grad_norm": 0.29724815487861633,
      "learning_rate": 9.91020563835152e-05,
      "loss": 0.7977,
      "step": 49
    },
    {
      "epoch": 0.009961647656522389,
      "grad_norm": 0.3157656490802765,
      "learning_rate": 9.903926402016153e-05,
      "loss": 0.8148,
      "step": 50
    },
    {
      "epoch": 0.010160880609652836,
      "grad_norm": 0.23612086474895477,
      "learning_rate": 9.897437097795257e-05,
      "loss": 0.5049,
      "step": 51
    },
    {
      "epoch": 0.010360113562783284,
      "grad_norm": 0.33656045794487,
      "learning_rate": 9.890738003669029e-05,
      "loss": 0.7237,
      "step": 52
    },
    {
      "epoch": 0.010559346515913733,
      "grad_norm": 0.301807701587677,
      "learning_rate": 9.883829406604363e-05,
      "loss": 0.6572,
      "step": 53
    },
    {
      "epoch": 0.01075857946904418,
      "grad_norm": 0.2367606908082962,
      "learning_rate": 9.876711602542563e-05,
      "loss": 0.637,
      "step": 54
    },
    {
      "epoch": 0.010957812422174628,
      "grad_norm": 0.2271997034549713,
      "learning_rate": 9.869384896386668e-05,
      "loss": 0.6629,
      "step": 55
    },
    {
      "epoch": 0.011157045375305076,
      "grad_norm": 0.2069195657968521,
      "learning_rate": 9.861849601988383e-05,
      "loss": 0.6359,
      "step": 56
    },
    {
      "epoch": 0.011356278328435523,
      "grad_norm": 0.20633797347545624,
      "learning_rate": 9.854106042134641e-05,
      "loss": 0.6114,
      "step": 57
    },
    {
      "epoch": 0.011555511281565971,
      "grad_norm": 0.23585379123687744,
      "learning_rate": 9.846154548533773e-05,
      "loss": 0.6855,
      "step": 58
    },
    {
      "epoch": 0.01175474423469642,
      "grad_norm": 0.21923767030239105,
      "learning_rate": 9.837995461801299e-05,
      "loss": 0.6436,
      "step": 59
    },
    {
      "epoch": 0.011953977187826866,
      "grad_norm": 0.26621073484420776,
      "learning_rate": 9.829629131445342e-05,
      "loss": 0.7332,
      "step": 60
    },
    {
      "epoch": 0.012153210140957315,
      "grad_norm": 0.21650448441505432,
      "learning_rate": 9.821055915851647e-05,
      "loss": 0.6092,
      "step": 61
    },
    {
      "epoch": 0.012352443094087761,
      "grad_norm": 0.22831426560878754,
      "learning_rate": 9.812276182268236e-05,
      "loss": 0.7218,
      "step": 62
    },
    {
      "epoch": 0.01255167604721821,
      "grad_norm": 0.21681798994541168,
      "learning_rate": 9.803290306789676e-05,
      "loss": 0.7292,
      "step": 63
    },
    {
      "epoch": 0.012750909000348658,
      "grad_norm": 0.27890080213546753,
      "learning_rate": 9.794098674340965e-05,
      "loss": 0.7812,
      "step": 64
    },
    {
      "epoch": 0.012950141953479105,
      "grad_norm": 0.21760623157024384,
      "learning_rate": 9.784701678661045e-05,
      "loss": 0.6858,
      "step": 65
    },
    {
      "epoch": 0.013149374906609553,
      "grad_norm": 0.2346140742301941,
      "learning_rate": 9.775099722285935e-05,
      "loss": 0.6121,
      "step": 66
    },
    {
      "epoch": 0.013348607859740002,
      "grad_norm": 0.24936628341674805,
      "learning_rate": 9.765293216531486e-05,
      "loss": 0.7378,
      "step": 67
    },
    {
      "epoch": 0.013547840812870448,
      "grad_norm": 0.22810809314250946,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.718,
      "step": 68
    },
    {
      "epoch": 0.013747073766000897,
      "grad_norm": 0.2293727546930313,
      "learning_rate": 9.74506824594107e-05,
      "loss": 0.7069,
      "step": 69
    },
    {
      "epoch": 0.013946306719131345,
      "grad_norm": 0.23354655504226685,
      "learning_rate": 9.73465064747553e-05,
      "loss": 0.72,
      "step": 70
    },
    {
      "epoch": 0.014145539672261792,
      "grad_norm": 0.2324572503566742,
      "learning_rate": 9.724030232334391e-05,
      "loss": 0.6631,
      "step": 71
    },
    {
      "epoch": 0.01434477262539224,
      "grad_norm": 0.273137629032135,
      "learning_rate": 9.713207455460894e-05,
      "loss": 0.7187,
      "step": 72
    },
    {
      "epoch": 0.014544005578522687,
      "grad_norm": 0.25326138734817505,
      "learning_rate": 9.702182780466775e-05,
      "loss": 0.706,
      "step": 73
    },
    {
      "epoch": 0.014743238531653135,
      "grad_norm": 0.2347702831029892,
      "learning_rate": 9.690956679612421e-05,
      "loss": 0.7365,
      "step": 74
    },
    {
      "epoch": 0.014942471484783584,
      "grad_norm": 0.21022625267505646,
      "learning_rate": 9.67952963378663e-05,
      "loss": 0.6422,
      "step": 75
    },
    {
      "epoch": 0.01514170443791403,
      "grad_norm": 0.2170475721359253,
      "learning_rate": 9.667902132486009e-05,
      "loss": 0.685,
      "step": 76
    },
    {
      "epoch": 0.015340937391044479,
      "grad_norm": 0.27755966782569885,
      "learning_rate": 9.656074673794018e-05,
      "loss": 0.8184,
      "step": 77
    },
    {
      "epoch": 0.015540170344174927,
      "grad_norm": 0.22904907166957855,
      "learning_rate": 9.644047764359622e-05,
      "loss": 0.6883,
      "step": 78
    },
    {
      "epoch": 0.015739403297305374,
      "grad_norm": 0.23313535749912262,
      "learning_rate": 9.631821919375591e-05,
      "loss": 0.6408,
      "step": 79
    },
    {
      "epoch": 0.01593863625043582,
      "grad_norm": 0.2328289896249771,
      "learning_rate": 9.619397662556435e-05,
      "loss": 0.668,
      "step": 80
    },
    {
      "epoch": 0.01613786920356627,
      "grad_norm": 0.22517356276512146,
      "learning_rate": 9.606775526115963e-05,
      "loss": 0.6917,
      "step": 81
    },
    {
      "epoch": 0.016337102156696717,
      "grad_norm": 0.22593192756175995,
      "learning_rate": 9.593956050744492e-05,
      "loss": 0.6252,
      "step": 82
    },
    {
      "epoch": 0.016536335109827164,
      "grad_norm": 0.2786768078804016,
      "learning_rate": 9.580939785585681e-05,
      "loss": 0.847,
      "step": 83
    },
    {
      "epoch": 0.016735568062957614,
      "grad_norm": 0.25397562980651855,
      "learning_rate": 9.567727288213005e-05,
      "loss": 0.7431,
      "step": 84
    },
    {
      "epoch": 0.01693480101608806,
      "grad_norm": 0.26364976167678833,
      "learning_rate": 9.554319124605879e-05,
      "loss": 0.7749,
      "step": 85
    },
    {
      "epoch": 0.017134033969218507,
      "grad_norm": 0.301891952753067,
      "learning_rate": 9.540715869125407e-05,
      "loss": 0.6903,
      "step": 86
    },
    {
      "epoch": 0.017333266922348958,
      "grad_norm": 0.25629404187202454,
      "learning_rate": 9.526918104489777e-05,
      "loss": 0.7569,
      "step": 87
    },
    {
      "epoch": 0.017532499875479404,
      "grad_norm": 0.24842076003551483,
      "learning_rate": 9.512926421749304e-05,
      "loss": 0.7618,
      "step": 88
    },
    {
      "epoch": 0.01773173282860985,
      "grad_norm": 0.26040297746658325,
      "learning_rate": 9.498741420261108e-05,
      "loss": 0.8032,
      "step": 89
    },
    {
      "epoch": 0.0179309657817403,
      "grad_norm": 0.2626326084136963,
      "learning_rate": 9.484363707663442e-05,
      "loss": 0.7924,
      "step": 90
    },
    {
      "epoch": 0.018130198734870748,
      "grad_norm": 0.22737815976142883,
      "learning_rate": 9.469793899849661e-05,
      "loss": 0.6443,
      "step": 91
    },
    {
      "epoch": 0.018329431688001194,
      "grad_norm": 0.24532002210617065,
      "learning_rate": 9.45503262094184e-05,
      "loss": 0.7,
      "step": 92
    },
    {
      "epoch": 0.018528664641131645,
      "grad_norm": 0.28320467472076416,
      "learning_rate": 9.440080503264037e-05,
      "loss": 0.783,
      "step": 93
    },
    {
      "epoch": 0.01872789759426209,
      "grad_norm": 0.257158100605011,
      "learning_rate": 9.42493818731521e-05,
      "loss": 0.7044,
      "step": 94
    },
    {
      "epoch": 0.018927130547392538,
      "grad_norm": 0.23965127766132355,
      "learning_rate": 9.409606321741775e-05,
      "loss": 0.6876,
      "step": 95
    },
    {
      "epoch": 0.019126363500522988,
      "grad_norm": 0.29275670647621155,
      "learning_rate": 9.394085563309827e-05,
      "loss": 0.8103,
      "step": 96
    },
    {
      "epoch": 0.019325596453653435,
      "grad_norm": 0.27652475237846375,
      "learning_rate": 9.378376576876999e-05,
      "loss": 0.7313,
      "step": 97
    },
    {
      "epoch": 0.01952482940678388,
      "grad_norm": 0.2445044219493866,
      "learning_rate": 9.362480035363986e-05,
      "loss": 0.6936,
      "step": 98
    },
    {
      "epoch": 0.01972406235991433,
      "grad_norm": 0.2570478916168213,
      "learning_rate": 9.34639661972572e-05,
      "loss": 0.6871,
      "step": 99
    },
    {
      "epoch": 0.019923295313044778,
      "grad_norm": 0.28658804297447205,
      "learning_rate": 9.330127018922194e-05,
      "loss": 0.7063,
      "step": 100
    },
    {
      "epoch": 0.019923295313044778,
      "eval_loss": 0.7203736305236816,
      "eval_runtime": 264.9972,
      "eval_samples_per_second": 31.902,
      "eval_steps_per_second": 7.977,
      "step": 100
    },
    {
      "epoch": 0.020122528266175225,
      "grad_norm": 0.21482613682746887,
      "learning_rate": 9.31367192988896e-05,
      "loss": 0.5356,
      "step": 101
    },
    {
      "epoch": 0.02032176121930567,
      "grad_norm": 0.28820475935935974,
      "learning_rate": 9.297032057507264e-05,
      "loss": 0.6606,
      "step": 102
    },
    {
      "epoch": 0.02052099417243612,
      "grad_norm": 0.22919978201389313,
      "learning_rate": 9.280208114573859e-05,
      "loss": 0.6681,
      "step": 103
    },
    {
      "epoch": 0.02072022712556657,
      "grad_norm": 0.20591995120048523,
      "learning_rate": 9.263200821770461e-05,
      "loss": 0.6523,
      "step": 104
    },
    {
      "epoch": 0.020919460078697015,
      "grad_norm": 0.20084571838378906,
      "learning_rate": 9.246010907632895e-05,
      "loss": 0.629,
      "step": 105
    },
    {
      "epoch": 0.021118693031827465,
      "grad_norm": 0.25101473927497864,
      "learning_rate": 9.228639108519868e-05,
      "loss": 0.7384,
      "step": 106
    },
    {
      "epoch": 0.021317925984957912,
      "grad_norm": 0.23734250664710999,
      "learning_rate": 9.211086168581433e-05,
      "loss": 0.6895,
      "step": 107
    },
    {
      "epoch": 0.02151715893808836,
      "grad_norm": 0.22536040842533112,
      "learning_rate": 9.193352839727121e-05,
      "loss": 0.7137,
      "step": 108
    },
    {
      "epoch": 0.02171639189121881,
      "grad_norm": 0.2355283945798874,
      "learning_rate": 9.175439881593716e-05,
      "loss": 0.7662,
      "step": 109
    },
    {
      "epoch": 0.021915624844349255,
      "grad_norm": 0.2228918820619583,
      "learning_rate": 9.157348061512727e-05,
      "loss": 0.6534,
      "step": 110
    },
    {
      "epoch": 0.022114857797479702,
      "grad_norm": 0.2184191644191742,
      "learning_rate": 9.139078154477512e-05,
      "loss": 0.7054,
      "step": 111
    },
    {
      "epoch": 0.022314090750610152,
      "grad_norm": 0.21408338844776154,
      "learning_rate": 9.120630943110077e-05,
      "loss": 0.6103,
      "step": 112
    },
    {
      "epoch": 0.0225133237037406,
      "grad_norm": 0.25427061319351196,
      "learning_rate": 9.102007217627568e-05,
      "loss": 0.8366,
      "step": 113
    },
    {
      "epoch": 0.022712556656871045,
      "grad_norm": 0.21583066880702972,
      "learning_rate": 9.083207775808396e-05,
      "loss": 0.7165,
      "step": 114
    },
    {
      "epoch": 0.022911789610001496,
      "grad_norm": 0.2396174967288971,
      "learning_rate": 9.064233422958077e-05,
      "loss": 0.6794,
      "step": 115
    },
    {
      "epoch": 0.023111022563131942,
      "grad_norm": 0.2334282398223877,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.7945,
      "step": 116
    },
    {
      "epoch": 0.02331025551626239,
      "grad_norm": 0.20607277750968933,
      "learning_rate": 9.025763242814291e-05,
      "loss": 0.5739,
      "step": 117
    },
    {
      "epoch": 0.02350948846939284,
      "grad_norm": 0.2125677615404129,
      "learning_rate": 9.006269063455304e-05,
      "loss": 0.6976,
      "step": 118
    },
    {
      "epoch": 0.023708721422523286,
      "grad_norm": 0.24233980476856232,
      "learning_rate": 8.986603268863536e-05,
      "loss": 0.7768,
      "step": 119
    },
    {
      "epoch": 0.023907954375653732,
      "grad_norm": 0.23903363943099976,
      "learning_rate": 8.966766701456177e-05,
      "loss": 0.7925,
      "step": 120
    },
    {
      "epoch": 0.024107187328784183,
      "grad_norm": 0.2270466685295105,
      "learning_rate": 8.94676021096575e-05,
      "loss": 0.679,
      "step": 121
    },
    {
      "epoch": 0.02430642028191463,
      "grad_norm": 0.22134599089622498,
      "learning_rate": 8.926584654403724e-05,
      "loss": 0.7155,
      "step": 122
    },
    {
      "epoch": 0.024505653235045076,
      "grad_norm": 0.2539938986301422,
      "learning_rate": 8.906240896023794e-05,
      "loss": 0.7156,
      "step": 123
    },
    {
      "epoch": 0.024704886188175523,
      "grad_norm": 0.24332107603549957,
      "learning_rate": 8.885729807284856e-05,
      "loss": 0.6988,
      "step": 124
    },
    {
      "epoch": 0.024904119141305973,
      "grad_norm": 0.23084275424480438,
      "learning_rate": 8.865052266813685e-05,
      "loss": 0.7142,
      "step": 125
    },
    {
      "epoch": 0.02510335209443642,
      "grad_norm": 0.24665014445781708,
      "learning_rate": 8.844209160367299e-05,
      "loss": 0.7471,
      "step": 126
    },
    {
      "epoch": 0.025302585047566866,
      "grad_norm": 0.23389537632465363,
      "learning_rate": 8.823201380795001e-05,
      "loss": 0.676,
      "step": 127
    },
    {
      "epoch": 0.025501818000697316,
      "grad_norm": 0.2225302904844284,
      "learning_rate": 8.802029828000156e-05,
      "loss": 0.6702,
      "step": 128
    },
    {
      "epoch": 0.025701050953827763,
      "grad_norm": 0.2194124162197113,
      "learning_rate": 8.780695408901613e-05,
      "loss": 0.7173,
      "step": 129
    },
    {
      "epoch": 0.02590028390695821,
      "grad_norm": 0.22123487293720245,
      "learning_rate": 8.759199037394887e-05,
      "loss": 0.679,
      "step": 130
    },
    {
      "epoch": 0.02609951686008866,
      "grad_norm": 0.2465553730726242,
      "learning_rate": 8.737541634312985e-05,
      "loss": 0.7151,
      "step": 131
    },
    {
      "epoch": 0.026298749813219106,
      "grad_norm": 0.2150459736585617,
      "learning_rate": 8.715724127386972e-05,
      "loss": 0.7106,
      "step": 132
    },
    {
      "epoch": 0.026497982766349553,
      "grad_norm": 0.1966014802455902,
      "learning_rate": 8.693747451206232e-05,
      "loss": 0.7174,
      "step": 133
    },
    {
      "epoch": 0.026697215719480003,
      "grad_norm": 0.25692620873451233,
      "learning_rate": 8.671612547178428e-05,
      "loss": 0.6709,
      "step": 134
    },
    {
      "epoch": 0.02689644867261045,
      "grad_norm": 0.19534507393836975,
      "learning_rate": 8.649320363489179e-05,
      "loss": 0.5871,
      "step": 135
    },
    {
      "epoch": 0.027095681625740897,
      "grad_norm": 0.25263822078704834,
      "learning_rate": 8.626871855061438e-05,
      "loss": 0.7815,
      "step": 136
    },
    {
      "epoch": 0.027294914578871347,
      "grad_norm": 0.24761663377285004,
      "learning_rate": 8.604267983514594e-05,
      "loss": 0.7253,
      "step": 137
    },
    {
      "epoch": 0.027494147532001793,
      "grad_norm": 0.2336164116859436,
      "learning_rate": 8.581509717123273e-05,
      "loss": 0.6871,
      "step": 138
    },
    {
      "epoch": 0.02769338048513224,
      "grad_norm": 0.2253238558769226,
      "learning_rate": 8.558598030775857e-05,
      "loss": 0.6806,
      "step": 139
    },
    {
      "epoch": 0.02789261343826269,
      "grad_norm": 0.2473248988389969,
      "learning_rate": 8.535533905932738e-05,
      "loss": 0.8147,
      "step": 140
    },
    {
      "epoch": 0.028091846391393137,
      "grad_norm": 0.2515900731086731,
      "learning_rate": 8.51231833058426e-05,
      "loss": 0.8055,
      "step": 141
    },
    {
      "epoch": 0.028291079344523583,
      "grad_norm": 0.2372109442949295,
      "learning_rate": 8.488952299208401e-05,
      "loss": 0.6404,
      "step": 142
    },
    {
      "epoch": 0.028490312297654034,
      "grad_norm": 0.24354617297649384,
      "learning_rate": 8.46543681272818e-05,
      "loss": 0.7355,
      "step": 143
    },
    {
      "epoch": 0.02868954525078448,
      "grad_norm": 0.2426522821187973,
      "learning_rate": 8.44177287846877e-05,
      "loss": 0.7561,
      "step": 144
    },
    {
      "epoch": 0.028888778203914927,
      "grad_norm": 0.24764999747276306,
      "learning_rate": 8.417961510114356e-05,
      "loss": 0.7203,
      "step": 145
    },
    {
      "epoch": 0.029088011157045374,
      "grad_norm": 0.2602037191390991,
      "learning_rate": 8.39400372766471e-05,
      "loss": 0.7111,
      "step": 146
    },
    {
      "epoch": 0.029287244110175824,
      "grad_norm": 0.25291872024536133,
      "learning_rate": 8.36990055739149e-05,
      "loss": 0.7011,
      "step": 147
    },
    {
      "epoch": 0.02948647706330627,
      "grad_norm": 0.2726818323135376,
      "learning_rate": 8.345653031794292e-05,
      "loss": 0.8132,
      "step": 148
    },
    {
      "epoch": 0.029685710016436717,
      "grad_norm": 0.2830042839050293,
      "learning_rate": 8.321262189556409e-05,
      "loss": 0.7577,
      "step": 149
    },
    {
      "epoch": 0.029884942969567167,
      "grad_norm": 0.2831919193267822,
      "learning_rate": 8.296729075500344e-05,
      "loss": 0.7566,
      "step": 150
    },
    {
      "epoch": 0.030084175922697614,
      "grad_norm": 0.33581486344337463,
      "learning_rate": 8.272054740543052e-05,
      "loss": 0.7664,
      "step": 151
    },
    {
      "epoch": 0.03028340887582806,
      "grad_norm": 0.3191780149936676,
      "learning_rate": 8.247240241650918e-05,
      "loss": 0.6583,
      "step": 152
    },
    {
      "epoch": 0.03048264182895851,
      "grad_norm": 0.22133736312389374,
      "learning_rate": 8.222286641794488e-05,
      "loss": 0.6328,
      "step": 153
    },
    {
      "epoch": 0.030681874782088957,
      "grad_norm": 0.21831941604614258,
      "learning_rate": 8.197195009902924e-05,
      "loss": 0.698,
      "step": 154
    },
    {
      "epoch": 0.030881107735219404,
      "grad_norm": 0.1761123090982437,
      "learning_rate": 8.171966420818228e-05,
      "loss": 0.5441,
      "step": 155
    },
    {
      "epoch": 0.031080340688349854,
      "grad_norm": 0.20375514030456543,
      "learning_rate": 8.146601955249188e-05,
      "loss": 0.6309,
      "step": 156
    },
    {
      "epoch": 0.0312795736414803,
      "grad_norm": 0.2410995066165924,
      "learning_rate": 8.121102699725089e-05,
      "loss": 0.6337,
      "step": 157
    },
    {
      "epoch": 0.03147880659461075,
      "grad_norm": 0.2229624092578888,
      "learning_rate": 8.095469746549172e-05,
      "loss": 0.7014,
      "step": 158
    },
    {
      "epoch": 0.0316780395477412,
      "grad_norm": 0.23791897296905518,
      "learning_rate": 8.069704193751832e-05,
      "loss": 0.6582,
      "step": 159
    },
    {
      "epoch": 0.03187727250087164,
      "grad_norm": 0.22524884343147278,
      "learning_rate": 8.043807145043604e-05,
      "loss": 0.7178,
      "step": 160
    },
    {
      "epoch": 0.03207650545400209,
      "grad_norm": 0.21897757053375244,
      "learning_rate": 8.017779709767858e-05,
      "loss": 0.7032,
      "step": 161
    },
    {
      "epoch": 0.03227573840713254,
      "grad_norm": 0.19019927084445953,
      "learning_rate": 7.991623002853296e-05,
      "loss": 0.5842,
      "step": 162
    },
    {
      "epoch": 0.032474971360262984,
      "grad_norm": 0.21591834723949432,
      "learning_rate": 7.965338144766186e-05,
      "loss": 0.7243,
      "step": 163
    },
    {
      "epoch": 0.032674204313393435,
      "grad_norm": 0.20649899542331696,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.6814,
      "step": 164
    },
    {
      "epoch": 0.032873437266523885,
      "grad_norm": 0.24540702998638153,
      "learning_rate": 7.912388484339012e-05,
      "loss": 0.7356,
      "step": 165
    },
    {
      "epoch": 0.03307267021965433,
      "grad_norm": 0.22577622532844543,
      "learning_rate": 7.88572595018617e-05,
      "loss": 0.6468,
      "step": 166
    },
    {
      "epoch": 0.03327190317278478,
      "grad_norm": 0.2168670892715454,
      "learning_rate": 7.858939801138061e-05,
      "loss": 0.6448,
      "step": 167
    },
    {
      "epoch": 0.03347113612591523,
      "grad_norm": 0.22892935574054718,
      "learning_rate": 7.832031184624164e-05,
      "loss": 0.6375,
      "step": 168
    },
    {
      "epoch": 0.03367036907904567,
      "grad_norm": 0.2086174190044403,
      "learning_rate": 7.80500125332005e-05,
      "loss": 0.6993,
      "step": 169
    },
    {
      "epoch": 0.03386960203217612,
      "grad_norm": 0.19050797820091248,
      "learning_rate": 7.777851165098012e-05,
      "loss": 0.5909,
      "step": 170
    },
    {
      "epoch": 0.03406883498530657,
      "grad_norm": 0.22635716199874878,
      "learning_rate": 7.750582082977467e-05,
      "loss": 0.6799,
      "step": 171
    },
    {
      "epoch": 0.034268067938437015,
      "grad_norm": 0.2369690090417862,
      "learning_rate": 7.723195175075136e-05,
      "loss": 0.6823,
      "step": 172
    },
    {
      "epoch": 0.034467300891567465,
      "grad_norm": 0.2258961796760559,
      "learning_rate": 7.695691614555003e-05,
      "loss": 0.6768,
      "step": 173
    },
    {
      "epoch": 0.034666533844697915,
      "grad_norm": 0.2175053060054779,
      "learning_rate": 7.668072579578058e-05,
      "loss": 0.6501,
      "step": 174
    },
    {
      "epoch": 0.03486576679782836,
      "grad_norm": 0.23859256505966187,
      "learning_rate": 7.64033925325184e-05,
      "loss": 0.7198,
      "step": 175
    },
    {
      "epoch": 0.03506499975095881,
      "grad_norm": 0.2090621143579483,
      "learning_rate": 7.612492823579745e-05,
      "loss": 0.6075,
      "step": 176
    },
    {
      "epoch": 0.03526423270408926,
      "grad_norm": 0.23009976744651794,
      "learning_rate": 7.584534483410137e-05,
      "loss": 0.6993,
      "step": 177
    },
    {
      "epoch": 0.0354634656572197,
      "grad_norm": 0.19365736842155457,
      "learning_rate": 7.55646543038526e-05,
      "loss": 0.61,
      "step": 178
    },
    {
      "epoch": 0.03566269861035015,
      "grad_norm": 0.24506577849388123,
      "learning_rate": 7.528286866889924e-05,
      "loss": 0.68,
      "step": 179
    },
    {
      "epoch": 0.0358619315634806,
      "grad_norm": 0.2776792049407959,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.7224,
      "step": 180
    },
    {
      "epoch": 0.036061164516611045,
      "grad_norm": 0.21556320786476135,
      "learning_rate": 7.471606041430723e-05,
      "loss": 0.6742,
      "step": 181
    },
    {
      "epoch": 0.036260397469741495,
      "grad_norm": 0.2549598217010498,
      "learning_rate": 7.443106207484776e-05,
      "loss": 0.7866,
      "step": 182
    },
    {
      "epoch": 0.036459630422871946,
      "grad_norm": 0.2427287995815277,
      "learning_rate": 7.414501719000187e-05,
      "loss": 0.7006,
      "step": 183
    },
    {
      "epoch": 0.03665886337600239,
      "grad_norm": 0.2485671192407608,
      "learning_rate": 7.385793801298042e-05,
      "loss": 0.678,
      "step": 184
    },
    {
      "epoch": 0.03685809632913284,
      "grad_norm": 0.23628251254558563,
      "learning_rate": 7.35698368412999e-05,
      "loss": 0.7176,
      "step": 185
    },
    {
      "epoch": 0.03705732928226329,
      "grad_norm": 0.245437353849411,
      "learning_rate": 7.328072601625557e-05,
      "loss": 0.6784,
      "step": 186
    },
    {
      "epoch": 0.03725656223539373,
      "grad_norm": 0.26534438133239746,
      "learning_rate": 7.2990617922393e-05,
      "loss": 0.7389,
      "step": 187
    },
    {
      "epoch": 0.03745579518852418,
      "grad_norm": 0.26061758399009705,
      "learning_rate": 7.269952498697734e-05,
      "loss": 0.637,
      "step": 188
    },
    {
      "epoch": 0.03765502814165463,
      "grad_norm": 0.247264102101326,
      "learning_rate": 7.240745967946113e-05,
      "loss": 0.7007,
      "step": 189
    },
    {
      "epoch": 0.037854261094785076,
      "grad_norm": 0.2501027584075928,
      "learning_rate": 7.211443451095007e-05,
      "loss": 0.7552,
      "step": 190
    },
    {
      "epoch": 0.038053494047915526,
      "grad_norm": 0.24374301731586456,
      "learning_rate": 7.18204620336671e-05,
      "loss": 0.7168,
      "step": 191
    },
    {
      "epoch": 0.038252727001045976,
      "grad_norm": 0.2584417760372162,
      "learning_rate": 7.152555484041476e-05,
      "loss": 0.6993,
      "step": 192
    },
    {
      "epoch": 0.03845195995417642,
      "grad_norm": 0.2678215205669403,
      "learning_rate": 7.122972556403567e-05,
      "loss": 0.7069,
      "step": 193
    },
    {
      "epoch": 0.03865119290730687,
      "grad_norm": 0.27493688464164734,
      "learning_rate": 7.09329868768714e-05,
      "loss": 0.7271,
      "step": 194
    },
    {
      "epoch": 0.03885042586043732,
      "grad_norm": 0.28842246532440186,
      "learning_rate": 7.063535149021973e-05,
      "loss": 0.7254,
      "step": 195
    },
    {
      "epoch": 0.03904965881356776,
      "grad_norm": 0.26705753803253174,
      "learning_rate": 7.033683215379002e-05,
      "loss": 0.7115,
      "step": 196
    },
    {
      "epoch": 0.03924889176669821,
      "grad_norm": 0.25876060128211975,
      "learning_rate": 7.003744165515705e-05,
      "loss": 0.7228,
      "step": 197
    },
    {
      "epoch": 0.03944812471982866,
      "grad_norm": 0.30024516582489014,
      "learning_rate": 6.973719281921335e-05,
      "loss": 0.7615,
      "step": 198
    },
    {
      "epoch": 0.039647357672959106,
      "grad_norm": 0.27740225195884705,
      "learning_rate": 6.943609850761979e-05,
      "loss": 0.7928,
      "step": 199
    },
    {
      "epoch": 0.039846590626089556,
      "grad_norm": 0.3024348020553589,
      "learning_rate": 6.91341716182545e-05,
      "loss": 0.7479,
      "step": 200
    },
    {
      "epoch": 0.039846590626089556,
      "eval_loss": 0.7008334398269653,
      "eval_runtime": 265.1826,
      "eval_samples_per_second": 31.88,
      "eval_steps_per_second": 7.972,
      "step": 200
    },
    {
      "epoch": 0.04004582357922,
      "grad_norm": 0.17536719143390656,
      "learning_rate": 6.883142508466054e-05,
      "loss": 0.5068,
      "step": 201
    },
    {
      "epoch": 0.04024505653235045,
      "grad_norm": 0.2694966197013855,
      "learning_rate": 6.852787187549182e-05,
      "loss": 0.6759,
      "step": 202
    },
    {
      "epoch": 0.0404442894854809,
      "grad_norm": 0.23542173206806183,
      "learning_rate": 6.82235249939575e-05,
      "loss": 0.7302,
      "step": 203
    },
    {
      "epoch": 0.04064352243861134,
      "grad_norm": 0.2004457414150238,
      "learning_rate": 6.7918397477265e-05,
      "loss": 0.6047,
      "step": 204
    },
    {
      "epoch": 0.04084275539174179,
      "grad_norm": 0.217708557844162,
      "learning_rate": 6.761250239606169e-05,
      "loss": 0.5939,
      "step": 205
    },
    {
      "epoch": 0.04104198834487224,
      "grad_norm": 0.2038157731294632,
      "learning_rate": 6.730585285387465e-05,
      "loss": 0.6299,
      "step": 206
    },
    {
      "epoch": 0.04124122129800269,
      "grad_norm": 0.2198810577392578,
      "learning_rate": 6.699846198654971e-05,
      "loss": 0.6102,
      "step": 207
    },
    {
      "epoch": 0.04144045425113314,
      "grad_norm": 0.21492627263069153,
      "learning_rate": 6.669034296168855e-05,
      "loss": 0.7281,
      "step": 208
    },
    {
      "epoch": 0.04163968720426359,
      "grad_norm": 0.20070651173591614,
      "learning_rate": 6.638150897808468e-05,
      "loss": 0.6202,
      "step": 209
    },
    {
      "epoch": 0.04183892015739403,
      "grad_norm": 0.20803692936897278,
      "learning_rate": 6.607197326515808e-05,
      "loss": 0.6648,
      "step": 210
    },
    {
      "epoch": 0.04203815311052448,
      "grad_norm": 0.21368266642093658,
      "learning_rate": 6.57617490823885e-05,
      "loss": 0.6737,
      "step": 211
    },
    {
      "epoch": 0.04223738606365493,
      "grad_norm": 0.22453172504901886,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.71,
      "step": 212
    },
    {
      "epoch": 0.042436619016785374,
      "grad_norm": 0.20329400897026062,
      "learning_rate": 6.513928849212873e-05,
      "loss": 0.6541,
      "step": 213
    },
    {
      "epoch": 0.042635851969915824,
      "grad_norm": 0.18720655143260956,
      "learning_rate": 6.482707874877854e-05,
      "loss": 0.6628,
      "step": 214
    },
    {
      "epoch": 0.042835084923046274,
      "grad_norm": 0.20335181057453156,
      "learning_rate": 6.451423386272312e-05,
      "loss": 0.7075,
      "step": 215
    },
    {
      "epoch": 0.04303431787617672,
      "grad_norm": 0.21466787159442902,
      "learning_rate": 6.420076723519614e-05,
      "loss": 0.6608,
      "step": 216
    },
    {
      "epoch": 0.04323355082930717,
      "grad_norm": 0.19446536898612976,
      "learning_rate": 6.388669229406462e-05,
      "loss": 0.6451,
      "step": 217
    },
    {
      "epoch": 0.04343278378243762,
      "grad_norm": 0.202213317155838,
      "learning_rate": 6.357202249325371e-05,
      "loss": 0.631,
      "step": 218
    },
    {
      "epoch": 0.04363201673556806,
      "grad_norm": 0.20673999190330505,
      "learning_rate": 6.32567713121704e-05,
      "loss": 0.6313,
      "step": 219
    },
    {
      "epoch": 0.04383124968869851,
      "grad_norm": 0.22817325592041016,
      "learning_rate": 6.294095225512603e-05,
      "loss": 0.6956,
      "step": 220
    },
    {
      "epoch": 0.04403048264182896,
      "grad_norm": 0.23594947159290314,
      "learning_rate": 6.26245788507579e-05,
      "loss": 0.6675,
      "step": 221
    },
    {
      "epoch": 0.044229715594959404,
      "grad_norm": 0.2236422449350357,
      "learning_rate": 6.230766465144967e-05,
      "loss": 0.6337,
      "step": 222
    },
    {
      "epoch": 0.044428948548089854,
      "grad_norm": 0.19244728982448578,
      "learning_rate": 6.199022323275083e-05,
      "loss": 0.6323,
      "step": 223
    },
    {
      "epoch": 0.044628181501220304,
      "grad_norm": 0.23294833302497864,
      "learning_rate": 6.167226819279528e-05,
      "loss": 0.6741,
      "step": 224
    },
    {
      "epoch": 0.04482741445435075,
      "grad_norm": 0.2407350391149521,
      "learning_rate": 6.135381315171867e-05,
      "loss": 0.7436,
      "step": 225
    },
    {
      "epoch": 0.0450266474074812,
      "grad_norm": 0.21430183947086334,
      "learning_rate": 6.103487175107507e-05,
      "loss": 0.6578,
      "step": 226
    },
    {
      "epoch": 0.04522588036061165,
      "grad_norm": 0.23648200929164886,
      "learning_rate": 6.071545765325254e-05,
      "loss": 0.7116,
      "step": 227
    },
    {
      "epoch": 0.04542511331374209,
      "grad_norm": 0.22720520198345184,
      "learning_rate": 6.0395584540887963e-05,
      "loss": 0.6536,
      "step": 228
    },
    {
      "epoch": 0.04562434626687254,
      "grad_norm": 0.2226054072380066,
      "learning_rate": 6.007526611628086e-05,
      "loss": 0.684,
      "step": 229
    },
    {
      "epoch": 0.04582357922000299,
      "grad_norm": 0.25174853205680847,
      "learning_rate": 5.9754516100806423e-05,
      "loss": 0.6755,
      "step": 230
    },
    {
      "epoch": 0.046022812173133434,
      "grad_norm": 0.22537098824977875,
      "learning_rate": 5.9433348234327765e-05,
      "loss": 0.6145,
      "step": 231
    },
    {
      "epoch": 0.046222045126263885,
      "grad_norm": 0.26092636585235596,
      "learning_rate": 5.911177627460739e-05,
      "loss": 0.7068,
      "step": 232
    },
    {
      "epoch": 0.046421278079394335,
      "grad_norm": 0.24872590601444244,
      "learning_rate": 5.8789813996717736e-05,
      "loss": 0.7006,
      "step": 233
    },
    {
      "epoch": 0.04662051103252478,
      "grad_norm": 0.27163049578666687,
      "learning_rate": 5.8467475192451226e-05,
      "loss": 0.6737,
      "step": 234
    },
    {
      "epoch": 0.04681974398565523,
      "grad_norm": 0.24463686347007751,
      "learning_rate": 5.814477366972945e-05,
      "loss": 0.6807,
      "step": 235
    },
    {
      "epoch": 0.04701897693878568,
      "grad_norm": 0.2596883177757263,
      "learning_rate": 5.782172325201155e-05,
      "loss": 0.7027,
      "step": 236
    },
    {
      "epoch": 0.04721820989191612,
      "grad_norm": 0.24346858263015747,
      "learning_rate": 5.749833777770225e-05,
      "loss": 0.6946,
      "step": 237
    },
    {
      "epoch": 0.04741744284504657,
      "grad_norm": 0.21713268756866455,
      "learning_rate": 5.717463109955896e-05,
      "loss": 0.6445,
      "step": 238
    },
    {
      "epoch": 0.04761667579817702,
      "grad_norm": 0.22738954424858093,
      "learning_rate": 5.685061708409841e-05,
      "loss": 0.6333,
      "step": 239
    },
    {
      "epoch": 0.047815908751307465,
      "grad_norm": 0.27048802375793457,
      "learning_rate": 5.6526309611002594e-05,
      "loss": 0.7271,
      "step": 240
    },
    {
      "epoch": 0.048015141704437915,
      "grad_norm": 0.26063868403434753,
      "learning_rate": 5.6201722572524275e-05,
      "loss": 0.6432,
      "step": 241
    },
    {
      "epoch": 0.048214374657568365,
      "grad_norm": 0.25063076615333557,
      "learning_rate": 5.587686987289189e-05,
      "loss": 0.7654,
      "step": 242
    },
    {
      "epoch": 0.04841360761069881,
      "grad_norm": 0.2545402944087982,
      "learning_rate": 5.5551765427713884e-05,
      "loss": 0.683,
      "step": 243
    },
    {
      "epoch": 0.04861284056382926,
      "grad_norm": 0.26064079999923706,
      "learning_rate": 5.522642316338268e-05,
      "loss": 0.6294,
      "step": 244
    },
    {
      "epoch": 0.0488120735169597,
      "grad_norm": 0.2834579348564148,
      "learning_rate": 5.490085701647805e-05,
      "loss": 0.7894,
      "step": 245
    },
    {
      "epoch": 0.04901130647009015,
      "grad_norm": 0.27373987436294556,
      "learning_rate": 5.457508093317013e-05,
      "loss": 0.7169,
      "step": 246
    },
    {
      "epoch": 0.0492105394232206,
      "grad_norm": 0.24024784564971924,
      "learning_rate": 5.4249108868622086e-05,
      "loss": 0.7348,
      "step": 247
    },
    {
      "epoch": 0.049409772376351045,
      "grad_norm": 0.24063633382320404,
      "learning_rate": 5.392295478639225e-05,
      "loss": 0.6727,
      "step": 248
    },
    {
      "epoch": 0.049609005329481495,
      "grad_norm": 0.24858322739601135,
      "learning_rate": 5.359663265783598e-05,
      "loss": 0.6676,
      "step": 249
    },
    {
      "epoch": 0.049808238282611945,
      "grad_norm": 0.32603856921195984,
      "learning_rate": 5.327015646150716e-05,
      "loss": 0.8165,
      "step": 250
    },
    {
      "epoch": 0.05000747123574239,
      "grad_norm": 0.20693735778331757,
      "learning_rate": 5.294354018255945e-05,
      "loss": 0.7082,
      "step": 251
    },
    {
      "epoch": 0.05020670418887284,
      "grad_norm": 0.2583768367767334,
      "learning_rate": 5.26167978121472e-05,
      "loss": 0.6606,
      "step": 252
    },
    {
      "epoch": 0.05040593714200329,
      "grad_norm": 0.20723305642604828,
      "learning_rate": 5.228994334682604e-05,
      "loss": 0.5572,
      "step": 253
    },
    {
      "epoch": 0.05060517009513373,
      "grad_norm": 0.21328355371952057,
      "learning_rate": 5.196299078795344e-05,
      "loss": 0.6035,
      "step": 254
    },
    {
      "epoch": 0.05080440304826418,
      "grad_norm": 0.17065133154392242,
      "learning_rate": 5.1635954141088813e-05,
      "loss": 0.5368,
      "step": 255
    },
    {
      "epoch": 0.05100363600139463,
      "grad_norm": 0.1970246285200119,
      "learning_rate": 5.1308847415393666e-05,
      "loss": 0.6758,
      "step": 256
    },
    {
      "epoch": 0.051202868954525076,
      "grad_norm": 0.19214090704917908,
      "learning_rate": 5.0981684623031415e-05,
      "loss": 0.6448,
      "step": 257
    },
    {
      "epoch": 0.051402101907655526,
      "grad_norm": 0.2105257362127304,
      "learning_rate": 5.0654479778567223e-05,
      "loss": 0.6485,
      "step": 258
    },
    {
      "epoch": 0.051601334860785976,
      "grad_norm": 0.1960887312889099,
      "learning_rate": 5.0327246898367597e-05,
      "loss": 0.6057,
      "step": 259
    },
    {
      "epoch": 0.05180056781391642,
      "grad_norm": 0.16844411194324493,
      "learning_rate": 5e-05,
      "loss": 0.5688,
      "step": 260
    },
    {
      "epoch": 0.05199980076704687,
      "grad_norm": 0.20722134411334991,
      "learning_rate": 4.9672753101632415e-05,
      "loss": 0.5989,
      "step": 261
    },
    {
      "epoch": 0.05219903372017732,
      "grad_norm": 0.20455193519592285,
      "learning_rate": 4.934552022143279e-05,
      "loss": 0.6179,
      "step": 262
    },
    {
      "epoch": 0.05239826667330776,
      "grad_norm": 0.21068598330020905,
      "learning_rate": 4.901831537696859e-05,
      "loss": 0.6609,
      "step": 263
    },
    {
      "epoch": 0.05259749962643821,
      "grad_norm": 0.2080422341823578,
      "learning_rate": 4.869115258460635e-05,
      "loss": 0.6534,
      "step": 264
    },
    {
      "epoch": 0.05279673257956866,
      "grad_norm": 0.2010425478219986,
      "learning_rate": 4.83640458589112e-05,
      "loss": 0.6078,
      "step": 265
    },
    {
      "epoch": 0.052995965532699106,
      "grad_norm": 0.1816807985305786,
      "learning_rate": 4.8037009212046586e-05,
      "loss": 0.5583,
      "step": 266
    },
    {
      "epoch": 0.053195198485829556,
      "grad_norm": 0.21408969163894653,
      "learning_rate": 4.7710056653173976e-05,
      "loss": 0.6408,
      "step": 267
    },
    {
      "epoch": 0.053394431438960006,
      "grad_norm": 0.23079368472099304,
      "learning_rate": 4.738320218785281e-05,
      "loss": 0.692,
      "step": 268
    },
    {
      "epoch": 0.05359366439209045,
      "grad_norm": 0.24382252991199493,
      "learning_rate": 4.7056459817440544e-05,
      "loss": 0.7265,
      "step": 269
    },
    {
      "epoch": 0.0537928973452209,
      "grad_norm": 0.23294223845005035,
      "learning_rate": 4.6729843538492847e-05,
      "loss": 0.7042,
      "step": 270
    },
    {
      "epoch": 0.05399213029835135,
      "grad_norm": 0.22150427103042603,
      "learning_rate": 4.640336734216403e-05,
      "loss": 0.6907,
      "step": 271
    },
    {
      "epoch": 0.05419136325148179,
      "grad_norm": 0.21380844712257385,
      "learning_rate": 4.607704521360776e-05,
      "loss": 0.6596,
      "step": 272
    },
    {
      "epoch": 0.05439059620461224,
      "grad_norm": 0.22134555876255035,
      "learning_rate": 4.575089113137792e-05,
      "loss": 0.7081,
      "step": 273
    },
    {
      "epoch": 0.05458982915774269,
      "grad_norm": 0.237199068069458,
      "learning_rate": 4.542491906682989e-05,
      "loss": 0.677,
      "step": 274
    },
    {
      "epoch": 0.05478906211087314,
      "grad_norm": 0.2207615077495575,
      "learning_rate": 4.509914298352197e-05,
      "loss": 0.6395,
      "step": 275
    },
    {
      "epoch": 0.05498829506400359,
      "grad_norm": 0.24687650799751282,
      "learning_rate": 4.477357683661734e-05,
      "loss": 0.6936,
      "step": 276
    },
    {
      "epoch": 0.05518752801713404,
      "grad_norm": 0.271902859210968,
      "learning_rate": 4.444823457228612e-05,
      "loss": 0.7742,
      "step": 277
    },
    {
      "epoch": 0.05538676097026448,
      "grad_norm": 0.23451142013072968,
      "learning_rate": 4.412313012710813e-05,
      "loss": 0.7369,
      "step": 278
    },
    {
      "epoch": 0.05558599392339493,
      "grad_norm": 0.24319756031036377,
      "learning_rate": 4.379827742747575e-05,
      "loss": 0.6709,
      "step": 279
    },
    {
      "epoch": 0.05578522687652538,
      "grad_norm": 0.24758942425251007,
      "learning_rate": 4.347369038899744e-05,
      "loss": 0.6954,
      "step": 280
    },
    {
      "epoch": 0.055984459829655824,
      "grad_norm": 0.20816007256507874,
      "learning_rate": 4.3149382915901606e-05,
      "loss": 0.6465,
      "step": 281
    },
    {
      "epoch": 0.056183692782786274,
      "grad_norm": 0.22767412662506104,
      "learning_rate": 4.282536890044104e-05,
      "loss": 0.6318,
      "step": 282
    },
    {
      "epoch": 0.056382925735916724,
      "grad_norm": 0.24410390853881836,
      "learning_rate": 4.250166222229774e-05,
      "loss": 0.7362,
      "step": 283
    },
    {
      "epoch": 0.05658215868904717,
      "grad_norm": 0.25331801176071167,
      "learning_rate": 4.2178276747988446e-05,
      "loss": 0.7286,
      "step": 284
    },
    {
      "epoch": 0.05678139164217762,
      "grad_norm": 0.2666231393814087,
      "learning_rate": 4.185522633027057e-05,
      "loss": 0.681,
      "step": 285
    },
    {
      "epoch": 0.05698062459530807,
      "grad_norm": 0.2353515923023224,
      "learning_rate": 4.153252480754877e-05,
      "loss": 0.7308,
      "step": 286
    },
    {
      "epoch": 0.05717985754843851,
      "grad_norm": 0.23132577538490295,
      "learning_rate": 4.1210186003282275e-05,
      "loss": 0.6479,
      "step": 287
    },
    {
      "epoch": 0.05737909050156896,
      "grad_norm": 0.23580260574817657,
      "learning_rate": 4.088822372539263e-05,
      "loss": 0.638,
      "step": 288
    },
    {
      "epoch": 0.057578323454699404,
      "grad_norm": 0.2322821170091629,
      "learning_rate": 4.0566651765672246e-05,
      "loss": 0.6389,
      "step": 289
    },
    {
      "epoch": 0.057777556407829854,
      "grad_norm": 0.269782692193985,
      "learning_rate": 4.0245483899193595e-05,
      "loss": 0.7522,
      "step": 290
    },
    {
      "epoch": 0.057976789360960304,
      "grad_norm": 0.24970482289791107,
      "learning_rate": 3.992473388371915e-05,
      "loss": 0.7166,
      "step": 291
    },
    {
      "epoch": 0.05817602231409075,
      "grad_norm": 0.25823989510536194,
      "learning_rate": 3.960441545911204e-05,
      "loss": 0.7155,
      "step": 292
    },
    {
      "epoch": 0.0583752552672212,
      "grad_norm": 0.23043105006217957,
      "learning_rate": 3.928454234674747e-05,
      "loss": 0.6239,
      "step": 293
    },
    {
      "epoch": 0.05857448822035165,
      "grad_norm": 0.26879456639289856,
      "learning_rate": 3.896512824892495e-05,
      "loss": 0.7149,
      "step": 294
    },
    {
      "epoch": 0.05877372117348209,
      "grad_norm": 0.26333490014076233,
      "learning_rate": 3.864618684828134e-05,
      "loss": 0.7298,
      "step": 295
    },
    {
      "epoch": 0.05897295412661254,
      "grad_norm": 0.26092275977134705,
      "learning_rate": 3.832773180720475e-05,
      "loss": 0.6897,
      "step": 296
    },
    {
      "epoch": 0.05917218707974299,
      "grad_norm": 0.2566492557525635,
      "learning_rate": 3.800977676724919e-05,
      "loss": 0.6571,
      "step": 297
    },
    {
      "epoch": 0.059371420032873434,
      "grad_norm": 0.27104929089546204,
      "learning_rate": 3.769233534855035e-05,
      "loss": 0.7303,
      "step": 298
    },
    {
      "epoch": 0.059570652986003884,
      "grad_norm": 0.2734347879886627,
      "learning_rate": 3.73754211492421e-05,
      "loss": 0.7918,
      "step": 299
    },
    {
      "epoch": 0.059769885939134335,
      "grad_norm": 0.24449193477630615,
      "learning_rate": 3.705904774487396e-05,
      "loss": 0.6002,
      "step": 300
    },
    {
      "epoch": 0.059769885939134335,
      "eval_loss": 0.6857987642288208,
      "eval_runtime": 265.3385,
      "eval_samples_per_second": 31.861,
      "eval_steps_per_second": 7.967,
      "step": 300
    },
    {
      "epoch": 0.05996911889226478,
      "grad_norm": 0.1829785406589508,
      "learning_rate": 3.6743228687829595e-05,
      "loss": 0.4575,
      "step": 301
    },
    {
      "epoch": 0.06016835184539523,
      "grad_norm": 0.2460087090730667,
      "learning_rate": 3.642797750674629e-05,
      "loss": 0.5853,
      "step": 302
    },
    {
      "epoch": 0.06036758479852568,
      "grad_norm": 0.25549712777137756,
      "learning_rate": 3.6113307705935396e-05,
      "loss": 0.6608,
      "step": 303
    },
    {
      "epoch": 0.06056681775165612,
      "grad_norm": 0.25257304310798645,
      "learning_rate": 3.579923276480387e-05,
      "loss": 0.6838,
      "step": 304
    },
    {
      "epoch": 0.06076605070478657,
      "grad_norm": 0.27871885895729065,
      "learning_rate": 3.5485766137276894e-05,
      "loss": 0.7465,
      "step": 305
    },
    {
      "epoch": 0.06096528365791702,
      "grad_norm": 0.21810035407543182,
      "learning_rate": 3.5172921251221455e-05,
      "loss": 0.6733,
      "step": 306
    },
    {
      "epoch": 0.061164516611047465,
      "grad_norm": 0.19901888072490692,
      "learning_rate": 3.486071150787128e-05,
      "loss": 0.6081,
      "step": 307
    },
    {
      "epoch": 0.061363749564177915,
      "grad_norm": 0.21920450031757355,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.6168,
      "step": 308
    },
    {
      "epoch": 0.061562982517308365,
      "grad_norm": 0.22061915695667267,
      "learning_rate": 3.423825091761153e-05,
      "loss": 0.6026,
      "step": 309
    },
    {
      "epoch": 0.06176221547043881,
      "grad_norm": 0.22976446151733398,
      "learning_rate": 3.392802673484193e-05,
      "loss": 0.7188,
      "step": 310
    },
    {
      "epoch": 0.06196144842356926,
      "grad_norm": 0.23608386516571045,
      "learning_rate": 3.361849102191533e-05,
      "loss": 0.6482,
      "step": 311
    },
    {
      "epoch": 0.06216068137669971,
      "grad_norm": 0.24407218396663666,
      "learning_rate": 3.330965703831146e-05,
      "loss": 0.6767,
      "step": 312
    },
    {
      "epoch": 0.06235991432983015,
      "grad_norm": 0.2779025137424469,
      "learning_rate": 3.300153801345028e-05,
      "loss": 0.7787,
      "step": 313
    },
    {
      "epoch": 0.0625591472829606,
      "grad_norm": 0.21833088994026184,
      "learning_rate": 3.2694147146125345e-05,
      "loss": 0.5775,
      "step": 314
    },
    {
      "epoch": 0.06275838023609105,
      "grad_norm": 0.2857174277305603,
      "learning_rate": 3.2387497603938326e-05,
      "loss": 0.7358,
      "step": 315
    },
    {
      "epoch": 0.0629576131892215,
      "grad_norm": 0.2404899299144745,
      "learning_rate": 3.2081602522734986e-05,
      "loss": 0.6752,
      "step": 316
    },
    {
      "epoch": 0.06315684614235194,
      "grad_norm": 0.25618627667427063,
      "learning_rate": 3.177647500604252e-05,
      "loss": 0.7334,
      "step": 317
    },
    {
      "epoch": 0.0633560790954824,
      "grad_norm": 0.21781964600086212,
      "learning_rate": 3.147212812450819e-05,
      "loss": 0.5937,
      "step": 318
    },
    {
      "epoch": 0.06355531204861284,
      "grad_norm": 0.21672700345516205,
      "learning_rate": 3.116857491533947e-05,
      "loss": 0.6095,
      "step": 319
    },
    {
      "epoch": 0.06375454500174328,
      "grad_norm": 0.2422553449869156,
      "learning_rate": 3.086582838174551e-05,
      "loss": 0.6426,
      "step": 320
    },
    {
      "epoch": 0.06395377795487374,
      "grad_norm": 0.2380169779062271,
      "learning_rate": 3.056390149238022e-05,
      "loss": 0.6648,
      "step": 321
    },
    {
      "epoch": 0.06415301090800418,
      "grad_norm": 0.224848210811615,
      "learning_rate": 3.0262807180786647e-05,
      "loss": 0.6811,
      "step": 322
    },
    {
      "epoch": 0.06435224386113463,
      "grad_norm": 0.2206374555826187,
      "learning_rate": 2.996255834484296e-05,
      "loss": 0.6629,
      "step": 323
    },
    {
      "epoch": 0.06455147681426508,
      "grad_norm": 0.22571969032287598,
      "learning_rate": 2.9663167846209998e-05,
      "loss": 0.641,
      "step": 324
    },
    {
      "epoch": 0.06475070976739553,
      "grad_norm": 0.22498847544193268,
      "learning_rate": 2.936464850978027e-05,
      "loss": 0.642,
      "step": 325
    },
    {
      "epoch": 0.06494994272052597,
      "grad_norm": 0.26433423161506653,
      "learning_rate": 2.9067013123128613e-05,
      "loss": 0.7015,
      "step": 326
    },
    {
      "epoch": 0.06514917567365643,
      "grad_norm": 0.2343292236328125,
      "learning_rate": 2.8770274435964355e-05,
      "loss": 0.6088,
      "step": 327
    },
    {
      "epoch": 0.06534840862678687,
      "grad_norm": 0.2766936123371124,
      "learning_rate": 2.8474445159585235e-05,
      "loss": 0.6763,
      "step": 328
    },
    {
      "epoch": 0.06554764157991731,
      "grad_norm": 0.2363848239183426,
      "learning_rate": 2.8179537966332887e-05,
      "loss": 0.7461,
      "step": 329
    },
    {
      "epoch": 0.06574687453304777,
      "grad_norm": 0.2353663295507431,
      "learning_rate": 2.7885565489049946e-05,
      "loss": 0.7216,
      "step": 330
    },
    {
      "epoch": 0.06594610748617821,
      "grad_norm": 0.2601001262664795,
      "learning_rate": 2.759254032053888e-05,
      "loss": 0.7722,
      "step": 331
    },
    {
      "epoch": 0.06614534043930866,
      "grad_norm": 0.23950740694999695,
      "learning_rate": 2.7300475013022663e-05,
      "loss": 0.5964,
      "step": 332
    },
    {
      "epoch": 0.06634457339243911,
      "grad_norm": 0.23393948376178741,
      "learning_rate": 2.700938207760701e-05,
      "loss": 0.6738,
      "step": 333
    },
    {
      "epoch": 0.06654380634556956,
      "grad_norm": 0.2648164629936218,
      "learning_rate": 2.671927398374443e-05,
      "loss": 0.6643,
      "step": 334
    },
    {
      "epoch": 0.0667430392987,
      "grad_norm": 0.263930082321167,
      "learning_rate": 2.6430163158700115e-05,
      "loss": 0.6688,
      "step": 335
    },
    {
      "epoch": 0.06694227225183046,
      "grad_norm": 0.2163466513156891,
      "learning_rate": 2.6142061987019577e-05,
      "loss": 0.5839,
      "step": 336
    },
    {
      "epoch": 0.0671415052049609,
      "grad_norm": 0.23970170319080353,
      "learning_rate": 2.5854982809998153e-05,
      "loss": 0.6826,
      "step": 337
    },
    {
      "epoch": 0.06734073815809134,
      "grad_norm": 0.26360487937927246,
      "learning_rate": 2.556893792515227e-05,
      "loss": 0.7177,
      "step": 338
    },
    {
      "epoch": 0.0675399711112218,
      "grad_norm": 0.2591302692890167,
      "learning_rate": 2.5283939585692783e-05,
      "loss": 0.6781,
      "step": 339
    },
    {
      "epoch": 0.06773920406435224,
      "grad_norm": 0.26808708906173706,
      "learning_rate": 2.500000000000001e-05,
      "loss": 0.6558,
      "step": 340
    },
    {
      "epoch": 0.06793843701748269,
      "grad_norm": 0.2927316725254059,
      "learning_rate": 2.471713133110078e-05,
      "loss": 0.715,
      "step": 341
    },
    {
      "epoch": 0.06813766997061314,
      "grad_norm": 0.25598007440567017,
      "learning_rate": 2.4435345696147403e-05,
      "loss": 0.6857,
      "step": 342
    },
    {
      "epoch": 0.06833690292374359,
      "grad_norm": 0.27012765407562256,
      "learning_rate": 2.4154655165898627e-05,
      "loss": 0.6704,
      "step": 343
    },
    {
      "epoch": 0.06853613587687403,
      "grad_norm": 0.27995193004608154,
      "learning_rate": 2.3875071764202563e-05,
      "loss": 0.6795,
      "step": 344
    },
    {
      "epoch": 0.06873536883000449,
      "grad_norm": 0.2844606935977936,
      "learning_rate": 2.3596607467481603e-05,
      "loss": 0.7176,
      "step": 345
    },
    {
      "epoch": 0.06893460178313493,
      "grad_norm": 0.3271996080875397,
      "learning_rate": 2.3319274204219428e-05,
      "loss": 0.7862,
      "step": 346
    },
    {
      "epoch": 0.06913383473626537,
      "grad_norm": 0.24104076623916626,
      "learning_rate": 2.3043083854449988e-05,
      "loss": 0.6574,
      "step": 347
    },
    {
      "epoch": 0.06933306768939583,
      "grad_norm": 0.27659985423088074,
      "learning_rate": 2.2768048249248648e-05,
      "loss": 0.7289,
      "step": 348
    },
    {
      "epoch": 0.06953230064252627,
      "grad_norm": 0.29960310459136963,
      "learning_rate": 2.2494179170225333e-05,
      "loss": 0.6818,
      "step": 349
    },
    {
      "epoch": 0.06973153359565672,
      "grad_norm": 0.3256206810474396,
      "learning_rate": 2.2221488349019903e-05,
      "loss": 0.7484,
      "step": 350
    },
    {
      "epoch": 0.06993076654878717,
      "grad_norm": 0.21653087437152863,
      "learning_rate": 2.194998746679952e-05,
      "loss": 0.6332,
      "step": 351
    },
    {
      "epoch": 0.07012999950191762,
      "grad_norm": 0.24491187930107117,
      "learning_rate": 2.167968815375837e-05,
      "loss": 0.6758,
      "step": 352
    },
    {
      "epoch": 0.07032923245504806,
      "grad_norm": 0.2615233361721039,
      "learning_rate": 2.1410601988619394e-05,
      "loss": 0.6832,
      "step": 353
    },
    {
      "epoch": 0.07052846540817852,
      "grad_norm": 0.3030376136302948,
      "learning_rate": 2.1142740498138324e-05,
      "loss": 0.7022,
      "step": 354
    },
    {
      "epoch": 0.07072769836130896,
      "grad_norm": 0.24537107348442078,
      "learning_rate": 2.08761151566099e-05,
      "loss": 0.6833,
      "step": 355
    },
    {
      "epoch": 0.0709269313144394,
      "grad_norm": 0.22852526605129242,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.6045,
      "step": 356
    },
    {
      "epoch": 0.07112616426756986,
      "grad_norm": 0.2126518040895462,
      "learning_rate": 2.034661855233815e-05,
      "loss": 0.6393,
      "step": 357
    },
    {
      "epoch": 0.0713253972207003,
      "grad_norm": 0.2170926332473755,
      "learning_rate": 2.008376997146705e-05,
      "loss": 0.636,
      "step": 358
    },
    {
      "epoch": 0.07152463017383075,
      "grad_norm": 0.24603933095932007,
      "learning_rate": 1.982220290232143e-05,
      "loss": 0.6854,
      "step": 359
    },
    {
      "epoch": 0.0717238631269612,
      "grad_norm": 0.25937581062316895,
      "learning_rate": 1.9561928549563968e-05,
      "loss": 0.6989,
      "step": 360
    },
    {
      "epoch": 0.07192309608009165,
      "grad_norm": 0.2045435905456543,
      "learning_rate": 1.9302958062481673e-05,
      "loss": 0.5754,
      "step": 361
    },
    {
      "epoch": 0.07212232903322209,
      "grad_norm": 0.24368344247341156,
      "learning_rate": 1.9045302534508297e-05,
      "loss": 0.6537,
      "step": 362
    },
    {
      "epoch": 0.07232156198635255,
      "grad_norm": 0.2470693588256836,
      "learning_rate": 1.8788973002749112e-05,
      "loss": 0.675,
      "step": 363
    },
    {
      "epoch": 0.07252079493948299,
      "grad_norm": 0.23468363285064697,
      "learning_rate": 1.8533980447508137e-05,
      "loss": 0.675,
      "step": 364
    },
    {
      "epoch": 0.07272002789261343,
      "grad_norm": 0.23341692984104156,
      "learning_rate": 1.8280335791817733e-05,
      "loss": 0.6899,
      "step": 365
    },
    {
      "epoch": 0.07291926084574389,
      "grad_norm": 0.24296964704990387,
      "learning_rate": 1.8028049900970767e-05,
      "loss": 0.6898,
      "step": 366
    },
    {
      "epoch": 0.07311849379887433,
      "grad_norm": 0.21605587005615234,
      "learning_rate": 1.777713358205514e-05,
      "loss": 0.615,
      "step": 367
    },
    {
      "epoch": 0.07331772675200478,
      "grad_norm": 0.23035632073879242,
      "learning_rate": 1.7527597583490822e-05,
      "loss": 0.7135,
      "step": 368
    },
    {
      "epoch": 0.07351695970513523,
      "grad_norm": 0.24987879395484924,
      "learning_rate": 1.7279452594569483e-05,
      "loss": 0.6646,
      "step": 369
    },
    {
      "epoch": 0.07371619265826568,
      "grad_norm": 0.23275192081928253,
      "learning_rate": 1.703270924499656e-05,
      "loss": 0.6405,
      "step": 370
    },
    {
      "epoch": 0.07391542561139612,
      "grad_norm": 0.2441992163658142,
      "learning_rate": 1.678737810443593e-05,
      "loss": 0.7028,
      "step": 371
    },
    {
      "epoch": 0.07411465856452658,
      "grad_norm": 0.21933166682720184,
      "learning_rate": 1.6543469682057106e-05,
      "loss": 0.5545,
      "step": 372
    },
    {
      "epoch": 0.07431389151765702,
      "grad_norm": 0.24288050830364227,
      "learning_rate": 1.6300994426085103e-05,
      "loss": 0.6971,
      "step": 373
    },
    {
      "epoch": 0.07451312447078746,
      "grad_norm": 0.2461194097995758,
      "learning_rate": 1.605996272335291e-05,
      "loss": 0.6359,
      "step": 374
    },
    {
      "epoch": 0.07471235742391792,
      "grad_norm": 0.25669926404953003,
      "learning_rate": 1.5820384898856434e-05,
      "loss": 0.6917,
      "step": 375
    },
    {
      "epoch": 0.07491159037704836,
      "grad_norm": 0.23437441885471344,
      "learning_rate": 1.5582271215312294e-05,
      "loss": 0.6353,
      "step": 376
    },
    {
      "epoch": 0.07511082333017881,
      "grad_norm": 0.2659025192260742,
      "learning_rate": 1.5345631872718214e-05,
      "loss": 0.7663,
      "step": 377
    },
    {
      "epoch": 0.07531005628330927,
      "grad_norm": 0.23349608480930328,
      "learning_rate": 1.5110477007916001e-05,
      "loss": 0.6564,
      "step": 378
    },
    {
      "epoch": 0.07550928923643971,
      "grad_norm": 0.24196575582027435,
      "learning_rate": 1.4876816694157419e-05,
      "loss": 0.6607,
      "step": 379
    },
    {
      "epoch": 0.07570852218957015,
      "grad_norm": 0.23814047873020172,
      "learning_rate": 1.4644660940672627e-05,
      "loss": 0.6354,
      "step": 380
    },
    {
      "epoch": 0.07590775514270061,
      "grad_norm": 0.23850150406360626,
      "learning_rate": 1.4414019692241437e-05,
      "loss": 0.6643,
      "step": 381
    },
    {
      "epoch": 0.07610698809583105,
      "grad_norm": 0.22468359768390656,
      "learning_rate": 1.4184902828767287e-05,
      "loss": 0.6246,
      "step": 382
    },
    {
      "epoch": 0.0763062210489615,
      "grad_norm": 0.23000527918338776,
      "learning_rate": 1.3957320164854059e-05,
      "loss": 0.6283,
      "step": 383
    },
    {
      "epoch": 0.07650545400209195,
      "grad_norm": 0.2583581507205963,
      "learning_rate": 1.373128144938563e-05,
      "loss": 0.6927,
      "step": 384
    },
    {
      "epoch": 0.0767046869552224,
      "grad_norm": 0.2393674999475479,
      "learning_rate": 1.3506796365108232e-05,
      "loss": 0.668,
      "step": 385
    },
    {
      "epoch": 0.07690391990835284,
      "grad_norm": 0.2238411009311676,
      "learning_rate": 1.3283874528215733e-05,
      "loss": 0.6291,
      "step": 386
    },
    {
      "epoch": 0.0771031528614833,
      "grad_norm": 0.22692270576953888,
      "learning_rate": 1.3062525487937699e-05,
      "loss": 0.6488,
      "step": 387
    },
    {
      "epoch": 0.07730238581461374,
      "grad_norm": 0.29837989807128906,
      "learning_rate": 1.2842758726130283e-05,
      "loss": 0.7459,
      "step": 388
    },
    {
      "epoch": 0.07750161876774418,
      "grad_norm": 0.26298198103904724,
      "learning_rate": 1.2624583656870154e-05,
      "loss": 0.6305,
      "step": 389
    },
    {
      "epoch": 0.07770085172087464,
      "grad_norm": 0.2501063644886017,
      "learning_rate": 1.2408009626051137e-05,
      "loss": 0.6475,
      "step": 390
    },
    {
      "epoch": 0.07790008467400508,
      "grad_norm": 0.26658689975738525,
      "learning_rate": 1.2193045910983863e-05,
      "loss": 0.6945,
      "step": 391
    },
    {
      "epoch": 0.07809931762713553,
      "grad_norm": 0.27355584502220154,
      "learning_rate": 1.1979701719998453e-05,
      "loss": 0.7023,
      "step": 392
    },
    {
      "epoch": 0.07829855058026598,
      "grad_norm": 0.24372565746307373,
      "learning_rate": 1.1767986192049984e-05,
      "loss": 0.6686,
      "step": 393
    },
    {
      "epoch": 0.07849778353339643,
      "grad_norm": 0.28914278745651245,
      "learning_rate": 1.1557908396327028e-05,
      "loss": 0.7444,
      "step": 394
    },
    {
      "epoch": 0.07869701648652687,
      "grad_norm": 0.26138660311698914,
      "learning_rate": 1.134947733186315e-05,
      "loss": 0.6624,
      "step": 395
    },
    {
      "epoch": 0.07889624943965733,
      "grad_norm": 0.25570860505104065,
      "learning_rate": 1.1142701927151456e-05,
      "loss": 0.6746,
      "step": 396
    },
    {
      "epoch": 0.07909548239278777,
      "grad_norm": 0.2897097170352936,
      "learning_rate": 1.0937591039762085e-05,
      "loss": 0.7515,
      "step": 397
    },
    {
      "epoch": 0.07929471534591821,
      "grad_norm": 0.2693999111652374,
      "learning_rate": 1.0734153455962765e-05,
      "loss": 0.7385,
      "step": 398
    },
    {
      "epoch": 0.07949394829904867,
      "grad_norm": 0.2895209491252899,
      "learning_rate": 1.0532397890342505e-05,
      "loss": 0.7491,
      "step": 399
    },
    {
      "epoch": 0.07969318125217911,
      "grad_norm": 0.2964557707309723,
      "learning_rate": 1.0332332985438248e-05,
      "loss": 0.7342,
      "step": 400
    },
    {
      "epoch": 0.07969318125217911,
      "eval_loss": 0.6768385171890259,
      "eval_runtime": 265.4226,
      "eval_samples_per_second": 31.851,
      "eval_steps_per_second": 7.965,
      "step": 400
    },
    {
      "epoch": 0.07989241420530956,
      "grad_norm": 0.19331474602222443,
      "learning_rate": 1.013396731136465e-05,
      "loss": 0.5627,
      "step": 401
    },
    {
      "epoch": 0.08009164715844,
      "grad_norm": 0.24819611012935638,
      "learning_rate": 9.937309365446973e-06,
      "loss": 0.7396,
      "step": 402
    },
    {
      "epoch": 0.08029088011157046,
      "grad_norm": 0.2520431578159332,
      "learning_rate": 9.742367571857091e-06,
      "loss": 0.6631,
      "step": 403
    },
    {
      "epoch": 0.0804901130647009,
      "grad_norm": 0.21120384335517883,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.6727,
      "step": 404
    },
    {
      "epoch": 0.08068934601783134,
      "grad_norm": 0.20902977883815765,
      "learning_rate": 9.357665770419244e-06,
      "loss": 0.6242,
      "step": 405
    },
    {
      "epoch": 0.0808885789709618,
      "grad_norm": 0.18309302628040314,
      "learning_rate": 9.167922241916055e-06,
      "loss": 0.5031,
      "step": 406
    },
    {
      "epoch": 0.08108781192409224,
      "grad_norm": 0.21523945033550262,
      "learning_rate": 8.97992782372432e-06,
      "loss": 0.6427,
      "step": 407
    },
    {
      "epoch": 0.08128704487722269,
      "grad_norm": 0.22867245972156525,
      "learning_rate": 8.793690568899216e-06,
      "loss": 0.6857,
      "step": 408
    },
    {
      "epoch": 0.08148627783035314,
      "grad_norm": 0.2331218272447586,
      "learning_rate": 8.609218455224893e-06,
      "loss": 0.6583,
      "step": 409
    },
    {
      "epoch": 0.08168551078348359,
      "grad_norm": 0.2502421736717224,
      "learning_rate": 8.426519384872733e-06,
      "loss": 0.7734,
      "step": 410
    },
    {
      "epoch": 0.08188474373661403,
      "grad_norm": 0.22778432071208954,
      "learning_rate": 8.245601184062852e-06,
      "loss": 0.6216,
      "step": 411
    },
    {
      "epoch": 0.08208397668974449,
      "grad_norm": 0.20863741636276245,
      "learning_rate": 8.066471602728803e-06,
      "loss": 0.5912,
      "step": 412
    },
    {
      "epoch": 0.08228320964287493,
      "grad_norm": 0.22159984707832336,
      "learning_rate": 7.889138314185678e-06,
      "loss": 0.6247,
      "step": 413
    },
    {
      "epoch": 0.08248244259600537,
      "grad_norm": 0.2666146457195282,
      "learning_rate": 7.71360891480134e-06,
      "loss": 0.748,
      "step": 414
    },
    {
      "epoch": 0.08268167554913583,
      "grad_norm": 0.25675085186958313,
      "learning_rate": 7.539890923671062e-06,
      "loss": 0.6865,
      "step": 415
    },
    {
      "epoch": 0.08288090850226627,
      "grad_norm": 0.24603363871574402,
      "learning_rate": 7.367991782295391e-06,
      "loss": 0.6717,
      "step": 416
    },
    {
      "epoch": 0.08308014145539672,
      "grad_norm": 0.31044501066207886,
      "learning_rate": 7.197918854261432e-06,
      "loss": 0.662,
      "step": 417
    },
    {
      "epoch": 0.08327937440852717,
      "grad_norm": 0.22573980689048767,
      "learning_rate": 7.029679424927365e-06,
      "loss": 0.6391,
      "step": 418
    },
    {
      "epoch": 0.08347860736165762,
      "grad_norm": 0.22419226169586182,
      "learning_rate": 6.863280701110408e-06,
      "loss": 0.6233,
      "step": 419
    },
    {
      "epoch": 0.08367784031478806,
      "grad_norm": 0.2326180338859558,
      "learning_rate": 6.698729810778065e-06,
      "loss": 0.7135,
      "step": 420
    },
    {
      "epoch": 0.08387707326791852,
      "grad_norm": 0.2318958342075348,
      "learning_rate": 6.536033802742813e-06,
      "loss": 0.6337,
      "step": 421
    },
    {
      "epoch": 0.08407630622104896,
      "grad_norm": 0.24011459946632385,
      "learning_rate": 6.375199646360142e-06,
      "loss": 0.6822,
      "step": 422
    },
    {
      "epoch": 0.0842755391741794,
      "grad_norm": 0.25653573870658875,
      "learning_rate": 6.216234231230012e-06,
      "loss": 0.6508,
      "step": 423
    },
    {
      "epoch": 0.08447477212730986,
      "grad_norm": 0.22322338819503784,
      "learning_rate": 6.059144366901736e-06,
      "loss": 0.6078,
      "step": 424
    },
    {
      "epoch": 0.0846740050804403,
      "grad_norm": 0.2577781677246094,
      "learning_rate": 5.903936782582253e-06,
      "loss": 0.683,
      "step": 425
    },
    {
      "epoch": 0.08487323803357075,
      "grad_norm": 0.22077085077762604,
      "learning_rate": 5.750618126847912e-06,
      "loss": 0.6902,
      "step": 426
    },
    {
      "epoch": 0.0850724709867012,
      "grad_norm": 0.2472268044948578,
      "learning_rate": 5.599194967359639e-06,
      "loss": 0.6837,
      "step": 427
    },
    {
      "epoch": 0.08527170393983165,
      "grad_norm": 0.22269569337368011,
      "learning_rate": 5.449673790581611e-06,
      "loss": 0.6202,
      "step": 428
    },
    {
      "epoch": 0.08547093689296209,
      "grad_norm": 0.21259906888008118,
      "learning_rate": 5.302061001503394e-06,
      "loss": 0.6024,
      "step": 429
    },
    {
      "epoch": 0.08567016984609255,
      "grad_norm": 0.24221090972423553,
      "learning_rate": 5.156362923365588e-06,
      "loss": 0.6671,
      "step": 430
    },
    {
      "epoch": 0.08586940279922299,
      "grad_norm": 0.24661476910114288,
      "learning_rate": 5.012585797388936e-06,
      "loss": 0.658,
      "step": 431
    },
    {
      "epoch": 0.08606863575235343,
      "grad_norm": 0.2605218291282654,
      "learning_rate": 4.87073578250698e-06,
      "loss": 0.63,
      "step": 432
    },
    {
      "epoch": 0.08626786870548389,
      "grad_norm": 0.2187577337026596,
      "learning_rate": 4.730818955102234e-06,
      "loss": 0.5845,
      "step": 433
    },
    {
      "epoch": 0.08646710165861433,
      "grad_norm": 0.2221374809741974,
      "learning_rate": 4.592841308745932e-06,
      "loss": 0.5466,
      "step": 434
    },
    {
      "epoch": 0.08666633461174478,
      "grad_norm": 0.2580986022949219,
      "learning_rate": 4.456808753941205e-06,
      "loss": 0.6753,
      "step": 435
    },
    {
      "epoch": 0.08686556756487523,
      "grad_norm": 0.2541551887989044,
      "learning_rate": 4.322727117869951e-06,
      "loss": 0.655,
      "step": 436
    },
    {
      "epoch": 0.08706480051800568,
      "grad_norm": 0.31278523802757263,
      "learning_rate": 4.190602144143207e-06,
      "loss": 0.7489,
      "step": 437
    },
    {
      "epoch": 0.08726403347113612,
      "grad_norm": 0.2648221552371979,
      "learning_rate": 4.06043949255509e-06,
      "loss": 0.6957,
      "step": 438
    },
    {
      "epoch": 0.08746326642426658,
      "grad_norm": 0.2753226161003113,
      "learning_rate": 3.932244738840379e-06,
      "loss": 0.6294,
      "step": 439
    },
    {
      "epoch": 0.08766249937739702,
      "grad_norm": 0.26904359459877014,
      "learning_rate": 3.8060233744356633e-06,
      "loss": 0.7311,
      "step": 440
    },
    {
      "epoch": 0.08786173233052746,
      "grad_norm": 0.26399528980255127,
      "learning_rate": 3.681780806244095e-06,
      "loss": 0.6948,
      "step": 441
    },
    {
      "epoch": 0.08806096528365792,
      "grad_norm": 0.2822622060775757,
      "learning_rate": 3.5595223564037884e-06,
      "loss": 0.6786,
      "step": 442
    },
    {
      "epoch": 0.08826019823678836,
      "grad_norm": 0.265007346868515,
      "learning_rate": 3.4392532620598216e-06,
      "loss": 0.7124,
      "step": 443
    },
    {
      "epoch": 0.08845943118991881,
      "grad_norm": 0.257730633020401,
      "learning_rate": 3.3209786751399187e-06,
      "loss": 0.6245,
      "step": 444
    },
    {
      "epoch": 0.08865866414304927,
      "grad_norm": 0.30108872056007385,
      "learning_rate": 3.2047036621337236e-06,
      "loss": 0.7926,
      "step": 445
    },
    {
      "epoch": 0.08885789709617971,
      "grad_norm": 0.3035549819469452,
      "learning_rate": 3.0904332038757977e-06,
      "loss": 0.8031,
      "step": 446
    },
    {
      "epoch": 0.08905713004931015,
      "grad_norm": 0.2673462927341461,
      "learning_rate": 2.978172195332263e-06,
      "loss": 0.6643,
      "step": 447
    },
    {
      "epoch": 0.08925636300244061,
      "grad_norm": 0.3156924545764923,
      "learning_rate": 2.8679254453910785e-06,
      "loss": 0.7175,
      "step": 448
    },
    {
      "epoch": 0.08945559595557105,
      "grad_norm": 0.29648149013519287,
      "learning_rate": 2.759697676656098e-06,
      "loss": 0.7006,
      "step": 449
    },
    {
      "epoch": 0.0896548289087015,
      "grad_norm": 0.304087370634079,
      "learning_rate": 2.653493525244721e-06,
      "loss": 0.6956,
      "step": 450
    },
    {
      "epoch": 0.08985406186183195,
      "grad_norm": 0.18861311674118042,
      "learning_rate": 2.549317540589308e-06,
      "loss": 0.5796,
      "step": 451
    },
    {
      "epoch": 0.0900532948149624,
      "grad_norm": 0.20030918717384338,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.5735,
      "step": 452
    },
    {
      "epoch": 0.09025252776809284,
      "grad_norm": 0.20259323716163635,
      "learning_rate": 2.3470678346851518e-06,
      "loss": 0.5841,
      "step": 453
    },
    {
      "epoch": 0.0904517607212233,
      "grad_norm": 0.22177210450172424,
      "learning_rate": 2.2490027771406687e-06,
      "loss": 0.6743,
      "step": 454
    },
    {
      "epoch": 0.09065099367435374,
      "grad_norm": 0.20536138117313385,
      "learning_rate": 2.152983213389559e-06,
      "loss": 0.6299,
      "step": 455
    },
    {
      "epoch": 0.09085022662748418,
      "grad_norm": 0.21932297945022583,
      "learning_rate": 2.0590132565903476e-06,
      "loss": 0.7102,
      "step": 456
    },
    {
      "epoch": 0.09104945958061464,
      "grad_norm": 0.1917608678340912,
      "learning_rate": 1.9670969321032407e-06,
      "loss": 0.5939,
      "step": 457
    },
    {
      "epoch": 0.09124869253374508,
      "grad_norm": 0.21632328629493713,
      "learning_rate": 1.8772381773176417e-06,
      "loss": 0.6413,
      "step": 458
    },
    {
      "epoch": 0.09144792548687553,
      "grad_norm": 0.23702724277973175,
      "learning_rate": 1.7894408414835362e-06,
      "loss": 0.6321,
      "step": 459
    },
    {
      "epoch": 0.09164715844000598,
      "grad_norm": 0.2379865050315857,
      "learning_rate": 1.70370868554659e-06,
      "loss": 0.6566,
      "step": 460
    },
    {
      "epoch": 0.09184639139313643,
      "grad_norm": 0.23803111910820007,
      "learning_rate": 1.620045381987012e-06,
      "loss": 0.6265,
      "step": 461
    },
    {
      "epoch": 0.09204562434626687,
      "grad_norm": 0.23217985033988953,
      "learning_rate": 1.5384545146622852e-06,
      "loss": 0.7032,
      "step": 462
    },
    {
      "epoch": 0.09224485729939733,
      "grad_norm": 0.2249220609664917,
      "learning_rate": 1.4589395786535953e-06,
      "loss": 0.608,
      "step": 463
    },
    {
      "epoch": 0.09244409025252777,
      "grad_norm": 0.21133165061473846,
      "learning_rate": 1.3815039801161721e-06,
      "loss": 0.6084,
      "step": 464
    },
    {
      "epoch": 0.09264332320565821,
      "grad_norm": 0.22552044689655304,
      "learning_rate": 1.3061510361333185e-06,
      "loss": 0.6371,
      "step": 465
    },
    {
      "epoch": 0.09284255615878867,
      "grad_norm": 0.2244846075773239,
      "learning_rate": 1.232883974574367e-06,
      "loss": 0.654,
      "step": 466
    },
    {
      "epoch": 0.09304178911191911,
      "grad_norm": 0.24191239476203918,
      "learning_rate": 1.1617059339563807e-06,
      "loss": 0.7616,
      "step": 467
    },
    {
      "epoch": 0.09324102206504956,
      "grad_norm": 0.22811263799667358,
      "learning_rate": 1.0926199633097157e-06,
      "loss": 0.6378,
      "step": 468
    },
    {
      "epoch": 0.09344025501818001,
      "grad_norm": 0.22340433299541473,
      "learning_rate": 1.0256290220474307e-06,
      "loss": 0.596,
      "step": 469
    },
    {
      "epoch": 0.09363948797131046,
      "grad_norm": 0.24107715487480164,
      "learning_rate": 9.607359798384785e-07,
      "loss": 0.6633,
      "step": 470
    },
    {
      "epoch": 0.0938387209244409,
      "grad_norm": 0.24450042843818665,
      "learning_rate": 8.979436164848088e-07,
      "loss": 0.706,
      "step": 471
    },
    {
      "epoch": 0.09403795387757136,
      "grad_norm": 0.24326194822788239,
      "learning_rate": 8.372546218022747e-07,
      "loss": 0.6342,
      "step": 472
    },
    {
      "epoch": 0.0942371868307018,
      "grad_norm": 0.25381046533584595,
      "learning_rate": 7.786715955054203e-07,
      "loss": 0.6603,
      "step": 473
    },
    {
      "epoch": 0.09443641978383224,
      "grad_norm": 0.24382087588310242,
      "learning_rate": 7.221970470961125e-07,
      "loss": 0.638,
      "step": 474
    },
    {
      "epoch": 0.0946356527369627,
      "grad_norm": 0.27060163021087646,
      "learning_rate": 6.678333957560512e-07,
      "loss": 0.7178,
      "step": 475
    },
    {
      "epoch": 0.09483488569009314,
      "grad_norm": 0.24051162600517273,
      "learning_rate": 6.15582970243117e-07,
      "loss": 0.5669,
      "step": 476
    },
    {
      "epoch": 0.09503411864322359,
      "grad_norm": 0.2567567527294159,
      "learning_rate": 5.654480087916303e-07,
      "loss": 0.6944,
      "step": 477
    },
    {
      "epoch": 0.09523335159635404,
      "grad_norm": 0.2464078813791275,
      "learning_rate": 5.174306590164879e-07,
      "loss": 0.7134,
      "step": 478
    },
    {
      "epoch": 0.09543258454948449,
      "grad_norm": 0.2505365014076233,
      "learning_rate": 4.715329778211375e-07,
      "loss": 0.7042,
      "step": 479
    },
    {
      "epoch": 0.09563181750261493,
      "grad_norm": 0.253229022026062,
      "learning_rate": 4.277569313094809e-07,
      "loss": 0.6158,
      "step": 480
    },
    {
      "epoch": 0.09583105045574539,
      "grad_norm": 0.23370255529880524,
      "learning_rate": 3.8610439470164737e-07,
      "loss": 0.6129,
      "step": 481
    },
    {
      "epoch": 0.09603028340887583,
      "grad_norm": 0.23140022158622742,
      "learning_rate": 3.465771522536854e-07,
      "loss": 0.5864,
      "step": 482
    },
    {
      "epoch": 0.09622951636200627,
      "grad_norm": 0.2647562026977539,
      "learning_rate": 3.09176897181096e-07,
      "loss": 0.6768,
      "step": 483
    },
    {
      "epoch": 0.09642874931513673,
      "grad_norm": 0.2301555573940277,
      "learning_rate": 2.7390523158633554e-07,
      "loss": 0.6862,
      "step": 484
    },
    {
      "epoch": 0.09662798226826717,
      "grad_norm": 0.25709620118141174,
      "learning_rate": 2.407636663901591e-07,
      "loss": 0.6935,
      "step": 485
    },
    {
      "epoch": 0.09682721522139762,
      "grad_norm": 0.27146580815315247,
      "learning_rate": 2.0975362126691712e-07,
      "loss": 0.7363,
      "step": 486
    },
    {
      "epoch": 0.09702644817452807,
      "grad_norm": 0.25563889741897583,
      "learning_rate": 1.8087642458373134e-07,
      "loss": 0.6381,
      "step": 487
    },
    {
      "epoch": 0.09722568112765852,
      "grad_norm": 0.2607404291629791,
      "learning_rate": 1.5413331334360182e-07,
      "loss": 0.6381,
      "step": 488
    },
    {
      "epoch": 0.09742491408078896,
      "grad_norm": 0.25664660334587097,
      "learning_rate": 1.2952543313240472e-07,
      "loss": 0.7206,
      "step": 489
    },
    {
      "epoch": 0.0976241470339194,
      "grad_norm": 0.26443812251091003,
      "learning_rate": 1.0705383806982606e-07,
      "loss": 0.6878,
      "step": 490
    },
    {
      "epoch": 0.09782337998704986,
      "grad_norm": 0.24642571806907654,
      "learning_rate": 8.671949076420882e-08,
      "loss": 0.6992,
      "step": 491
    },
    {
      "epoch": 0.0980226129401803,
      "grad_norm": 0.25187572836875916,
      "learning_rate": 6.852326227130834e-08,
      "loss": 0.7137,
      "step": 492
    },
    {
      "epoch": 0.09822184589331075,
      "grad_norm": 0.3068501651287079,
      "learning_rate": 5.246593205699424e-08,
      "loss": 0.7202,
      "step": 493
    },
    {
      "epoch": 0.0984210788464412,
      "grad_norm": 0.2589758634567261,
      "learning_rate": 3.8548187963854956e-08,
      "loss": 0.6706,
      "step": 494
    },
    {
      "epoch": 0.09862031179957165,
      "grad_norm": 0.2692635953426361,
      "learning_rate": 2.6770626181715773e-08,
      "loss": 0.6816,
      "step": 495
    },
    {
      "epoch": 0.09881954475270209,
      "grad_norm": 0.2632291615009308,
      "learning_rate": 1.7133751222137007e-08,
      "loss": 0.6855,
      "step": 496
    },
    {
      "epoch": 0.09901877770583255,
      "grad_norm": 0.27418622374534607,
      "learning_rate": 9.637975896759077e-09,
      "loss": 0.7329,
      "step": 497
    },
    {
      "epoch": 0.09921801065896299,
      "grad_norm": 0.2903919816017151,
      "learning_rate": 4.2836212996499865e-09,
      "loss": 0.7565,
      "step": 498
    },
    {
      "epoch": 0.09941724361209343,
      "grad_norm": 0.3054603636264801,
      "learning_rate": 1.0709167935385455e-09,
      "loss": 0.7183,
      "step": 499
    },
    {
      "epoch": 0.09961647656522389,
      "grad_norm": 0.30139267444610596,
      "learning_rate": 0.0,
      "loss": 0.6969,
      "step": 500
    },
    {
      "epoch": 0.09961647656522389,
      "eval_loss": 0.6748951077461243,
      "eval_runtime": 265.2168,
      "eval_samples_per_second": 31.876,
      "eval_steps_per_second": 7.971,
      "step": 500
    }
  ],
  "logging_steps": 1,
  "max_steps": 500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.6627926527115264e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}