{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 12.0,
  "eval_steps": 500,
  "global_step": 609492,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001968852749502865,
      "grad_norm": 1.9081757068634033,
      "learning_rate": 9.999999655665828e-07,
      "loss": 1.9557,
      "num_input_tokens_seen": 102400,
      "step": 100
    },
    {
      "epoch": 0.00393770549900573,
      "grad_norm": 2.0245535373687744,
      "learning_rate": 9.999998465374926e-07,
      "loss": 2.0835,
      "num_input_tokens_seen": 202416,
      "step": 200
    },
    {
      "epoch": 0.005906558248508594,
      "grad_norm": 1.8482612371444702,
      "learning_rate": 9.999996424876457e-07,
      "loss": 1.9022,
      "num_input_tokens_seen": 304816,
      "step": 300
    },
    {
      "epoch": 0.00787541099801146,
      "grad_norm": 1.8051942586898804,
      "learning_rate": 9.99999353417077e-07,
      "loss": 1.9903,
      "num_input_tokens_seen": 407216,
      "step": 400
    },
    {
      "epoch": 0.009844263747514324,
      "grad_norm": 2.291351795196533,
      "learning_rate": 9.999989793258351e-07,
      "loss": 1.9607,
      "num_input_tokens_seen": 509368,
      "step": 500
    },
    {
      "epoch": 0.011813116497017188,
      "grad_norm": 1.9627857208251953,
      "learning_rate": 9.999985202139843e-07,
      "loss": 1.9747,
      "num_input_tokens_seen": 611152,
      "step": 600
    },
    {
      "epoch": 0.013781969246520052,
      "grad_norm": 1.911093831062317,
      "learning_rate": 9.99997976081602e-07,
      "loss": 1.9834,
      "num_input_tokens_seen": 712304,
      "step": 700
    },
    {
      "epoch": 0.01575082199602292,
      "grad_norm": 1.9051035642623901,
      "learning_rate": 9.999973469287812e-07,
      "loss": 1.9282,
      "num_input_tokens_seen": 814040,
      "step": 800
    },
    {
      "epoch": 0.01771967474552578,
      "grad_norm": 1.8340057134628296,
      "learning_rate": 9.999966327556286e-07,
      "loss": 1.9803,
      "num_input_tokens_seen": 915376,
      "step": 900
    },
    {
      "epoch": 0.019688527495028647,
      "grad_norm": 2.0401248931884766,
      "learning_rate": 9.99995833562266e-07,
      "loss": 2.0006,
      "num_input_tokens_seen": 1017776,
      "step": 1000
    },
    {
      "epoch": 0.02165738024453151,
      "grad_norm": 2.080766201019287,
      "learning_rate": 9.999949493488289e-07,
      "loss": 2.0001,
      "num_input_tokens_seen": 1120176,
      "step": 1100
    },
    {
      "epoch": 0.023626232994034376,
      "grad_norm": 1.8966131210327148,
      "learning_rate": 9.999939801154677e-07,
      "loss": 2.0123,
      "num_input_tokens_seen": 1221976,
      "step": 1200
    },
    {
      "epoch": 0.025595085743537242,
      "grad_norm": 2.087946653366089,
      "learning_rate": 9.999929258623475e-07,
      "loss": 1.9841,
      "num_input_tokens_seen": 1324376,
      "step": 1300
    },
    {
      "epoch": 0.027563938493040105,
      "grad_norm": 4.0202460289001465,
      "learning_rate": 9.999917865896474e-07,
      "loss": 1.9815,
      "num_input_tokens_seen": 1425480,
      "step": 1400
    },
    {
      "epoch": 0.02953279124254297,
      "grad_norm": 1.7611056566238403,
      "learning_rate": 9.99990562297561e-07,
      "loss": 2.0264,
      "num_input_tokens_seen": 1525512,
      "step": 1500
    },
    {
      "epoch": 0.03150164399204584,
      "grad_norm": 2.0431771278381348,
      "learning_rate": 9.999892529862967e-07,
      "loss": 2.0026,
      "num_input_tokens_seen": 1627912,
      "step": 1600
    },
    {
      "epoch": 0.0334704967415487,
      "grad_norm": 1.9018654823303223,
      "learning_rate": 9.999878586560772e-07,
      "loss": 2.0291,
      "num_input_tokens_seen": 1728800,
      "step": 1700
    },
    {
      "epoch": 0.03543934949105156,
      "grad_norm": 1.7411730289459229,
      "learning_rate": 9.999863793071392e-07,
      "loss": 2.0356,
      "num_input_tokens_seen": 1828824,
      "step": 1800
    },
    {
      "epoch": 0.03740820224055443,
      "grad_norm": 1.8061546087265015,
      "learning_rate": 9.999848149397345e-07,
      "loss": 1.9708,
      "num_input_tokens_seen": 1929944,
      "step": 1900
    },
    {
      "epoch": 0.039377054990057295,
      "grad_norm": 1.8095059394836426,
      "learning_rate": 9.999831655541291e-07,
      "loss": 1.969,
      "num_input_tokens_seen": 2032256,
      "step": 2000
    },
    {
      "epoch": 0.04134590773956016,
      "grad_norm": 1.8762112855911255,
      "learning_rate": 9.999814311506033e-07,
      "loss": 1.9973,
      "num_input_tokens_seen": 2134656,
      "step": 2100
    },
    {
      "epoch": 0.04331476048906302,
      "grad_norm": 1.7770034074783325,
      "learning_rate": 9.999796117294525e-07,
      "loss": 1.9728,
      "num_input_tokens_seen": 2237056,
      "step": 2200
    },
    {
      "epoch": 0.04528361323856589,
      "grad_norm": 8.828059196472168,
      "learning_rate": 9.999777072909855e-07,
      "loss": 1.9541,
      "num_input_tokens_seen": 2338688,
      "step": 2300
    },
    {
      "epoch": 0.04725246598806875,
      "grad_norm": 1.8785701990127563,
      "learning_rate": 9.999757178355265e-07,
      "loss": 1.9632,
      "num_input_tokens_seen": 2441088,
      "step": 2400
    },
    {
      "epoch": 0.049221318737571615,
      "grad_norm": 1.9118993282318115,
      "learning_rate": 9.999736433634135e-07,
      "loss": 2.0193,
      "num_input_tokens_seen": 2543488,
      "step": 2500
    },
    {
      "epoch": 0.051190171487074485,
      "grad_norm": 2.069748878479004,
      "learning_rate": 9.999714838749996e-07,
      "loss": 1.9967,
      "num_input_tokens_seen": 2645664,
      "step": 2600
    },
    {
      "epoch": 0.05315902423657735,
      "grad_norm": 1.9387198686599731,
      "learning_rate": 9.999692393706517e-07,
      "loss": 1.9902,
      "num_input_tokens_seen": 2748064,
      "step": 2700
    },
    {
      "epoch": 0.05512787698608021,
      "grad_norm": 1.8751716613769531,
      "learning_rate": 9.999669098507517e-07,
      "loss": 1.9793,
      "num_input_tokens_seen": 2850464,
      "step": 2800
    },
    {
      "epoch": 0.05709672973558307,
      "grad_norm": 1.972404956817627,
      "learning_rate": 9.999644953156955e-07,
      "loss": 2.0275,
      "num_input_tokens_seen": 2952040,
      "step": 2900
    },
    {
      "epoch": 0.05906558248508594,
      "grad_norm": 1.9006932973861694,
      "learning_rate": 9.999619957658938e-07,
      "loss": 1.9634,
      "num_input_tokens_seen": 3054440,
      "step": 3000
    },
    {
      "epoch": 0.061034435234588805,
      "grad_norm": 2.1948349475860596,
      "learning_rate": 9.999594112017717e-07,
      "loss": 2.0436,
      "num_input_tokens_seen": 3155968,
      "step": 3100
    },
    {
      "epoch": 0.06300328798409167,
      "grad_norm": 1.8219056129455566,
      "learning_rate": 9.999567416237685e-07,
      "loss": 2.0317,
      "num_input_tokens_seen": 3256848,
      "step": 3200
    },
    {
      "epoch": 0.06497214073359453,
      "grad_norm": 2.4099738597869873,
      "learning_rate": 9.99953987032338e-07,
      "loss": 1.9925,
      "num_input_tokens_seen": 3359048,
      "step": 3300
    },
    {
      "epoch": 0.0669409934830974,
      "grad_norm": 2.0764834880828857,
      "learning_rate": 9.99951147427949e-07,
      "loss": 2.0083,
      "num_input_tokens_seen": 3459880,
      "step": 3400
    },
    {
      "epoch": 0.06890984623260027,
      "grad_norm": 1.985681176185608,
      "learning_rate": 9.999482228110844e-07,
      "loss": 1.9774,
      "num_input_tokens_seen": 3561432,
      "step": 3500
    },
    {
      "epoch": 0.07087869898210312,
      "grad_norm": 2.1401822566986084,
      "learning_rate": 9.99945213182241e-07,
      "loss": 2.0003,
      "num_input_tokens_seen": 3663832,
      "step": 3600
    },
    {
      "epoch": 0.072847551731606,
      "grad_norm": 1.6803908348083496,
      "learning_rate": 9.999421185419309e-07,
      "loss": 2.0152,
      "num_input_tokens_seen": 3765496,
      "step": 3700
    },
    {
      "epoch": 0.07481640448110886,
      "grad_norm": 1.7712147235870361,
      "learning_rate": 9.9993893889068e-07,
      "loss": 1.9871,
      "num_input_tokens_seen": 3866968,
      "step": 3800
    },
    {
      "epoch": 0.07678525723061172,
      "grad_norm": 1.8583042621612549,
      "learning_rate": 9.999356742290296e-07,
      "loss": 1.9997,
      "num_input_tokens_seen": 3968728,
      "step": 3900
    },
    {
      "epoch": 0.07875410998011459,
      "grad_norm": 1.827921986579895,
      "learning_rate": 9.999323245575343e-07,
      "loss": 1.9609,
      "num_input_tokens_seen": 4071128,
      "step": 4000
    },
    {
      "epoch": 0.08072296272961745,
      "grad_norm": 1.7819452285766602,
      "learning_rate": 9.999288898767638e-07,
      "loss": 2.0073,
      "num_input_tokens_seen": 4172968,
      "step": 4100
    },
    {
      "epoch": 0.08269181547912031,
      "grad_norm": 2.0420584678649902,
      "learning_rate": 9.999253701873024e-07,
      "loss": 1.9616,
      "num_input_tokens_seen": 4275368,
      "step": 4200
    },
    {
      "epoch": 0.08466066822862318,
      "grad_norm": 1.8458151817321777,
      "learning_rate": 9.999217654897481e-07,
      "loss": 1.976,
      "num_input_tokens_seen": 4377768,
      "step": 4300
    },
    {
      "epoch": 0.08662952097812604,
      "grad_norm": 1.8586608171463013,
      "learning_rate": 9.999180757847142e-07,
      "loss": 1.9941,
      "num_input_tokens_seen": 4478712,
      "step": 4400
    },
    {
      "epoch": 0.08859837372762891,
      "grad_norm": 2.039126396179199,
      "learning_rate": 9.999143010728277e-07,
      "loss": 1.981,
      "num_input_tokens_seen": 4581112,
      "step": 4500
    },
    {
      "epoch": 0.09056722647713178,
      "grad_norm": 1.9401798248291016,
      "learning_rate": 9.999104413547313e-07,
      "loss": 1.9281,
      "num_input_tokens_seen": 4683512,
      "step": 4600
    },
    {
      "epoch": 0.09253607922663463,
      "grad_norm": 1.9561400413513184,
      "learning_rate": 9.999064966310804e-07,
      "loss": 2.0088,
      "num_input_tokens_seen": 4785088,
      "step": 4700
    },
    {
      "epoch": 0.0945049319761375,
      "grad_norm": 2.128077268600464,
      "learning_rate": 9.999024669025464e-07,
      "loss": 2.0149,
      "num_input_tokens_seen": 4886864,
      "step": 4800
    },
    {
      "epoch": 0.09647378472564037,
      "grad_norm": 2.0094833374023438,
      "learning_rate": 9.998983521698141e-07,
      "loss": 2.034,
      "num_input_tokens_seen": 4989264,
      "step": 4900
    },
    {
      "epoch": 0.09844263747514323,
      "grad_norm": 1.8435900211334229,
      "learning_rate": 9.998941524335834e-07,
      "loss": 1.9854,
      "num_input_tokens_seen": 5091664,
      "step": 5000
    },
    {
      "epoch": 0.1004114902246461,
      "grad_norm": 1.8683404922485352,
      "learning_rate": 9.998898676945684e-07,
      "loss": 1.9672,
      "num_input_tokens_seen": 5192240,
      "step": 5100
    },
    {
      "epoch": 0.10238034297414897,
      "grad_norm": 1.8303996324539185,
      "learning_rate": 9.998854979534977e-07,
      "loss": 1.9863,
      "num_input_tokens_seen": 5293912,
      "step": 5200
    },
    {
      "epoch": 0.10434919572365182,
      "grad_norm": 1.9261983633041382,
      "learning_rate": 9.998810432111144e-07,
      "loss": 1.9907,
      "num_input_tokens_seen": 5395552,
      "step": 5300
    },
    {
      "epoch": 0.1063180484731547,
      "grad_norm": 1.674626350402832,
      "learning_rate": 9.998765034681758e-07,
      "loss": 2.031,
      "num_input_tokens_seen": 5496816,
      "step": 5400
    },
    {
      "epoch": 0.10828690122265756,
      "grad_norm": 1.9371134042739868,
      "learning_rate": 9.998718787254539e-07,
      "loss": 1.9305,
      "num_input_tokens_seen": 5598768,
      "step": 5500
    },
    {
      "epoch": 0.11025575397216042,
      "grad_norm": 5.228818893432617,
      "learning_rate": 9.99867168983735e-07,
      "loss": 1.9867,
      "num_input_tokens_seen": 5701168,
      "step": 5600
    },
    {
      "epoch": 0.11222460672166329,
      "grad_norm": 1.9057998657226562,
      "learning_rate": 9.998623742438202e-07,
      "loss": 1.9319,
      "num_input_tokens_seen": 5803568,
      "step": 5700
    },
    {
      "epoch": 0.11419345947116614,
      "grad_norm": 2.1413800716400146,
      "learning_rate": 9.998574945065248e-07,
      "loss": 1.9946,
      "num_input_tokens_seen": 5905968,
      "step": 5800
    },
    {
      "epoch": 0.11616231222066901,
      "grad_norm": 2.0873007774353027,
      "learning_rate": 9.998525297726783e-07,
      "loss": 1.938,
      "num_input_tokens_seen": 6008368,
      "step": 5900
    },
    {
      "epoch": 0.11813116497017188,
      "grad_norm": 1.8770489692687988,
      "learning_rate": 9.998474800431252e-07,
      "loss": 2.0306,
      "num_input_tokens_seen": 6109176,
      "step": 6000
    },
    {
      "epoch": 0.12010001771967474,
      "grad_norm": 2.004056692123413,
      "learning_rate": 9.99842345318724e-07,
      "loss": 1.9941,
      "num_input_tokens_seen": 6210776,
      "step": 6100
    },
    {
      "epoch": 0.12206887046917761,
      "grad_norm": 2.201301336288452,
      "learning_rate": 9.998371256003478e-07,
      "loss": 1.9467,
      "num_input_tokens_seen": 6312432,
      "step": 6200
    },
    {
      "epoch": 0.12403772321868048,
      "grad_norm": 1.8141396045684814,
      "learning_rate": 9.998318208888844e-07,
      "loss": 2.0164,
      "num_input_tokens_seen": 6413424,
      "step": 6300
    },
    {
      "epoch": 0.12600657596818335,
      "grad_norm": 1.8870656490325928,
      "learning_rate": 9.998264311852353e-07,
      "loss": 1.9963,
      "num_input_tokens_seen": 6515200,
      "step": 6400
    },
    {
      "epoch": 0.1279754287176862,
      "grad_norm": 2.036982774734497,
      "learning_rate": 9.998209564903175e-07,
      "loss": 1.9512,
      "num_input_tokens_seen": 6616832,
      "step": 6500
    },
    {
      "epoch": 0.12994428146718906,
      "grad_norm": 2.0326342582702637,
      "learning_rate": 9.998153968050617e-07,
      "loss": 2.0194,
      "num_input_tokens_seen": 6719232,
      "step": 6600
    },
    {
      "epoch": 0.13191313421669193,
      "grad_norm": 1.8858219385147095,
      "learning_rate": 9.998097521304132e-07,
      "loss": 1.9661,
      "num_input_tokens_seen": 6821632,
      "step": 6700
    },
    {
      "epoch": 0.1338819869661948,
      "grad_norm": 2.1061222553253174,
      "learning_rate": 9.998040224673321e-07,
      "loss": 2.0127,
      "num_input_tokens_seen": 6924032,
      "step": 6800
    },
    {
      "epoch": 0.13585083971569767,
      "grad_norm": 2.0696656703948975,
      "learning_rate": 9.997982078167925e-07,
      "loss": 1.9279,
      "num_input_tokens_seen": 7025856,
      "step": 6900
    },
    {
      "epoch": 0.13781969246520054,
      "grad_norm": 1.8722087144851685,
      "learning_rate": 9.997923081797832e-07,
      "loss": 2.0707,
      "num_input_tokens_seen": 7127656,
      "step": 7000
    },
    {
      "epoch": 0.13978854521470338,
      "grad_norm": 1.7816176414489746,
      "learning_rate": 9.997863235573072e-07,
      "loss": 1.9551,
      "num_input_tokens_seen": 7230056,
      "step": 7100
    },
    {
      "epoch": 0.14175739796420625,
      "grad_norm": 8.757242202758789,
      "learning_rate": 9.997802539503824e-07,
      "loss": 1.9781,
      "num_input_tokens_seen": 7331696,
      "step": 7200
    },
    {
      "epoch": 0.14372625071370912,
      "grad_norm": 2.2203147411346436,
      "learning_rate": 9.997740993600406e-07,
      "loss": 2.0103,
      "num_input_tokens_seen": 7433472,
      "step": 7300
    },
    {
      "epoch": 0.145695103463212,
      "grad_norm": 1.7229341268539429,
      "learning_rate": 9.997678597873285e-07,
      "loss": 2.0444,
      "num_input_tokens_seen": 7534736,
      "step": 7400
    },
    {
      "epoch": 0.14766395621271486,
      "grad_norm": 2.069223165512085,
      "learning_rate": 9.997615352333072e-07,
      "loss": 1.9925,
      "num_input_tokens_seen": 7637136,
      "step": 7500
    },
    {
      "epoch": 0.14963280896221773,
      "grad_norm": 2.1618645191192627,
      "learning_rate": 9.99755125699052e-07,
      "loss": 1.9466,
      "num_input_tokens_seen": 7739536,
      "step": 7600
    },
    {
      "epoch": 0.15160166171172057,
      "grad_norm": 1.8824418783187866,
      "learning_rate": 9.99748631185653e-07,
      "loss": 1.9693,
      "num_input_tokens_seen": 7841640,
      "step": 7700
    },
    {
      "epoch": 0.15357051446122344,
      "grad_norm": 1.7198357582092285,
      "learning_rate": 9.99742051694214e-07,
      "loss": 2.0451,
      "num_input_tokens_seen": 7943032,
      "step": 7800
    },
    {
      "epoch": 0.1555393672107263,
      "grad_norm": 1.7849814891815186,
      "learning_rate": 9.997353872258542e-07,
      "loss": 1.9811,
      "num_input_tokens_seen": 8044224,
      "step": 7900
    },
    {
      "epoch": 0.15750821996022918,
      "grad_norm": 1.9055421352386475,
      "learning_rate": 9.997286377817067e-07,
      "loss": 2.0199,
      "num_input_tokens_seen": 8145968,
      "step": 8000
    },
    {
      "epoch": 0.15947707270973205,
      "grad_norm": 2.1569740772247314,
      "learning_rate": 9.997218033629195e-07,
      "loss": 2.0,
      "num_input_tokens_seen": 8247792,
      "step": 8100
    },
    {
      "epoch": 0.1614459254592349,
      "grad_norm": 2.0451645851135254,
      "learning_rate": 9.997148839706545e-07,
      "loss": 1.9879,
      "num_input_tokens_seen": 8349264,
      "step": 8200
    },
    {
      "epoch": 0.16341477820873776,
      "grad_norm": 1.8271421194076538,
      "learning_rate": 9.99707879606088e-07,
      "loss": 1.9729,
      "num_input_tokens_seen": 8451152,
      "step": 8300
    },
    {
      "epoch": 0.16538363095824063,
      "grad_norm": 1.8655251264572144,
      "learning_rate": 9.997007902704115e-07,
      "loss": 2.0167,
      "num_input_tokens_seen": 8553040,
      "step": 8400
    },
    {
      "epoch": 0.1673524837077435,
      "grad_norm": 1.8250237703323364,
      "learning_rate": 9.996936159648302e-07,
      "loss": 2.0309,
      "num_input_tokens_seen": 8655440,
      "step": 8500
    },
    {
      "epoch": 0.16932133645724637,
      "grad_norm": 2.0580437183380127,
      "learning_rate": 9.99686356690564e-07,
      "loss": 1.9358,
      "num_input_tokens_seen": 8757840,
      "step": 8600
    },
    {
      "epoch": 0.17129018920674924,
      "grad_norm": 2.1230578422546387,
      "learning_rate": 9.996790124488477e-07,
      "loss": 1.9934,
      "num_input_tokens_seen": 8858664,
      "step": 8700
    },
    {
      "epoch": 0.17325904195625208,
      "grad_norm": 1.9461112022399902,
      "learning_rate": 9.996715832409295e-07,
      "loss": 1.9581,
      "num_input_tokens_seen": 8961064,
      "step": 8800
    },
    {
      "epoch": 0.17522789470575495,
      "grad_norm": 1.7925480604171753,
      "learning_rate": 9.996640690680732e-07,
      "loss": 1.9722,
      "num_input_tokens_seen": 9062776,
      "step": 8900
    },
    {
      "epoch": 0.17719674745525782,
      "grad_norm": 1.9623992443084717,
      "learning_rate": 9.996564699315561e-07,
      "loss": 1.9504,
      "num_input_tokens_seen": 9165176,
      "step": 9000
    },
    {
      "epoch": 0.1791656002047607,
      "grad_norm": 1.958788514137268,
      "learning_rate": 9.996487858326708e-07,
      "loss": 1.9981,
      "num_input_tokens_seen": 9266312,
      "step": 9100
    },
    {
      "epoch": 0.18113445295426356,
      "grad_norm": 1.8302428722381592,
      "learning_rate": 9.996410167727236e-07,
      "loss": 1.9342,
      "num_input_tokens_seen": 9367936,
      "step": 9200
    },
    {
      "epoch": 0.18310330570376643,
      "grad_norm": 1.993533730506897,
      "learning_rate": 9.99633162753036e-07,
      "loss": 1.9561,
      "num_input_tokens_seen": 9470336,
      "step": 9300
    },
    {
      "epoch": 0.18507215845326927,
      "grad_norm": 2.04597544670105,
      "learning_rate": 9.996252237749427e-07,
      "loss": 1.9991,
      "num_input_tokens_seen": 9572736,
      "step": 9400
    },
    {
      "epoch": 0.18704101120277214,
      "grad_norm": 1.8397719860076904,
      "learning_rate": 9.996171998397944e-07,
      "loss": 1.9686,
      "num_input_tokens_seen": 9675136,
      "step": 9500
    },
    {
      "epoch": 0.189009863952275,
      "grad_norm": 1.842245101928711,
      "learning_rate": 9.996090909489552e-07,
      "loss": 1.97,
      "num_input_tokens_seen": 9777536,
      "step": 9600
    },
    {
      "epoch": 0.19097871670177788,
      "grad_norm": 2.032658576965332,
      "learning_rate": 9.99600897103804e-07,
      "loss": 1.9829,
      "num_input_tokens_seen": 9879392,
      "step": 9700
    },
    {
      "epoch": 0.19294756945128075,
      "grad_norm": 1.9740447998046875,
      "learning_rate": 9.995926183057338e-07,
      "loss": 1.9965,
      "num_input_tokens_seen": 9980648,
      "step": 9800
    },
    {
      "epoch": 0.1949164222007836,
      "grad_norm": 1.9238474369049072,
      "learning_rate": 9.995842545561527e-07,
      "loss": 1.981,
      "num_input_tokens_seen": 10083048,
      "step": 9900
    },
    {
      "epoch": 0.19688527495028646,
      "grad_norm": 2.00785493850708,
      "learning_rate": 9.995758058564832e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 10184864,
      "step": 10000
    },
    {
      "epoch": 0.19885412769978933,
      "grad_norm": 2.054298162460327,
      "learning_rate": 9.99567272208161e-07,
      "loss": 1.9848,
      "num_input_tokens_seen": 10287264,
      "step": 10100
    },
    {
      "epoch": 0.2008229804492922,
      "grad_norm": 1.957642912864685,
      "learning_rate": 9.995586536126381e-07,
      "loss": 1.9503,
      "num_input_tokens_seen": 10389664,
      "step": 10200
    },
    {
      "epoch": 0.20279183319879507,
      "grad_norm": 1.7643107175827026,
      "learning_rate": 9.995499500713795e-07,
      "loss": 1.9629,
      "num_input_tokens_seen": 10491440,
      "step": 10300
    },
    {
      "epoch": 0.20476068594829794,
      "grad_norm": 3.147120714187622,
      "learning_rate": 9.995411615858654e-07,
      "loss": 1.9909,
      "num_input_tokens_seen": 10593840,
      "step": 10400
    },
    {
      "epoch": 0.20672953869780078,
      "grad_norm": 1.8852829933166504,
      "learning_rate": 9.995322881575898e-07,
      "loss": 2.0632,
      "num_input_tokens_seen": 10694928,
      "step": 10500
    },
    {
      "epoch": 0.20869839144730365,
      "grad_norm": 1.8616101741790771,
      "learning_rate": 9.99523329788062e-07,
      "loss": 2.0156,
      "num_input_tokens_seen": 10796552,
      "step": 10600
    },
    {
      "epoch": 0.21066724419680652,
      "grad_norm": 1.8704726696014404,
      "learning_rate": 9.995142864788051e-07,
      "loss": 1.9922,
      "num_input_tokens_seen": 10898160,
      "step": 10700
    },
    {
      "epoch": 0.2126360969463094,
      "grad_norm": 2.0497450828552246,
      "learning_rate": 9.99505158231357e-07,
      "loss": 1.9887,
      "num_input_tokens_seen": 11000560,
      "step": 10800
    },
    {
      "epoch": 0.21460494969581226,
      "grad_norm": 1.9596359729766846,
      "learning_rate": 9.994959450472697e-07,
      "loss": 1.9892,
      "num_input_tokens_seen": 11101736,
      "step": 10900
    },
    {
      "epoch": 0.21657380244531513,
      "grad_norm": 1.9173911809921265,
      "learning_rate": 9.9948664692811e-07,
      "loss": 1.999,
      "num_input_tokens_seen": 11203232,
      "step": 11000
    },
    {
      "epoch": 0.21854265519481797,
      "grad_norm": 1.9784644842147827,
      "learning_rate": 9.994772638754587e-07,
      "loss": 2.0029,
      "num_input_tokens_seen": 11305632,
      "step": 11100
    },
    {
      "epoch": 0.22051150794432084,
      "grad_norm": 2.05433988571167,
      "learning_rate": 9.994677958909117e-07,
      "loss": 1.9623,
      "num_input_tokens_seen": 11408032,
      "step": 11200
    },
    {
      "epoch": 0.2224803606938237,
      "grad_norm": 2.146061658859253,
      "learning_rate": 9.994582429760785e-07,
      "loss": 1.9939,
      "num_input_tokens_seen": 11510016,
      "step": 11300
    },
    {
      "epoch": 0.22444921344332658,
      "grad_norm": 1.7213598489761353,
      "learning_rate": 9.994486051325837e-07,
      "loss": 1.9569,
      "num_input_tokens_seen": 11611544,
      "step": 11400
    },
    {
      "epoch": 0.22641806619282945,
      "grad_norm": 1.723879337310791,
      "learning_rate": 9.99438882362066e-07,
      "loss": 2.0075,
      "num_input_tokens_seen": 11713528,
      "step": 11500
    },
    {
      "epoch": 0.2283869189423323,
      "grad_norm": 2.2458531856536865,
      "learning_rate": 9.99429074666179e-07,
      "loss": 1.9548,
      "num_input_tokens_seen": 11814312,
      "step": 11600
    },
    {
      "epoch": 0.23035577169183516,
      "grad_norm": 2.29547381401062,
      "learning_rate": 9.994191820465902e-07,
      "loss": 2.0178,
      "num_input_tokens_seen": 11916712,
      "step": 11700
    },
    {
      "epoch": 0.23232462444133803,
      "grad_norm": 1.716240644454956,
      "learning_rate": 9.99409204504982e-07,
      "loss": 1.9757,
      "num_input_tokens_seen": 12019112,
      "step": 11800
    },
    {
      "epoch": 0.2342934771908409,
      "grad_norm": 1.7362383604049683,
      "learning_rate": 9.993991420430506e-07,
      "loss": 1.9657,
      "num_input_tokens_seen": 12120656,
      "step": 11900
    },
    {
      "epoch": 0.23626232994034377,
      "grad_norm": 1.9518760442733765,
      "learning_rate": 9.993889946625073e-07,
      "loss": 1.9828,
      "num_input_tokens_seen": 12223056,
      "step": 12000
    },
    {
      "epoch": 0.23823118268984664,
      "grad_norm": 1.9001251459121704,
      "learning_rate": 9.993787623650773e-07,
      "loss": 1.9723,
      "num_input_tokens_seen": 12324856,
      "step": 12100
    },
    {
      "epoch": 0.24020003543934948,
      "grad_norm": 2.2148642539978027,
      "learning_rate": 9.993684451525008e-07,
      "loss": 2.0026,
      "num_input_tokens_seen": 12427256,
      "step": 12200
    },
    {
      "epoch": 0.24216888818885235,
      "grad_norm": 2.0185770988464355,
      "learning_rate": 9.993580430265321e-07,
      "loss": 1.9573,
      "num_input_tokens_seen": 12529656,
      "step": 12300
    },
    {
      "epoch": 0.24413774093835522,
      "grad_norm": 1.9938510656356812,
      "learning_rate": 9.993475559889401e-07,
      "loss": 1.9974,
      "num_input_tokens_seen": 12631136,
      "step": 12400
    },
    {
      "epoch": 0.2461065936878581,
      "grad_norm": 1.89698326587677,
      "learning_rate": 9.993369840415076e-07,
      "loss": 1.9587,
      "num_input_tokens_seen": 12732168,
      "step": 12500
    },
    {
      "epoch": 0.24807544643736096,
      "grad_norm": 2.3082284927368164,
      "learning_rate": 9.993263271860328e-07,
      "loss": 1.9721,
      "num_input_tokens_seen": 12833992,
      "step": 12600
    },
    {
      "epoch": 0.2500442991868638,
      "grad_norm": 2.106632947921753,
      "learning_rate": 9.993155854243275e-07,
      "loss": 2.0185,
      "num_input_tokens_seen": 12934888,
      "step": 12700
    },
    {
      "epoch": 0.2520131519363667,
      "grad_norm": 2.0504987239837646,
      "learning_rate": 9.993047587582184e-07,
      "loss": 1.9909,
      "num_input_tokens_seen": 13035984,
      "step": 12800
    },
    {
      "epoch": 0.25398200468586957,
      "grad_norm": 2.0314905643463135,
      "learning_rate": 9.992938471895462e-07,
      "loss": 2.1097,
      "num_input_tokens_seen": 13137528,
      "step": 12900
    },
    {
      "epoch": 0.2559508574353724,
      "grad_norm": 2.0496342182159424,
      "learning_rate": 9.992828507201666e-07,
      "loss": 1.9904,
      "num_input_tokens_seen": 13239928,
      "step": 13000
    },
    {
      "epoch": 0.25791971018487525,
      "grad_norm": 1.7611814737319946,
      "learning_rate": 9.992717693519497e-07,
      "loss": 1.9373,
      "num_input_tokens_seen": 13342328,
      "step": 13100
    },
    {
      "epoch": 0.2598885629343781,
      "grad_norm": 1.7378196716308594,
      "learning_rate": 9.992606030867792e-07,
      "loss": 1.974,
      "num_input_tokens_seen": 13443976,
      "step": 13200
    },
    {
      "epoch": 0.261857415683881,
      "grad_norm": 1.9088518619537354,
      "learning_rate": 9.99249351926554e-07,
      "loss": 1.9461,
      "num_input_tokens_seen": 13546376,
      "step": 13300
    },
    {
      "epoch": 0.26382626843338386,
      "grad_norm": 1.9952328205108643,
      "learning_rate": 9.992380158731875e-07,
      "loss": 1.9667,
      "num_input_tokens_seen": 13648776,
      "step": 13400
    },
    {
      "epoch": 0.26579512118288673,
      "grad_norm": 2.2141008377075195,
      "learning_rate": 9.99226594928607e-07,
      "loss": 1.9328,
      "num_input_tokens_seen": 13751176,
      "step": 13500
    },
    {
      "epoch": 0.2677639739323896,
      "grad_norm": 2.5217039585113525,
      "learning_rate": 9.992150890947551e-07,
      "loss": 1.9817,
      "num_input_tokens_seen": 13853576,
      "step": 13600
    },
    {
      "epoch": 0.26973282668189247,
      "grad_norm": 1.8611009120941162,
      "learning_rate": 9.992034983735876e-07,
      "loss": 1.9639,
      "num_input_tokens_seen": 13955480,
      "step": 13700
    },
    {
      "epoch": 0.27170167943139534,
      "grad_norm": 2.1035237312316895,
      "learning_rate": 9.991918227670757e-07,
      "loss": 1.9584,
      "num_input_tokens_seen": 14057880,
      "step": 13800
    },
    {
      "epoch": 0.2736705321808982,
      "grad_norm": 1.857049822807312,
      "learning_rate": 9.991800622772047e-07,
      "loss": 1.9333,
      "num_input_tokens_seen": 14160280,
      "step": 13900
    },
    {
      "epoch": 0.2756393849304011,
      "grad_norm": 1.952817678451538,
      "learning_rate": 9.991682169059744e-07,
      "loss": 1.9996,
      "num_input_tokens_seen": 14262680,
      "step": 14000
    },
    {
      "epoch": 0.27760823767990395,
      "grad_norm": 1.6179795265197754,
      "learning_rate": 9.99156286655399e-07,
      "loss": 1.9756,
      "num_input_tokens_seen": 14364768,
      "step": 14100
    },
    {
      "epoch": 0.27957709042940676,
      "grad_norm": 2.3380112648010254,
      "learning_rate": 9.991442715275072e-07,
      "loss": 1.9927,
      "num_input_tokens_seen": 14467168,
      "step": 14200
    },
    {
      "epoch": 0.28154594317890963,
      "grad_norm": 1.9956783056259155,
      "learning_rate": 9.99132171524342e-07,
      "loss": 2.0169,
      "num_input_tokens_seen": 14569568,
      "step": 14300
    },
    {
      "epoch": 0.2835147959284125,
      "grad_norm": 2.460737466812134,
      "learning_rate": 9.991199866479607e-07,
      "loss": 1.9812,
      "num_input_tokens_seen": 14671968,
      "step": 14400
    },
    {
      "epoch": 0.28548364867791537,
      "grad_norm": 1.8070951700210571,
      "learning_rate": 9.991077169004355e-07,
      "loss": 1.9598,
      "num_input_tokens_seen": 14773656,
      "step": 14500
    },
    {
      "epoch": 0.28745250142741824,
      "grad_norm": 2.0932130813598633,
      "learning_rate": 9.99095362283853e-07,
      "loss": 1.9954,
      "num_input_tokens_seen": 14875680,
      "step": 14600
    },
    {
      "epoch": 0.2894213541769211,
      "grad_norm": 1.738515853881836,
      "learning_rate": 9.990829228003136e-07,
      "loss": 1.9532,
      "num_input_tokens_seen": 14978080,
      "step": 14700
    },
    {
      "epoch": 0.291390206926424,
      "grad_norm": 1.9239839315414429,
      "learning_rate": 9.990703984519324e-07,
      "loss": 2.0163,
      "num_input_tokens_seen": 15080480,
      "step": 14800
    },
    {
      "epoch": 0.29335905967592685,
      "grad_norm": 1.837448000907898,
      "learning_rate": 9.990577892408396e-07,
      "loss": 1.9416,
      "num_input_tokens_seen": 15182336,
      "step": 14900
    },
    {
      "epoch": 0.2953279124254297,
      "grad_norm": 1.738576889038086,
      "learning_rate": 9.990450951691786e-07,
      "loss": 2.0071,
      "num_input_tokens_seen": 15283712,
      "step": 15000
    },
    {
      "epoch": 0.2972967651749326,
      "grad_norm": 2.089157819747925,
      "learning_rate": 9.990323162391087e-07,
      "loss": 1.9592,
      "num_input_tokens_seen": 15385432,
      "step": 15100
    },
    {
      "epoch": 0.29926561792443546,
      "grad_norm": 1.9610596895217896,
      "learning_rate": 9.990194524528024e-07,
      "loss": 1.8992,
      "num_input_tokens_seen": 15487832,
      "step": 15200
    },
    {
      "epoch": 0.30123447067393827,
      "grad_norm": 1.824532151222229,
      "learning_rate": 9.99006503812447e-07,
      "loss": 1.9654,
      "num_input_tokens_seen": 15589736,
      "step": 15300
    },
    {
      "epoch": 0.30320332342344114,
      "grad_norm": 1.8911042213439941,
      "learning_rate": 9.989934703202444e-07,
      "loss": 1.9673,
      "num_input_tokens_seen": 15692136,
      "step": 15400
    },
    {
      "epoch": 0.305172176172944,
      "grad_norm": 2.0679547786712646,
      "learning_rate": 9.98980351978411e-07,
      "loss": 2.0,
      "num_input_tokens_seen": 15793344,
      "step": 15500
    },
    {
      "epoch": 0.3071410289224469,
      "grad_norm": 2.1341922283172607,
      "learning_rate": 9.989671487891773e-07,
      "loss": 2.0583,
      "num_input_tokens_seen": 15894016,
      "step": 15600
    },
    {
      "epoch": 0.30910988167194975,
      "grad_norm": 2.13081955909729,
      "learning_rate": 9.989538607547883e-07,
      "loss": 1.9946,
      "num_input_tokens_seen": 15994488,
      "step": 15700
    },
    {
      "epoch": 0.3110787344214526,
      "grad_norm": 1.6867297887802124,
      "learning_rate": 9.989404878775038e-07,
      "loss": 1.9935,
      "num_input_tokens_seen": 16096888,
      "step": 15800
    },
    {
      "epoch": 0.3130475871709555,
      "grad_norm": 2.1156623363494873,
      "learning_rate": 9.989270301595976e-07,
      "loss": 1.981,
      "num_input_tokens_seen": 16198688,
      "step": 15900
    },
    {
      "epoch": 0.31501643992045836,
      "grad_norm": 1.995002031326294,
      "learning_rate": 9.98913487603358e-07,
      "loss": 1.9702,
      "num_input_tokens_seen": 16301088,
      "step": 16000
    },
    {
      "epoch": 0.3169852926699612,
      "grad_norm": 3.548793077468872,
      "learning_rate": 9.98899860211088e-07,
      "loss": 1.9911,
      "num_input_tokens_seen": 16402144,
      "step": 16100
    },
    {
      "epoch": 0.3189541454194641,
      "grad_norm": 1.8882970809936523,
      "learning_rate": 9.988861479851045e-07,
      "loss": 1.9966,
      "num_input_tokens_seen": 16503728,
      "step": 16200
    },
    {
      "epoch": 0.32092299816896697,
      "grad_norm": 2.009093999862671,
      "learning_rate": 9.988723509277395e-07,
      "loss": 1.9795,
      "num_input_tokens_seen": 16606128,
      "step": 16300
    },
    {
      "epoch": 0.3228918509184698,
      "grad_norm": 1.9955816268920898,
      "learning_rate": 9.98858469041339e-07,
      "loss": 1.9247,
      "num_input_tokens_seen": 16708528,
      "step": 16400
    },
    {
      "epoch": 0.32486070366797265,
      "grad_norm": 2.0711095333099365,
      "learning_rate": 9.988445023282631e-07,
      "loss": 2.0131,
      "num_input_tokens_seen": 16810928,
      "step": 16500
    },
    {
      "epoch": 0.3268295564174755,
      "grad_norm": 1.8371134996414185,
      "learning_rate": 9.988304507908872e-07,
      "loss": 1.9906,
      "num_input_tokens_seen": 16912496,
      "step": 16600
    },
    {
      "epoch": 0.3287984091669784,
      "grad_norm": 1.6996644735336304,
      "learning_rate": 9.988163144316005e-07,
      "loss": 1.972,
      "num_input_tokens_seen": 17014288,
      "step": 16700
    },
    {
      "epoch": 0.33076726191648126,
      "grad_norm": 1.89493989944458,
      "learning_rate": 9.988020932528065e-07,
      "loss": 1.9729,
      "num_input_tokens_seen": 17115536,
      "step": 16800
    },
    {
      "epoch": 0.33273611466598413,
      "grad_norm": 2.2083075046539307,
      "learning_rate": 9.987877872569239e-07,
      "loss": 2.0161,
      "num_input_tokens_seen": 17216168,
      "step": 16900
    },
    {
      "epoch": 0.334704967415487,
      "grad_norm": 1.9821158647537231,
      "learning_rate": 9.98773396446385e-07,
      "loss": 1.9248,
      "num_input_tokens_seen": 17318072,
      "step": 17000
    },
    {
      "epoch": 0.33667382016498987,
      "grad_norm": 2.1424949169158936,
      "learning_rate": 9.987589208236368e-07,
      "loss": 2.002,
      "num_input_tokens_seen": 17420472,
      "step": 17100
    },
    {
      "epoch": 0.33864267291449274,
      "grad_norm": 1.6785787343978882,
      "learning_rate": 9.987443603911408e-07,
      "loss": 1.997,
      "num_input_tokens_seen": 17521296,
      "step": 17200
    },
    {
      "epoch": 0.3406115256639956,
      "grad_norm": 3.9235410690307617,
      "learning_rate": 9.98729715151373e-07,
      "loss": 1.9538,
      "num_input_tokens_seen": 17623696,
      "step": 17300
    },
    {
      "epoch": 0.3425803784134985,
      "grad_norm": 1.839826226234436,
      "learning_rate": 9.987149851068237e-07,
      "loss": 2.0244,
      "num_input_tokens_seen": 17725640,
      "step": 17400
    },
    {
      "epoch": 0.34454923116300135,
      "grad_norm": 1.8692381381988525,
      "learning_rate": 9.987001702599974e-07,
      "loss": 2.046,
      "num_input_tokens_seen": 17828040,
      "step": 17500
    },
    {
      "epoch": 0.34651808391250416,
      "grad_norm": 1.8370298147201538,
      "learning_rate": 9.986852706134134e-07,
      "loss": 1.9298,
      "num_input_tokens_seen": 17930440,
      "step": 17600
    },
    {
      "epoch": 0.34848693666200703,
      "grad_norm": 2.1793007850646973,
      "learning_rate": 9.986702861696053e-07,
      "loss": 2.0293,
      "num_input_tokens_seen": 18031296,
      "step": 17700
    },
    {
      "epoch": 0.3504557894115099,
      "grad_norm": 2.0108096599578857,
      "learning_rate": 9.986552169311211e-07,
      "loss": 1.9718,
      "num_input_tokens_seen": 18132232,
      "step": 17800
    },
    {
      "epoch": 0.35242464216101277,
      "grad_norm": 1.7110337018966675,
      "learning_rate": 9.98640062900523e-07,
      "loss": 2.0083,
      "num_input_tokens_seen": 18233696,
      "step": 17900
    },
    {
      "epoch": 0.35439349491051564,
      "grad_norm": 1.9672855138778687,
      "learning_rate": 9.986248240803878e-07,
      "loss": 1.9991,
      "num_input_tokens_seen": 18335176,
      "step": 18000
    },
    {
      "epoch": 0.3563623476600185,
      "grad_norm": 1.8208098411560059,
      "learning_rate": 9.98609500473307e-07,
      "loss": 1.9941,
      "num_input_tokens_seen": 18437576,
      "step": 18100
    },
    {
      "epoch": 0.3583312004095214,
      "grad_norm": 2.0128719806671143,
      "learning_rate": 9.985940920818863e-07,
      "loss": 1.9779,
      "num_input_tokens_seen": 18539976,
      "step": 18200
    },
    {
      "epoch": 0.36030005315902425,
      "grad_norm": 2.181436777114868,
      "learning_rate": 9.985785989087454e-07,
      "loss": 1.9569,
      "num_input_tokens_seen": 18641568,
      "step": 18300
    },
    {
      "epoch": 0.3622689059085271,
      "grad_norm": 1.764965534210205,
      "learning_rate": 9.985630209565187e-07,
      "loss": 2.0087,
      "num_input_tokens_seen": 18742960,
      "step": 18400
    },
    {
      "epoch": 0.36423775865803,
      "grad_norm": 1.8703725337982178,
      "learning_rate": 9.985473582278558e-07,
      "loss": 2.0024,
      "num_input_tokens_seen": 18843736,
      "step": 18500
    },
    {
      "epoch": 0.36620661140753286,
      "grad_norm": 2.4369924068450928,
      "learning_rate": 9.985316107254193e-07,
      "loss": 2.0421,
      "num_input_tokens_seen": 18945800,
      "step": 18600
    },
    {
      "epoch": 0.36817546415703567,
      "grad_norm": 2.4326465129852295,
      "learning_rate": 9.985157784518873e-07,
      "loss": 2.0023,
      "num_input_tokens_seen": 19047632,
      "step": 18700
    },
    {
      "epoch": 0.37014431690653854,
      "grad_norm": 2.0959532260894775,
      "learning_rate": 9.98499861409952e-07,
      "loss": 1.9879,
      "num_input_tokens_seen": 19148216,
      "step": 18800
    },
    {
      "epoch": 0.3721131696560414,
      "grad_norm": 2.0876073837280273,
      "learning_rate": 9.984838596023195e-07,
      "loss": 1.9447,
      "num_input_tokens_seen": 19250616,
      "step": 18900
    },
    {
      "epoch": 0.3740820224055443,
      "grad_norm": 1.9188069105148315,
      "learning_rate": 9.984677730317112e-07,
      "loss": 1.9803,
      "num_input_tokens_seen": 19353016,
      "step": 19000
    },
    {
      "epoch": 0.37605087515504715,
      "grad_norm": 2.220883846282959,
      "learning_rate": 9.984516017008623e-07,
      "loss": 1.9828,
      "num_input_tokens_seen": 19455416,
      "step": 19100
    },
    {
      "epoch": 0.37801972790455,
      "grad_norm": 1.742920160293579,
      "learning_rate": 9.98435345612523e-07,
      "loss": 1.9757,
      "num_input_tokens_seen": 19557144,
      "step": 19200
    },
    {
      "epoch": 0.3799885806540529,
      "grad_norm": 2.3675317764282227,
      "learning_rate": 9.98419004769457e-07,
      "loss": 1.9925,
      "num_input_tokens_seen": 19659544,
      "step": 19300
    },
    {
      "epoch": 0.38195743340355576,
      "grad_norm": 1.9253923892974854,
      "learning_rate": 9.98402579174443e-07,
      "loss": 1.9984,
      "num_input_tokens_seen": 19760632,
      "step": 19400
    },
    {
      "epoch": 0.3839262861530586,
      "grad_norm": 1.7223613262176514,
      "learning_rate": 9.98386068830274e-07,
      "loss": 1.9942,
      "num_input_tokens_seen": 19861480,
      "step": 19500
    },
    {
      "epoch": 0.3858951389025615,
      "grad_norm": 1.9526335000991821,
      "learning_rate": 9.983694737397579e-07,
      "loss": 1.9479,
      "num_input_tokens_seen": 19963304,
      "step": 19600
    },
    {
      "epoch": 0.38786399165206437,
      "grad_norm": 1.8235361576080322,
      "learning_rate": 9.98352793905716e-07,
      "loss": 2.0048,
      "num_input_tokens_seen": 20065704,
      "step": 19700
    },
    {
      "epoch": 0.3898328444015672,
      "grad_norm": 1.9568926095962524,
      "learning_rate": 9.983360293309849e-07,
      "loss": 1.9596,
      "num_input_tokens_seen": 20166448,
      "step": 19800
    },
    {
      "epoch": 0.39180169715107005,
      "grad_norm": 1.8626855611801147,
      "learning_rate": 9.983191800184153e-07,
      "loss": 1.9505,
      "num_input_tokens_seen": 20268848,
      "step": 19900
    },
    {
      "epoch": 0.3937705499005729,
      "grad_norm": 3.4124538898468018,
      "learning_rate": 9.98302245970872e-07,
      "loss": 1.9489,
      "num_input_tokens_seen": 20370352,
      "step": 20000
    },
    {
      "epoch": 0.3957394026500758,
      "grad_norm": 1.766021966934204,
      "learning_rate": 9.982852271912345e-07,
      "loss": 1.9811,
      "num_input_tokens_seen": 20472752,
      "step": 20100
    },
    {
      "epoch": 0.39770825539957866,
      "grad_norm": 1.6870958805084229,
      "learning_rate": 9.982681236823972e-07,
      "loss": 1.9737,
      "num_input_tokens_seen": 20575152,
      "step": 20200
    },
    {
      "epoch": 0.39967710814908153,
      "grad_norm": 1.963392972946167,
      "learning_rate": 9.982509354472677e-07,
      "loss": 1.9698,
      "num_input_tokens_seen": 20677552,
      "step": 20300
    },
    {
      "epoch": 0.4016459608985844,
      "grad_norm": 2.3354506492614746,
      "learning_rate": 9.982336624887693e-07,
      "loss": 1.9336,
      "num_input_tokens_seen": 20779952,
      "step": 20400
    },
    {
      "epoch": 0.40361481364808727,
      "grad_norm": 1.7728350162506104,
      "learning_rate": 9.982163048098388e-07,
      "loss": 2.0186,
      "num_input_tokens_seen": 20881720,
      "step": 20500
    },
    {
      "epoch": 0.40558366639759014,
      "grad_norm": 2.3466122150421143,
      "learning_rate": 9.981988624134278e-07,
      "loss": 2.0098,
      "num_input_tokens_seen": 20983992,
      "step": 20600
    },
    {
      "epoch": 0.407552519147093,
      "grad_norm": 1.976973533630371,
      "learning_rate": 9.981813353025023e-07,
      "loss": 1.9684,
      "num_input_tokens_seen": 21084976,
      "step": 20700
    },
    {
      "epoch": 0.4095213718965959,
      "grad_norm": 2.03411602973938,
      "learning_rate": 9.981637234800426e-07,
      "loss": 2.0056,
      "num_input_tokens_seen": 21186616,
      "step": 20800
    },
    {
      "epoch": 0.41149022464609875,
      "grad_norm": 2.321899890899658,
      "learning_rate": 9.981460269490434e-07,
      "loss": 1.9566,
      "num_input_tokens_seen": 21289016,
      "step": 20900
    },
    {
      "epoch": 0.41345907739560156,
      "grad_norm": 1.7294496297836304,
      "learning_rate": 9.981282457125138e-07,
      "loss": 1.9389,
      "num_input_tokens_seen": 21390608,
      "step": 21000
    },
    {
      "epoch": 0.41542793014510443,
      "grad_norm": 1.7769101858139038,
      "learning_rate": 9.981103797734775e-07,
      "loss": 1.964,
      "num_input_tokens_seen": 21490552,
      "step": 21100
    },
    {
      "epoch": 0.4173967828946073,
      "grad_norm": 2.246415376663208,
      "learning_rate": 9.980924291349724e-07,
      "loss": 1.9817,
      "num_input_tokens_seen": 21592952,
      "step": 21200
    },
    {
      "epoch": 0.41936563564411017,
      "grad_norm": 2.132946491241455,
      "learning_rate": 9.980743938000508e-07,
      "loss": 1.9842,
      "num_input_tokens_seen": 21695056,
      "step": 21300
    },
    {
      "epoch": 0.42133448839361304,
      "grad_norm": 1.9394687414169312,
      "learning_rate": 9.980562737717797e-07,
      "loss": 2.0426,
      "num_input_tokens_seen": 21796200,
      "step": 21400
    },
    {
      "epoch": 0.4233033411431159,
      "grad_norm": 1.7560745477676392,
      "learning_rate": 9.980380690532398e-07,
      "loss": 1.9813,
      "num_input_tokens_seen": 21898600,
      "step": 21500
    },
    {
      "epoch": 0.4252721938926188,
      "grad_norm": 1.9573732614517212,
      "learning_rate": 9.98019779647527e-07,
      "loss": 1.9945,
      "num_input_tokens_seen": 22000320,
      "step": 21600
    },
    {
      "epoch": 0.42724104664212165,
      "grad_norm": 1.7684223651885986,
      "learning_rate": 9.98001405557751e-07,
      "loss": 1.9862,
      "num_input_tokens_seen": 22101792,
      "step": 21700
    },
    {
      "epoch": 0.4292098993916245,
      "grad_norm": 2.01031231880188,
      "learning_rate": 9.979829467870365e-07,
      "loss": 1.9401,
      "num_input_tokens_seen": 22204192,
      "step": 21800
    },
    {
      "epoch": 0.4311787521411274,
      "grad_norm": 2.198564052581787,
      "learning_rate": 9.97964403338522e-07,
      "loss": 1.9513,
      "num_input_tokens_seen": 22306048,
      "step": 21900
    },
    {
      "epoch": 0.43314760489063026,
      "grad_norm": 1.8433729410171509,
      "learning_rate": 9.979457752153606e-07,
      "loss": 2.0143,
      "num_input_tokens_seen": 22407904,
      "step": 22000
    },
    {
      "epoch": 0.43511645764013307,
      "grad_norm": 2.1233456134796143,
      "learning_rate": 9.979270624207201e-07,
      "loss": 1.9838,
      "num_input_tokens_seen": 22510208,
      "step": 22100
    },
    {
      "epoch": 0.43708531038963594,
      "grad_norm": 1.9405741691589355,
      "learning_rate": 9.979082649577826e-07,
      "loss": 1.9756,
      "num_input_tokens_seen": 22611880,
      "step": 22200
    },
    {
      "epoch": 0.4390541631391388,
      "grad_norm": 1.918004035949707,
      "learning_rate": 9.97889382829744e-07,
      "loss": 1.9765,
      "num_input_tokens_seen": 22714280,
      "step": 22300
    },
    {
      "epoch": 0.4410230158886417,
      "grad_norm": 1.7071990966796875,
      "learning_rate": 9.97870416039815e-07,
      "loss": 1.9983,
      "num_input_tokens_seen": 22815312,
      "step": 22400
    },
    {
      "epoch": 0.44299186863814455,
      "grad_norm": 1.6820582151412964,
      "learning_rate": 9.978513645912214e-07,
      "loss": 1.9532,
      "num_input_tokens_seen": 22917712,
      "step": 22500
    },
    {
      "epoch": 0.4449607213876474,
      "grad_norm": 2.0030429363250732,
      "learning_rate": 9.978322284872021e-07,
      "loss": 1.9757,
      "num_input_tokens_seen": 23019560,
      "step": 22600
    },
    {
      "epoch": 0.4469295741371503,
      "grad_norm": 2.0032713413238525,
      "learning_rate": 9.978130077310113e-07,
      "loss": 1.973,
      "num_input_tokens_seen": 23121960,
      "step": 22700
    },
    {
      "epoch": 0.44889842688665316,
      "grad_norm": 1.9425948858261108,
      "learning_rate": 9.977937023259173e-07,
      "loss": 1.9391,
      "num_input_tokens_seen": 23224360,
      "step": 22800
    },
    {
      "epoch": 0.450867279636156,
      "grad_norm": 2.022223472595215,
      "learning_rate": 9.97774312275203e-07,
      "loss": 1.9796,
      "num_input_tokens_seen": 23326288,
      "step": 22900
    },
    {
      "epoch": 0.4528361323856589,
      "grad_norm": 10.623699188232422,
      "learning_rate": 9.977548375821649e-07,
      "loss": 1.9697,
      "num_input_tokens_seen": 23428688,
      "step": 23000
    },
    {
      "epoch": 0.45480498513516177,
      "grad_norm": 1.8131344318389893,
      "learning_rate": 9.977352782501151e-07,
      "loss": 1.9518,
      "num_input_tokens_seen": 23531088,
      "step": 23100
    },
    {
      "epoch": 0.4567738378846646,
      "grad_norm": 2.295611619949341,
      "learning_rate": 9.977156342823794e-07,
      "loss": 1.9559,
      "num_input_tokens_seen": 23633488,
      "step": 23200
    },
    {
      "epoch": 0.45874269063416745,
      "grad_norm": 1.9369465112686157,
      "learning_rate": 9.97695905682298e-07,
      "loss": 2.0245,
      "num_input_tokens_seen": 23735888,
      "step": 23300
    },
    {
      "epoch": 0.4607115433836703,
      "grad_norm": 1.5936909914016724,
      "learning_rate": 9.976760924532254e-07,
      "loss": 1.9733,
      "num_input_tokens_seen": 23837088,
      "step": 23400
    },
    {
      "epoch": 0.4626803961331732,
      "grad_norm": 2.748304605484009,
      "learning_rate": 9.976561945985312e-07,
      "loss": 1.9581,
      "num_input_tokens_seen": 23939488,
      "step": 23500
    },
    {
      "epoch": 0.46464924888267606,
      "grad_norm": 2.1762402057647705,
      "learning_rate": 9.976362121215984e-07,
      "loss": 2.0525,
      "num_input_tokens_seen": 24039880,
      "step": 23600
    },
    {
      "epoch": 0.4666181016321789,
      "grad_norm": 2.3380510807037354,
      "learning_rate": 9.97616145025825e-07,
      "loss": 1.9868,
      "num_input_tokens_seen": 24141800,
      "step": 23700
    },
    {
      "epoch": 0.4685869543816818,
      "grad_norm": 11.509876251220703,
      "learning_rate": 9.975959933146231e-07,
      "loss": 2.006,
      "num_input_tokens_seen": 24242464,
      "step": 23800
    },
    {
      "epoch": 0.47055580713118467,
      "grad_norm": 1.9859744310379028,
      "learning_rate": 9.975757569914196e-07,
      "loss": 1.9781,
      "num_input_tokens_seen": 24344864,
      "step": 23900
    },
    {
      "epoch": 0.47252465988068754,
      "grad_norm": 1.9662309885025024,
      "learning_rate": 9.975554360596554e-07,
      "loss": 2.0087,
      "num_input_tokens_seen": 24447264,
      "step": 24000
    },
    {
      "epoch": 0.4744935126301904,
      "grad_norm": 2.0596296787261963,
      "learning_rate": 9.975350305227858e-07,
      "loss": 1.9695,
      "num_input_tokens_seen": 24549664,
      "step": 24100
    },
    {
      "epoch": 0.4764623653796933,
      "grad_norm": 1.8109757900238037,
      "learning_rate": 9.975145403842806e-07,
      "loss": 2.0153,
      "num_input_tokens_seen": 24651336,
      "step": 24200
    },
    {
      "epoch": 0.4784312181291961,
      "grad_norm": 2.031705617904663,
      "learning_rate": 9.97493965647624e-07,
      "loss": 1.9615,
      "num_input_tokens_seen": 24753216,
      "step": 24300
    },
    {
      "epoch": 0.48040007087869896,
      "grad_norm": 1.7945663928985596,
      "learning_rate": 9.974733063163148e-07,
      "loss": 1.9781,
      "num_input_tokens_seen": 24854888,
      "step": 24400
    },
    {
      "epoch": 0.48236892362820183,
      "grad_norm": 1.979754090309143,
      "learning_rate": 9.974525623938657e-07,
      "loss": 2.0,
      "num_input_tokens_seen": 24957288,
      "step": 24500
    },
    {
      "epoch": 0.4843377763777047,
      "grad_norm": 1.6545555591583252,
      "learning_rate": 9.97431733883804e-07,
      "loss": 1.985,
      "num_input_tokens_seen": 25059480,
      "step": 24600
    },
    {
      "epoch": 0.48630662912720757,
      "grad_norm": 1.8531856536865234,
      "learning_rate": 9.974108207896715e-07,
      "loss": 1.962,
      "num_input_tokens_seen": 25161104,
      "step": 24700
    },
    {
      "epoch": 0.48827548187671044,
      "grad_norm": 1.9446638822555542,
      "learning_rate": 9.973898231150243e-07,
      "loss": 1.9843,
      "num_input_tokens_seen": 25262976,
      "step": 24800
    },
    {
      "epoch": 0.4902443346262133,
      "grad_norm": 3.837829351425171,
      "learning_rate": 9.973687408634329e-07,
      "loss": 1.9635,
      "num_input_tokens_seen": 25365376,
      "step": 24900
    },
    {
      "epoch": 0.4922131873757162,
      "grad_norm": 2.8814587593078613,
      "learning_rate": 9.97347574038482e-07,
      "loss": 1.9557,
      "num_input_tokens_seen": 25466552,
      "step": 25000
    },
    {
      "epoch": 0.49418204012521905,
      "grad_norm": 2.385894298553467,
      "learning_rate": 9.97326322643771e-07,
      "loss": 1.9476,
      "num_input_tokens_seen": 25567552,
      "step": 25100
    },
    {
      "epoch": 0.4961508928747219,
      "grad_norm": 1.806183934211731,
      "learning_rate": 9.973049866829136e-07,
      "loss": 2.011,
      "num_input_tokens_seen": 25669424,
      "step": 25200
    },
    {
      "epoch": 0.4981197456242248,
      "grad_norm": 2.0874178409576416,
      "learning_rate": 9.972835661595377e-07,
      "loss": 1.9997,
      "num_input_tokens_seen": 25771736,
      "step": 25300
    },
    {
      "epoch": 0.5000885983737277,
      "grad_norm": 1.8951822519302368,
      "learning_rate": 9.972620610772853e-07,
      "loss": 2.0229,
      "num_input_tokens_seen": 25872328,
      "step": 25400
    },
    {
      "epoch": 0.5020574511232305,
      "grad_norm": 1.9708080291748047,
      "learning_rate": 9.972404714398137e-07,
      "loss": 2.0027,
      "num_input_tokens_seen": 25972904,
      "step": 25500
    },
    {
      "epoch": 0.5040263038727334,
      "grad_norm": 2.000120162963867,
      "learning_rate": 9.97218797250794e-07,
      "loss": 1.9755,
      "num_input_tokens_seen": 26075304,
      "step": 25600
    },
    {
      "epoch": 0.5059951566222363,
      "grad_norm": 2.2590060234069824,
      "learning_rate": 9.971970385139114e-07,
      "loss": 1.9947,
      "num_input_tokens_seen": 26176432,
      "step": 25700
    },
    {
      "epoch": 0.5079640093717391,
      "grad_norm": 1.7969183921813965,
      "learning_rate": 9.971751952328662e-07,
      "loss": 2.0154,
      "num_input_tokens_seen": 26277160,
      "step": 25800
    },
    {
      "epoch": 0.509932862121242,
      "grad_norm": 1.7295571565628052,
      "learning_rate": 9.97153267411372e-07,
      "loss": 2.0067,
      "num_input_tokens_seen": 26378984,
      "step": 25900
    },
    {
      "epoch": 0.5119017148707448,
      "grad_norm": 1.9042713642120361,
      "learning_rate": 9.971312550531582e-07,
      "loss": 2.0285,
      "num_input_tokens_seen": 26480568,
      "step": 26000
    },
    {
      "epoch": 0.5138705676202476,
      "grad_norm": 2.275377035140991,
      "learning_rate": 9.971091581619674e-07,
      "loss": 1.9685,
      "num_input_tokens_seen": 26582968,
      "step": 26100
    },
    {
      "epoch": 0.5158394203697505,
      "grad_norm": 1.9059438705444336,
      "learning_rate": 9.970869767415572e-07,
      "loss": 1.9568,
      "num_input_tokens_seen": 26685368,
      "step": 26200
    },
    {
      "epoch": 0.5178082731192534,
      "grad_norm": 1.9786062240600586,
      "learning_rate": 9.97064710795699e-07,
      "loss": 2.0229,
      "num_input_tokens_seen": 26787328,
      "step": 26300
    },
    {
      "epoch": 0.5197771258687562,
      "grad_norm": 2.0805323123931885,
      "learning_rate": 9.970423603281792e-07,
      "loss": 1.9997,
      "num_input_tokens_seen": 26888392,
      "step": 26400
    },
    {
      "epoch": 0.5217459786182591,
      "grad_norm": 2.0707738399505615,
      "learning_rate": 9.970199253427984e-07,
      "loss": 2.0388,
      "num_input_tokens_seen": 26990144,
      "step": 26500
    },
    {
      "epoch": 0.523714831367762,
      "grad_norm": 2.0716702938079834,
      "learning_rate": 9.969974058433712e-07,
      "loss": 2.0231,
      "num_input_tokens_seen": 27091712,
      "step": 26600
    },
    {
      "epoch": 0.5256836841172648,
      "grad_norm": 2.0474820137023926,
      "learning_rate": 9.96974801833727e-07,
      "loss": 1.9902,
      "num_input_tokens_seen": 27194112,
      "step": 26700
    },
    {
      "epoch": 0.5276525368667677,
      "grad_norm": 1.650191068649292,
      "learning_rate": 9.969521133177095e-07,
      "loss": 2.0088,
      "num_input_tokens_seen": 27296512,
      "step": 26800
    },
    {
      "epoch": 0.5296213896162706,
      "grad_norm": 1.7658859491348267,
      "learning_rate": 9.969293402991768e-07,
      "loss": 1.9835,
      "num_input_tokens_seen": 27398144,
      "step": 26900
    },
    {
      "epoch": 0.5315902423657735,
      "grad_norm": 2.167694091796875,
      "learning_rate": 9.969064827820009e-07,
      "loss": 1.9739,
      "num_input_tokens_seen": 27500544,
      "step": 27000
    },
    {
      "epoch": 0.5335590951152763,
      "grad_norm": 2.014336347579956,
      "learning_rate": 9.968835407700687e-07,
      "loss": 1.9942,
      "num_input_tokens_seen": 27602944,
      "step": 27100
    },
    {
      "epoch": 0.5355279478647792,
      "grad_norm": 1.7608187198638916,
      "learning_rate": 9.968605142672813e-07,
      "loss": 1.9802,
      "num_input_tokens_seen": 27704704,
      "step": 27200
    },
    {
      "epoch": 0.5374968006142821,
      "grad_norm": 1.9191299676895142,
      "learning_rate": 9.968374032775542e-07,
      "loss": 2.0222,
      "num_input_tokens_seen": 27805696,
      "step": 27300
    },
    {
      "epoch": 0.5394656533637849,
      "grad_norm": 2.027268171310425,
      "learning_rate": 9.968142078048171e-07,
      "loss": 1.986,
      "num_input_tokens_seen": 27907048,
      "step": 27400
    },
    {
      "epoch": 0.5414345061132878,
      "grad_norm": 1.9456969499588013,
      "learning_rate": 9.967909278530144e-07,
      "loss": 1.9341,
      "num_input_tokens_seen": 28009448,
      "step": 27500
    },
    {
      "epoch": 0.5434033588627907,
      "grad_norm": 1.9179428815841675,
      "learning_rate": 9.967675634261043e-07,
      "loss": 1.9452,
      "num_input_tokens_seen": 28111848,
      "step": 27600
    },
    {
      "epoch": 0.5453722116122935,
      "grad_norm": 4.113171100616455,
      "learning_rate": 9.967441145280603e-07,
      "loss": 1.9426,
      "num_input_tokens_seen": 28213720,
      "step": 27700
    },
    {
      "epoch": 0.5473410643617964,
      "grad_norm": 1.9656105041503906,
      "learning_rate": 9.967205811628691e-07,
      "loss": 1.9545,
      "num_input_tokens_seen": 28316120,
      "step": 27800
    },
    {
      "epoch": 0.5493099171112993,
      "grad_norm": 1.8778409957885742,
      "learning_rate": 9.966969633345327e-07,
      "loss": 1.9895,
      "num_input_tokens_seen": 28417872,
      "step": 27900
    },
    {
      "epoch": 0.5512787698608022,
      "grad_norm": 1.7477798461914062,
      "learning_rate": 9.96673261047067e-07,
      "loss": 1.9813,
      "num_input_tokens_seen": 28519480,
      "step": 28000
    },
    {
      "epoch": 0.553247622610305,
      "grad_norm": 1.9219067096710205,
      "learning_rate": 9.966494743045026e-07,
      "loss": 1.9445,
      "num_input_tokens_seen": 28621056,
      "step": 28100
    },
    {
      "epoch": 0.5552164753598079,
      "grad_norm": 1.8571618795394897,
      "learning_rate": 9.96625603110884e-07,
      "loss": 2.0336,
      "num_input_tokens_seen": 28721304,
      "step": 28200
    },
    {
      "epoch": 0.5571853281093107,
      "grad_norm": 2.0712876319885254,
      "learning_rate": 9.9660164747027e-07,
      "loss": 1.9648,
      "num_input_tokens_seen": 28823704,
      "step": 28300
    },
    {
      "epoch": 0.5591541808588135,
      "grad_norm": 2.039795398712158,
      "learning_rate": 9.965776073867346e-07,
      "loss": 1.9886,
      "num_input_tokens_seen": 28926104,
      "step": 28400
    },
    {
      "epoch": 0.5611230336083164,
      "grad_norm": 2.1456973552703857,
      "learning_rate": 9.965534828643655e-07,
      "loss": 2.0234,
      "num_input_tokens_seen": 29027024,
      "step": 28500
    },
    {
      "epoch": 0.5630918863578193,
      "grad_norm": 1.9275834560394287,
      "learning_rate": 9.965292739072645e-07,
      "loss": 1.9691,
      "num_input_tokens_seen": 29128792,
      "step": 28600
    },
    {
      "epoch": 0.5650607391073221,
      "grad_norm": 2.1054723262786865,
      "learning_rate": 9.965049805195486e-07,
      "loss": 2.0016,
      "num_input_tokens_seen": 29230640,
      "step": 28700
    },
    {
      "epoch": 0.567029591856825,
      "grad_norm": 1.7965164184570312,
      "learning_rate": 9.964806027053485e-07,
      "loss": 1.9457,
      "num_input_tokens_seen": 29332304,
      "step": 28800
    },
    {
      "epoch": 0.5689984446063279,
      "grad_norm": 2.277695417404175,
      "learning_rate": 9.964561404688095e-07,
      "loss": 2.1029,
      "num_input_tokens_seen": 29432944,
      "step": 28900
    },
    {
      "epoch": 0.5709672973558307,
      "grad_norm": 1.9059792757034302,
      "learning_rate": 9.964315938140908e-07,
      "loss": 1.9671,
      "num_input_tokens_seen": 29533832,
      "step": 29000
    },
    {
      "epoch": 0.5729361501053336,
      "grad_norm": 1.7866215705871582,
      "learning_rate": 9.96406962745367e-07,
      "loss": 1.9647,
      "num_input_tokens_seen": 29636232,
      "step": 29100
    },
    {
      "epoch": 0.5749050028548365,
      "grad_norm": 1.9816657304763794,
      "learning_rate": 9.963822472668257e-07,
      "loss": 1.973,
      "num_input_tokens_seen": 29738632,
      "step": 29200
    },
    {
      "epoch": 0.5768738556043393,
      "grad_norm": 2.256709575653076,
      "learning_rate": 9.963574473826702e-07,
      "loss": 1.9592,
      "num_input_tokens_seen": 29840240,
      "step": 29300
    },
    {
      "epoch": 0.5788427083538422,
      "grad_norm": 2.0529892444610596,
      "learning_rate": 9.963325630971172e-07,
      "loss": 1.9705,
      "num_input_tokens_seen": 29942640,
      "step": 29400
    },
    {
      "epoch": 0.5808115611033451,
      "grad_norm": 1.9970955848693848,
      "learning_rate": 9.96307594414398e-07,
      "loss": 2.0044,
      "num_input_tokens_seen": 30043048,
      "step": 29500
    },
    {
      "epoch": 0.582780413852848,
      "grad_norm": 2.6988253593444824,
      "learning_rate": 9.962825413387585e-07,
      "loss": 1.9797,
      "num_input_tokens_seen": 30145448,
      "step": 29600
    },
    {
      "epoch": 0.5847492666023508,
      "grad_norm": 1.9380062818527222,
      "learning_rate": 9.962574038744587e-07,
      "loss": 1.9933,
      "num_input_tokens_seen": 30247384,
      "step": 29700
    },
    {
      "epoch": 0.5867181193518537,
      "grad_norm": 1.6916238069534302,
      "learning_rate": 9.96232182025773e-07,
      "loss": 1.9392,
      "num_input_tokens_seen": 30349784,
      "step": 29800
    },
    {
      "epoch": 0.5886869721013566,
      "grad_norm": 1.9196768999099731,
      "learning_rate": 9.962068757969902e-07,
      "loss": 2.0357,
      "num_input_tokens_seen": 30452184,
      "step": 29900
    },
    {
      "epoch": 0.5906558248508594,
      "grad_norm": 1.998913049697876,
      "learning_rate": 9.961814851924132e-07,
      "loss": 1.9804,
      "num_input_tokens_seen": 30553720,
      "step": 30000
    },
    {
      "epoch": 0.5926246776003623,
      "grad_norm": 2.203202247619629,
      "learning_rate": 9.9615601021636e-07,
      "loss": 1.9781,
      "num_input_tokens_seen": 30656120,
      "step": 30100
    },
    {
      "epoch": 0.5945935303498652,
      "grad_norm": 1.7600692510604858,
      "learning_rate": 9.961304508731616e-07,
      "loss": 2.001,
      "num_input_tokens_seen": 30757728,
      "step": 30200
    },
    {
      "epoch": 0.596562383099368,
      "grad_norm": 1.7541885375976562,
      "learning_rate": 9.961048071671648e-07,
      "loss": 2.0252,
      "num_input_tokens_seen": 30860128,
      "step": 30300
    },
    {
      "epoch": 0.5985312358488709,
      "grad_norm": 2.0382676124572754,
      "learning_rate": 9.960790791027297e-07,
      "loss": 1.9911,
      "num_input_tokens_seen": 30960928,
      "step": 30400
    },
    {
      "epoch": 0.6005000885983738,
      "grad_norm": 3.8343398571014404,
      "learning_rate": 9.960532666842317e-07,
      "loss": 1.953,
      "num_input_tokens_seen": 31062480,
      "step": 30500
    },
    {
      "epoch": 0.6024689413478765,
      "grad_norm": 1.8821525573730469,
      "learning_rate": 9.960273699160593e-07,
      "loss": 1.9692,
      "num_input_tokens_seen": 31163536,
      "step": 30600
    },
    {
      "epoch": 0.6044377940973794,
      "grad_norm": 1.9925546646118164,
      "learning_rate": 9.960013888026165e-07,
      "loss": 2.0457,
      "num_input_tokens_seen": 31264296,
      "step": 30700
    },
    {
      "epoch": 0.6064066468468823,
      "grad_norm": 2.1350507736206055,
      "learning_rate": 9.95975323348321e-07,
      "loss": 1.9794,
      "num_input_tokens_seen": 31366200,
      "step": 30800
    },
    {
      "epoch": 0.6083754995963851,
      "grad_norm": 2.237239360809326,
      "learning_rate": 9.959491735576048e-07,
      "loss": 1.9788,
      "num_input_tokens_seen": 31468056,
      "step": 30900
    },
    {
      "epoch": 0.610344352345888,
      "grad_norm": 1.8590543270111084,
      "learning_rate": 9.95922939434915e-07,
      "loss": 2.0007,
      "num_input_tokens_seen": 31570024,
      "step": 31000
    },
    {
      "epoch": 0.6123132050953909,
      "grad_norm": 1.8755176067352295,
      "learning_rate": 9.958966209847119e-07,
      "loss": 2.0011,
      "num_input_tokens_seen": 31670456,
      "step": 31100
    },
    {
      "epoch": 0.6142820578448938,
      "grad_norm": 1.9030488729476929,
      "learning_rate": 9.95870218211471e-07,
      "loss": 2.0265,
      "num_input_tokens_seen": 31772304,
      "step": 31200
    },
    {
      "epoch": 0.6162509105943966,
      "grad_norm": 1.8745536804199219,
      "learning_rate": 9.95843731119682e-07,
      "loss": 2.0047,
      "num_input_tokens_seen": 31874224,
      "step": 31300
    },
    {
      "epoch": 0.6182197633438995,
      "grad_norm": 1.827952265739441,
      "learning_rate": 9.958171597138484e-07,
      "loss": 1.9865,
      "num_input_tokens_seen": 31975768,
      "step": 31400
    },
    {
      "epoch": 0.6201886160934024,
      "grad_norm": 2.2605669498443604,
      "learning_rate": 9.957905039984888e-07,
      "loss": 1.9703,
      "num_input_tokens_seen": 32077064,
      "step": 31500
    },
    {
      "epoch": 0.6221574688429052,
      "grad_norm": 1.8941019773483276,
      "learning_rate": 9.957637639781359e-07,
      "loss": 1.9939,
      "num_input_tokens_seen": 32178832,
      "step": 31600
    },
    {
      "epoch": 0.6241263215924081,
      "grad_norm": 1.839310646057129,
      "learning_rate": 9.957369396573362e-07,
      "loss": 1.9873,
      "num_input_tokens_seen": 32281232,
      "step": 31700
    },
    {
      "epoch": 0.626095174341911,
      "grad_norm": 1.820748209953308,
      "learning_rate": 9.957100310406511e-07,
      "loss": 2.0094,
      "num_input_tokens_seen": 32383632,
      "step": 31800
    },
    {
      "epoch": 0.6280640270914138,
      "grad_norm": 1.7682918310165405,
      "learning_rate": 9.956830381326565e-07,
      "loss": 1.9753,
      "num_input_tokens_seen": 32486032,
      "step": 31900
    },
    {
      "epoch": 0.6300328798409167,
      "grad_norm": 1.772078514099121,
      "learning_rate": 9.956559609379418e-07,
      "loss": 2.0039,
      "num_input_tokens_seen": 32588432,
      "step": 32000
    },
    {
      "epoch": 0.6320017325904196,
      "grad_norm": 1.7432925701141357,
      "learning_rate": 9.956287994611116e-07,
      "loss": 2.0084,
      "num_input_tokens_seen": 32688968,
      "step": 32100
    },
    {
      "epoch": 0.6339705853399225,
      "grad_norm": 2.1200695037841797,
      "learning_rate": 9.956015537067843e-07,
      "loss": 2.0633,
      "num_input_tokens_seen": 32790728,
      "step": 32200
    },
    {
      "epoch": 0.6359394380894253,
      "grad_norm": 1.9778767824172974,
      "learning_rate": 9.95574223679593e-07,
      "loss": 2.0204,
      "num_input_tokens_seen": 32892320,
      "step": 32300
    },
    {
      "epoch": 0.6379082908389282,
      "grad_norm": 1.8613898754119873,
      "learning_rate": 9.955468093841848e-07,
      "loss": 1.9224,
      "num_input_tokens_seen": 32994720,
      "step": 32400
    },
    {
      "epoch": 0.6398771435884311,
      "grad_norm": 1.9930576086044312,
      "learning_rate": 9.955193108252214e-07,
      "loss": 2.0055,
      "num_input_tokens_seen": 33095600,
      "step": 32500
    },
    {
      "epoch": 0.6418459963379339,
      "grad_norm": 1.9638782739639282,
      "learning_rate": 9.954917280073784e-07,
      "loss": 1.993,
      "num_input_tokens_seen": 33197152,
      "step": 32600
    },
    {
      "epoch": 0.6438148490874368,
      "grad_norm": 1.9950863122940063,
      "learning_rate": 9.954640609353462e-07,
      "loss": 1.9665,
      "num_input_tokens_seen": 33298512,
      "step": 32700
    },
    {
      "epoch": 0.6457837018369396,
      "grad_norm": 2.18471360206604,
      "learning_rate": 9.954363096138297e-07,
      "loss": 2.0437,
      "num_input_tokens_seen": 33397424,
      "step": 32800
    },
    {
      "epoch": 0.6477525545864424,
      "grad_norm": 1.8377265930175781,
      "learning_rate": 9.95408474047547e-07,
      "loss": 1.9945,
      "num_input_tokens_seen": 33499384,
      "step": 32900
    },
    {
      "epoch": 0.6497214073359453,
      "grad_norm": 1.9662699699401855,
      "learning_rate": 9.95380554241232e-07,
      "loss": 1.966,
      "num_input_tokens_seen": 33601552,
      "step": 33000
    },
    {
      "epoch": 0.6516902600854482,
      "grad_norm": 1.7178562879562378,
      "learning_rate": 9.95352550199632e-07,
      "loss": 1.9928,
      "num_input_tokens_seen": 33703952,
      "step": 33100
    },
    {
      "epoch": 0.653659112834951,
      "grad_norm": 1.94487464427948,
      "learning_rate": 9.953244619275088e-07,
      "loss": 2.0221,
      "num_input_tokens_seen": 33806352,
      "step": 33200
    },
    {
      "epoch": 0.6556279655844539,
      "grad_norm": 2.074824810028076,
      "learning_rate": 9.952962894296386e-07,
      "loss": 1.9933,
      "num_input_tokens_seen": 33908752,
      "step": 33300
    },
    {
      "epoch": 0.6575968183339568,
      "grad_norm": 2.118398427963257,
      "learning_rate": 9.952680327108122e-07,
      "loss": 1.9703,
      "num_input_tokens_seen": 34011152,
      "step": 33400
    },
    {
      "epoch": 0.6595656710834596,
      "grad_norm": 1.9463146924972534,
      "learning_rate": 9.952396917758336e-07,
      "loss": 1.9908,
      "num_input_tokens_seen": 34113048,
      "step": 33500
    },
    {
      "epoch": 0.6615345238329625,
      "grad_norm": 1.757574439048767,
      "learning_rate": 9.952112666295227e-07,
      "loss": 2.0082,
      "num_input_tokens_seen": 34215152,
      "step": 33600
    },
    {
      "epoch": 0.6635033765824654,
      "grad_norm": 1.937621831893921,
      "learning_rate": 9.951827572767128e-07,
      "loss": 1.975,
      "num_input_tokens_seen": 34316464,
      "step": 33700
    },
    {
      "epoch": 0.6654722293319683,
      "grad_norm": 1.856847882270813,
      "learning_rate": 9.951541637222513e-07,
      "loss": 2.0031,
      "num_input_tokens_seen": 34418312,
      "step": 33800
    },
    {
      "epoch": 0.6674410820814711,
      "grad_norm": 2.0302321910858154,
      "learning_rate": 9.95125485971001e-07,
      "loss": 1.9705,
      "num_input_tokens_seen": 34519792,
      "step": 33900
    },
    {
      "epoch": 0.669409934830974,
      "grad_norm": 1.937517762184143,
      "learning_rate": 9.950967240278377e-07,
      "loss": 1.9463,
      "num_input_tokens_seen": 34622192,
      "step": 34000
    },
    {
      "epoch": 0.6713787875804769,
      "grad_norm": 2.2435901165008545,
      "learning_rate": 9.950678778976523e-07,
      "loss": 1.9799,
      "num_input_tokens_seen": 34724592,
      "step": 34100
    },
    {
      "epoch": 0.6733476403299797,
      "grad_norm": 1.9277533292770386,
      "learning_rate": 9.950389475853499e-07,
      "loss": 1.9848,
      "num_input_tokens_seen": 34826992,
      "step": 34200
    },
    {
      "epoch": 0.6753164930794826,
      "grad_norm": 2.025343418121338,
      "learning_rate": 9.950099330958495e-07,
      "loss": 1.9814,
      "num_input_tokens_seen": 34929392,
      "step": 34300
    },
    {
      "epoch": 0.6772853458289855,
      "grad_norm": 2.034280300140381,
      "learning_rate": 9.949808344340855e-07,
      "loss": 2.0129,
      "num_input_tokens_seen": 35031448,
      "step": 34400
    },
    {
      "epoch": 0.6792541985784883,
      "grad_norm": 1.7865476608276367,
      "learning_rate": 9.949516516050052e-07,
      "loss": 1.9997,
      "num_input_tokens_seen": 35133128,
      "step": 34500
    },
    {
      "epoch": 0.6812230513279912,
      "grad_norm": 2.143674373626709,
      "learning_rate": 9.949223846135713e-07,
      "loss": 1.9308,
      "num_input_tokens_seen": 35234960,
      "step": 34600
    },
    {
      "epoch": 0.6831919040774941,
      "grad_norm": 2.0520286560058594,
      "learning_rate": 9.948930334647602e-07,
      "loss": 2.0636,
      "num_input_tokens_seen": 35334576,
      "step": 34700
    },
    {
      "epoch": 0.685160756826997,
      "grad_norm": 1.8532273769378662,
      "learning_rate": 9.94863598163563e-07,
      "loss": 1.9685,
      "num_input_tokens_seen": 35436176,
      "step": 34800
    },
    {
      "epoch": 0.6871296095764998,
      "grad_norm": 2.1574697494506836,
      "learning_rate": 9.948340787149847e-07,
      "loss": 1.9843,
      "num_input_tokens_seen": 35538576,
      "step": 34900
    },
    {
      "epoch": 0.6890984623260027,
      "grad_norm": 1.7824547290802002,
      "learning_rate": 9.94804475124045e-07,
      "loss": 1.9959,
      "num_input_tokens_seen": 35640288,
      "step": 35000
    },
    {
      "epoch": 0.6910673150755055,
      "grad_norm": 1.9223896265029907,
      "learning_rate": 9.947747873957775e-07,
      "loss": 1.988,
      "num_input_tokens_seen": 35741992,
      "step": 35100
    },
    {
      "epoch": 0.6930361678250083,
      "grad_norm": 2.1041834354400635,
      "learning_rate": 9.94745015535231e-07,
      "loss": 1.9787,
      "num_input_tokens_seen": 35844392,
      "step": 35200
    },
    {
      "epoch": 0.6950050205745112,
      "grad_norm": 1.9088863134384155,
      "learning_rate": 9.947151595474668e-07,
      "loss": 2.0372,
      "num_input_tokens_seen": 35946208,
      "step": 35300
    },
    {
      "epoch": 0.6969738733240141,
      "grad_norm": 1.6192907094955444,
      "learning_rate": 9.946852194375629e-07,
      "loss": 1.9878,
      "num_input_tokens_seen": 36048040,
      "step": 35400
    },
    {
      "epoch": 0.6989427260735169,
      "grad_norm": 2.0870442390441895,
      "learning_rate": 9.946551952106094e-07,
      "loss": 2.0269,
      "num_input_tokens_seen": 36150440,
      "step": 35500
    },
    {
      "epoch": 0.7009115788230198,
      "grad_norm": 2.09017014503479,
      "learning_rate": 9.946250868717122e-07,
      "loss": 2.0078,
      "num_input_tokens_seen": 36251920,
      "step": 35600
    },
    {
      "epoch": 0.7028804315725227,
      "grad_norm": 2.113693952560425,
      "learning_rate": 9.94594894425991e-07,
      "loss": 1.9861,
      "num_input_tokens_seen": 36354320,
      "step": 35700
    },
    {
      "epoch": 0.7048492843220255,
      "grad_norm": 1.7313716411590576,
      "learning_rate": 9.945646178785797e-07,
      "loss": 1.9537,
      "num_input_tokens_seen": 36456720,
      "step": 35800
    },
    {
      "epoch": 0.7068181370715284,
      "grad_norm": 2.220529556274414,
      "learning_rate": 9.94534257234626e-07,
      "loss": 2.0293,
      "num_input_tokens_seen": 36557424,
      "step": 35900
    },
    {
      "epoch": 0.7087869898210313,
      "grad_norm": 2.143380880355835,
      "learning_rate": 9.945038124992933e-07,
      "loss": 1.9949,
      "num_input_tokens_seen": 36658560,
      "step": 36000
    },
    {
      "epoch": 0.7107558425705341,
      "grad_norm": 1.8441015481948853,
      "learning_rate": 9.94473283677758e-07,
      "loss": 1.9858,
      "num_input_tokens_seen": 36760960,
      "step": 36100
    },
    {
      "epoch": 0.712724695320037,
      "grad_norm": 2.3321261405944824,
      "learning_rate": 9.944426707752117e-07,
      "loss": 2.0326,
      "num_input_tokens_seen": 36861200,
      "step": 36200
    },
    {
      "epoch": 0.7146935480695399,
      "grad_norm": 1.9875354766845703,
      "learning_rate": 9.944119737968592e-07,
      "loss": 1.9999,
      "num_input_tokens_seen": 36963600,
      "step": 36300
    },
    {
      "epoch": 0.7166624008190428,
      "grad_norm": 2.2651355266571045,
      "learning_rate": 9.943811927479207e-07,
      "loss": 1.9843,
      "num_input_tokens_seen": 37066000,
      "step": 36400
    },
    {
      "epoch": 0.7186312535685456,
      "grad_norm": 2.0598232746124268,
      "learning_rate": 9.9435032763363e-07,
      "loss": 1.9995,
      "num_input_tokens_seen": 37168400,
      "step": 36500
    },
    {
      "epoch": 0.7206001063180485,
      "grad_norm": 3.320465087890625,
      "learning_rate": 9.94319378459236e-07,
      "loss": 1.9962,
      "num_input_tokens_seen": 37269864,
      "step": 36600
    },
    {
      "epoch": 0.7225689590675514,
      "grad_norm": 2.06569766998291,
      "learning_rate": 9.94288345230001e-07,
      "loss": 1.9829,
      "num_input_tokens_seen": 37371496,
      "step": 36700
    },
    {
      "epoch": 0.7245378118170542,
      "grad_norm": 1.7494834661483765,
      "learning_rate": 9.942572279512014e-07,
      "loss": 2.007,
      "num_input_tokens_seen": 37473080,
      "step": 36800
    },
    {
      "epoch": 0.7265066645665571,
      "grad_norm": 1.762199878692627,
      "learning_rate": 9.942260266281295e-07,
      "loss": 2.008,
      "num_input_tokens_seen": 37574736,
      "step": 36900
    },
    {
      "epoch": 0.72847551731606,
      "grad_norm": 2.2232725620269775,
      "learning_rate": 9.9419474126609e-07,
      "loss": 1.9557,
      "num_input_tokens_seen": 37676088,
      "step": 37000
    },
    {
      "epoch": 0.7304443700655628,
      "grad_norm": 1.9961750507354736,
      "learning_rate": 9.94163371870403e-07,
      "loss": 2.0327,
      "num_input_tokens_seen": 37775664,
      "step": 37100
    },
    {
      "epoch": 0.7324132228150657,
      "grad_norm": 1.8107763528823853,
      "learning_rate": 9.941319184464026e-07,
      "loss": 1.9494,
      "num_input_tokens_seen": 37878064,
      "step": 37200
    },
    {
      "epoch": 0.7343820755645685,
      "grad_norm": 2.0820484161376953,
      "learning_rate": 9.94100380999437e-07,
      "loss": 1.9297,
      "num_input_tokens_seen": 37980464,
      "step": 37300
    },
    {
      "epoch": 0.7363509283140713,
      "grad_norm": 1.7529029846191406,
      "learning_rate": 9.940687595348693e-07,
      "loss": 2.0081,
      "num_input_tokens_seen": 38082072,
      "step": 37400
    },
    {
      "epoch": 0.7383197810635742,
      "grad_norm": 1.797738790512085,
      "learning_rate": 9.94037054058076e-07,
      "loss": 1.9588,
      "num_input_tokens_seen": 38184472,
      "step": 37500
    },
    {
      "epoch": 0.7402886338130771,
      "grad_norm": 1.732525110244751,
      "learning_rate": 9.940052645744488e-07,
      "loss": 1.9768,
      "num_input_tokens_seen": 38286736,
      "step": 37600
    },
    {
      "epoch": 0.74225748656258,
      "grad_norm": 1.833274006843567,
      "learning_rate": 9.939733910893928e-07,
      "loss": 2.005,
      "num_input_tokens_seen": 38388616,
      "step": 37700
    },
    {
      "epoch": 0.7442263393120828,
      "grad_norm": 1.9808897972106934,
      "learning_rate": 9.93941433608328e-07,
      "loss": 1.9749,
      "num_input_tokens_seen": 38490312,
      "step": 37800
    },
    {
      "epoch": 0.7461951920615857,
      "grad_norm": 1.959140658378601,
      "learning_rate": 9.939093921366888e-07,
      "loss": 1.9948,
      "num_input_tokens_seen": 38591968,
      "step": 37900
    },
    {
      "epoch": 0.7481640448110886,
      "grad_norm": 1.8730968236923218,
      "learning_rate": 9.93877266679923e-07,
      "loss": 1.9451,
      "num_input_tokens_seen": 38693520,
      "step": 38000
    },
    {
      "epoch": 0.7501328975605914,
      "grad_norm": 1.9512994289398193,
      "learning_rate": 9.938450572434936e-07,
      "loss": 2.0305,
      "num_input_tokens_seen": 38792232,
      "step": 38100
    },
    {
      "epoch": 0.7521017503100943,
      "grad_norm": 2.2991583347320557,
      "learning_rate": 9.938127638328775e-07,
      "loss": 1.9864,
      "num_input_tokens_seen": 38893096,
      "step": 38200
    },
    {
      "epoch": 0.7540706030595972,
      "grad_norm": 1.5798085927963257,
      "learning_rate": 9.93780386453566e-07,
      "loss": 1.9588,
      "num_input_tokens_seen": 38995496,
      "step": 38300
    },
    {
      "epoch": 0.7560394558091,
      "grad_norm": 2.174484968185425,
      "learning_rate": 9.937479251110642e-07,
      "loss": 1.9728,
      "num_input_tokens_seen": 39097896,
      "step": 38400
    },
    {
      "epoch": 0.7580083085586029,
      "grad_norm": 6.0786452293396,
      "learning_rate": 9.937153798108927e-07,
      "loss": 1.9651,
      "num_input_tokens_seen": 39200296,
      "step": 38500
    },
    {
      "epoch": 0.7599771613081058,
      "grad_norm": 2.087667465209961,
      "learning_rate": 9.936827505585848e-07,
      "loss": 2.0322,
      "num_input_tokens_seen": 39302696,
      "step": 38600
    },
    {
      "epoch": 0.7619460140576086,
      "grad_norm": 1.6887339353561401,
      "learning_rate": 9.93650037359689e-07,
      "loss": 1.9569,
      "num_input_tokens_seen": 39405096,
      "step": 38700
    },
    {
      "epoch": 0.7639148668071115,
      "grad_norm": 1.8706046342849731,
      "learning_rate": 9.936172402197682e-07,
      "loss": 1.9536,
      "num_input_tokens_seen": 39507496,
      "step": 38800
    },
    {
      "epoch": 0.7658837195566144,
      "grad_norm": 1.905015230178833,
      "learning_rate": 9.935843591443988e-07,
      "loss": 1.9576,
      "num_input_tokens_seen": 39608984,
      "step": 38900
    },
    {
      "epoch": 0.7678525723061173,
      "grad_norm": 1.635048508644104,
      "learning_rate": 9.935513941391724e-07,
      "loss": 1.983,
      "num_input_tokens_seen": 39711384,
      "step": 39000
    },
    {
      "epoch": 0.7698214250556201,
      "grad_norm": 2.011270761489868,
      "learning_rate": 9.935183452096943e-07,
      "loss": 1.9447,
      "num_input_tokens_seen": 39813312,
      "step": 39100
    },
    {
      "epoch": 0.771790277805123,
      "grad_norm": 2.384460210800171,
      "learning_rate": 9.934852123615838e-07,
      "loss": 1.9493,
      "num_input_tokens_seen": 39915712,
      "step": 39200
    },
    {
      "epoch": 0.7737591305546259,
      "grad_norm": 2.0290377140045166,
      "learning_rate": 9.934519956004753e-07,
      "loss": 2.0189,
      "num_input_tokens_seen": 40017192,
      "step": 39300
    },
    {
      "epoch": 0.7757279833041287,
      "grad_norm": 2.2202324867248535,
      "learning_rate": 9.934186949320172e-07,
      "loss": 1.9848,
      "num_input_tokens_seen": 40118776,
      "step": 39400
    },
    {
      "epoch": 0.7776968360536316,
      "grad_norm": 2.19101881980896,
      "learning_rate": 9.933853103618715e-07,
      "loss": 2.0028,
      "num_input_tokens_seen": 40220000,
      "step": 39500
    },
    {
      "epoch": 0.7796656888031344,
      "grad_norm": 2.1909310817718506,
      "learning_rate": 9.93351841895715e-07,
      "loss": 1.9879,
      "num_input_tokens_seen": 40322400,
      "step": 39600
    },
    {
      "epoch": 0.7816345415526372,
      "grad_norm": 1.8135439157485962,
      "learning_rate": 9.933182895392392e-07,
      "loss": 1.9514,
      "num_input_tokens_seen": 40424096,
      "step": 39700
    },
    {
      "epoch": 0.7836033943021401,
      "grad_norm": 2.112919569015503,
      "learning_rate": 9.93284653298149e-07,
      "loss": 2.0307,
      "num_input_tokens_seen": 40526264,
      "step": 39800
    },
    {
      "epoch": 0.785572247051643,
      "grad_norm": 1.9405063390731812,
      "learning_rate": 9.932509331781641e-07,
      "loss": 1.9797,
      "num_input_tokens_seen": 40628664,
      "step": 39900
    },
    {
      "epoch": 0.7875410998011458,
      "grad_norm": 1.945940375328064,
      "learning_rate": 9.93217129185018e-07,
      "loss": 2.0123,
      "num_input_tokens_seen": 40731064,
      "step": 40000
    },
    {
      "epoch": 0.7895099525506487,
      "grad_norm": 1.8466817140579224,
      "learning_rate": 9.931832413244595e-07,
      "loss": 1.9777,
      "num_input_tokens_seen": 40831424,
      "step": 40100
    },
    {
      "epoch": 0.7914788053001516,
      "grad_norm": 2.042346715927124,
      "learning_rate": 9.931492696022503e-07,
      "loss": 1.9964,
      "num_input_tokens_seen": 40933824,
      "step": 40200
    },
    {
      "epoch": 0.7934476580496544,
      "grad_norm": 1.955869436264038,
      "learning_rate": 9.931152140241673e-07,
      "loss": 1.9658,
      "num_input_tokens_seen": 41035632,
      "step": 40300
    },
    {
      "epoch": 0.7954165107991573,
      "grad_norm": 1.9215490818023682,
      "learning_rate": 9.930810745960012e-07,
      "loss": 1.9834,
      "num_input_tokens_seen": 41138032,
      "step": 40400
    },
    {
      "epoch": 0.7973853635486602,
      "grad_norm": 1.7350525856018066,
      "learning_rate": 9.930468513235573e-07,
      "loss": 2.0175,
      "num_input_tokens_seen": 41239848,
      "step": 40500
    },
    {
      "epoch": 0.7993542162981631,
      "grad_norm": 2.2450482845306396,
      "learning_rate": 9.930125442126546e-07,
      "loss": 1.9772,
      "num_input_tokens_seen": 41341712,
      "step": 40600
    },
    {
      "epoch": 0.8013230690476659,
      "grad_norm": 1.9311119318008423,
      "learning_rate": 9.929781532691275e-07,
      "loss": 1.9549,
      "num_input_tokens_seen": 41442864,
      "step": 40700
    },
    {
      "epoch": 0.8032919217971688,
      "grad_norm": 1.9055299758911133,
      "learning_rate": 9.929436784988229e-07,
      "loss": 2.0071,
      "num_input_tokens_seen": 41544624,
      "step": 40800
    },
    {
      "epoch": 0.8052607745466717,
      "grad_norm": 1.8988803625106812,
      "learning_rate": 9.929091199076038e-07,
      "loss": 2.0175,
      "num_input_tokens_seen": 41645312,
      "step": 40900
    },
    {
      "epoch": 0.8072296272961745,
      "grad_norm": 2.0395917892456055,
      "learning_rate": 9.928744775013462e-07,
      "loss": 1.9737,
      "num_input_tokens_seen": 41745664,
      "step": 41000
    },
    {
      "epoch": 0.8091984800456774,
      "grad_norm": 1.9754116535186768,
      "learning_rate": 9.928397512859405e-07,
      "loss": 1.9739,
      "num_input_tokens_seen": 41847232,
      "step": 41100
    },
    {
      "epoch": 0.8111673327951803,
      "grad_norm": 2.7534637451171875,
      "learning_rate": 9.928049412672923e-07,
      "loss": 1.969,
      "num_input_tokens_seen": 41949632,
      "step": 41200
    },
    {
      "epoch": 0.8131361855446831,
      "grad_norm": 1.875912070274353,
      "learning_rate": 9.9277004745132e-07,
      "loss": 1.9624,
      "num_input_tokens_seen": 42052032,
      "step": 41300
    },
    {
      "epoch": 0.815105038294186,
      "grad_norm": 1.896838903427124,
      "learning_rate": 9.927350698439577e-07,
      "loss": 1.9597,
      "num_input_tokens_seen": 42154432,
      "step": 41400
    },
    {
      "epoch": 0.8170738910436889,
      "grad_norm": 1.8952997922897339,
      "learning_rate": 9.927000084511525e-07,
      "loss": 2.0168,
      "num_input_tokens_seen": 42255664,
      "step": 41500
    },
    {
      "epoch": 0.8190427437931918,
      "grad_norm": 1.8672600984573364,
      "learning_rate": 9.926648632788663e-07,
      "loss": 1.9418,
      "num_input_tokens_seen": 42357248,
      "step": 41600
    },
    {
      "epoch": 0.8210115965426946,
      "grad_norm": 2.009817123413086,
      "learning_rate": 9.926296343330758e-07,
      "loss": 1.9218,
      "num_input_tokens_seen": 42459648,
      "step": 41700
    },
    {
      "epoch": 0.8229804492921975,
      "grad_norm": 1.8217511177062988,
      "learning_rate": 9.925943216197707e-07,
      "loss": 2.0096,
      "num_input_tokens_seen": 42561272,
      "step": 41800
    },
    {
      "epoch": 0.8249493020417002,
      "grad_norm": 1.804321050643921,
      "learning_rate": 9.925589251449561e-07,
      "loss": 2.0198,
      "num_input_tokens_seen": 42663672,
      "step": 41900
    },
    {
      "epoch": 0.8269181547912031,
      "grad_norm": 1.753313422203064,
      "learning_rate": 9.925234449146507e-07,
      "loss": 1.9451,
      "num_input_tokens_seen": 42764968,
      "step": 42000
    },
    {
      "epoch": 0.828887007540706,
      "grad_norm": 1.8761632442474365,
      "learning_rate": 9.924878809348875e-07,
      "loss": 2.0011,
      "num_input_tokens_seen": 42866600,
      "step": 42100
    },
    {
      "epoch": 0.8308558602902089,
      "grad_norm": 1.9971232414245605,
      "learning_rate": 9.924522332117143e-07,
      "loss": 1.9953,
      "num_input_tokens_seen": 42968352,
      "step": 42200
    },
    {
      "epoch": 0.8328247130397117,
      "grad_norm": 1.9807604551315308,
      "learning_rate": 9.92416501751192e-07,
      "loss": 1.964,
      "num_input_tokens_seen": 43070752,
      "step": 42300
    },
    {
      "epoch": 0.8347935657892146,
      "grad_norm": 1.793778419494629,
      "learning_rate": 9.92380686559397e-07,
      "loss": 1.9883,
      "num_input_tokens_seen": 43173152,
      "step": 42400
    },
    {
      "epoch": 0.8367624185387175,
      "grad_norm": 1.7874850034713745,
      "learning_rate": 9.923447876424192e-07,
      "loss": 1.928,
      "num_input_tokens_seen": 43274384,
      "step": 42500
    },
    {
      "epoch": 0.8387312712882203,
      "grad_norm": 1.9487388134002686,
      "learning_rate": 9.923088050063628e-07,
      "loss": 1.9831,
      "num_input_tokens_seen": 43376784,
      "step": 42600
    },
    {
      "epoch": 0.8407001240377232,
      "grad_norm": 3.431633234024048,
      "learning_rate": 9.922727386573465e-07,
      "loss": 1.9724,
      "num_input_tokens_seen": 43479184,
      "step": 42700
    },
    {
      "epoch": 0.8426689767872261,
      "grad_norm": 1.9606448411941528,
      "learning_rate": 9.92236588601503e-07,
      "loss": 1.9748,
      "num_input_tokens_seen": 43580352,
      "step": 42800
    },
    {
      "epoch": 0.844637829536729,
      "grad_norm": 2.18243145942688,
      "learning_rate": 9.922003548449793e-07,
      "loss": 2.0376,
      "num_input_tokens_seen": 43680896,
      "step": 42900
    },
    {
      "epoch": 0.8466066822862318,
      "grad_norm": 1.7839654684066772,
      "learning_rate": 9.921640373939367e-07,
      "loss": 1.9521,
      "num_input_tokens_seen": 43783296,
      "step": 43000
    },
    {
      "epoch": 0.8485755350357347,
      "grad_norm": 1.9653984308242798,
      "learning_rate": 9.921276362545506e-07,
      "loss": 1.9842,
      "num_input_tokens_seen": 43884520,
      "step": 43100
    },
    {
      "epoch": 0.8505443877852376,
      "grad_norm": 6.457769870758057,
      "learning_rate": 9.920911514330108e-07,
      "loss": 1.9927,
      "num_input_tokens_seen": 43986920,
      "step": 43200
    },
    {
      "epoch": 0.8525132405347404,
      "grad_norm": 2.368502616882324,
      "learning_rate": 9.920545829355212e-07,
      "loss": 1.9525,
      "num_input_tokens_seen": 44089320,
      "step": 43300
    },
    {
      "epoch": 0.8544820932842433,
      "grad_norm": 1.9824861288070679,
      "learning_rate": 9.920179307682998e-07,
      "loss": 2.0064,
      "num_input_tokens_seen": 44190960,
      "step": 43400
    },
    {
      "epoch": 0.8564509460337462,
      "grad_norm": 1.8569347858428955,
      "learning_rate": 9.919811949375795e-07,
      "loss": 2.0172,
      "num_input_tokens_seen": 44293360,
      "step": 43500
    },
    {
      "epoch": 0.858419798783249,
      "grad_norm": 1.8042532205581665,
      "learning_rate": 9.919443754496063e-07,
      "loss": 2.0037,
      "num_input_tokens_seen": 44395216,
      "step": 43600
    },
    {
      "epoch": 0.8603886515327519,
      "grad_norm": 2.066889762878418,
      "learning_rate": 9.91907472310641e-07,
      "loss": 1.9806,
      "num_input_tokens_seen": 44496800,
      "step": 43700
    },
    {
      "epoch": 0.8623575042822548,
      "grad_norm": 2.8663549423217773,
      "learning_rate": 9.918704855269595e-07,
      "loss": 2.0107,
      "num_input_tokens_seen": 44598504,
      "step": 43800
    },
    {
      "epoch": 0.8643263570317576,
      "grad_norm": 1.7114691734313965,
      "learning_rate": 9.918334151048504e-07,
      "loss": 1.9491,
      "num_input_tokens_seen": 44700320,
      "step": 43900
    },
    {
      "epoch": 0.8662952097812605,
      "grad_norm": 1.7910566329956055,
      "learning_rate": 9.917962610506173e-07,
      "loss": 1.9544,
      "num_input_tokens_seen": 44802720,
      "step": 44000
    },
    {
      "epoch": 0.8682640625307633,
      "grad_norm": 2.052602767944336,
      "learning_rate": 9.917590233705779e-07,
      "loss": 2.0316,
      "num_input_tokens_seen": 44904528,
      "step": 44100
    },
    {
      "epoch": 0.8702329152802661,
      "grad_norm": 1.9590590000152588,
      "learning_rate": 9.917217020710645e-07,
      "loss": 1.9944,
      "num_input_tokens_seen": 45005728,
      "step": 44200
    },
    {
      "epoch": 0.872201768029769,
      "grad_norm": 1.757790207862854,
      "learning_rate": 9.91684297158423e-07,
      "loss": 2.0238,
      "num_input_tokens_seen": 45108128,
      "step": 44300
    },
    {
      "epoch": 0.8741706207792719,
      "grad_norm": 1.8873578310012817,
      "learning_rate": 9.916468086390137e-07,
      "loss": 2.0045,
      "num_input_tokens_seen": 45209224,
      "step": 44400
    },
    {
      "epoch": 0.8761394735287747,
      "grad_norm": 1.9909436702728271,
      "learning_rate": 9.916092365192115e-07,
      "loss": 1.9804,
      "num_input_tokens_seen": 45311072,
      "step": 44500
    },
    {
      "epoch": 0.8781083262782776,
      "grad_norm": 2.1602702140808105,
      "learning_rate": 9.915715808054049e-07,
      "loss": 2.0682,
      "num_input_tokens_seen": 45410968,
      "step": 44600
    },
    {
      "epoch": 0.8800771790277805,
      "grad_norm": 1.7149327993392944,
      "learning_rate": 9.91533841503997e-07,
      "loss": 1.9653,
      "num_input_tokens_seen": 45513368,
      "step": 44700
    },
    {
      "epoch": 0.8820460317772834,
      "grad_norm": 2.3716652393341064,
      "learning_rate": 9.914960186214055e-07,
      "loss": 1.9645,
      "num_input_tokens_seen": 45615768,
      "step": 44800
    },
    {
      "epoch": 0.8840148845267862,
      "grad_norm": 2.006587266921997,
      "learning_rate": 9.914581121640612e-07,
      "loss": 1.9974,
      "num_input_tokens_seen": 45717544,
      "step": 44900
    },
    {
      "epoch": 0.8859837372762891,
      "grad_norm": 1.9556728601455688,
      "learning_rate": 9.914201221384101e-07,
      "loss": 1.9571,
      "num_input_tokens_seen": 45819128,
      "step": 45000
    },
    {
      "epoch": 0.887952590025792,
      "grad_norm": 2.1543567180633545,
      "learning_rate": 9.913820485509122e-07,
      "loss": 1.9756,
      "num_input_tokens_seen": 45920952,
      "step": 45100
    },
    {
      "epoch": 0.8899214427752948,
      "grad_norm": 1.9320228099822998,
      "learning_rate": 9.913438914080414e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 46023352,
      "step": 45200
    },
    {
      "epoch": 0.8918902955247977,
      "grad_norm": 1.7192232608795166,
      "learning_rate": 9.91305650716286e-07,
      "loss": 2.0042,
      "num_input_tokens_seen": 46124944,
      "step": 45300
    },
    {
      "epoch": 0.8938591482743006,
      "grad_norm": 1.710333228111267,
      "learning_rate": 9.912673264821485e-07,
      "loss": 2.0368,
      "num_input_tokens_seen": 46226720,
      "step": 45400
    },
    {
      "epoch": 0.8958280010238034,
      "grad_norm": 1.861194133758545,
      "learning_rate": 9.91228918712146e-07,
      "loss": 1.9908,
      "num_input_tokens_seen": 46329120,
      "step": 45500
    },
    {
      "epoch": 0.8977968537733063,
      "grad_norm": 2.087844133377075,
      "learning_rate": 9.911904274128087e-07,
      "loss": 2.0278,
      "num_input_tokens_seen": 46430336,
      "step": 45600
    },
    {
      "epoch": 0.8997657065228092,
      "grad_norm": 1.9030890464782715,
      "learning_rate": 9.911518525906822e-07,
      "loss": 1.9771,
      "num_input_tokens_seen": 46532080,
      "step": 45700
    },
    {
      "epoch": 0.901734559272312,
      "grad_norm": 1.9622066020965576,
      "learning_rate": 9.911131942523257e-07,
      "loss": 1.976,
      "num_input_tokens_seen": 46632704,
      "step": 45800
    },
    {
      "epoch": 0.9037034120218149,
      "grad_norm": 1.8562543392181396,
      "learning_rate": 9.91074452404313e-07,
      "loss": 2.0071,
      "num_input_tokens_seen": 46734312,
      "step": 45900
    },
    {
      "epoch": 0.9056722647713178,
      "grad_norm": 1.9798778295516968,
      "learning_rate": 9.910356270532314e-07,
      "loss": 1.9931,
      "num_input_tokens_seen": 46835240,
      "step": 46000
    },
    {
      "epoch": 0.9076411175208207,
      "grad_norm": 2.1422736644744873,
      "learning_rate": 9.90996718205683e-07,
      "loss": 1.9681,
      "num_input_tokens_seen": 46937640,
      "step": 46100
    },
    {
      "epoch": 0.9096099702703235,
      "grad_norm": 2.19767427444458,
      "learning_rate": 9.90957725868284e-07,
      "loss": 1.9566,
      "num_input_tokens_seen": 47040040,
      "step": 46200
    },
    {
      "epoch": 0.9115788230198264,
      "grad_norm": 2.0844364166259766,
      "learning_rate": 9.909186500476645e-07,
      "loss": 1.9897,
      "num_input_tokens_seen": 47141048,
      "step": 46300
    },
    {
      "epoch": 0.9135476757693292,
      "grad_norm": 1.9962643384933472,
      "learning_rate": 9.908794907504693e-07,
      "loss": 1.9685,
      "num_input_tokens_seen": 47243448,
      "step": 46400
    },
    {
      "epoch": 0.915516528518832,
      "grad_norm": 1.8777378797531128,
      "learning_rate": 9.908402479833569e-07,
      "loss": 1.9752,
      "num_input_tokens_seen": 47345848,
      "step": 46500
    },
    {
      "epoch": 0.9174853812683349,
      "grad_norm": 2.2894251346588135,
      "learning_rate": 9.908009217530004e-07,
      "loss": 1.9997,
      "num_input_tokens_seen": 47447496,
      "step": 46600
    },
    {
      "epoch": 0.9194542340178378,
      "grad_norm": 1.9015779495239258,
      "learning_rate": 9.907615120660866e-07,
      "loss": 1.9873,
      "num_input_tokens_seen": 47549320,
      "step": 46700
    },
    {
      "epoch": 0.9214230867673406,
      "grad_norm": 1.8924967050552368,
      "learning_rate": 9.90722018929317e-07,
      "loss": 2.0095,
      "num_input_tokens_seen": 47649912,
      "step": 46800
    },
    {
      "epoch": 0.9233919395168435,
      "grad_norm": 2.0390703678131104,
      "learning_rate": 9.90682442349407e-07,
      "loss": 1.9489,
      "num_input_tokens_seen": 47752312,
      "step": 46900
    },
    {
      "epoch": 0.9253607922663464,
      "grad_norm": 1.8609728813171387,
      "learning_rate": 9.906427823330862e-07,
      "loss": 2.0041,
      "num_input_tokens_seen": 47854272,
      "step": 47000
    },
    {
      "epoch": 0.9273296450158492,
      "grad_norm": 1.909618616104126,
      "learning_rate": 9.906030388870988e-07,
      "loss": 1.9714,
      "num_input_tokens_seen": 47955944,
      "step": 47100
    },
    {
      "epoch": 0.9292984977653521,
      "grad_norm": 1.7780113220214844,
      "learning_rate": 9.905632120182024e-07,
      "loss": 1.9818,
      "num_input_tokens_seen": 48057888,
      "step": 47200
    },
    {
      "epoch": 0.931267350514855,
      "grad_norm": 1.7854924201965332,
      "learning_rate": 9.905233017331695e-07,
      "loss": 1.9912,
      "num_input_tokens_seen": 48158632,
      "step": 47300
    },
    {
      "epoch": 0.9332362032643579,
      "grad_norm": 3.2184438705444336,
      "learning_rate": 9.904833080387862e-07,
      "loss": 2.0233,
      "num_input_tokens_seen": 48261032,
      "step": 47400
    },
    {
      "epoch": 0.9352050560138607,
      "grad_norm": 1.8930283784866333,
      "learning_rate": 9.904432309418534e-07,
      "loss": 1.99,
      "num_input_tokens_seen": 48363432,
      "step": 47500
    },
    {
      "epoch": 0.9371739087633636,
      "grad_norm": 2.1618216037750244,
      "learning_rate": 9.90403070449186e-07,
      "loss": 1.9754,
      "num_input_tokens_seen": 48465832,
      "step": 47600
    },
    {
      "epoch": 0.9391427615128665,
      "grad_norm": 1.9905320405960083,
      "learning_rate": 9.903628265676124e-07,
      "loss": 1.9668,
      "num_input_tokens_seen": 48566944,
      "step": 47700
    },
    {
      "epoch": 0.9411116142623693,
      "grad_norm": 2.0114645957946777,
      "learning_rate": 9.90322499303976e-07,
      "loss": 1.9741,
      "num_input_tokens_seen": 48668720,
      "step": 47800
    },
    {
      "epoch": 0.9430804670118722,
      "grad_norm": 1.9229475259780884,
      "learning_rate": 9.902820886651344e-07,
      "loss": 1.9926,
      "num_input_tokens_seen": 48770432,
      "step": 47900
    },
    {
      "epoch": 0.9450493197613751,
      "grad_norm": 2.2934682369232178,
      "learning_rate": 9.902415946579589e-07,
      "loss": 1.9601,
      "num_input_tokens_seen": 48872832,
      "step": 48000
    },
    {
      "epoch": 0.9470181725108779,
      "grad_norm": 1.8845819234848022,
      "learning_rate": 9.902010172893349e-07,
      "loss": 1.9797,
      "num_input_tokens_seen": 48975232,
      "step": 48100
    },
    {
      "epoch": 0.9489870252603808,
      "grad_norm": 1.9820157289505005,
      "learning_rate": 9.901603565661626e-07,
      "loss": 2.003,
      "num_input_tokens_seen": 49077320,
      "step": 48200
    },
    {
      "epoch": 0.9509558780098837,
      "grad_norm": 1.7189642190933228,
      "learning_rate": 9.901196124953557e-07,
      "loss": 1.9997,
      "num_input_tokens_seen": 49178888,
      "step": 48300
    },
    {
      "epoch": 0.9529247307593866,
      "grad_norm": 1.933113694190979,
      "learning_rate": 9.900787850838428e-07,
      "loss": 1.9996,
      "num_input_tokens_seen": 49281288,
      "step": 48400
    },
    {
      "epoch": 0.9548935835088894,
      "grad_norm": 1.7269684076309204,
      "learning_rate": 9.900378743385659e-07,
      "loss": 2.0065,
      "num_input_tokens_seen": 49383168,
      "step": 48500
    },
    {
      "epoch": 0.9568624362583922,
      "grad_norm": 1.9899290800094604,
      "learning_rate": 9.899968802664816e-07,
      "loss": 1.9805,
      "num_input_tokens_seen": 49484008,
      "step": 48600
    },
    {
      "epoch": 0.958831289007895,
      "grad_norm": 1.9168274402618408,
      "learning_rate": 9.899558028745607e-07,
      "loss": 1.9843,
      "num_input_tokens_seen": 49585552,
      "step": 48700
    },
    {
      "epoch": 0.9608001417573979,
      "grad_norm": 1.8895981311798096,
      "learning_rate": 9.89914642169788e-07,
      "loss": 2.0373,
      "num_input_tokens_seen": 49687952,
      "step": 48800
    },
    {
      "epoch": 0.9627689945069008,
      "grad_norm": 1.8340901136398315,
      "learning_rate": 9.898733981591625e-07,
      "loss": 1.9435,
      "num_input_tokens_seen": 49790352,
      "step": 48900
    },
    {
      "epoch": 0.9647378472564037,
      "grad_norm": 1.5985980033874512,
      "learning_rate": 9.898320708496974e-07,
      "loss": 1.9343,
      "num_input_tokens_seen": 49891936,
      "step": 49000
    },
    {
      "epoch": 0.9667067000059065,
      "grad_norm": 1.835910677909851,
      "learning_rate": 9.8979066024842e-07,
      "loss": 2.025,
      "num_input_tokens_seen": 49994336,
      "step": 49100
    },
    {
      "epoch": 0.9686755527554094,
      "grad_norm": 1.802293062210083,
      "learning_rate": 9.89749166362372e-07,
      "loss": 1.9972,
      "num_input_tokens_seen": 50096256,
      "step": 49200
    },
    {
      "epoch": 0.9706444055049123,
      "grad_norm": 3.131337881088257,
      "learning_rate": 9.89707589198609e-07,
      "loss": 1.9958,
      "num_input_tokens_seen": 50198656,
      "step": 49300
    },
    {
      "epoch": 0.9726132582544151,
      "grad_norm": 1.742756962776184,
      "learning_rate": 9.896659287642008e-07,
      "loss": 1.9598,
      "num_input_tokens_seen": 50301056,
      "step": 49400
    },
    {
      "epoch": 0.974582111003918,
      "grad_norm": 2.2476494312286377,
      "learning_rate": 9.896241850662316e-07,
      "loss": 1.9685,
      "num_input_tokens_seen": 50402888,
      "step": 49500
    },
    {
      "epoch": 0.9765509637534209,
      "grad_norm": 1.8031234741210938,
      "learning_rate": 9.895823581117992e-07,
      "loss": 1.9915,
      "num_input_tokens_seen": 50505288,
      "step": 49600
    },
    {
      "epoch": 0.9785198165029237,
      "grad_norm": 1.8810124397277832,
      "learning_rate": 9.895404479080162e-07,
      "loss": 2.0079,
      "num_input_tokens_seen": 50607688,
      "step": 49700
    },
    {
      "epoch": 0.9804886692524266,
      "grad_norm": 2.075939416885376,
      "learning_rate": 9.894984544620092e-07,
      "loss": 2.0239,
      "num_input_tokens_seen": 50709168,
      "step": 49800
    },
    {
      "epoch": 0.9824575220019295,
      "grad_norm": 1.7710177898406982,
      "learning_rate": 9.894563777809185e-07,
      "loss": 2.003,
      "num_input_tokens_seen": 50809968,
      "step": 49900
    },
    {
      "epoch": 0.9844263747514324,
      "grad_norm": 2.261514186859131,
      "learning_rate": 9.894142178718989e-07,
      "loss": 1.9824,
      "num_input_tokens_seen": 50911664,
      "step": 50000
    },
    {
      "epoch": 0.9863952275009352,
      "grad_norm": 2.0110349655151367,
      "learning_rate": 9.893719747421196e-07,
      "loss": 1.9648,
      "num_input_tokens_seen": 51013200,
      "step": 50100
    },
    {
      "epoch": 0.9883640802504381,
      "grad_norm": 1.8570513725280762,
      "learning_rate": 9.893296483987636e-07,
      "loss": 1.9823,
      "num_input_tokens_seen": 51115104,
      "step": 50200
    },
    {
      "epoch": 0.990332932999941,
      "grad_norm": 1.7586266994476318,
      "learning_rate": 9.89287238849028e-07,
      "loss": 1.9897,
      "num_input_tokens_seen": 51217504,
      "step": 50300
    },
    {
      "epoch": 0.9923017857494438,
      "grad_norm": 2.0321013927459717,
      "learning_rate": 9.892447461001243e-07,
      "loss": 2.0414,
      "num_input_tokens_seen": 51318288,
      "step": 50400
    },
    {
      "epoch": 0.9942706384989467,
      "grad_norm": 2.01430606842041,
      "learning_rate": 9.892021701592778e-07,
      "loss": 1.9811,
      "num_input_tokens_seen": 51420136,
      "step": 50500
    },
    {
      "epoch": 0.9962394912484496,
      "grad_norm": 1.980430245399475,
      "learning_rate": 9.891595110337288e-07,
      "loss": 1.9351,
      "num_input_tokens_seen": 51522536,
      "step": 50600
    },
    {
      "epoch": 0.9982083439979524,
      "grad_norm": 7.373106956481934,
      "learning_rate": 9.891167687307303e-07,
      "loss": 2.0245,
      "num_input_tokens_seen": 51622928,
      "step": 50700
    },
    {
      "epoch": 1.0001771967474553,
      "grad_norm": 1.8478271961212158,
      "learning_rate": 9.89073943257551e-07,
      "loss": 1.9652,
      "num_input_tokens_seen": 51725328,
      "step": 50800
    },
    {
      "epoch": 1.002146049496958,
      "grad_norm": 1.8001004457473755,
      "learning_rate": 9.890310346214726e-07,
      "loss": 1.9659,
      "num_input_tokens_seen": 51827728,
      "step": 50900
    },
    {
      "epoch": 1.004114902246461,
      "grad_norm": 2.1868462562561035,
      "learning_rate": 9.889880428297912e-07,
      "loss": 1.9379,
      "num_input_tokens_seen": 51930128,
      "step": 51000
    },
    {
      "epoch": 1.0060837549959638,
      "grad_norm": 2.034771680831909,
      "learning_rate": 9.889449678898176e-07,
      "loss": 1.9795,
      "num_input_tokens_seen": 52032528,
      "step": 51100
    },
    {
      "epoch": 1.0080526077454668,
      "grad_norm": 2.0096757411956787,
      "learning_rate": 9.889018098088763e-07,
      "loss": 1.9782,
      "num_input_tokens_seen": 52133120,
      "step": 51200
    },
    {
      "epoch": 1.0100214604949695,
      "grad_norm": 2.168287754058838,
      "learning_rate": 9.888585685943057e-07,
      "loss": 1.9832,
      "num_input_tokens_seen": 52235000,
      "step": 51300
    },
    {
      "epoch": 1.0119903132444725,
      "grad_norm": 1.8824621438980103,
      "learning_rate": 9.888152442534587e-07,
      "loss": 1.9645,
      "num_input_tokens_seen": 52337400,
      "step": 51400
    },
    {
      "epoch": 1.0139591659939753,
      "grad_norm": 1.8408710956573486,
      "learning_rate": 9.887718367937023e-07,
      "loss": 2.0113,
      "num_input_tokens_seen": 52438976,
      "step": 51500
    },
    {
      "epoch": 1.0159280187434783,
      "grad_norm": 2.1709954738616943,
      "learning_rate": 9.887283462224176e-07,
      "loss": 1.9937,
      "num_input_tokens_seen": 52541376,
      "step": 51600
    },
    {
      "epoch": 1.017896871492981,
      "grad_norm": 2.017909288406372,
      "learning_rate": 9.886847725469999e-07,
      "loss": 1.9855,
      "num_input_tokens_seen": 52643776,
      "step": 51700
    },
    {
      "epoch": 1.019865724242484,
      "grad_norm": 2.0906968116760254,
      "learning_rate": 9.88641115774858e-07,
      "loss": 1.974,
      "num_input_tokens_seen": 52746176,
      "step": 51800
    },
    {
      "epoch": 1.0218345769919868,
      "grad_norm": 2.0039424896240234,
      "learning_rate": 9.88597375913416e-07,
      "loss": 1.9614,
      "num_input_tokens_seen": 52847936,
      "step": 51900
    },
    {
      "epoch": 1.0238034297414895,
      "grad_norm": 1.7447197437286377,
      "learning_rate": 9.885535529701113e-07,
      "loss": 1.996,
      "num_input_tokens_seen": 52950248,
      "step": 52000
    },
    {
      "epoch": 1.0257722824909925,
      "grad_norm": 1.8214547634124756,
      "learning_rate": 9.885096469523954e-07,
      "loss": 1.9267,
      "num_input_tokens_seen": 53052648,
      "step": 52100
    },
    {
      "epoch": 1.0277411352404953,
      "grad_norm": 2.217625617980957,
      "learning_rate": 9.884656578677344e-07,
      "loss": 1.958,
      "num_input_tokens_seen": 53155048,
      "step": 52200
    },
    {
      "epoch": 1.0297099879899982,
      "grad_norm": 1.7927933931350708,
      "learning_rate": 9.884215857236082e-07,
      "loss": 1.9878,
      "num_input_tokens_seen": 53256672,
      "step": 52300
    },
    {
      "epoch": 1.031678840739501,
      "grad_norm": 1.8615748882293701,
      "learning_rate": 9.883774305275107e-07,
      "loss": 2.0078,
      "num_input_tokens_seen": 53358264,
      "step": 52400
    },
    {
      "epoch": 1.033647693489004,
      "grad_norm": 2.009561777114868,
      "learning_rate": 9.883331922869505e-07,
      "loss": 1.9489,
      "num_input_tokens_seen": 53460664,
      "step": 52500
    },
    {
      "epoch": 1.0356165462385067,
      "grad_norm": 1.8303207159042358,
      "learning_rate": 9.882888710094498e-07,
      "loss": 2.0028,
      "num_input_tokens_seen": 53562144,
      "step": 52600
    },
    {
      "epoch": 1.0375853989880097,
      "grad_norm": 1.85591721534729,
      "learning_rate": 9.882444667025449e-07,
      "loss": 2.0363,
      "num_input_tokens_seen": 53664544,
      "step": 52700
    },
    {
      "epoch": 1.0395542517375125,
      "grad_norm": 1.7349711656570435,
      "learning_rate": 9.881999793737865e-07,
      "loss": 1.9919,
      "num_input_tokens_seen": 53766376,
      "step": 52800
    },
    {
      "epoch": 1.0415231044870155,
      "grad_norm": 1.818564772605896,
      "learning_rate": 9.881554090307393e-07,
      "loss": 1.9611,
      "num_input_tokens_seen": 53868776,
      "step": 52900
    },
    {
      "epoch": 1.0434919572365182,
      "grad_norm": 1.9757059812545776,
      "learning_rate": 9.88110755680982e-07,
      "loss": 1.9805,
      "num_input_tokens_seen": 53971176,
      "step": 53000
    },
    {
      "epoch": 1.0454608099860212,
      "grad_norm": 1.6863590478897095,
      "learning_rate": 9.880660193321078e-07,
      "loss": 1.9512,
      "num_input_tokens_seen": 54073424,
      "step": 53100
    },
    {
      "epoch": 1.047429662735524,
      "grad_norm": 1.8998291492462158,
      "learning_rate": 9.880211999917234e-07,
      "loss": 1.9927,
      "num_input_tokens_seen": 54175104,
      "step": 53200
    },
    {
      "epoch": 1.049398515485027,
      "grad_norm": 2.458225727081299,
      "learning_rate": 9.8797629766745e-07,
      "loss": 1.9625,
      "num_input_tokens_seen": 54276800,
      "step": 53300
    },
    {
      "epoch": 1.0513673682345297,
      "grad_norm": 1.8298643827438354,
      "learning_rate": 9.87931312366923e-07,
      "loss": 1.9786,
      "num_input_tokens_seen": 54378432,
      "step": 53400
    },
    {
      "epoch": 1.0533362209840327,
      "grad_norm": 2.184784412384033,
      "learning_rate": 9.87886244097792e-07,
      "loss": 1.9977,
      "num_input_tokens_seen": 54480152,
      "step": 53500
    },
    {
      "epoch": 1.0553050737335354,
      "grad_norm": 1.840516209602356,
      "learning_rate": 9.8784109286772e-07,
      "loss": 1.9702,
      "num_input_tokens_seen": 54582256,
      "step": 53600
    },
    {
      "epoch": 1.0572739264830384,
      "grad_norm": 1.9477344751358032,
      "learning_rate": 9.87795858684385e-07,
      "loss": 1.9578,
      "num_input_tokens_seen": 54683912,
      "step": 53700
    },
    {
      "epoch": 1.0592427792325412,
      "grad_norm": 1.9940179586410522,
      "learning_rate": 9.87750541555478e-07,
      "loss": 1.9477,
      "num_input_tokens_seen": 54786312,
      "step": 53800
    },
    {
      "epoch": 1.0612116319820442,
      "grad_norm": 1.9116097688674927,
      "learning_rate": 9.877051414887058e-07,
      "loss": 2.022,
      "num_input_tokens_seen": 54888712,
      "step": 53900
    },
    {
      "epoch": 1.063180484731547,
      "grad_norm": 2.0709564685821533,
      "learning_rate": 9.876596584917876e-07,
      "loss": 1.9789,
      "num_input_tokens_seen": 54991112,
      "step": 54000
    },
    {
      "epoch": 1.06514933748105,
      "grad_norm": 1.8915525674819946,
      "learning_rate": 9.876140925724574e-07,
      "loss": 2.003,
      "num_input_tokens_seen": 55092752,
      "step": 54100
    },
    {
      "epoch": 1.0671181902305527,
      "grad_norm": 1.904774785041809,
      "learning_rate": 9.875684437384637e-07,
      "loss": 1.966,
      "num_input_tokens_seen": 55194512,
      "step": 54200
    },
    {
      "epoch": 1.0690870429800556,
      "grad_norm": 1.8894705772399902,
      "learning_rate": 9.875227119975685e-07,
      "loss": 2.0077,
      "num_input_tokens_seen": 55296912,
      "step": 54300
    },
    {
      "epoch": 1.0710558957295584,
      "grad_norm": 1.9152696132659912,
      "learning_rate": 9.87476897357548e-07,
      "loss": 1.9686,
      "num_input_tokens_seen": 55398832,
      "step": 54400
    },
    {
      "epoch": 1.0730247484790612,
      "grad_norm": 2.045538902282715,
      "learning_rate": 9.874309998261927e-07,
      "loss": 2.0322,
      "num_input_tokens_seen": 55501232,
      "step": 54500
    },
    {
      "epoch": 1.0749936012285641,
      "grad_norm": 2.1518824100494385,
      "learning_rate": 9.873850194113072e-07,
      "loss": 2.0826,
      "num_input_tokens_seen": 55602496,
      "step": 54600
    },
    {
      "epoch": 1.076962453978067,
      "grad_norm": 2.073474168777466,
      "learning_rate": 9.873389561207097e-07,
      "loss": 1.9722,
      "num_input_tokens_seen": 55703048,
      "step": 54700
    },
    {
      "epoch": 1.0789313067275699,
      "grad_norm": 1.8059463500976562,
      "learning_rate": 9.872928099622334e-07,
      "loss": 1.9814,
      "num_input_tokens_seen": 55804592,
      "step": 54800
    },
    {
      "epoch": 1.0809001594770726,
      "grad_norm": 2.14548397064209,
      "learning_rate": 9.872465809437245e-07,
      "loss": 1.9392,
      "num_input_tokens_seen": 55906248,
      "step": 54900
    },
    {
      "epoch": 1.0828690122265756,
      "grad_norm": 1.8822154998779297,
      "learning_rate": 9.872002690730444e-07,
      "loss": 2.0209,
      "num_input_tokens_seen": 56007496,
      "step": 55000
    },
    {
      "epoch": 1.0848378649760784,
      "grad_norm": 1.85554039478302,
      "learning_rate": 9.871538743580675e-07,
      "loss": 1.986,
      "num_input_tokens_seen": 56109896,
      "step": 55100
    },
    {
      "epoch": 1.0868067177255814,
      "grad_norm": 1.8365095853805542,
      "learning_rate": 9.871073968066832e-07,
      "loss": 1.9862,
      "num_input_tokens_seen": 56212296,
      "step": 55200
    },
    {
      "epoch": 1.088775570475084,
      "grad_norm": 2.052393913269043,
      "learning_rate": 9.870608364267946e-07,
      "loss": 1.9599,
      "num_input_tokens_seen": 56314696,
      "step": 55300
    },
    {
      "epoch": 1.090744423224587,
      "grad_norm": 1.9801706075668335,
      "learning_rate": 9.87014193226319e-07,
      "loss": 2.0252,
      "num_input_tokens_seen": 56417096,
      "step": 55400
    },
    {
      "epoch": 1.0927132759740898,
      "grad_norm": 1.9569658041000366,
      "learning_rate": 9.86967467213187e-07,
      "loss": 1.9421,
      "num_input_tokens_seen": 56519496,
      "step": 55500
    },
    {
      "epoch": 1.0946821287235928,
      "grad_norm": 1.944030523300171,
      "learning_rate": 9.869206583953448e-07,
      "loss": 1.9723,
      "num_input_tokens_seen": 56621896,
      "step": 55600
    },
    {
      "epoch": 1.0966509814730956,
      "grad_norm": 2.0947165489196777,
      "learning_rate": 9.868737667807517e-07,
      "loss": 2.0142,
      "num_input_tokens_seen": 56724296,
      "step": 55700
    },
    {
      "epoch": 1.0986198342225986,
      "grad_norm": 1.771150827407837,
      "learning_rate": 9.868267923773807e-07,
      "loss": 1.9717,
      "num_input_tokens_seen": 56826384,
      "step": 55800
    },
    {
      "epoch": 1.1005886869721013,
      "grad_norm": 1.905934453010559,
      "learning_rate": 9.8677973519322e-07,
      "loss": 1.9431,
      "num_input_tokens_seen": 56928784,
      "step": 55900
    },
    {
      "epoch": 1.1025575397216043,
      "grad_norm": 2.3866124153137207,
      "learning_rate": 9.86732595236271e-07,
      "loss": 2.0366,
      "num_input_tokens_seen": 57031184,
      "step": 56000
    },
    {
      "epoch": 1.104526392471107,
      "grad_norm": 1.8602744340896606,
      "learning_rate": 9.866853725145491e-07,
      "loss": 2.0002,
      "num_input_tokens_seen": 57133096,
      "step": 56100
    },
    {
      "epoch": 1.10649524522061,
      "grad_norm": 2.094759941101074,
      "learning_rate": 9.866380670360848e-07,
      "loss": 1.9963,
      "num_input_tokens_seen": 57233904,
      "step": 56200
    },
    {
      "epoch": 1.1084640979701128,
      "grad_norm": 1.7452400922775269,
      "learning_rate": 9.865906788089218e-07,
      "loss": 2.0254,
      "num_input_tokens_seen": 57336304,
      "step": 56300
    },
    {
      "epoch": 1.1104329507196158,
      "grad_norm": 3.6265110969543457,
      "learning_rate": 9.865432078411178e-07,
      "loss": 1.9778,
      "num_input_tokens_seen": 57438096,
      "step": 56400
    },
    {
      "epoch": 1.1124018034691185,
      "grad_norm": 2.027275562286377,
      "learning_rate": 9.86495654140745e-07,
      "loss": 2.0114,
      "num_input_tokens_seen": 57539928,
      "step": 56500
    },
    {
      "epoch": 1.1143706562186213,
      "grad_norm": 1.9589576721191406,
      "learning_rate": 9.864480177158896e-07,
      "loss": 1.9805,
      "num_input_tokens_seen": 57641136,
      "step": 56600
    },
    {
      "epoch": 1.1163395089681243,
      "grad_norm": 2.1725780963897705,
      "learning_rate": 9.864002985746515e-07,
      "loss": 1.9595,
      "num_input_tokens_seen": 57743536,
      "step": 56700
    },
    {
      "epoch": 1.118308361717627,
      "grad_norm": 2.1328485012054443,
      "learning_rate": 9.863524967251454e-07,
      "loss": 2.0062,
      "num_input_tokens_seen": 57845104,
      "step": 56800
    },
    {
      "epoch": 1.12027721446713,
      "grad_norm": 1.820858120918274,
      "learning_rate": 9.86304612175499e-07,
      "loss": 1.9794,
      "num_input_tokens_seen": 57946904,
      "step": 56900
    },
    {
      "epoch": 1.1222460672166328,
      "grad_norm": 2.196723222732544,
      "learning_rate": 9.862566449338551e-07,
      "loss": 2.0205,
      "num_input_tokens_seen": 58048648,
      "step": 57000
    },
    {
      "epoch": 1.1242149199661358,
      "grad_norm": 2.0164146423339844,
      "learning_rate": 9.8620859500837e-07,
      "loss": 2.025,
      "num_input_tokens_seen": 58151048,
      "step": 57100
    },
    {
      "epoch": 1.1261837727156385,
      "grad_norm": 2.214961528778076,
      "learning_rate": 9.861604624072144e-07,
      "loss": 2.0139,
      "num_input_tokens_seen": 58252152,
      "step": 57200
    },
    {
      "epoch": 1.1281526254651415,
      "grad_norm": 2.091853380203247,
      "learning_rate": 9.861122471385725e-07,
      "loss": 1.9569,
      "num_input_tokens_seen": 58354552,
      "step": 57300
    },
    {
      "epoch": 1.1301214782146443,
      "grad_norm": 1.9607510566711426,
      "learning_rate": 9.86063949210643e-07,
      "loss": 1.986,
      "num_input_tokens_seen": 58456952,
      "step": 57400
    },
    {
      "epoch": 1.1320903309641472,
      "grad_norm": 2.089627265930176,
      "learning_rate": 9.860155686316385e-07,
      "loss": 2.0413,
      "num_input_tokens_seen": 58559352,
      "step": 57500
    },
    {
      "epoch": 1.13405918371365,
      "grad_norm": 2.056424140930176,
      "learning_rate": 9.85967105409786e-07,
      "loss": 2.0551,
      "num_input_tokens_seen": 58660832,
      "step": 57600
    },
    {
      "epoch": 1.136028036463153,
      "grad_norm": 2.1005749702453613,
      "learning_rate": 9.85918559553326e-07,
      "loss": 2.0195,
      "num_input_tokens_seen": 58763232,
      "step": 57700
    },
    {
      "epoch": 1.1379968892126557,
      "grad_norm": 2.014408826828003,
      "learning_rate": 9.858699310705132e-07,
      "loss": 2.0333,
      "num_input_tokens_seen": 58863984,
      "step": 57800
    },
    {
      "epoch": 1.1399657419621587,
      "grad_norm": 1.7907952070236206,
      "learning_rate": 9.858212199696168e-07,
      "loss": 1.9875,
      "num_input_tokens_seen": 58965472,
      "step": 57900
    },
    {
      "epoch": 1.1419345947116615,
      "grad_norm": 1.7765934467315674,
      "learning_rate": 9.857724262589196e-07,
      "loss": 1.9172,
      "num_input_tokens_seen": 59067872,
      "step": 58000
    },
    {
      "epoch": 1.1439034474611645,
      "grad_norm": 2.1868581771850586,
      "learning_rate": 9.857235499467183e-07,
      "loss": 1.9401,
      "num_input_tokens_seen": 59170272,
      "step": 58100
    },
    {
      "epoch": 1.1458723002106672,
      "grad_norm": 2.226532220840454,
      "learning_rate": 9.856745910413244e-07,
      "loss": 1.9258,
      "num_input_tokens_seen": 59272672,
      "step": 58200
    },
    {
      "epoch": 1.1478411529601702,
      "grad_norm": 1.9788908958435059,
      "learning_rate": 9.856255495510624e-07,
      "loss": 2.0125,
      "num_input_tokens_seen": 59375072,
      "step": 58300
    },
    {
      "epoch": 1.149810005709673,
      "grad_norm": 2.0415232181549072,
      "learning_rate": 9.855764254842717e-07,
      "loss": 1.9584,
      "num_input_tokens_seen": 59477472,
      "step": 58400
    },
    {
      "epoch": 1.151778858459176,
      "grad_norm": 1.9657243490219116,
      "learning_rate": 9.855272188493055e-07,
      "loss": 1.9672,
      "num_input_tokens_seen": 59579064,
      "step": 58500
    },
    {
      "epoch": 1.1537477112086787,
      "grad_norm": 2.114288091659546,
      "learning_rate": 9.854779296545307e-07,
      "loss": 1.9999,
      "num_input_tokens_seen": 59680728,
      "step": 58600
    },
    {
      "epoch": 1.1557165639581815,
      "grad_norm": 2.0081071853637695,
      "learning_rate": 9.854285579083288e-07,
      "loss": 2.0145,
      "num_input_tokens_seen": 59783128,
      "step": 58700
    },
    {
      "epoch": 1.1576854167076844,
      "grad_norm": 1.8314753770828247,
      "learning_rate": 9.853791036190951e-07,
      "loss": 1.9645,
      "num_input_tokens_seen": 59884232,
      "step": 58800
    },
    {
      "epoch": 1.1596542694571874,
      "grad_norm": 1.7582591772079468,
      "learning_rate": 9.853295667952384e-07,
      "loss": 1.9537,
      "num_input_tokens_seen": 59986632,
      "step": 58900
    },
    {
      "epoch": 1.1616231222066902,
      "grad_norm": 2.2304999828338623,
      "learning_rate": 9.852799474451824e-07,
      "loss": 2.0467,
      "num_input_tokens_seen": 60085184,
      "step": 59000
    },
    {
      "epoch": 1.163591974956193,
      "grad_norm": 1.8509920835494995,
      "learning_rate": 9.852302455773644e-07,
      "loss": 1.9878,
      "num_input_tokens_seen": 60187240,
      "step": 59100
    },
    {
      "epoch": 1.165560827705696,
      "grad_norm": 1.9994957447052002,
      "learning_rate": 9.851804612002357e-07,
      "loss": 2.027,
      "num_input_tokens_seen": 60289048,
      "step": 59200
    },
    {
      "epoch": 1.1675296804551987,
      "grad_norm": 1.8670504093170166,
      "learning_rate": 9.851305943222619e-07,
      "loss": 1.9659,
      "num_input_tokens_seen": 60391448,
      "step": 59300
    },
    {
      "epoch": 1.1694985332047017,
      "grad_norm": 1.867996335029602,
      "learning_rate": 9.850806449519223e-07,
      "loss": 1.9666,
      "num_input_tokens_seen": 60493848,
      "step": 59400
    },
    {
      "epoch": 1.1714673859542044,
      "grad_norm": 2.133732318878174,
      "learning_rate": 9.850306130977102e-07,
      "loss": 1.9786,
      "num_input_tokens_seen": 60594760,
      "step": 59500
    },
    {
      "epoch": 1.1734362387037074,
      "grad_norm": 1.793951392173767,
      "learning_rate": 9.849804987681334e-07,
      "loss": 1.9985,
      "num_input_tokens_seen": 60696576,
      "step": 59600
    },
    {
      "epoch": 1.1754050914532101,
      "grad_norm": 1.9974703788757324,
      "learning_rate": 9.849303019717134e-07,
      "loss": 2.0163,
      "num_input_tokens_seen": 60798120,
      "step": 59700
    },
    {
      "epoch": 1.1773739442027131,
      "grad_norm": 1.9553022384643555,
      "learning_rate": 9.848800227169855e-07,
      "loss": 1.9601,
      "num_input_tokens_seen": 60900520,
      "step": 59800
    },
    {
      "epoch": 1.1793427969522159,
      "grad_norm": 2.0108377933502197,
      "learning_rate": 9.848296610124995e-07,
      "loss": 1.9743,
      "num_input_tokens_seen": 61002920,
      "step": 59900
    },
    {
      "epoch": 1.1813116497017189,
      "grad_norm": 1.959549069404602,
      "learning_rate": 9.847792168668189e-07,
      "loss": 2.024,
      "num_input_tokens_seen": 61103056,
      "step": 60000
    },
    {
      "epoch": 1.1832805024512216,
      "grad_norm": 2.4373843669891357,
      "learning_rate": 9.847286902885212e-07,
      "loss": 1.9976,
      "num_input_tokens_seen": 61205456,
      "step": 60100
    },
    {
      "epoch": 1.1852493552007246,
      "grad_norm": 2.2283754348754883,
      "learning_rate": 9.846780812861981e-07,
      "loss": 1.9978,
      "num_input_tokens_seen": 61307856,
      "step": 60200
    },
    {
      "epoch": 1.1872182079502274,
      "grad_norm": 1.4838213920593262,
      "learning_rate": 9.846273898684555e-07,
      "loss": 2.0157,
      "num_input_tokens_seen": 61409440,
      "step": 60300
    },
    {
      "epoch": 1.1891870606997303,
      "grad_norm": 1.7456793785095215,
      "learning_rate": 9.845766160439125e-07,
      "loss": 2.0098,
      "num_input_tokens_seen": 61509456,
      "step": 60400
    },
    {
      "epoch": 1.191155913449233,
      "grad_norm": 1.8974699974060059,
      "learning_rate": 9.845257598212031e-07,
      "loss": 2.0143,
      "num_input_tokens_seen": 61610824,
      "step": 60500
    },
    {
      "epoch": 1.193124766198736,
      "grad_norm": 1.9952293634414673,
      "learning_rate": 9.844748212089748e-07,
      "loss": 1.9789,
      "num_input_tokens_seen": 61712376,
      "step": 60600
    },
    {
      "epoch": 1.1950936189482388,
      "grad_norm": 2.1045644283294678,
      "learning_rate": 9.844238002158896e-07,
      "loss": 2.0057,
      "num_input_tokens_seen": 61813528,
      "step": 60700
    },
    {
      "epoch": 1.1970624716977416,
      "grad_norm": 1.774985909461975,
      "learning_rate": 9.84372696850623e-07,
      "loss": 2.0387,
      "num_input_tokens_seen": 61915296,
      "step": 60800
    },
    {
      "epoch": 1.1990313244472446,
      "grad_norm": 2.020932912826538,
      "learning_rate": 9.843215111218646e-07,
      "loss": 1.9788,
      "num_input_tokens_seen": 62017384,
      "step": 60900
    },
    {
      "epoch": 1.2010001771967476,
      "grad_norm": 1.9226939678192139,
      "learning_rate": 9.842702430383183e-07,
      "loss": 1.9791,
      "num_input_tokens_seen": 62119056,
      "step": 61000
    },
    {
      "epoch": 1.2029690299462503,
      "grad_norm": 1.7677568197250366,
      "learning_rate": 9.842188926087016e-07,
      "loss": 1.965,
      "num_input_tokens_seen": 62220832,
      "step": 61100
    },
    {
      "epoch": 1.204937882695753,
      "grad_norm": 2.3421859741210938,
      "learning_rate": 9.841674598417462e-07,
      "loss": 1.9089,
      "num_input_tokens_seen": 62322504,
      "step": 61200
    },
    {
      "epoch": 1.206906735445256,
      "grad_norm": 1.7903199195861816,
      "learning_rate": 9.841159447461982e-07,
      "loss": 1.9467,
      "num_input_tokens_seen": 62423800,
      "step": 61300
    },
    {
      "epoch": 1.2088755881947588,
      "grad_norm": 4.110795974731445,
      "learning_rate": 9.840643473308167e-07,
      "loss": 1.9633,
      "num_input_tokens_seen": 62526200,
      "step": 61400
    },
    {
      "epoch": 1.2108444409442618,
      "grad_norm": 1.9007803201675415,
      "learning_rate": 9.84012667604376e-07,
      "loss": 1.9583,
      "num_input_tokens_seen": 62626904,
      "step": 61500
    },
    {
      "epoch": 1.2128132936937646,
      "grad_norm": 1.7440636157989502,
      "learning_rate": 9.839609055756633e-07,
      "loss": 1.9895,
      "num_input_tokens_seen": 62727760,
      "step": 61600
    },
    {
      "epoch": 1.2147821464432675,
      "grad_norm": 1.9405783414840698,
      "learning_rate": 9.839090612534807e-07,
      "loss": 2.0228,
      "num_input_tokens_seen": 62830160,
      "step": 61700
    },
    {
      "epoch": 1.2167509991927703,
      "grad_norm": 1.877098560333252,
      "learning_rate": 9.838571346466435e-07,
      "loss": 2.0125,
      "num_input_tokens_seen": 62931632,
      "step": 61800
    },
    {
      "epoch": 1.2187198519422733,
      "grad_norm": 1.8819992542266846,
      "learning_rate": 9.838051257639816e-07,
      "loss": 1.9714,
      "num_input_tokens_seen": 63033176,
      "step": 61900
    },
    {
      "epoch": 1.220688704691776,
      "grad_norm": 2.1795432567596436,
      "learning_rate": 9.83753034614339e-07,
      "loss": 2.0194,
      "num_input_tokens_seen": 63135576,
      "step": 62000
    },
    {
      "epoch": 1.222657557441279,
      "grad_norm": 1.9468737840652466,
      "learning_rate": 9.837008612065728e-07,
      "loss": 1.9567,
      "num_input_tokens_seen": 63237376,
      "step": 62100
    },
    {
      "epoch": 1.2246264101907818,
      "grad_norm": 1.988161563873291,
      "learning_rate": 9.836486055495548e-07,
      "loss": 1.9964,
      "num_input_tokens_seen": 63339776,
      "step": 62200
    },
    {
      "epoch": 1.2265952629402848,
      "grad_norm": 1.9543042182922363,
      "learning_rate": 9.835962676521708e-07,
      "loss": 1.9864,
      "num_input_tokens_seen": 63439912,
      "step": 62300
    },
    {
      "epoch": 1.2285641156897875,
      "grad_norm": 1.9151384830474854,
      "learning_rate": 9.835438475233204e-07,
      "loss": 2.0018,
      "num_input_tokens_seen": 63541744,
      "step": 62400
    },
    {
      "epoch": 1.2305329684392905,
      "grad_norm": 1.9542925357818604,
      "learning_rate": 9.834913451719168e-07,
      "loss": 2.0296,
      "num_input_tokens_seen": 63643704,
      "step": 62500
    },
    {
      "epoch": 1.2325018211887933,
      "grad_norm": 1.7564761638641357,
      "learning_rate": 9.834387606068883e-07,
      "loss": 1.9982,
      "num_input_tokens_seen": 63745576,
      "step": 62600
    },
    {
      "epoch": 1.2344706739382962,
      "grad_norm": 1.9793672561645508,
      "learning_rate": 9.833860938371758e-07,
      "loss": 1.9754,
      "num_input_tokens_seen": 63847888,
      "step": 62700
    },
    {
      "epoch": 1.236439526687799,
      "grad_norm": 1.9184461832046509,
      "learning_rate": 9.833333448717354e-07,
      "loss": 2.0246,
      "num_input_tokens_seen": 63949704,
      "step": 62800
    },
    {
      "epoch": 1.238408379437302,
      "grad_norm": 1.7936235666275024,
      "learning_rate": 9.832805137195362e-07,
      "loss": 1.9625,
      "num_input_tokens_seen": 64051544,
      "step": 62900
    },
    {
      "epoch": 1.2403772321868047,
      "grad_norm": 1.685490369796753,
      "learning_rate": 9.832276003895616e-07,
      "loss": 1.9919,
      "num_input_tokens_seen": 64153944,
      "step": 63000
    },
    {
      "epoch": 1.2423460849363077,
      "grad_norm": 1.8245015144348145,
      "learning_rate": 9.831746048908095e-07,
      "loss": 1.9627,
      "num_input_tokens_seen": 64255584,
      "step": 63100
    },
    {
      "epoch": 1.2443149376858105,
      "grad_norm": 3.798107624053955,
      "learning_rate": 9.83121527232291e-07,
      "loss": 1.9801,
      "num_input_tokens_seen": 64356992,
      "step": 63200
    },
    {
      "epoch": 1.2462837904353132,
      "grad_norm": 1.8981205224990845,
      "learning_rate": 9.830683674230318e-07,
      "loss": 2.0256,
      "num_input_tokens_seen": 64456856,
      "step": 63300
    },
    {
      "epoch": 1.2482526431848162,
      "grad_norm": 1.7309049367904663,
      "learning_rate": 9.83015125472071e-07,
      "loss": 1.9839,
      "num_input_tokens_seen": 64559256,
      "step": 63400
    },
    {
      "epoch": 1.2502214959343192,
      "grad_norm": 1.970591425895691,
      "learning_rate": 9.82961801388462e-07,
      "loss": 2.0219,
      "num_input_tokens_seen": 64661656,
      "step": 63500
    },
    {
      "epoch": 1.252190348683822,
      "grad_norm": 2.072432279586792,
      "learning_rate": 9.829083951812723e-07,
      "loss": 2.0168,
      "num_input_tokens_seen": 64763960,
      "step": 63600
    },
    {
      "epoch": 1.2541592014333247,
      "grad_norm": 1.9331374168395996,
      "learning_rate": 9.828549068595829e-07,
      "loss": 1.9511,
      "num_input_tokens_seen": 64865648,
      "step": 63700
    },
    {
      "epoch": 1.2561280541828277,
      "grad_norm": 1.9606348276138306,
      "learning_rate": 9.828013364324892e-07,
      "loss": 1.9926,
      "num_input_tokens_seen": 64967240,
      "step": 63800
    },
    {
      "epoch": 1.2580969069323304,
      "grad_norm": 1.8813121318817139,
      "learning_rate": 9.827476839091006e-07,
      "loss": 1.9805,
      "num_input_tokens_seen": 65068880,
      "step": 63900
    },
    {
      "epoch": 1.2600657596818334,
      "grad_norm": 1.8504828214645386,
      "learning_rate": 9.826939492985398e-07,
      "loss": 1.9444,
      "num_input_tokens_seen": 65171280,
      "step": 64000
    },
    {
      "epoch": 1.2620346124313362,
      "grad_norm": 2.1883697509765625,
      "learning_rate": 9.826401326099442e-07,
      "loss": 1.9323,
      "num_input_tokens_seen": 65273680,
      "step": 64100
    },
    {
      "epoch": 1.2640034651808392,
      "grad_norm": 2.226407527923584,
      "learning_rate": 9.825862338524648e-07,
      "loss": 2.0457,
      "num_input_tokens_seen": 65375520,
      "step": 64200
    },
    {
      "epoch": 1.265972317930342,
      "grad_norm": 2.0757224559783936,
      "learning_rate": 9.825322530352666e-07,
      "loss": 1.9929,
      "num_input_tokens_seen": 65477072,
      "step": 64300
    },
    {
      "epoch": 1.267941170679845,
      "grad_norm": 2.087914228439331,
      "learning_rate": 9.824781901675287e-07,
      "loss": 1.9284,
      "num_input_tokens_seen": 65578792,
      "step": 64400
    },
    {
      "epoch": 1.2699100234293477,
      "grad_norm": 2.0337281227111816,
      "learning_rate": 9.82424045258444e-07,
      "loss": 1.9872,
      "num_input_tokens_seen": 65681192,
      "step": 64500
    },
    {
      "epoch": 1.2718788761788506,
      "grad_norm": 2.0307414531707764,
      "learning_rate": 9.823698183172192e-07,
      "loss": 2.009,
      "num_input_tokens_seen": 65782744,
      "step": 64600
    },
    {
      "epoch": 1.2738477289283534,
      "grad_norm": 1.97044038772583,
      "learning_rate": 9.823155093530755e-07,
      "loss": 1.9632,
      "num_input_tokens_seen": 65885144,
      "step": 64700
    },
    {
      "epoch": 1.2758165816778564,
      "grad_norm": 2.425098419189453,
      "learning_rate": 9.822611183752473e-07,
      "loss": 1.9498,
      "num_input_tokens_seen": 65987544,
      "step": 64800
    },
    {
      "epoch": 1.2777854344273591,
      "grad_norm": 1.8895459175109863,
      "learning_rate": 9.822066453929839e-07,
      "loss": 2.0062,
      "num_input_tokens_seen": 66089416,
      "step": 64900
    },
    {
      "epoch": 1.279754287176862,
      "grad_norm": 2.0372626781463623,
      "learning_rate": 9.821520904155471e-07,
      "loss": 1.9604,
      "num_input_tokens_seen": 66191056,
      "step": 65000
    },
    {
      "epoch": 1.2817231399263649,
      "grad_norm": 2.016897439956665,
      "learning_rate": 9.820974534522142e-07,
      "loss": 1.9932,
      "num_input_tokens_seen": 66293456,
      "step": 65100
    },
    {
      "epoch": 1.2836919926758679,
      "grad_norm": 1.8881981372833252,
      "learning_rate": 9.820427345122755e-07,
      "loss": 1.9576,
      "num_input_tokens_seen": 66394368,
      "step": 65200
    },
    {
      "epoch": 1.2856608454253706,
      "grad_norm": 2.034299612045288,
      "learning_rate": 9.819879336050355e-07,
      "loss": 1.9723,
      "num_input_tokens_seen": 66496664,
      "step": 65300
    },
    {
      "epoch": 1.2876296981748734,
      "grad_norm": 2.127408981323242,
      "learning_rate": 9.819330507398127e-07,
      "loss": 1.9823,
      "num_input_tokens_seen": 66598288,
      "step": 65400
    },
    {
      "epoch": 1.2895985509243764,
      "grad_norm": 2.232555866241455,
      "learning_rate": 9.818780859259395e-07,
      "loss": 1.9796,
      "num_input_tokens_seen": 66700688,
      "step": 65500
    },
    {
      "epoch": 1.2915674036738793,
      "grad_norm": 1.929018259048462,
      "learning_rate": 9.81823039172762e-07,
      "loss": 2.0197,
      "num_input_tokens_seen": 66802256,
      "step": 65600
    },
    {
      "epoch": 1.293536256423382,
      "grad_norm": 2.0691864490509033,
      "learning_rate": 9.817679104896407e-07,
      "loss": 1.9878,
      "num_input_tokens_seen": 66904656,
      "step": 65700
    },
    {
      "epoch": 1.2955051091728849,
      "grad_norm": 2.009652853012085,
      "learning_rate": 9.817126998859493e-07,
      "loss": 1.9737,
      "num_input_tokens_seen": 67005536,
      "step": 65800
    },
    {
      "epoch": 1.2974739619223878,
      "grad_norm": 1.7023661136627197,
      "learning_rate": 9.816574073710767e-07,
      "loss": 1.9983,
      "num_input_tokens_seen": 67105736,
      "step": 65900
    },
    {
      "epoch": 1.2994428146718908,
      "grad_norm": 1.9258289337158203,
      "learning_rate": 9.81602032954424e-07,
      "loss": 1.9838,
      "num_input_tokens_seen": 67208136,
      "step": 66000
    },
    {
      "epoch": 1.3014116674213936,
      "grad_norm": 1.9641227722167969,
      "learning_rate": 9.81546576645408e-07,
      "loss": 1.9792,
      "num_input_tokens_seen": 67310536,
      "step": 66100
    },
    {
      "epoch": 1.3033805201708963,
      "grad_norm": 1.8868162631988525,
      "learning_rate": 9.81491038453458e-07,
      "loss": 1.9768,
      "num_input_tokens_seen": 67412160,
      "step": 66200
    },
    {
      "epoch": 1.3053493729203993,
      "grad_norm": 1.6703177690505981,
      "learning_rate": 9.814354183880182e-07,
      "loss": 1.9902,
      "num_input_tokens_seen": 67513384,
      "step": 66300
    },
    {
      "epoch": 1.307318225669902,
      "grad_norm": 1.9915131330490112,
      "learning_rate": 9.81379716458546e-07,
      "loss": 1.979,
      "num_input_tokens_seen": 67614232,
      "step": 66400
    },
    {
      "epoch": 1.309287078419405,
      "grad_norm": 1.8469550609588623,
      "learning_rate": 9.813239326745131e-07,
      "loss": 2.0128,
      "num_input_tokens_seen": 67716632,
      "step": 66500
    },
    {
      "epoch": 1.3112559311689078,
      "grad_norm": 1.702362298965454,
      "learning_rate": 9.812680670454051e-07,
      "loss": 2.0334,
      "num_input_tokens_seen": 67818328,
      "step": 66600
    },
    {
      "epoch": 1.3132247839184108,
      "grad_norm": 2.065275192260742,
      "learning_rate": 9.812121195807216e-07,
      "loss": 1.9849,
      "num_input_tokens_seen": 67919528,
      "step": 66700
    },
    {
      "epoch": 1.3151936366679136,
      "grad_norm": 1.8248300552368164,
      "learning_rate": 9.811560902899758e-07,
      "loss": 2.0113,
      "num_input_tokens_seen": 68021928,
      "step": 66800
    },
    {
      "epoch": 1.3171624894174165,
      "grad_norm": 1.6874494552612305,
      "learning_rate": 9.810999791826953e-07,
      "loss": 1.9536,
      "num_input_tokens_seen": 68124328,
      "step": 66900
    },
    {
      "epoch": 1.3191313421669193,
      "grad_norm": 1.9437017440795898,
      "learning_rate": 9.81043786268421e-07,
      "loss": 1.9535,
      "num_input_tokens_seen": 68225256,
      "step": 67000
    },
    {
      "epoch": 1.3211001949164223,
      "grad_norm": 1.8744845390319824,
      "learning_rate": 9.80987511556708e-07,
      "loss": 1.9658,
      "num_input_tokens_seen": 68326312,
      "step": 67100
    },
    {
      "epoch": 1.323069047665925,
      "grad_norm": 1.9006444215774536,
      "learning_rate": 9.809311550571259e-07,
      "loss": 1.9966,
      "num_input_tokens_seen": 68428712,
      "step": 67200
    },
    {
      "epoch": 1.325037900415428,
      "grad_norm": 2.078761339187622,
      "learning_rate": 9.808747167792567e-07,
      "loss": 1.983,
      "num_input_tokens_seen": 68529312,
      "step": 67300
    },
    {
      "epoch": 1.3270067531649308,
      "grad_norm": 2.2462334632873535,
      "learning_rate": 9.808181967326983e-07,
      "loss": 1.9474,
      "num_input_tokens_seen": 68631264,
      "step": 67400
    },
    {
      "epoch": 1.3289756059144335,
      "grad_norm": 2.1578781604766846,
      "learning_rate": 9.807615949270606e-07,
      "loss": 1.9404,
      "num_input_tokens_seen": 68733664,
      "step": 67500
    },
    {
      "epoch": 1.3309444586639365,
      "grad_norm": 1.954037070274353,
      "learning_rate": 9.807049113719687e-07,
      "loss": 1.9482,
      "num_input_tokens_seen": 68835272,
      "step": 67600
    },
    {
      "epoch": 1.3329133114134395,
      "grad_norm": 2.041626453399658,
      "learning_rate": 9.806481460770613e-07,
      "loss": 1.9334,
      "num_input_tokens_seen": 68937672,
      "step": 67700
    },
    {
      "epoch": 1.3348821641629423,
      "grad_norm": 1.780411720275879,
      "learning_rate": 9.805912990519903e-07,
      "loss": 1.95,
      "num_input_tokens_seen": 69039592,
      "step": 67800
    },
    {
      "epoch": 1.336851016912445,
      "grad_norm": 7.53485631942749,
      "learning_rate": 9.805343703064226e-07,
      "loss": 1.97,
      "num_input_tokens_seen": 69141136,
      "step": 67900
    },
    {
      "epoch": 1.338819869661948,
      "grad_norm": 1.8268909454345703,
      "learning_rate": 9.804773598500381e-07,
      "loss": 1.9966,
      "num_input_tokens_seen": 69243536,
      "step": 68000
    },
    {
      "epoch": 1.340788722411451,
      "grad_norm": 1.770240068435669,
      "learning_rate": 9.804202676925313e-07,
      "loss": 1.981,
      "num_input_tokens_seen": 69345808,
      "step": 68100
    },
    {
      "epoch": 1.3427575751609537,
      "grad_norm": 1.959314227104187,
      "learning_rate": 9.8036309384361e-07,
      "loss": 1.9632,
      "num_input_tokens_seen": 69447728,
      "step": 68200
    },
    {
      "epoch": 1.3447264279104565,
      "grad_norm": 1.9582931995391846,
      "learning_rate": 9.803058383129958e-07,
      "loss": 1.9684,
      "num_input_tokens_seen": 69548456,
      "step": 68300
    },
    {
      "epoch": 1.3466952806599595,
      "grad_norm": 1.813268780708313,
      "learning_rate": 9.802485011104254e-07,
      "loss": 1.9948,
      "num_input_tokens_seen": 69649928,
      "step": 68400
    },
    {
      "epoch": 1.3486641334094622,
      "grad_norm": 2.0228004455566406,
      "learning_rate": 9.801910822456476e-07,
      "loss": 1.9943,
      "num_input_tokens_seen": 69750912,
      "step": 68500
    },
    {
      "epoch": 1.3506329861589652,
      "grad_norm": 1.9284993410110474,
      "learning_rate": 9.801335817284266e-07,
      "loss": 1.9669,
      "num_input_tokens_seen": 69852200,
      "step": 68600
    },
    {
      "epoch": 1.352601838908468,
      "grad_norm": 2.1699578762054443,
      "learning_rate": 9.800759995685395e-07,
      "loss": 1.9891,
      "num_input_tokens_seen": 69952848,
      "step": 68700
    },
    {
      "epoch": 1.354570691657971,
      "grad_norm": 2.152777910232544,
      "learning_rate": 9.800183357757778e-07,
      "loss": 2.0238,
      "num_input_tokens_seen": 70054280,
      "step": 68800
    },
    {
      "epoch": 1.3565395444074737,
      "grad_norm": 1.7112054824829102,
      "learning_rate": 9.79960590359947e-07,
      "loss": 1.9838,
      "num_input_tokens_seen": 70155976,
      "step": 68900
    },
    {
      "epoch": 1.3585083971569767,
      "grad_norm": 1.9984151124954224,
      "learning_rate": 9.799027633308658e-07,
      "loss": 1.9918,
      "num_input_tokens_seen": 70258376,
      "step": 69000
    },
    {
      "epoch": 1.3604772499064794,
      "grad_norm": 2.0464658737182617,
      "learning_rate": 9.798448546983674e-07,
      "loss": 1.9936,
      "num_input_tokens_seen": 70358560,
      "step": 69100
    },
    {
      "epoch": 1.3624461026559824,
      "grad_norm": 1.952075719833374,
      "learning_rate": 9.797868644722987e-07,
      "loss": 1.9968,
      "num_input_tokens_seen": 70460960,
      "step": 69200
    },
    {
      "epoch": 1.3644149554054852,
      "grad_norm": 1.8766202926635742,
      "learning_rate": 9.797287926625203e-07,
      "loss": 1.9406,
      "num_input_tokens_seen": 70562768,
      "step": 69300
    },
    {
      "epoch": 1.3663838081549882,
      "grad_norm": 1.9930871725082397,
      "learning_rate": 9.796706392789072e-07,
      "loss": 1.9732,
      "num_input_tokens_seen": 70663960,
      "step": 69400
    },
    {
      "epoch": 1.368352660904491,
      "grad_norm": 1.9259198904037476,
      "learning_rate": 9.796124043313474e-07,
      "loss": 1.9615,
      "num_input_tokens_seen": 70765448,
      "step": 69500
    },
    {
      "epoch": 1.3703215136539937,
      "grad_norm": 1.7779191732406616,
      "learning_rate": 9.795540878297434e-07,
      "loss": 1.9313,
      "num_input_tokens_seen": 70867144,
      "step": 69600
    },
    {
      "epoch": 1.3722903664034967,
      "grad_norm": 1.705250859260559,
      "learning_rate": 9.794956897840118e-07,
      "loss": 1.9658,
      "num_input_tokens_seen": 70968928,
      "step": 69700
    },
    {
      "epoch": 1.3742592191529996,
      "grad_norm": 1.8308429718017578,
      "learning_rate": 9.794372102040822e-07,
      "loss": 1.9558,
      "num_input_tokens_seen": 71071328,
      "step": 69800
    },
    {
      "epoch": 1.3762280719025024,
      "grad_norm": 1.8280293941497803,
      "learning_rate": 9.793786490998987e-07,
      "loss": 1.9873,
      "num_input_tokens_seen": 71171512,
      "step": 69900
    },
    {
      "epoch": 1.3781969246520052,
      "grad_norm": 2.1454601287841797,
      "learning_rate": 9.793200064814193e-07,
      "loss": 1.9997,
      "num_input_tokens_seen": 71273208,
      "step": 70000
    },
    {
      "epoch": 1.3801657774015081,
      "grad_norm": 2.003838539123535,
      "learning_rate": 9.792612823586158e-07,
      "loss": 1.9791,
      "num_input_tokens_seen": 71375608,
      "step": 70100
    },
    {
      "epoch": 1.3821346301510111,
      "grad_norm": 1.8229576349258423,
      "learning_rate": 9.792024767414731e-07,
      "loss": 1.9928,
      "num_input_tokens_seen": 71477360,
      "step": 70200
    },
    {
      "epoch": 1.3841034829005139,
      "grad_norm": 1.9113233089447021,
      "learning_rate": 9.791435896399913e-07,
      "loss": 1.9577,
      "num_input_tokens_seen": 71579760,
      "step": 70300
    },
    {
      "epoch": 1.3860723356500166,
      "grad_norm": 1.8677958250045776,
      "learning_rate": 9.790846210641832e-07,
      "loss": 1.9734,
      "num_input_tokens_seen": 71682160,
      "step": 70400
    },
    {
      "epoch": 1.3880411883995196,
      "grad_norm": 2.3678808212280273,
      "learning_rate": 9.79025571024076e-07,
      "loss": 2.0064,
      "num_input_tokens_seen": 71783792,
      "step": 70500
    },
    {
      "epoch": 1.3900100411490226,
      "grad_norm": 2.2890989780426025,
      "learning_rate": 9.789664395297109e-07,
      "loss": 2.0239,
      "num_input_tokens_seen": 71885384,
      "step": 70600
    },
    {
      "epoch": 1.3919788938985254,
      "grad_norm": 1.7234455347061157,
      "learning_rate": 9.789072265911426e-07,
      "loss": 1.9708,
      "num_input_tokens_seen": 71987128,
      "step": 70700
    },
    {
      "epoch": 1.3939477466480281,
      "grad_norm": 1.7842670679092407,
      "learning_rate": 9.788479322184395e-07,
      "loss": 1.9854,
      "num_input_tokens_seen": 72088096,
      "step": 70800
    },
    {
      "epoch": 1.395916599397531,
      "grad_norm": 1.779493808746338,
      "learning_rate": 9.787885564216844e-07,
      "loss": 2.0041,
      "num_input_tokens_seen": 72189944,
      "step": 70900
    },
    {
      "epoch": 1.3978854521470339,
      "grad_norm": 1.9403914213180542,
      "learning_rate": 9.787290992109737e-07,
      "loss": 1.9884,
      "num_input_tokens_seen": 72290592,
      "step": 71000
    },
    {
      "epoch": 1.3998543048965368,
      "grad_norm": 2.0947275161743164,
      "learning_rate": 9.786695605964175e-07,
      "loss": 1.9893,
      "num_input_tokens_seen": 72391536,
      "step": 71100
    },
    {
      "epoch": 1.4018231576460396,
      "grad_norm": 1.8270337581634521,
      "learning_rate": 9.786099405881395e-07,
      "loss": 1.9986,
      "num_input_tokens_seen": 72492576,
      "step": 71200
    },
    {
      "epoch": 1.4037920103955426,
      "grad_norm": 2.048389196395874,
      "learning_rate": 9.785502391962781e-07,
      "loss": 1.9589,
      "num_input_tokens_seen": 72593528,
      "step": 71300
    },
    {
      "epoch": 1.4057608631450453,
      "grad_norm": 1.9211962223052979,
      "learning_rate": 9.78490456430985e-07,
      "loss": 1.9909,
      "num_input_tokens_seen": 72695928,
      "step": 71400
    },
    {
      "epoch": 1.4077297158945483,
      "grad_norm": 1.9742146730422974,
      "learning_rate": 9.78430592302425e-07,
      "loss": 2.0325,
      "num_input_tokens_seen": 72797424,
      "step": 71500
    },
    {
      "epoch": 1.409698568644051,
      "grad_norm": 1.8823710680007935,
      "learning_rate": 9.783706468207784e-07,
      "loss": 1.9457,
      "num_input_tokens_seen": 72899824,
      "step": 71600
    },
    {
      "epoch": 1.411667421393554,
      "grad_norm": 1.8808926343917847,
      "learning_rate": 9.783106199962381e-07,
      "loss": 2.0035,
      "num_input_tokens_seen": 73000784,
      "step": 71700
    },
    {
      "epoch": 1.4136362741430568,
      "grad_norm": 2.0258290767669678,
      "learning_rate": 9.782505118390109e-07,
      "loss": 1.9815,
      "num_input_tokens_seen": 73102680,
      "step": 71800
    },
    {
      "epoch": 1.4156051268925598,
      "grad_norm": 1.8572781085968018,
      "learning_rate": 9.781903223593182e-07,
      "loss": 1.9955,
      "num_input_tokens_seen": 73205080,
      "step": 71900
    },
    {
      "epoch": 1.4175739796420626,
      "grad_norm": 2.2411041259765625,
      "learning_rate": 9.781300515673942e-07,
      "loss": 2.0017,
      "num_input_tokens_seen": 73306728,
      "step": 72000
    },
    {
      "epoch": 1.4195428323915653,
      "grad_norm": 2.064722776412964,
      "learning_rate": 9.78069699473488e-07,
      "loss": 1.9744,
      "num_input_tokens_seen": 73409128,
      "step": 72100
    },
    {
      "epoch": 1.4215116851410683,
      "grad_norm": 2.029839515686035,
      "learning_rate": 9.780092660878614e-07,
      "loss": 1.9744,
      "num_input_tokens_seen": 73511032,
      "step": 72200
    },
    {
      "epoch": 1.4234805378905713,
      "grad_norm": 1.7096360921859741,
      "learning_rate": 9.779487514207906e-07,
      "loss": 2.0243,
      "num_input_tokens_seen": 73612768,
      "step": 72300
    },
    {
      "epoch": 1.425449390640074,
      "grad_norm": 1.8959747552871704,
      "learning_rate": 9.778881554825662e-07,
      "loss": 1.9732,
      "num_input_tokens_seen": 73714232,
      "step": 72400
    },
    {
      "epoch": 1.4274182433895768,
      "grad_norm": 2.193753957748413,
      "learning_rate": 9.778274782834915e-07,
      "loss": 1.9609,
      "num_input_tokens_seen": 73816232,
      "step": 72500
    },
    {
      "epoch": 1.4293870961390798,
      "grad_norm": 1.890655755996704,
      "learning_rate": 9.777667198338843e-07,
      "loss": 1.9909,
      "num_input_tokens_seen": 73918632,
      "step": 72600
    },
    {
      "epoch": 1.4313559488885828,
      "grad_norm": 1.971684455871582,
      "learning_rate": 9.77705880144076e-07,
      "loss": 2.0129,
      "num_input_tokens_seen": 74020240,
      "step": 72700
    },
    {
      "epoch": 1.4333248016380855,
      "grad_norm": 1.9532390832901,
      "learning_rate": 9.77644959224412e-07,
      "loss": 1.9846,
      "num_input_tokens_seen": 74119088,
      "step": 72800
    },
    {
      "epoch": 1.4352936543875883,
      "grad_norm": 2.263477325439453,
      "learning_rate": 9.775839570852512e-07,
      "loss": 2.0401,
      "num_input_tokens_seen": 74221488,
      "step": 72900
    },
    {
      "epoch": 1.4372625071370912,
      "grad_norm": 1.9588102102279663,
      "learning_rate": 9.77522873736967e-07,
      "loss": 2.0336,
      "num_input_tokens_seen": 74322552,
      "step": 73000
    },
    {
      "epoch": 1.439231359886594,
      "grad_norm": 2.2284955978393555,
      "learning_rate": 9.774617091899455e-07,
      "loss": 1.9896,
      "num_input_tokens_seen": 74424288,
      "step": 73100
    },
    {
      "epoch": 1.441200212636097,
      "grad_norm": 1.8370540142059326,
      "learning_rate": 9.774004634545874e-07,
      "loss": 2.0052,
      "num_input_tokens_seen": 74526688,
      "step": 73200
    },
    {
      "epoch": 1.4431690653855997,
      "grad_norm": 2.0275652408599854,
      "learning_rate": 9.77339136541307e-07,
      "loss": 2.0013,
      "num_input_tokens_seen": 74628320,
      "step": 73300
    },
    {
      "epoch": 1.4451379181351027,
      "grad_norm": 2.0400390625,
      "learning_rate": 9.772777284605327e-07,
      "loss": 1.9748,
      "num_input_tokens_seen": 74729792,
      "step": 73400
    },
    {
      "epoch": 1.4471067708846055,
      "grad_norm": 1.8525439500808716,
      "learning_rate": 9.772162392227062e-07,
      "loss": 2.0092,
      "num_input_tokens_seen": 74831344,
      "step": 73500
    },
    {
      "epoch": 1.4490756236341085,
      "grad_norm": 1.9950172901153564,
      "learning_rate": 9.771546688382831e-07,
      "loss": 2.0175,
      "num_input_tokens_seen": 74931784,
      "step": 73600
    },
    {
      "epoch": 1.4510444763836112,
      "grad_norm": 1.8894357681274414,
      "learning_rate": 9.770930173177332e-07,
      "loss": 1.9752,
      "num_input_tokens_seen": 75034184,
      "step": 73700
    },
    {
      "epoch": 1.4530133291331142,
      "grad_norm": 2.0317158699035645,
      "learning_rate": 9.770312846715397e-07,
      "loss": 1.9799,
      "num_input_tokens_seen": 75135936,
      "step": 73800
    },
    {
      "epoch": 1.454982181882617,
      "grad_norm": 1.9283066987991333,
      "learning_rate": 9.769694709101997e-07,
      "loss": 1.9754,
      "num_input_tokens_seen": 75237440,
      "step": 73900
    },
    {
      "epoch": 1.45695103463212,
      "grad_norm": 1.9150584936141968,
      "learning_rate": 9.76907576044224e-07,
      "loss": 1.9905,
      "num_input_tokens_seen": 75338424,
      "step": 74000
    },
    {
      "epoch": 1.4589198873816227,
      "grad_norm": 1.8755513429641724,
      "learning_rate": 9.768456000841374e-07,
      "loss": 1.9839,
      "num_input_tokens_seen": 75440200,
      "step": 74100
    },
    {
      "epoch": 1.4608887401311255,
      "grad_norm": 2.5257606506347656,
      "learning_rate": 9.767835430404785e-07,
      "loss": 1.9775,
      "num_input_tokens_seen": 75542600,
      "step": 74200
    },
    {
      "epoch": 1.4628575928806284,
      "grad_norm": 2.086940050125122,
      "learning_rate": 9.767214049237996e-07,
      "loss": 1.9546,
      "num_input_tokens_seen": 75644952,
      "step": 74300
    },
    {
      "epoch": 1.4648264456301314,
      "grad_norm": 1.6934620141983032,
      "learning_rate": 9.766591857446664e-07,
      "loss": 1.9341,
      "num_input_tokens_seen": 75747352,
      "step": 74400
    },
    {
      "epoch": 1.4667952983796342,
      "grad_norm": 1.8574316501617432,
      "learning_rate": 9.765968855136593e-07,
      "loss": 1.9894,
      "num_input_tokens_seen": 75849752,
      "step": 74500
    },
    {
      "epoch": 1.468764151129137,
      "grad_norm": 2.30375075340271,
      "learning_rate": 9.765345042413713e-07,
      "loss": 1.9596,
      "num_input_tokens_seen": 75951408,
      "step": 74600
    },
    {
      "epoch": 1.47073300387864,
      "grad_norm": 1.8743842840194702,
      "learning_rate": 9.764720419384102e-07,
      "loss": 2.0778,
      "num_input_tokens_seen": 76052984,
      "step": 74700
    },
    {
      "epoch": 1.472701856628143,
      "grad_norm": 2.6566689014434814,
      "learning_rate": 9.764094986153972e-07,
      "loss": 2.046,
      "num_input_tokens_seen": 76155208,
      "step": 74800
    },
    {
      "epoch": 1.4746707093776457,
      "grad_norm": 2.362420082092285,
      "learning_rate": 9.763468742829673e-07,
      "loss": 1.9592,
      "num_input_tokens_seen": 76257608,
      "step": 74900
    },
    {
      "epoch": 1.4766395621271484,
      "grad_norm": 2.2277305126190186,
      "learning_rate": 9.76284168951769e-07,
      "loss": 1.9825,
      "num_input_tokens_seen": 76360008,
      "step": 75000
    },
    {
      "epoch": 1.4786084148766514,
      "grad_norm": 2.019000768661499,
      "learning_rate": 9.76221382632465e-07,
      "loss": 1.9805,
      "num_input_tokens_seen": 76461784,
      "step": 75100
    },
    {
      "epoch": 1.4805772676261542,
      "grad_norm": 2.016059160232544,
      "learning_rate": 9.761585153357315e-07,
      "loss": 1.9806,
      "num_input_tokens_seen": 76563816,
      "step": 75200
    },
    {
      "epoch": 1.4825461203756571,
      "grad_norm": 2.0543510913848877,
      "learning_rate": 9.760955670722584e-07,
      "loss": 1.9819,
      "num_input_tokens_seen": 76666216,
      "step": 75300
    },
    {
      "epoch": 1.48451497312516,
      "grad_norm": 1.8296061754226685,
      "learning_rate": 9.760325378527498e-07,
      "loss": 1.944,
      "num_input_tokens_seen": 76768616,
      "step": 75400
    },
    {
      "epoch": 1.4864838258746629,
      "grad_norm": 1.9528096914291382,
      "learning_rate": 9.759694276879233e-07,
      "loss": 1.974,
      "num_input_tokens_seen": 76871016,
      "step": 75500
    },
    {
      "epoch": 1.4884526786241656,
      "grad_norm": 1.9242920875549316,
      "learning_rate": 9.7590623658851e-07,
      "loss": 2.0285,
      "num_input_tokens_seen": 76973416,
      "step": 75600
    },
    {
      "epoch": 1.4904215313736686,
      "grad_norm": 2.241938829421997,
      "learning_rate": 9.758429645652552e-07,
      "loss": 1.9852,
      "num_input_tokens_seen": 77074376,
      "step": 75700
    },
    {
      "epoch": 1.4923903841231714,
      "grad_norm": 2.608829975128174,
      "learning_rate": 9.757796116289176e-07,
      "loss": 1.9513,
      "num_input_tokens_seen": 77175984,
      "step": 75800
    },
    {
      "epoch": 1.4943592368726744,
      "grad_norm": 2.105858087539673,
      "learning_rate": 9.757161777902703e-07,
      "loss": 1.9967,
      "num_input_tokens_seen": 77277296,
      "step": 75900
    },
    {
      "epoch": 1.4963280896221771,
      "grad_norm": 2.0412871837615967,
      "learning_rate": 9.75652663060099e-07,
      "loss": 2.0007,
      "num_input_tokens_seen": 77379448,
      "step": 76000
    },
    {
      "epoch": 1.49829694237168,
      "grad_norm": 1.908644199371338,
      "learning_rate": 9.755890674492043e-07,
      "loss": 2.004,
      "num_input_tokens_seen": 77480992,
      "step": 76100
    },
    {
      "epoch": 1.5002657951211829,
      "grad_norm": 1.8958549499511719,
      "learning_rate": 9.755253909683999e-07,
      "loss": 2.0009,
      "num_input_tokens_seen": 77583392,
      "step": 76200
    },
    {
      "epoch": 1.5022346478706856,
      "grad_norm": 1.7821376323699951,
      "learning_rate": 9.754616336285136e-07,
      "loss": 1.973,
      "num_input_tokens_seen": 77684488,
      "step": 76300
    },
    {
      "epoch": 1.5042035006201886,
      "grad_norm": 2.1679296493530273,
      "learning_rate": 9.753977954403868e-07,
      "loss": 1.9916,
      "num_input_tokens_seen": 77785016,
      "step": 76400
    },
    {
      "epoch": 1.5061723533696916,
      "grad_norm": 1.8659836053848267,
      "learning_rate": 9.753338764148745e-07,
      "loss": 1.9824,
      "num_input_tokens_seen": 77887416,
      "step": 76500
    },
    {
      "epoch": 1.5081412061191943,
      "grad_norm": 1.8900649547576904,
      "learning_rate": 9.752698765628456e-07,
      "loss": 1.9843,
      "num_input_tokens_seen": 77989264,
      "step": 76600
    },
    {
      "epoch": 1.510110058868697,
      "grad_norm": 2.197361946105957,
      "learning_rate": 9.75205795895183e-07,
      "loss": 1.9854,
      "num_input_tokens_seen": 78090520,
      "step": 76700
    },
    {
      "epoch": 1.5120789116182,
      "grad_norm": 1.8914963006973267,
      "learning_rate": 9.751416344227828e-07,
      "loss": 1.9851,
      "num_input_tokens_seen": 78192232,
      "step": 76800
    },
    {
      "epoch": 1.514047764367703,
      "grad_norm": 2.121875762939453,
      "learning_rate": 9.750773921565551e-07,
      "loss": 2.0401,
      "num_input_tokens_seen": 78294632,
      "step": 76900
    },
    {
      "epoch": 1.5160166171172058,
      "grad_norm": 1.6164422035217285,
      "learning_rate": 9.75013069107424e-07,
      "loss": 1.9996,
      "num_input_tokens_seen": 78396368,
      "step": 77000
    },
    {
      "epoch": 1.5179854698667086,
      "grad_norm": 1.7804019451141357,
      "learning_rate": 9.749486652863268e-07,
      "loss": 1.9095,
      "num_input_tokens_seen": 78498216,
      "step": 77100
    },
    {
      "epoch": 1.5199543226162116,
      "grad_norm": 1.8734009265899658,
      "learning_rate": 9.74884180704215e-07,
      "loss": 2.0188,
      "num_input_tokens_seen": 78599792,
      "step": 77200
    },
    {
      "epoch": 1.5219231753657145,
      "grad_norm": 1.8652548789978027,
      "learning_rate": 9.748196153720536e-07,
      "loss": 2.0122,
      "num_input_tokens_seen": 78702192,
      "step": 77300
    },
    {
      "epoch": 1.5238920281152173,
      "grad_norm": 1.9243494272232056,
      "learning_rate": 9.747549693008214e-07,
      "loss": 1.9853,
      "num_input_tokens_seen": 78804592,
      "step": 77400
    },
    {
      "epoch": 1.52586088086472,
      "grad_norm": 2.356846332550049,
      "learning_rate": 9.746902425015112e-07,
      "loss": 2.0327,
      "num_input_tokens_seen": 78906064,
      "step": 77500
    },
    {
      "epoch": 1.527829733614223,
      "grad_norm": 10.839113235473633,
      "learning_rate": 9.746254349851286e-07,
      "loss": 2.0012,
      "num_input_tokens_seen": 79008024,
      "step": 77600
    },
    {
      "epoch": 1.529798586363726,
      "grad_norm": 1.8322460651397705,
      "learning_rate": 9.745605467626943e-07,
      "loss": 2.0063,
      "num_input_tokens_seen": 79110120,
      "step": 77700
    },
    {
      "epoch": 1.5317674391132288,
      "grad_norm": 1.9234142303466797,
      "learning_rate": 9.744955778452414e-07,
      "loss": 1.947,
      "num_input_tokens_seen": 79211944,
      "step": 77800
    },
    {
      "epoch": 1.5337362918627315,
      "grad_norm": 1.7628679275512695,
      "learning_rate": 9.744305282438177e-07,
      "loss": 1.9934,
      "num_input_tokens_seen": 79312744,
      "step": 77900
    },
    {
      "epoch": 1.5357051446122343,
      "grad_norm": 2.5412709712982178,
      "learning_rate": 9.743653979694841e-07,
      "loss": 1.9765,
      "num_input_tokens_seen": 79415144,
      "step": 78000
    },
    {
      "epoch": 1.5376739973617373,
      "grad_norm": 2.0403952598571777,
      "learning_rate": 9.743001870333156e-07,
      "loss": 2.0687,
      "num_input_tokens_seen": 79517256,
      "step": 78100
    },
    {
      "epoch": 1.5396428501112402,
      "grad_norm": 2.046607255935669,
      "learning_rate": 9.742348954464007e-07,
      "loss": 1.9899,
      "num_input_tokens_seen": 79619112,
      "step": 78200
    },
    {
      "epoch": 1.541611702860743,
      "grad_norm": 2.3528730869293213,
      "learning_rate": 9.74169523219842e-07,
      "loss": 1.9849,
      "num_input_tokens_seen": 79720952,
      "step": 78300
    },
    {
      "epoch": 1.5435805556102458,
      "grad_norm": 1.8831902742385864,
      "learning_rate": 9.74104070364755e-07,
      "loss": 1.9927,
      "num_input_tokens_seen": 79821640,
      "step": 78400
    },
    {
      "epoch": 1.5455494083597487,
      "grad_norm": 2.3101558685302734,
      "learning_rate": 9.740385368922695e-07,
      "loss": 1.9666,
      "num_input_tokens_seen": 79924040,
      "step": 78500
    },
    {
      "epoch": 1.5475182611092517,
      "grad_norm": 2.005356550216675,
      "learning_rate": 9.739729228135291e-07,
      "loss": 1.9533,
      "num_input_tokens_seen": 80026440,
      "step": 78600
    },
    {
      "epoch": 1.5494871138587545,
      "grad_norm": 1.7179080247879028,
      "learning_rate": 9.739072281396908e-07,
      "loss": 1.9791,
      "num_input_tokens_seen": 80128840,
      "step": 78700
    },
    {
      "epoch": 1.5514559666082572,
      "grad_norm": 2.230632781982422,
      "learning_rate": 9.738414528819257e-07,
      "loss": 2.0118,
      "num_input_tokens_seen": 80229680,
      "step": 78800
    },
    {
      "epoch": 1.5534248193577602,
      "grad_norm": 1.7507121562957764,
      "learning_rate": 9.737755970514179e-07,
      "loss": 1.9749,
      "num_input_tokens_seen": 80331384,
      "step": 78900
    },
    {
      "epoch": 1.5553936721072632,
      "grad_norm": 1.8396021127700806,
      "learning_rate": 9.737096606593658e-07,
      "loss": 1.9282,
      "num_input_tokens_seen": 80433784,
      "step": 79000
    },
    {
      "epoch": 1.557362524856766,
      "grad_norm": 2.0760533809661865,
      "learning_rate": 9.736436437169813e-07,
      "loss": 2.0078,
      "num_input_tokens_seen": 80535416,
      "step": 79100
    },
    {
      "epoch": 1.5593313776062687,
      "grad_norm": 1.8407785892486572,
      "learning_rate": 9.735775462354904e-07,
      "loss": 2.0361,
      "num_input_tokens_seen": 80637360,
      "step": 79200
    },
    {
      "epoch": 1.5613002303557717,
      "grad_norm": 1.8386290073394775,
      "learning_rate": 9.735113682261318e-07,
      "loss": 1.9458,
      "num_input_tokens_seen": 80739160,
      "step": 79300
    },
    {
      "epoch": 1.5632690831052747,
      "grad_norm": 2.1957356929779053,
      "learning_rate": 9.734451097001588e-07,
      "loss": 2.0063,
      "num_input_tokens_seen": 80840392,
      "step": 79400
    },
    {
      "epoch": 1.5652379358547774,
      "grad_norm": 2.0933775901794434,
      "learning_rate": 9.733787706688383e-07,
      "loss": 1.9894,
      "num_input_tokens_seen": 80942792,
      "step": 79500
    },
    {
      "epoch": 1.5672067886042802,
      "grad_norm": 1.7478200197219849,
      "learning_rate": 9.733123511434504e-07,
      "loss": 2.0121,
      "num_input_tokens_seen": 81044760,
      "step": 79600
    },
    {
      "epoch": 1.5691756413537832,
      "grad_norm": 1.8533891439437866,
      "learning_rate": 9.732458511352894e-07,
      "loss": 2.0458,
      "num_input_tokens_seen": 81146360,
      "step": 79700
    },
    {
      "epoch": 1.5711444941032862,
      "grad_norm": 1.640206217765808,
      "learning_rate": 9.73179270655663e-07,
      "loss": 1.9348,
      "num_input_tokens_seen": 81248760,
      "step": 79800
    },
    {
      "epoch": 1.573113346852789,
      "grad_norm": 2.1212682723999023,
      "learning_rate": 9.731126097158923e-07,
      "loss": 1.9619,
      "num_input_tokens_seen": 81351160,
      "step": 79900
    },
    {
      "epoch": 1.5750821996022917,
      "grad_norm": 1.9808043241500854,
      "learning_rate": 9.73045868327313e-07,
      "loss": 2.0226,
      "num_input_tokens_seen": 81452800,
      "step": 80000
    },
    {
      "epoch": 1.5770510523517947,
      "grad_norm": 2.1366357803344727,
      "learning_rate": 9.729790465012735e-07,
      "loss": 1.9838,
      "num_input_tokens_seen": 81554072,
      "step": 80100
    },
    {
      "epoch": 1.5790199051012976,
      "grad_norm": 1.896931767463684,
      "learning_rate": 9.729121442491366e-07,
      "loss": 1.9746,
      "num_input_tokens_seen": 81656472,
      "step": 80200
    },
    {
      "epoch": 1.5809887578508004,
      "grad_norm": 1.8604445457458496,
      "learning_rate": 9.728451615822782e-07,
      "loss": 1.9403,
      "num_input_tokens_seen": 81758424,
      "step": 80300
    },
    {
      "epoch": 1.5829576106003032,
      "grad_norm": 1.9297837018966675,
      "learning_rate": 9.727780985120882e-07,
      "loss": 1.9945,
      "num_input_tokens_seen": 81860824,
      "step": 80400
    },
    {
      "epoch": 1.584926463349806,
      "grad_norm": 2.1828677654266357,
      "learning_rate": 9.727109550499701e-07,
      "loss": 2.0118,
      "num_input_tokens_seen": 81962864,
      "step": 80500
    },
    {
      "epoch": 1.586895316099309,
      "grad_norm": 2.0296082496643066,
      "learning_rate": 9.726437312073413e-07,
      "loss": 2.0029,
      "num_input_tokens_seen": 82065264,
      "step": 80600
    },
    {
      "epoch": 1.5888641688488119,
      "grad_norm": 1.9530014991760254,
      "learning_rate": 9.725764269956322e-07,
      "loss": 2.0214,
      "num_input_tokens_seen": 82166888,
      "step": 80700
    },
    {
      "epoch": 1.5908330215983146,
      "grad_norm": 1.8273735046386719,
      "learning_rate": 9.725090424262877e-07,
      "loss": 2.0482,
      "num_input_tokens_seen": 82268104,
      "step": 80800
    },
    {
      "epoch": 1.5928018743478174,
      "grad_norm": 21.396066665649414,
      "learning_rate": 9.72441577510766e-07,
      "loss": 1.9227,
      "num_input_tokens_seen": 82370504,
      "step": 80900
    },
    {
      "epoch": 1.5947707270973204,
      "grad_norm": 1.795525074005127,
      "learning_rate": 9.723740322605387e-07,
      "loss": 1.9769,
      "num_input_tokens_seen": 82472904,
      "step": 81000
    },
    {
      "epoch": 1.5967395798468234,
      "grad_norm": 1.9250332117080688,
      "learning_rate": 9.723064066870915e-07,
      "loss": 1.9725,
      "num_input_tokens_seen": 82575304,
      "step": 81100
    },
    {
      "epoch": 1.5987084325963261,
      "grad_norm": 2.055537223815918,
      "learning_rate": 9.722387008019233e-07,
      "loss": 1.9381,
      "num_input_tokens_seen": 82677088,
      "step": 81200
    },
    {
      "epoch": 1.6006772853458289,
      "grad_norm": 2.0581607818603516,
      "learning_rate": 9.721709146165472e-07,
      "loss": 2.0336,
      "num_input_tokens_seen": 82778960,
      "step": 81300
    },
    {
      "epoch": 1.6026461380953319,
      "grad_norm": 1.6808435916900635,
      "learning_rate": 9.721030481424894e-07,
      "loss": 1.9552,
      "num_input_tokens_seen": 82881360,
      "step": 81400
    },
    {
      "epoch": 1.6046149908448348,
      "grad_norm": 1.754149317741394,
      "learning_rate": 9.720351013912904e-07,
      "loss": 1.9766,
      "num_input_tokens_seen": 82983016,
      "step": 81500
    },
    {
      "epoch": 1.6065838435943376,
      "grad_norm": 1.7373837232589722,
      "learning_rate": 9.719670743745035e-07,
      "loss": 2.0009,
      "num_input_tokens_seen": 83084648,
      "step": 81600
    },
    {
      "epoch": 1.6085526963438403,
      "grad_norm": 2.476461172103882,
      "learning_rate": 9.718989671036964e-07,
      "loss": 2.0358,
      "num_input_tokens_seen": 83184536,
      "step": 81700
    },
    {
      "epoch": 1.6105215490933433,
      "grad_norm": 2.0237913131713867,
      "learning_rate": 9.7183077959045e-07,
      "loss": 1.9746,
      "num_input_tokens_seen": 83285576,
      "step": 81800
    },
    {
      "epoch": 1.6124904018428463,
      "grad_norm": 1.7231385707855225,
      "learning_rate": 9.717625118463596e-07,
      "loss": 2.0006,
      "num_input_tokens_seen": 83387664,
      "step": 81900
    },
    {
      "epoch": 1.614459254592349,
      "grad_norm": 2.1230030059814453,
      "learning_rate": 9.716941638830327e-07,
      "loss": 1.9574,
      "num_input_tokens_seen": 83489336,
      "step": 82000
    },
    {
      "epoch": 1.6164281073418518,
      "grad_norm": 2.5109426975250244,
      "learning_rate": 9.716257357120917e-07,
      "loss": 1.985,
      "num_input_tokens_seen": 83589680,
      "step": 82100
    },
    {
      "epoch": 1.6183969600913548,
      "grad_norm": 2.1775271892547607,
      "learning_rate": 9.715572273451724e-07,
      "loss": 1.9733,
      "num_input_tokens_seen": 83692080,
      "step": 82200
    },
    {
      "epoch": 1.6203658128408578,
      "grad_norm": 1.7170239686965942,
      "learning_rate": 9.714886387939236e-07,
      "loss": 1.9695,
      "num_input_tokens_seen": 83793120,
      "step": 82300
    },
    {
      "epoch": 1.6223346655903605,
      "grad_norm": 1.8938959836959839,
      "learning_rate": 9.714199700700087e-07,
      "loss": 1.9293,
      "num_input_tokens_seen": 83894968,
      "step": 82400
    },
    {
      "epoch": 1.6243035183398633,
      "grad_norm": 1.83748459815979,
      "learning_rate": 9.713512211851041e-07,
      "loss": 1.9694,
      "num_input_tokens_seen": 83996896,
      "step": 82500
    },
    {
      "epoch": 1.626272371089366,
      "grad_norm": 3.0680124759674072,
      "learning_rate": 9.712823921508998e-07,
      "loss": 2.0562,
      "num_input_tokens_seen": 84096456,
      "step": 82600
    },
    {
      "epoch": 1.628241223838869,
      "grad_norm": 1.8784260749816895,
      "learning_rate": 9.712134829790997e-07,
      "loss": 1.9984,
      "num_input_tokens_seen": 84198856,
      "step": 82700
    },
    {
      "epoch": 1.630210076588372,
      "grad_norm": 2.3644661903381348,
      "learning_rate": 9.711444936814211e-07,
      "loss": 1.9305,
      "num_input_tokens_seen": 84300632,
      "step": 82800
    },
    {
      "epoch": 1.6321789293378748,
      "grad_norm": 1.8629422187805176,
      "learning_rate": 9.710754242695952e-07,
      "loss": 1.9551,
      "num_input_tokens_seen": 84402584,
      "step": 82900
    },
    {
      "epoch": 1.6341477820873775,
      "grad_norm": 1.981223464012146,
      "learning_rate": 9.710062747553668e-07,
      "loss": 1.9489,
      "num_input_tokens_seen": 84504984,
      "step": 83000
    },
    {
      "epoch": 1.6361166348368805,
      "grad_norm": 1.9939093589782715,
      "learning_rate": 9.70937045150494e-07,
      "loss": 1.9794,
      "num_input_tokens_seen": 84606920,
      "step": 83100
    },
    {
      "epoch": 1.6380854875863835,
      "grad_norm": 1.9648970365524292,
      "learning_rate": 9.708677354667487e-07,
      "loss": 2.0041,
      "num_input_tokens_seen": 84708232,
      "step": 83200
    },
    {
      "epoch": 1.6400543403358863,
      "grad_norm": 2.2339165210723877,
      "learning_rate": 9.707983457159164e-07,
      "loss": 2.0034,
      "num_input_tokens_seen": 84809944,
      "step": 83300
    },
    {
      "epoch": 1.642023193085389,
      "grad_norm": 13.141552925109863,
      "learning_rate": 9.707288759097963e-07,
      "loss": 2.0143,
      "num_input_tokens_seen": 84911480,
      "step": 83400
    },
    {
      "epoch": 1.643992045834892,
      "grad_norm": 1.8389010429382324,
      "learning_rate": 9.70659326060201e-07,
      "loss": 1.9945,
      "num_input_tokens_seen": 85013152,
      "step": 83500
    },
    {
      "epoch": 1.645960898584395,
      "grad_norm": 1.8468313217163086,
      "learning_rate": 9.705896961789575e-07,
      "loss": 1.9712,
      "num_input_tokens_seen": 85115552,
      "step": 83600
    },
    {
      "epoch": 1.6479297513338977,
      "grad_norm": 2.0421788692474365,
      "learning_rate": 9.70519986277905e-07,
      "loss": 2.0293,
      "num_input_tokens_seen": 85217256,
      "step": 83700
    },
    {
      "epoch": 1.6498986040834005,
      "grad_norm": 2.0900497436523438,
      "learning_rate": 9.704501963688973e-07,
      "loss": 1.978,
      "num_input_tokens_seen": 85319104,
      "step": 83800
    },
    {
      "epoch": 1.6518674568329035,
      "grad_norm": 1.9232817888259888,
      "learning_rate": 9.703803264638018e-07,
      "loss": 2.0701,
      "num_input_tokens_seen": 85421296,
      "step": 83900
    },
    {
      "epoch": 1.6538363095824065,
      "grad_norm": 1.8633781671524048,
      "learning_rate": 9.703103765744992e-07,
      "loss": 1.9357,
      "num_input_tokens_seen": 85523144,
      "step": 84000
    },
    {
      "epoch": 1.6558051623319092,
      "grad_norm": 1.8194736242294312,
      "learning_rate": 9.702403467128838e-07,
      "loss": 2.0163,
      "num_input_tokens_seen": 85623960,
      "step": 84100
    },
    {
      "epoch": 1.657774015081412,
      "grad_norm": 1.9421623945236206,
      "learning_rate": 9.701702368908636e-07,
      "loss": 2.0152,
      "num_input_tokens_seen": 85725696,
      "step": 84200
    },
    {
      "epoch": 1.659742867830915,
      "grad_norm": 1.9922300577163696,
      "learning_rate": 9.7010004712036e-07,
      "loss": 1.9914,
      "num_input_tokens_seen": 85828096,
      "step": 84300
    },
    {
      "epoch": 1.661711720580418,
      "grad_norm": 2.143540143966675,
      "learning_rate": 9.700297774133088e-07,
      "loss": 2.0373,
      "num_input_tokens_seen": 85928912,
      "step": 84400
    },
    {
      "epoch": 1.6636805733299207,
      "grad_norm": 1.841282606124878,
      "learning_rate": 9.699594277816582e-07,
      "loss": 2.0367,
      "num_input_tokens_seen": 86031312,
      "step": 84500
    },
    {
      "epoch": 1.6656494260794235,
      "grad_norm": 1.9451500177383423,
      "learning_rate": 9.698889982373707e-07,
      "loss": 1.9781,
      "num_input_tokens_seen": 86133192,
      "step": 84600
    },
    {
      "epoch": 1.6676182788289264,
      "grad_norm": 1.8159606456756592,
      "learning_rate": 9.698184887924224e-07,
      "loss": 2.015,
      "num_input_tokens_seen": 86233248,
      "step": 84700
    },
    {
      "epoch": 1.6695871315784292,
      "grad_norm": 2.2536284923553467,
      "learning_rate": 9.697478994588026e-07,
      "loss": 2.0173,
      "num_input_tokens_seen": 86335648,
      "step": 84800
    },
    {
      "epoch": 1.6715559843279322,
      "grad_norm": 2.063523530960083,
      "learning_rate": 9.696772302485146e-07,
      "loss": 1.9713,
      "num_input_tokens_seen": 86438048,
      "step": 84900
    },
    {
      "epoch": 1.673524837077435,
      "grad_norm": 1.7738423347473145,
      "learning_rate": 9.69606481173575e-07,
      "loss": 1.9183,
      "num_input_tokens_seen": 86539944,
      "step": 85000
    },
    {
      "epoch": 1.6754936898269377,
      "grad_norm": 2.0155112743377686,
      "learning_rate": 9.69535652246014e-07,
      "loss": 1.9736,
      "num_input_tokens_seen": 86642344,
      "step": 85100
    },
    {
      "epoch": 1.6774625425764407,
      "grad_norm": 2.012538433074951,
      "learning_rate": 9.694647434778758e-07,
      "loss": 1.9544,
      "num_input_tokens_seen": 86744064,
      "step": 85200
    },
    {
      "epoch": 1.6794313953259437,
      "grad_norm": 2.0843119621276855,
      "learning_rate": 9.693937548812175e-07,
      "loss": 1.9852,
      "num_input_tokens_seen": 86846464,
      "step": 85300
    },
    {
      "epoch": 1.6814002480754464,
      "grad_norm": 2.1301987171173096,
      "learning_rate": 9.693226864681101e-07,
      "loss": 1.9742,
      "num_input_tokens_seen": 86948360,
      "step": 85400
    },
    {
      "epoch": 1.6833691008249492,
      "grad_norm": 2.021069288253784,
      "learning_rate": 9.692515382506385e-07,
      "loss": 1.9824,
      "num_input_tokens_seen": 87050168,
      "step": 85500
    },
    {
      "epoch": 1.6853379535744522,
      "grad_norm": 2.3929224014282227,
      "learning_rate": 9.691803102409007e-07,
      "loss": 1.9663,
      "num_input_tokens_seen": 87152568,
      "step": 85600
    },
    {
      "epoch": 1.6873068063239551,
      "grad_norm": 1.986000418663025,
      "learning_rate": 9.691090024510082e-07,
      "loss": 2.0377,
      "num_input_tokens_seen": 87253352,
      "step": 85700
    },
    {
      "epoch": 1.689275659073458,
      "grad_norm": 1.9733489751815796,
      "learning_rate": 9.690376148930866e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 87354512,
      "step": 85800
    },
    {
      "epoch": 1.6912445118229606,
      "grad_norm": 21.12444305419922,
      "learning_rate": 9.689661475792747e-07,
      "loss": 1.9621,
      "num_input_tokens_seen": 87455960,
      "step": 85900
    },
    {
      "epoch": 1.6932133645724636,
      "grad_norm": 2.2019286155700684,
      "learning_rate": 9.688946005217247e-07,
      "loss": 2.0407,
      "num_input_tokens_seen": 87556448,
      "step": 86000
    },
    {
      "epoch": 1.6951822173219666,
      "grad_norm": 2.0862855911254883,
      "learning_rate": 9.688229737326028e-07,
      "loss": 2.0029,
      "num_input_tokens_seen": 87658168,
      "step": 86100
    },
    {
      "epoch": 1.6971510700714694,
      "grad_norm": 1.8821500539779663,
      "learning_rate": 9.687512672240886e-07,
      "loss": 2.0032,
      "num_input_tokens_seen": 87759112,
      "step": 86200
    },
    {
      "epoch": 1.6991199228209721,
      "grad_norm": 1.727913737297058,
      "learning_rate": 9.686794810083748e-07,
      "loss": 1.9832,
      "num_input_tokens_seen": 87861512,
      "step": 86300
    },
    {
      "epoch": 1.701088775570475,
      "grad_norm": 5.58336067199707,
      "learning_rate": 9.686076150976683e-07,
      "loss": 1.9768,
      "num_input_tokens_seen": 87961408,
      "step": 86400
    },
    {
      "epoch": 1.703057628319978,
      "grad_norm": 4.310601234436035,
      "learning_rate": 9.685356695041894e-07,
      "loss": 1.9896,
      "num_input_tokens_seen": 88062232,
      "step": 86500
    },
    {
      "epoch": 1.7050264810694808,
      "grad_norm": 1.8383711576461792,
      "learning_rate": 9.684636442401717e-07,
      "loss": 2.0618,
      "num_input_tokens_seen": 88162504,
      "step": 86600
    },
    {
      "epoch": 1.7069953338189836,
      "grad_norm": 1.9602752923965454,
      "learning_rate": 9.683915393178623e-07,
      "loss": 2.0026,
      "num_input_tokens_seen": 88264904,
      "step": 86700
    },
    {
      "epoch": 1.7089641865684866,
      "grad_norm": 1.879712462425232,
      "learning_rate": 9.683193547495225e-07,
      "loss": 1.9409,
      "num_input_tokens_seen": 88367304,
      "step": 86800
    },
    {
      "epoch": 1.7109330393179896,
      "grad_norm": 2.159916400909424,
      "learning_rate": 9.682470905474262e-07,
      "loss": 1.9363,
      "num_input_tokens_seen": 88469704,
      "step": 86900
    },
    {
      "epoch": 1.7129018920674923,
      "grad_norm": 2.1803338527679443,
      "learning_rate": 9.681747467238616e-07,
      "loss": 2.0092,
      "num_input_tokens_seen": 88570816,
      "step": 87000
    },
    {
      "epoch": 1.714870744816995,
      "grad_norm": 2.102480411529541,
      "learning_rate": 9.681023232911303e-07,
      "loss": 2.0043,
      "num_input_tokens_seen": 88673216,
      "step": 87100
    },
    {
      "epoch": 1.7168395975664978,
      "grad_norm": 2.1512107849121094,
      "learning_rate": 9.680298202615469e-07,
      "loss": 1.9723,
      "num_input_tokens_seen": 88775616,
      "step": 87200
    },
    {
      "epoch": 1.7188084503160008,
      "grad_norm": 1.8901636600494385,
      "learning_rate": 9.6795723764744e-07,
      "loss": 2.0005,
      "num_input_tokens_seen": 88878016,
      "step": 87300
    },
    {
      "epoch": 1.7207773030655038,
      "grad_norm": 1.9403070211410522,
      "learning_rate": 9.67884575461152e-07,
      "loss": 1.9812,
      "num_input_tokens_seen": 88979880,
      "step": 87400
    },
    {
      "epoch": 1.7227461558150066,
      "grad_norm": 2.147320032119751,
      "learning_rate": 9.67811833715038e-07,
      "loss": 1.9936,
      "num_input_tokens_seen": 89080840,
      "step": 87500
    },
    {
      "epoch": 1.7247150085645093,
      "grad_norm": 1.8907344341278076,
      "learning_rate": 9.677390124214676e-07,
      "loss": 1.9728,
      "num_input_tokens_seen": 89181928,
      "step": 87600
    },
    {
      "epoch": 1.7266838613140123,
      "grad_norm": 3.7446982860565186,
      "learning_rate": 9.676661115928232e-07,
      "loss": 2.0041,
      "num_input_tokens_seen": 89284328,
      "step": 87700
    },
    {
      "epoch": 1.7286527140635153,
      "grad_norm": 1.849755048751831,
      "learning_rate": 9.67593131241501e-07,
      "loss": 2.0008,
      "num_input_tokens_seen": 89386432,
      "step": 87800
    },
    {
      "epoch": 1.730621566813018,
      "grad_norm": 1.9824796915054321,
      "learning_rate": 9.675200713799105e-07,
      "loss": 1.9439,
      "num_input_tokens_seen": 89488832,
      "step": 87900
    },
    {
      "epoch": 1.7325904195625208,
      "grad_norm": 2.4449574947357178,
      "learning_rate": 9.674469320204754e-07,
      "loss": 1.9409,
      "num_input_tokens_seen": 89590416,
      "step": 88000
    },
    {
      "epoch": 1.7345592723120238,
      "grad_norm": 1.7975555658340454,
      "learning_rate": 9.67373713175632e-07,
      "loss": 1.9803,
      "num_input_tokens_seen": 89691896,
      "step": 88100
    },
    {
      "epoch": 1.7365281250615268,
      "grad_norm": 4.588106632232666,
      "learning_rate": 9.673004148578308e-07,
      "loss": 1.9907,
      "num_input_tokens_seen": 89792872,
      "step": 88200
    },
    {
      "epoch": 1.7384969778110295,
      "grad_norm": 2.0367724895477295,
      "learning_rate": 9.672270370795353e-07,
      "loss": 1.9816,
      "num_input_tokens_seen": 89894496,
      "step": 88300
    },
    {
      "epoch": 1.7404658305605323,
      "grad_norm": 1.5424094200134277,
      "learning_rate": 9.67153579853223e-07,
      "loss": 2.0191,
      "num_input_tokens_seen": 89996896,
      "step": 88400
    },
    {
      "epoch": 1.7424346833100353,
      "grad_norm": 2.260645627975464,
      "learning_rate": 9.670800431913843e-07,
      "loss": 1.9791,
      "num_input_tokens_seen": 90097968,
      "step": 88500
    },
    {
      "epoch": 1.7444035360595382,
      "grad_norm": 1.9642016887664795,
      "learning_rate": 9.67006427106524e-07,
      "loss": 1.9712,
      "num_input_tokens_seen": 90199632,
      "step": 88600
    },
    {
      "epoch": 1.746372388809041,
      "grad_norm": 2.2560601234436035,
      "learning_rate": 9.6693273161116e-07,
      "loss": 2.022,
      "num_input_tokens_seen": 90302032,
      "step": 88700
    },
    {
      "epoch": 1.7483412415585438,
      "grad_norm": 1.8089711666107178,
      "learning_rate": 9.66858956717823e-07,
      "loss": 1.9656,
      "num_input_tokens_seen": 90404432,
      "step": 88800
    },
    {
      "epoch": 1.7503100943080467,
      "grad_norm": 1.9336212873458862,
      "learning_rate": 9.66785102439058e-07,
      "loss": 1.9988,
      "num_input_tokens_seen": 90506832,
      "step": 88900
    },
    {
      "epoch": 1.7522789470575497,
      "grad_norm": 2.186123847961426,
      "learning_rate": 9.667111687874235e-07,
      "loss": 2.0029,
      "num_input_tokens_seen": 90608080,
      "step": 89000
    },
    {
      "epoch": 1.7542477998070525,
      "grad_norm": 2.1508913040161133,
      "learning_rate": 9.66637155775491e-07,
      "loss": 1.9678,
      "num_input_tokens_seen": 90710480,
      "step": 89100
    },
    {
      "epoch": 1.7562166525565552,
      "grad_norm": 2.0526251792907715,
      "learning_rate": 9.665630634158463e-07,
      "loss": 1.9458,
      "num_input_tokens_seen": 90812880,
      "step": 89200
    },
    {
      "epoch": 1.758185505306058,
      "grad_norm": 1.6457970142364502,
      "learning_rate": 9.664888917210876e-07,
      "loss": 2.0069,
      "num_input_tokens_seen": 90914512,
      "step": 89300
    },
    {
      "epoch": 1.760154358055561,
      "grad_norm": 2.02527117729187,
      "learning_rate": 9.664146407038276e-07,
      "loss": 2.0099,
      "num_input_tokens_seen": 91016912,
      "step": 89400
    },
    {
      "epoch": 1.762123210805064,
      "grad_norm": 1.8556376695632935,
      "learning_rate": 9.663403103766915e-07,
      "loss": 1.9936,
      "num_input_tokens_seen": 91118384,
      "step": 89500
    },
    {
      "epoch": 1.7640920635545667,
      "grad_norm": 2.298250436782837,
      "learning_rate": 9.662659007523194e-07,
      "loss": 1.9865,
      "num_input_tokens_seen": 91219864,
      "step": 89600
    },
    {
      "epoch": 1.7660609163040695,
      "grad_norm": 1.9188224077224731,
      "learning_rate": 9.661914118433632e-07,
      "loss": 1.9887,
      "num_input_tokens_seen": 91322264,
      "step": 89700
    },
    {
      "epoch": 1.7680297690535725,
      "grad_norm": 1.9386297464370728,
      "learning_rate": 9.661168436624896e-07,
      "loss": 1.9671,
      "num_input_tokens_seen": 91423704,
      "step": 89800
    },
    {
      "epoch": 1.7699986218030754,
      "grad_norm": 2.0172412395477295,
      "learning_rate": 9.66042196222378e-07,
      "loss": 1.9824,
      "num_input_tokens_seen": 91525368,
      "step": 89900
    },
    {
      "epoch": 1.7719674745525782,
      "grad_norm": 2.1074607372283936,
      "learning_rate": 9.659674695357219e-07,
      "loss": 1.9883,
      "num_input_tokens_seen": 91626848,
      "step": 90000
    },
    {
      "epoch": 1.773936327302081,
      "grad_norm": 1.7277181148529053,
      "learning_rate": 9.658926636152277e-07,
      "loss": 1.984,
      "num_input_tokens_seen": 91729248,
      "step": 90100
    },
    {
      "epoch": 1.775905180051584,
      "grad_norm": 2.026073932647705,
      "learning_rate": 9.658177784736155e-07,
      "loss": 1.9806,
      "num_input_tokens_seen": 91830856,
      "step": 90200
    },
    {
      "epoch": 1.777874032801087,
      "grad_norm": 2.0041587352752686,
      "learning_rate": 9.65742814123619e-07,
      "loss": 1.9766,
      "num_input_tokens_seen": 91932720,
      "step": 90300
    },
    {
      "epoch": 1.7798428855505897,
      "grad_norm": 1.86166250705719,
      "learning_rate": 9.656677705779851e-07,
      "loss": 1.9995,
      "num_input_tokens_seen": 92035120,
      "step": 90400
    },
    {
      "epoch": 1.7818117383000924,
      "grad_norm": 1.8686820268630981,
      "learning_rate": 9.655926478494745e-07,
      "loss": 1.9849,
      "num_input_tokens_seen": 92136656,
      "step": 90500
    },
    {
      "epoch": 1.7837805910495954,
      "grad_norm": 2.093017101287842,
      "learning_rate": 9.655174459508612e-07,
      "loss": 1.97,
      "num_input_tokens_seen": 92239056,
      "step": 90600
    },
    {
      "epoch": 1.7857494437990984,
      "grad_norm": 2.0866103172302246,
      "learning_rate": 9.654421648949326e-07,
      "loss": 2.0135,
      "num_input_tokens_seen": 92340088,
      "step": 90700
    },
    {
      "epoch": 1.7877182965486011,
      "grad_norm": 1.6865144968032837,
      "learning_rate": 9.653668046944895e-07,
      "loss": 1.9846,
      "num_input_tokens_seen": 92442488,
      "step": 90800
    },
    {
      "epoch": 1.789687149298104,
      "grad_norm": 1.776648759841919,
      "learning_rate": 9.652913653623463e-07,
      "loss": 2.0015,
      "num_input_tokens_seen": 92544456,
      "step": 90900
    },
    {
      "epoch": 1.7916560020476069,
      "grad_norm": 1.8553825616836548,
      "learning_rate": 9.65215846911331e-07,
      "loss": 1.9725,
      "num_input_tokens_seen": 92645928,
      "step": 91000
    },
    {
      "epoch": 1.7936248547971099,
      "grad_norm": 1.9707273244857788,
      "learning_rate": 9.651402493542845e-07,
      "loss": 1.9466,
      "num_input_tokens_seen": 92748328,
      "step": 91100
    },
    {
      "epoch": 1.7955937075466126,
      "grad_norm": 1.7398895025253296,
      "learning_rate": 9.650645727040622e-07,
      "loss": 1.9408,
      "num_input_tokens_seen": 92849856,
      "step": 91200
    },
    {
      "epoch": 1.7975625602961154,
      "grad_norm": 1.9941409826278687,
      "learning_rate": 9.649888169735316e-07,
      "loss": 2.0098,
      "num_input_tokens_seen": 92951760,
      "step": 91300
    },
    {
      "epoch": 1.7995314130456184,
      "grad_norm": 2.0445821285247803,
      "learning_rate": 9.649129821755744e-07,
      "loss": 1.9681,
      "num_input_tokens_seen": 93053248,
      "step": 91400
    },
    {
      "epoch": 1.8015002657951213,
      "grad_norm": 1.999802589416504,
      "learning_rate": 9.648370683230862e-07,
      "loss": 2.0062,
      "num_input_tokens_seen": 93155264,
      "step": 91500
    },
    {
      "epoch": 1.803469118544624,
      "grad_norm": 1.9985079765319824,
      "learning_rate": 9.64761075428975e-07,
      "loss": 1.9573,
      "num_input_tokens_seen": 93257664,
      "step": 91600
    },
    {
      "epoch": 1.8054379712941269,
      "grad_norm": 1.9642248153686523,
      "learning_rate": 9.646850035061627e-07,
      "loss": 1.9518,
      "num_input_tokens_seen": 93359472,
      "step": 91700
    },
    {
      "epoch": 1.8074068240436296,
      "grad_norm": 2.118866205215454,
      "learning_rate": 9.64608852567585e-07,
      "loss": 1.953,
      "num_input_tokens_seen": 93461872,
      "step": 91800
    },
    {
      "epoch": 1.8093756767931326,
      "grad_norm": 2.0875632762908936,
      "learning_rate": 9.645326226261907e-07,
      "loss": 2.0197,
      "num_input_tokens_seen": 93564272,
      "step": 91900
    },
    {
      "epoch": 1.8113445295426356,
      "grad_norm": 2.062621593475342,
      "learning_rate": 9.64456313694942e-07,
      "loss": 1.9998,
      "num_input_tokens_seen": 93666152,
      "step": 92000
    },
    {
      "epoch": 1.8133133822921383,
      "grad_norm": 1.7994866371154785,
      "learning_rate": 9.643799257868144e-07,
      "loss": 1.9895,
      "num_input_tokens_seen": 93768552,
      "step": 92100
    },
    {
      "epoch": 1.815282235041641,
      "grad_norm": 1.7963272333145142,
      "learning_rate": 9.643034589147974e-07,
      "loss": 1.9744,
      "num_input_tokens_seen": 93870192,
      "step": 92200
    },
    {
      "epoch": 1.817251087791144,
      "grad_norm": 2.3792521953582764,
      "learning_rate": 9.642269130918933e-07,
      "loss": 1.9813,
      "num_input_tokens_seen": 93971720,
      "step": 92300
    },
    {
      "epoch": 1.819219940540647,
      "grad_norm": 1.9264003038406372,
      "learning_rate": 9.64150288331118e-07,
      "loss": 1.9521,
      "num_input_tokens_seen": 94073592,
      "step": 92400
    },
    {
      "epoch": 1.8211887932901498,
      "grad_norm": 2.1435394287109375,
      "learning_rate": 9.640735846455008e-07,
      "loss": 1.9639,
      "num_input_tokens_seen": 94174896,
      "step": 92500
    },
    {
      "epoch": 1.8231576460396526,
      "grad_norm": 1.9926447868347168,
      "learning_rate": 9.639968020480848e-07,
      "loss": 2.0172,
      "num_input_tokens_seen": 94276528,
      "step": 92600
    },
    {
      "epoch": 1.8251264987891556,
      "grad_norm": 1.8766510486602783,
      "learning_rate": 9.639199405519264e-07,
      "loss": 1.9633,
      "num_input_tokens_seen": 94378928,
      "step": 92700
    },
    {
      "epoch": 1.8270953515386585,
      "grad_norm": 1.9852813482284546,
      "learning_rate": 9.638430001700947e-07,
      "loss": 1.9817,
      "num_input_tokens_seen": 94480736,
      "step": 92800
    },
    {
      "epoch": 1.8290642042881613,
      "grad_norm": 1.9052667617797852,
      "learning_rate": 9.637659809156734e-07,
      "loss": 1.9597,
      "num_input_tokens_seen": 94583136,
      "step": 92900
    },
    {
      "epoch": 1.831033057037664,
      "grad_norm": 1.8089063167572021,
      "learning_rate": 9.636888828017583e-07,
      "loss": 1.977,
      "num_input_tokens_seen": 94684968,
      "step": 93000
    },
    {
      "epoch": 1.833001909787167,
      "grad_norm": 2.035827159881592,
      "learning_rate": 9.636117058414597e-07,
      "loss": 2.0445,
      "num_input_tokens_seen": 94787368,
      "step": 93100
    },
    {
      "epoch": 1.83497076253667,
      "grad_norm": 2.105203151702881,
      "learning_rate": 9.635344500479008e-07,
      "loss": 2.0241,
      "num_input_tokens_seen": 94889768,
      "step": 93200
    },
    {
      "epoch": 1.8369396152861728,
      "grad_norm": 1.791733741760254,
      "learning_rate": 9.634571154342185e-07,
      "loss": 1.9712,
      "num_input_tokens_seen": 94991496,
      "step": 93300
    },
    {
      "epoch": 1.8389084680356755,
      "grad_norm": 1.906721830368042,
      "learning_rate": 9.633797020135623e-07,
      "loss": 1.9808,
      "num_input_tokens_seen": 95093896,
      "step": 93400
    },
    {
      "epoch": 1.8408773207851785,
      "grad_norm": 1.7625744342803955,
      "learning_rate": 9.633022097990964e-07,
      "loss": 2.0874,
      "num_input_tokens_seen": 95196296,
      "step": 93500
    },
    {
      "epoch": 1.8428461735346815,
      "grad_norm": 1.900176763534546,
      "learning_rate": 9.632246388039973e-07,
      "loss": 1.9437,
      "num_input_tokens_seen": 95298072,
      "step": 93600
    },
    {
      "epoch": 1.8448150262841843,
      "grad_norm": 2.072072982788086,
      "learning_rate": 9.631469890414555e-07,
      "loss": 1.9883,
      "num_input_tokens_seen": 95399152,
      "step": 93700
    },
    {
      "epoch": 1.846783879033687,
      "grad_norm": 1.7902374267578125,
      "learning_rate": 9.630692605246745e-07,
      "loss": 2.0166,
      "num_input_tokens_seen": 95501552,
      "step": 93800
    },
    {
      "epoch": 1.8487527317831898,
      "grad_norm": 1.8923031091690063,
      "learning_rate": 9.629914532668714e-07,
      "loss": 2.0062,
      "num_input_tokens_seen": 95602872,
      "step": 93900
    },
    {
      "epoch": 1.8507215845326928,
      "grad_norm": 1.8135712146759033,
      "learning_rate": 9.629135672812767e-07,
      "loss": 1.9453,
      "num_input_tokens_seen": 95704456,
      "step": 94000
    },
    {
      "epoch": 1.8526904372821957,
      "grad_norm": 2.003229856491089,
      "learning_rate": 9.628356025811344e-07,
      "loss": 1.9747,
      "num_input_tokens_seen": 95806856,
      "step": 94100
    },
    {
      "epoch": 1.8546592900316985,
      "grad_norm": 1.997440218925476,
      "learning_rate": 9.627575591797014e-07,
      "loss": 1.9613,
      "num_input_tokens_seen": 95909256,
      "step": 94200
    },
    {
      "epoch": 1.8566281427812013,
      "grad_norm": 2.084867477416992,
      "learning_rate": 9.626794370902488e-07,
      "loss": 1.9573,
      "num_input_tokens_seen": 96011280,
      "step": 94300
    },
    {
      "epoch": 1.8585969955307042,
      "grad_norm": 2.002774238586426,
      "learning_rate": 9.626012363260602e-07,
      "loss": 2.0107,
      "num_input_tokens_seen": 96112632,
      "step": 94400
    },
    {
      "epoch": 1.8605658482802072,
      "grad_norm": 1.995097279548645,
      "learning_rate": 9.625229569004332e-07,
      "loss": 2.0194,
      "num_input_tokens_seen": 96215032,
      "step": 94500
    },
    {
      "epoch": 1.86253470102971,
      "grad_norm": 2.048365592956543,
      "learning_rate": 9.624445988266784e-07,
      "loss": 1.9917,
      "num_input_tokens_seen": 96317432,
      "step": 94600
    },
    {
      "epoch": 1.8645035537792127,
      "grad_norm": 1.882153034210205,
      "learning_rate": 9.6236616211812e-07,
      "loss": 1.962,
      "num_input_tokens_seen": 96417968,
      "step": 94700
    },
    {
      "epoch": 1.8664724065287157,
      "grad_norm": 2.021857261657715,
      "learning_rate": 9.622876467880955e-07,
      "loss": 1.9406,
      "num_input_tokens_seen": 96519552,
      "step": 94800
    },
    {
      "epoch": 1.8684412592782187,
      "grad_norm": 1.8111686706542969,
      "learning_rate": 9.62209052849956e-07,
      "loss": 1.9959,
      "num_input_tokens_seen": 96621520,
      "step": 94900
    },
    {
      "epoch": 1.8704101120277214,
      "grad_norm": 1.9492944478988647,
      "learning_rate": 9.621303803170653e-07,
      "loss": 1.9945,
      "num_input_tokens_seen": 96722568,
      "step": 95000
    },
    {
      "epoch": 1.8723789647772242,
      "grad_norm": 2.197702407836914,
      "learning_rate": 9.620516292028013e-07,
      "loss": 2.0119,
      "num_input_tokens_seen": 96824128,
      "step": 95100
    },
    {
      "epoch": 1.8743478175267272,
      "grad_norm": 6.745215892791748,
      "learning_rate": 9.619727995205549e-07,
      "loss": 1.9748,
      "num_input_tokens_seen": 96925128,
      "step": 95200
    },
    {
      "epoch": 1.8763166702762302,
      "grad_norm": 1.797088861465454,
      "learning_rate": 9.618938912837302e-07,
      "loss": 1.9338,
      "num_input_tokens_seen": 97027528,
      "step": 95300
    },
    {
      "epoch": 1.878285523025733,
      "grad_norm": 2.1103646755218506,
      "learning_rate": 9.618149045057453e-07,
      "loss": 1.9623,
      "num_input_tokens_seen": 97129928,
      "step": 95400
    },
    {
      "epoch": 1.8802543757752357,
      "grad_norm": 1.8976510763168335,
      "learning_rate": 9.617358392000308e-07,
      "loss": 1.9466,
      "num_input_tokens_seen": 97232328,
      "step": 95500
    },
    {
      "epoch": 1.8822232285247387,
      "grad_norm": 1.9777162075042725,
      "learning_rate": 9.616566953800315e-07,
      "loss": 1.9678,
      "num_input_tokens_seen": 97334128,
      "step": 95600
    },
    {
      "epoch": 1.8841920812742416,
      "grad_norm": 0.0,
      "learning_rate": 9.615774730592048e-07,
      "loss": 1.9838,
      "num_input_tokens_seen": 97436528,
      "step": 95700
    },
    {
      "epoch": 1.8861609340237444,
      "grad_norm": 2.0475876331329346,
      "learning_rate": 9.61498172251022e-07,
      "loss": 1.9909,
      "num_input_tokens_seen": 97537648,
      "step": 95800
    },
    {
      "epoch": 1.8881297867732472,
      "grad_norm": 1.9707614183425903,
      "learning_rate": 9.614187929689676e-07,
      "loss": 1.9675,
      "num_input_tokens_seen": 97640048,
      "step": 95900
    },
    {
      "epoch": 1.8900986395227501,
      "grad_norm": 2.077984571456909,
      "learning_rate": 9.61339335226539e-07,
      "loss": 1.9872,
      "num_input_tokens_seen": 97742448,
      "step": 96000
    },
    {
      "epoch": 1.892067492272253,
      "grad_norm": 1.7685893774032593,
      "learning_rate": 9.612597990372478e-07,
      "loss": 1.9564,
      "num_input_tokens_seen": 97844272,
      "step": 96100
    },
    {
      "epoch": 1.8940363450217559,
      "grad_norm": 1.7292839288711548,
      "learning_rate": 9.61180184414618e-07,
      "loss": 1.9709,
      "num_input_tokens_seen": 97946672,
      "step": 96200
    },
    {
      "epoch": 1.8960051977712586,
      "grad_norm": 1.8539750576019287,
      "learning_rate": 9.611004913721875e-07,
      "loss": 2.0124,
      "num_input_tokens_seen": 98048792,
      "step": 96300
    },
    {
      "epoch": 1.8979740505207614,
      "grad_norm": 2.281482458114624,
      "learning_rate": 9.61020719923508e-07,
      "loss": 1.9263,
      "num_input_tokens_seen": 98151192,
      "step": 96400
    },
    {
      "epoch": 1.8999429032702644,
      "grad_norm": 2.0945627689361572,
      "learning_rate": 9.609408700821431e-07,
      "loss": 1.9979,
      "num_input_tokens_seen": 98253592,
      "step": 96500
    },
    {
      "epoch": 1.9019117560197674,
      "grad_norm": 2.0845375061035156,
      "learning_rate": 9.608609418616712e-07,
      "loss": 2.0247,
      "num_input_tokens_seen": 98355992,
      "step": 96600
    },
    {
      "epoch": 1.9038806087692701,
      "grad_norm": 1.7269103527069092,
      "learning_rate": 9.607809352756833e-07,
      "loss": 1.9279,
      "num_input_tokens_seen": 98457472,
      "step": 96700
    },
    {
      "epoch": 1.9058494615187729,
      "grad_norm": 1.9988739490509033,
      "learning_rate": 9.607008503377835e-07,
      "loss": 2.002,
      "num_input_tokens_seen": 98559072,
      "step": 96800
    },
    {
      "epoch": 1.9078183142682759,
      "grad_norm": 1.806113839149475,
      "learning_rate": 9.6062068706159e-07,
      "loss": 1.9937,
      "num_input_tokens_seen": 98661472,
      "step": 96900
    },
    {
      "epoch": 1.9097871670177788,
      "grad_norm": 2.1428043842315674,
      "learning_rate": 9.605404454607336e-07,
      "loss": 1.9797,
      "num_input_tokens_seen": 98763872,
      "step": 97000
    },
    {
      "epoch": 1.9117560197672816,
      "grad_norm": 1.758191466331482,
      "learning_rate": 9.60460125548859e-07,
      "loss": 1.9668,
      "num_input_tokens_seen": 98866272,
      "step": 97100
    },
    {
      "epoch": 1.9137248725167844,
      "grad_norm": 2.152453899383545,
      "learning_rate": 9.603797273396238e-07,
      "loss": 1.9813,
      "num_input_tokens_seen": 98967592,
      "step": 97200
    },
    {
      "epoch": 1.9156937252662873,
      "grad_norm": 1.9883763790130615,
      "learning_rate": 9.602992508466988e-07,
      "loss": 2.0022,
      "num_input_tokens_seen": 99069328,
      "step": 97300
    },
    {
      "epoch": 1.9176625780157903,
      "grad_norm": 2.1257636547088623,
      "learning_rate": 9.602186960837686e-07,
      "loss": 1.9448,
      "num_input_tokens_seen": 99171296,
      "step": 97400
    },
    {
      "epoch": 1.919631430765293,
      "grad_norm": 2.0925965309143066,
      "learning_rate": 9.601380630645306e-07,
      "loss": 1.9926,
      "num_input_tokens_seen": 99272304,
      "step": 97500
    },
    {
      "epoch": 1.9216002835147958,
      "grad_norm": 1.9280892610549927,
      "learning_rate": 9.600573518026962e-07,
      "loss": 1.9754,
      "num_input_tokens_seen": 99374704,
      "step": 97600
    },
    {
      "epoch": 1.9235691362642988,
      "grad_norm": 3.2464559078216553,
      "learning_rate": 9.599765623119894e-07,
      "loss": 1.9787,
      "num_input_tokens_seen": 99476232,
      "step": 97700
    },
    {
      "epoch": 1.9255379890138018,
      "grad_norm": 2.031052827835083,
      "learning_rate": 9.598956946061477e-07,
      "loss": 1.997,
      "num_input_tokens_seen": 99578632,
      "step": 97800
    },
    {
      "epoch": 1.9275068417633046,
      "grad_norm": 2.3694489002227783,
      "learning_rate": 9.59814748698922e-07,
      "loss": 2.0419,
      "num_input_tokens_seen": 99680224,
      "step": 97900
    },
    {
      "epoch": 1.9294756945128073,
      "grad_norm": 1.87174391746521,
      "learning_rate": 9.597337246040768e-07,
      "loss": 1.9977,
      "num_input_tokens_seen": 99781520,
      "step": 98000
    },
    {
      "epoch": 1.9314445472623103,
      "grad_norm": 2.0461645126342773,
      "learning_rate": 9.59652622335389e-07,
      "loss": 1.9952,
      "num_input_tokens_seen": 99883256,
      "step": 98100
    },
    {
      "epoch": 1.9334134000118133,
      "grad_norm": 1.7510236501693726,
      "learning_rate": 9.5957144190665e-07,
      "loss": 1.9985,
      "num_input_tokens_seen": 99985656,
      "step": 98200
    },
    {
      "epoch": 1.935382252761316,
      "grad_norm": 1.9866126775741577,
      "learning_rate": 9.594901833316633e-07,
      "loss": 1.9399,
      "num_input_tokens_seen": 100087608,
      "step": 98300
    },
    {
      "epoch": 1.9373511055108188,
      "grad_norm": 1.7863428592681885,
      "learning_rate": 9.594088466242465e-07,
      "loss": 1.9726,
      "num_input_tokens_seen": 100190008,
      "step": 98400
    },
    {
      "epoch": 1.9393199582603216,
      "grad_norm": 1.5792447328567505,
      "learning_rate": 9.593274317982302e-07,
      "loss": 2.0431,
      "num_input_tokens_seen": 100291576,
      "step": 98500
    },
    {
      "epoch": 1.9412888110098245,
      "grad_norm": 26.58302116394043,
      "learning_rate": 9.592459388674584e-07,
      "loss": 2.0123,
      "num_input_tokens_seen": 100393976,
      "step": 98600
    },
    {
      "epoch": 1.9432576637593275,
      "grad_norm": 1.784316897392273,
      "learning_rate": 9.59164367845788e-07,
      "loss": 1.9633,
      "num_input_tokens_seen": 100496376,
      "step": 98700
    },
    {
      "epoch": 1.9452265165088303,
      "grad_norm": 1.9430768489837646,
      "learning_rate": 9.590827187470895e-07,
      "loss": 2.0061,
      "num_input_tokens_seen": 100598016,
      "step": 98800
    },
    {
      "epoch": 1.947195369258333,
      "grad_norm": 1.8332784175872803,
      "learning_rate": 9.590009915852468e-07,
      "loss": 1.9904,
      "num_input_tokens_seen": 100700416,
      "step": 98900
    },
    {
      "epoch": 1.949164222007836,
      "grad_norm": 1.8319400548934937,
      "learning_rate": 9.589191863741569e-07,
      "loss": 1.9649,
      "num_input_tokens_seen": 100802816,
      "step": 99000
    },
    {
      "epoch": 1.951133074757339,
      "grad_norm": 2.0436296463012695,
      "learning_rate": 9.588373031277302e-07,
      "loss": 2.0067,
      "num_input_tokens_seen": 100904712,
      "step": 99100
    },
    {
      "epoch": 1.9531019275068418,
      "grad_norm": 1.6619454622268677,
      "learning_rate": 9.5875534185989e-07,
      "loss": 2.052,
      "num_input_tokens_seen": 101006576,
      "step": 99200
    },
    {
      "epoch": 1.9550707802563445,
      "grad_norm": 1.8762683868408203,
      "learning_rate": 9.586733025845732e-07,
      "loss": 1.9737,
      "num_input_tokens_seen": 101108408,
      "step": 99300
    },
    {
      "epoch": 1.9570396330058475,
      "grad_norm": 2.014247179031372,
      "learning_rate": 9.5859118531573e-07,
      "loss": 1.984,
      "num_input_tokens_seen": 101210200,
      "step": 99400
    },
    {
      "epoch": 1.9590084857553505,
      "grad_norm": 1.9944854974746704,
      "learning_rate": 9.585089900673235e-07,
      "loss": 2.0265,
      "num_input_tokens_seen": 101311160,
      "step": 99500
    },
    {
      "epoch": 1.9609773385048532,
      "grad_norm": 1.8577845096588135,
      "learning_rate": 9.584267168533306e-07,
      "loss": 1.9765,
      "num_input_tokens_seen": 101412824,
      "step": 99600
    },
    {
      "epoch": 1.962946191254356,
      "grad_norm": 1.9040420055389404,
      "learning_rate": 9.583443656877413e-07,
      "loss": 1.9512,
      "num_input_tokens_seen": 101515224,
      "step": 99700
    },
    {
      "epoch": 1.964915044003859,
      "grad_norm": 2.1334547996520996,
      "learning_rate": 9.582619365845582e-07,
      "loss": 2.0074,
      "num_input_tokens_seen": 101617624,
      "step": 99800
    },
    {
      "epoch": 1.966883896753362,
      "grad_norm": 1.7800005674362183,
      "learning_rate": 9.581794295577979e-07,
      "loss": 1.9828,
      "num_input_tokens_seen": 101720024,
      "step": 99900
    },
    {
      "epoch": 1.9688527495028647,
      "grad_norm": 1.7841966152191162,
      "learning_rate": 9.580968446214903e-07,
      "loss": 1.9944,
      "num_input_tokens_seen": 101821616,
      "step": 100000
    },
    {
      "epoch": 1.9708216022523675,
      "grad_norm": 1.8359742164611816,
      "learning_rate": 9.580141817896779e-07,
      "loss": 1.9757,
      "num_input_tokens_seen": 101923464,
      "step": 100100
    },
    {
      "epoch": 1.9727904550018704,
      "grad_norm": 2.160520553588867,
      "learning_rate": 9.579314410764168e-07,
      "loss": 2.0337,
      "num_input_tokens_seen": 102025192,
      "step": 100200
    },
    {
      "epoch": 1.9747593077513734,
      "grad_norm": 1.8562108278274536,
      "learning_rate": 9.578486224957768e-07,
      "loss": 2.0133,
      "num_input_tokens_seen": 102127592,
      "step": 100300
    },
    {
      "epoch": 1.9767281605008762,
      "grad_norm": 2.1131303310394287,
      "learning_rate": 9.577657260618401e-07,
      "loss": 1.9646,
      "num_input_tokens_seen": 102229992,
      "step": 100400
    },
    {
      "epoch": 1.978697013250379,
      "grad_norm": 1.9324053525924683,
      "learning_rate": 9.576827517887024e-07,
      "loss": 2.0036,
      "num_input_tokens_seen": 102331856,
      "step": 100500
    },
    {
      "epoch": 1.9806658659998817,
      "grad_norm": 3.1988070011138916,
      "learning_rate": 9.575996996904733e-07,
      "loss": 1.996,
      "num_input_tokens_seen": 102434256,
      "step": 100600
    },
    {
      "epoch": 1.9826347187493847,
      "grad_norm": 2.1580376625061035,
      "learning_rate": 9.57516569781275e-07,
      "loss": 1.9995,
      "num_input_tokens_seen": 102536296,
      "step": 100700
    },
    {
      "epoch": 1.9846035714988877,
      "grad_norm": 1.9465465545654297,
      "learning_rate": 9.574333620752426e-07,
      "loss": 2.0036,
      "num_input_tokens_seen": 102638080,
      "step": 100800
    },
    {
      "epoch": 1.9865724242483904,
      "grad_norm": 2.174964189529419,
      "learning_rate": 9.573500765865253e-07,
      "loss": 1.939,
      "num_input_tokens_seen": 102740480,
      "step": 100900
    },
    {
      "epoch": 1.9885412769978932,
      "grad_norm": 2.338625431060791,
      "learning_rate": 9.572667133292849e-07,
      "loss": 1.9403,
      "num_input_tokens_seen": 102842336,
      "step": 101000
    },
    {
      "epoch": 1.9905101297473962,
      "grad_norm": 1.8449496030807495,
      "learning_rate": 9.571832723176967e-07,
      "loss": 2.0347,
      "num_input_tokens_seen": 102942616,
      "step": 101100
    },
    {
      "epoch": 1.9924789824968991,
      "grad_norm": 1.9634342193603516,
      "learning_rate": 9.57099753565949e-07,
      "loss": 1.9701,
      "num_input_tokens_seen": 103044392,
      "step": 101200
    },
    {
      "epoch": 1.994447835246402,
      "grad_norm": 1.7099233865737915,
      "learning_rate": 9.570161570882436e-07,
      "loss": 1.9833,
      "num_input_tokens_seen": 103146792,
      "step": 101300
    },
    {
      "epoch": 1.9964166879959047,
      "grad_norm": 2.008478879928589,
      "learning_rate": 9.569324828987955e-07,
      "loss": 1.9858,
      "num_input_tokens_seen": 103247832,
      "step": 101400
    },
    {
      "epoch": 1.9983855407454076,
      "grad_norm": 1.8442299365997314,
      "learning_rate": 9.568487310118324e-07,
      "loss": 1.9665,
      "num_input_tokens_seen": 103348552,
      "step": 101500
    },
    {
      "epoch": 2.0003543934949106,
      "grad_norm": 1.7430541515350342,
      "learning_rate": 9.56764901441596e-07,
      "loss": 1.9641,
      "num_input_tokens_seen": 103450656,
      "step": 101600
    },
    {
      "epoch": 2.0023232462444134,
      "grad_norm": 2.0603220462799072,
      "learning_rate": 9.566809942023405e-07,
      "loss": 1.9251,
      "num_input_tokens_seen": 103553056,
      "step": 101700
    },
    {
      "epoch": 2.004292098993916,
      "grad_norm": 2.089635133743286,
      "learning_rate": 9.56597009308334e-07,
      "loss": 2.0007,
      "num_input_tokens_seen": 103654728,
      "step": 101800
    },
    {
      "epoch": 2.006260951743419,
      "grad_norm": 1.8020271062850952,
      "learning_rate": 9.56512946773857e-07,
      "loss": 1.9986,
      "num_input_tokens_seen": 103756528,
      "step": 101900
    },
    {
      "epoch": 2.008229804492922,
      "grad_norm": 1.8193224668502808,
      "learning_rate": 9.564288066132037e-07,
      "loss": 1.9309,
      "num_input_tokens_seen": 103858584,
      "step": 102000
    },
    {
      "epoch": 2.010198657242425,
      "grad_norm": 2.004086971282959,
      "learning_rate": 9.563445888406815e-07,
      "loss": 1.9526,
      "num_input_tokens_seen": 103960984,
      "step": 102100
    },
    {
      "epoch": 2.0121675099919276,
      "grad_norm": 1.7387619018554688,
      "learning_rate": 9.562602934706112e-07,
      "loss": 2.0099,
      "num_input_tokens_seen": 104062488,
      "step": 102200
    },
    {
      "epoch": 2.0141363627414304,
      "grad_norm": 1.8126006126403809,
      "learning_rate": 9.561759205173264e-07,
      "loss": 2.0142,
      "num_input_tokens_seen": 104163696,
      "step": 102300
    },
    {
      "epoch": 2.0161052154909336,
      "grad_norm": 2.117628335952759,
      "learning_rate": 9.560914699951736e-07,
      "loss": 1.9843,
      "num_input_tokens_seen": 104266096,
      "step": 102400
    },
    {
      "epoch": 2.0180740682404363,
      "grad_norm": 1.9752012491226196,
      "learning_rate": 9.560069419185131e-07,
      "loss": 1.9188,
      "num_input_tokens_seen": 104368496,
      "step": 102500
    },
    {
      "epoch": 2.020042920989939,
      "grad_norm": 1.9322516918182373,
      "learning_rate": 9.559223363017184e-07,
      "loss": 1.9917,
      "num_input_tokens_seen": 104467432,
      "step": 102600
    },
    {
      "epoch": 2.022011773739442,
      "grad_norm": 1.8220409154891968,
      "learning_rate": 9.558376531591758e-07,
      "loss": 2.0283,
      "num_input_tokens_seen": 104569368,
      "step": 102700
    },
    {
      "epoch": 2.023980626488945,
      "grad_norm": 2.0056328773498535,
      "learning_rate": 9.55752892505285e-07,
      "loss": 2.0213,
      "num_input_tokens_seen": 104669888,
      "step": 102800
    },
    {
      "epoch": 2.025949479238448,
      "grad_norm": 1.9241981506347656,
      "learning_rate": 9.556680543544591e-07,
      "loss": 2.0665,
      "num_input_tokens_seen": 104771376,
      "step": 102900
    },
    {
      "epoch": 2.0279183319879506,
      "grad_norm": 1.925260305404663,
      "learning_rate": 9.555831387211234e-07,
      "loss": 1.9477,
      "num_input_tokens_seen": 104873776,
      "step": 103000
    },
    {
      "epoch": 2.0298871847374533,
      "grad_norm": 1.9352415800094604,
      "learning_rate": 9.55498145619718e-07,
      "loss": 2.0178,
      "num_input_tokens_seen": 104976176,
      "step": 103100
    },
    {
      "epoch": 2.0318560374869565,
      "grad_norm": 1.9706671237945557,
      "learning_rate": 9.554130750646946e-07,
      "loss": 1.9767,
      "num_input_tokens_seen": 105078032,
      "step": 103200
    },
    {
      "epoch": 2.0338248902364593,
      "grad_norm": 1.8645918369293213,
      "learning_rate": 9.553279270705186e-07,
      "loss": 1.9914,
      "num_input_tokens_seen": 105180432,
      "step": 103300
    },
    {
      "epoch": 2.035793742985962,
      "grad_norm": 1.78374183177948,
      "learning_rate": 9.552427016516692e-07,
      "loss": 1.9623,
      "num_input_tokens_seen": 105282832,
      "step": 103400
    },
    {
      "epoch": 2.037762595735465,
      "grad_norm": 2.105135202407837,
      "learning_rate": 9.551573988226382e-07,
      "loss": 1.9699,
      "num_input_tokens_seen": 105385232,
      "step": 103500
    },
    {
      "epoch": 2.039731448484968,
      "grad_norm": 2.1146798133850098,
      "learning_rate": 9.550720185979306e-07,
      "loss": 2.0081,
      "num_input_tokens_seen": 105487632,
      "step": 103600
    },
    {
      "epoch": 2.0417003012344708,
      "grad_norm": 1.8652223348617554,
      "learning_rate": 9.549865609920642e-07,
      "loss": 2.0589,
      "num_input_tokens_seen": 105588200,
      "step": 103700
    },
    {
      "epoch": 2.0436691539839735,
      "grad_norm": 1.7666460275650024,
      "learning_rate": 9.54901026019571e-07,
      "loss": 1.9854,
      "num_input_tokens_seen": 105689824,
      "step": 103800
    },
    {
      "epoch": 2.0456380067334763,
      "grad_norm": 2.112034559249878,
      "learning_rate": 9.548154136949948e-07,
      "loss": 1.9903,
      "num_input_tokens_seen": 105791384,
      "step": 103900
    },
    {
      "epoch": 2.047606859482979,
      "grad_norm": 2.0934624671936035,
      "learning_rate": 9.547297240328937e-07,
      "loss": 2.0259,
      "num_input_tokens_seen": 105893784,
      "step": 104000
    },
    {
      "epoch": 2.0495757122324822,
      "grad_norm": 2.0104362964630127,
      "learning_rate": 9.546439570478383e-07,
      "loss": 1.9472,
      "num_input_tokens_seen": 105996184,
      "step": 104100
    },
    {
      "epoch": 2.051544564981985,
      "grad_norm": 1.9536254405975342,
      "learning_rate": 9.545581127544129e-07,
      "loss": 2.0228,
      "num_input_tokens_seen": 106097432,
      "step": 104200
    },
    {
      "epoch": 2.0535134177314878,
      "grad_norm": 1.8959099054336548,
      "learning_rate": 9.54472191167214e-07,
      "loss": 1.9946,
      "num_input_tokens_seen": 106197552,
      "step": 104300
    },
    {
      "epoch": 2.0554822704809905,
      "grad_norm": 1.8936536312103271,
      "learning_rate": 9.543861923008526e-07,
      "loss": 1.9822,
      "num_input_tokens_seen": 106299952,
      "step": 104400
    },
    {
      "epoch": 2.0574511232304937,
      "grad_norm": 1.8303178548812866,
      "learning_rate": 9.543001161699514e-07,
      "loss": 1.9786,
      "num_input_tokens_seen": 106401848,
      "step": 104500
    },
    {
      "epoch": 2.0594199759799965,
      "grad_norm": 1.9874995946884155,
      "learning_rate": 9.54213962789147e-07,
      "loss": 2.0032,
      "num_input_tokens_seen": 106502624,
      "step": 104600
    },
    {
      "epoch": 2.0613888287294992,
      "grad_norm": 1.8439503908157349,
      "learning_rate": 9.541277321730896e-07,
      "loss": 2.0316,
      "num_input_tokens_seen": 106604280,
      "step": 104700
    },
    {
      "epoch": 2.063357681479002,
      "grad_norm": 1.7990156412124634,
      "learning_rate": 9.540414243364415e-07,
      "loss": 1.9838,
      "num_input_tokens_seen": 106705936,
      "step": 104800
    },
    {
      "epoch": 2.065326534228505,
      "grad_norm": 2.1370999813079834,
      "learning_rate": 9.539550392938787e-07,
      "loss": 1.9563,
      "num_input_tokens_seen": 106808336,
      "step": 104900
    },
    {
      "epoch": 2.067295386978008,
      "grad_norm": 1.9992778301239014,
      "learning_rate": 9.538685770600906e-07,
      "loss": 2.0396,
      "num_input_tokens_seen": 106910296,
      "step": 105000
    },
    {
      "epoch": 2.0692642397275107,
      "grad_norm": 1.9784587621688843,
      "learning_rate": 9.537820376497787e-07,
      "loss": 1.9999,
      "num_input_tokens_seen": 107012696,
      "step": 105100
    },
    {
      "epoch": 2.0712330924770135,
      "grad_norm": 1.8854153156280518,
      "learning_rate": 9.536954210776588e-07,
      "loss": 1.9577,
      "num_input_tokens_seen": 107115096,
      "step": 105200
    },
    {
      "epoch": 2.0732019452265167,
      "grad_norm": 1.8609813451766968,
      "learning_rate": 9.536087273584592e-07,
      "loss": 2.0351,
      "num_input_tokens_seen": 107215840,
      "step": 105300
    },
    {
      "epoch": 2.0751707979760194,
      "grad_norm": 1.9522591829299927,
      "learning_rate": 9.535219565069215e-07,
      "loss": 1.9819,
      "num_input_tokens_seen": 107317464,
      "step": 105400
    },
    {
      "epoch": 2.077139650725522,
      "grad_norm": 2.1145567893981934,
      "learning_rate": 9.534351085378001e-07,
      "loss": 2.0,
      "num_input_tokens_seen": 107419056,
      "step": 105500
    },
    {
      "epoch": 2.079108503475025,
      "grad_norm": 2.0271804332733154,
      "learning_rate": 9.533481834658631e-07,
      "loss": 1.953,
      "num_input_tokens_seen": 107521456,
      "step": 105600
    },
    {
      "epoch": 2.081077356224528,
      "grad_norm": 1.906770944595337,
      "learning_rate": 9.532611813058912e-07,
      "loss": 1.9844,
      "num_input_tokens_seen": 107623856,
      "step": 105700
    },
    {
      "epoch": 2.083046208974031,
      "grad_norm": 2.354865074157715,
      "learning_rate": 9.531741020726783e-07,
      "loss": 2.0101,
      "num_input_tokens_seen": 107726256,
      "step": 105800
    },
    {
      "epoch": 2.0850150617235337,
      "grad_norm": 2.017036199569702,
      "learning_rate": 9.530869457810316e-07,
      "loss": 2.0208,
      "num_input_tokens_seen": 107827480,
      "step": 105900
    },
    {
      "epoch": 2.0869839144730364,
      "grad_norm": 2.106940269470215,
      "learning_rate": 9.529997124457713e-07,
      "loss": 1.9645,
      "num_input_tokens_seen": 107929880,
      "step": 106000
    },
    {
      "epoch": 2.0889527672225396,
      "grad_norm": 1.7717925310134888,
      "learning_rate": 9.529124020817308e-07,
      "loss": 1.9071,
      "num_input_tokens_seen": 108032280,
      "step": 106100
    },
    {
      "epoch": 2.0909216199720424,
      "grad_norm": 1.8631401062011719,
      "learning_rate": 9.528250147037563e-07,
      "loss": 2.0584,
      "num_input_tokens_seen": 108134680,
      "step": 106200
    },
    {
      "epoch": 2.092890472721545,
      "grad_norm": 1.8711248636245728,
      "learning_rate": 9.527375503267073e-07,
      "loss": 1.9705,
      "num_input_tokens_seen": 108236464,
      "step": 106300
    },
    {
      "epoch": 2.094859325471048,
      "grad_norm": 1.8259410858154297,
      "learning_rate": 9.526500089654565e-07,
      "loss": 1.9693,
      "num_input_tokens_seen": 108338864,
      "step": 106400
    },
    {
      "epoch": 2.0968281782205507,
      "grad_norm": 2.248173713684082,
      "learning_rate": 9.525623906348895e-07,
      "loss": 1.9823,
      "num_input_tokens_seen": 108440848,
      "step": 106500
    },
    {
      "epoch": 2.098797030970054,
      "grad_norm": 1.7875350713729858,
      "learning_rate": 9.52474695349905e-07,
      "loss": 2.0181,
      "num_input_tokens_seen": 108542304,
      "step": 106600
    },
    {
      "epoch": 2.1007658837195566,
      "grad_norm": 2.09037709236145,
      "learning_rate": 9.52386923125415e-07,
      "loss": 2.0001,
      "num_input_tokens_seen": 108644704,
      "step": 106700
    },
    {
      "epoch": 2.1027347364690594,
      "grad_norm": 1.9783692359924316,
      "learning_rate": 9.522990739763442e-07,
      "loss": 1.9933,
      "num_input_tokens_seen": 108747104,
      "step": 106800
    },
    {
      "epoch": 2.104703589218562,
      "grad_norm": 2.1536645889282227,
      "learning_rate": 9.522111479176307e-07,
      "loss": 2.0328,
      "num_input_tokens_seen": 108849504,
      "step": 106900
    },
    {
      "epoch": 2.1066724419680654,
      "grad_norm": 1.737360954284668,
      "learning_rate": 9.521231449642257e-07,
      "loss": 1.9932,
      "num_input_tokens_seen": 108950008,
      "step": 107000
    },
    {
      "epoch": 2.108641294717568,
      "grad_norm": 1.7101801633834839,
      "learning_rate": 9.520350651310934e-07,
      "loss": 1.9547,
      "num_input_tokens_seen": 109052408,
      "step": 107100
    },
    {
      "epoch": 2.110610147467071,
      "grad_norm": 1.784883737564087,
      "learning_rate": 9.519469084332108e-07,
      "loss": 1.9893,
      "num_input_tokens_seen": 109154376,
      "step": 107200
    },
    {
      "epoch": 2.1125790002165736,
      "grad_norm": 2.0286502838134766,
      "learning_rate": 9.518586748855683e-07,
      "loss": 1.9651,
      "num_input_tokens_seen": 109256208,
      "step": 107300
    },
    {
      "epoch": 2.114547852966077,
      "grad_norm": 1.8560724258422852,
      "learning_rate": 9.517703645031693e-07,
      "loss": 1.965,
      "num_input_tokens_seen": 109358608,
      "step": 107400
    },
    {
      "epoch": 2.1165167057155796,
      "grad_norm": 1.9718207120895386,
      "learning_rate": 9.516819773010303e-07,
      "loss": 2.0604,
      "num_input_tokens_seen": 109459968,
      "step": 107500
    },
    {
      "epoch": 2.1184855584650824,
      "grad_norm": 1.9123685359954834,
      "learning_rate": 9.515935132941806e-07,
      "loss": 1.9922,
      "num_input_tokens_seen": 109561296,
      "step": 107600
    },
    {
      "epoch": 2.120454411214585,
      "grad_norm": 1.8966621160507202,
      "learning_rate": 9.515049724976629e-07,
      "loss": 1.9399,
      "num_input_tokens_seen": 109663696,
      "step": 107700
    },
    {
      "epoch": 2.1224232639640883,
      "grad_norm": 1.7923712730407715,
      "learning_rate": 9.514163549265329e-07,
      "loss": 1.971,
      "num_input_tokens_seen": 109765912,
      "step": 107800
    },
    {
      "epoch": 2.124392116713591,
      "grad_norm": 2.306278944015503,
      "learning_rate": 9.513276605958592e-07,
      "loss": 1.9789,
      "num_input_tokens_seen": 109868312,
      "step": 107900
    },
    {
      "epoch": 2.126360969463094,
      "grad_norm": 1.8584436178207397,
      "learning_rate": 9.512388895207233e-07,
      "loss": 1.9907,
      "num_input_tokens_seen": 109970712,
      "step": 108000
    },
    {
      "epoch": 2.1283298222125966,
      "grad_norm": 2.0306830406188965,
      "learning_rate": 9.511500417162203e-07,
      "loss": 1.9896,
      "num_input_tokens_seen": 110071440,
      "step": 108100
    },
    {
      "epoch": 2.1302986749621,
      "grad_norm": 1.6810091733932495,
      "learning_rate": 9.510611171974579e-07,
      "loss": 2.0375,
      "num_input_tokens_seen": 110171600,
      "step": 108200
    },
    {
      "epoch": 2.1322675277116026,
      "grad_norm": 4.35719633102417,
      "learning_rate": 9.509721159795567e-07,
      "loss": 1.9479,
      "num_input_tokens_seen": 110272832,
      "step": 108300
    },
    {
      "epoch": 2.1342363804611053,
      "grad_norm": 2.484416961669922,
      "learning_rate": 9.508830380776512e-07,
      "loss": 2.0175,
      "num_input_tokens_seen": 110375232,
      "step": 108400
    },
    {
      "epoch": 2.136205233210608,
      "grad_norm": 1.8578368425369263,
      "learning_rate": 9.507938835068877e-07,
      "loss": 1.9634,
      "num_input_tokens_seen": 110477632,
      "step": 108500
    },
    {
      "epoch": 2.1381740859601113,
      "grad_norm": 1.989546537399292,
      "learning_rate": 9.507046522824267e-07,
      "loss": 2.0112,
      "num_input_tokens_seen": 110579904,
      "step": 108600
    },
    {
      "epoch": 2.140142938709614,
      "grad_norm": 2.319326162338257,
      "learning_rate": 9.506153444194409e-07,
      "loss": 1.9663,
      "num_input_tokens_seen": 110682304,
      "step": 108700
    },
    {
      "epoch": 2.142111791459117,
      "grad_norm": 1.910566806793213,
      "learning_rate": 9.505259599331164e-07,
      "loss": 1.9984,
      "num_input_tokens_seen": 110784704,
      "step": 108800
    },
    {
      "epoch": 2.1440806442086195,
      "grad_norm": 2.196389675140381,
      "learning_rate": 9.504364988386524e-07,
      "loss": 2.0161,
      "num_input_tokens_seen": 110885160,
      "step": 108900
    },
    {
      "epoch": 2.1460494969581223,
      "grad_norm": 2.400207757949829,
      "learning_rate": 9.503469611512609e-07,
      "loss": 2.0052,
      "num_input_tokens_seen": 110986720,
      "step": 109000
    },
    {
      "epoch": 2.1480183497076255,
      "grad_norm": 1.8510814905166626,
      "learning_rate": 9.50257346886167e-07,
      "loss": 2.0,
      "num_input_tokens_seen": 111087864,
      "step": 109100
    },
    {
      "epoch": 2.1499872024571283,
      "grad_norm": 2.4665424823760986,
      "learning_rate": 9.501676560586091e-07,
      "loss": 1.9725,
      "num_input_tokens_seen": 111188912,
      "step": 109200
    },
    {
      "epoch": 2.151956055206631,
      "grad_norm": 1.8784955739974976,
      "learning_rate": 9.50077888683838e-07,
      "loss": 1.9766,
      "num_input_tokens_seen": 111290312,
      "step": 109300
    },
    {
      "epoch": 2.153924907956134,
      "grad_norm": 2.3580844402313232,
      "learning_rate": 9.49988044777118e-07,
      "loss": 1.9844,
      "num_input_tokens_seen": 111392712,
      "step": 109400
    },
    {
      "epoch": 2.155893760705637,
      "grad_norm": 2.142921209335327,
      "learning_rate": 9.498981243537265e-07,
      "loss": 2.008,
      "num_input_tokens_seen": 111492632,
      "step": 109500
    },
    {
      "epoch": 2.1578626134551397,
      "grad_norm": 2.0246009826660156,
      "learning_rate": 9.498081274289535e-07,
      "loss": 1.9994,
      "num_input_tokens_seen": 111594104,
      "step": 109600
    },
    {
      "epoch": 2.1598314662046425,
      "grad_norm": 2.277479887008667,
      "learning_rate": 9.497180540181024e-07,
      "loss": 1.9804,
      "num_input_tokens_seen": 111695128,
      "step": 109700
    },
    {
      "epoch": 2.1618003189541453,
      "grad_norm": 1.8636360168457031,
      "learning_rate": 9.496279041364893e-07,
      "loss": 1.954,
      "num_input_tokens_seen": 111796760,
      "step": 109800
    },
    {
      "epoch": 2.1637691717036485,
      "grad_norm": 1.7724113464355469,
      "learning_rate": 9.495376777994434e-07,
      "loss": 1.9674,
      "num_input_tokens_seen": 111899160,
      "step": 109900
    },
    {
      "epoch": 2.1657380244531512,
      "grad_norm": 2.1405045986175537,
      "learning_rate": 9.494473750223069e-07,
      "loss": 2.0086,
      "num_input_tokens_seen": 112000976,
      "step": 110000
    },
    {
      "epoch": 2.167706877202654,
      "grad_norm": 1.9437774419784546,
      "learning_rate": 9.493569958204352e-07,
      "loss": 1.9913,
      "num_input_tokens_seen": 112103376,
      "step": 110100
    },
    {
      "epoch": 2.1696757299521567,
      "grad_norm": 1.9000684022903442,
      "learning_rate": 9.492665402091964e-07,
      "loss": 1.9962,
      "num_input_tokens_seen": 112205776,
      "step": 110200
    },
    {
      "epoch": 2.17164458270166,
      "grad_norm": 2.1248345375061035,
      "learning_rate": 9.491760082039716e-07,
      "loss": 1.9823,
      "num_input_tokens_seen": 112307224,
      "step": 110300
    },
    {
      "epoch": 2.1736134354511627,
      "grad_norm": 2.0565896034240723,
      "learning_rate": 9.490853998201551e-07,
      "loss": 1.9862,
      "num_input_tokens_seen": 112409624,
      "step": 110400
    },
    {
      "epoch": 2.1755822882006655,
      "grad_norm": 2.0068960189819336,
      "learning_rate": 9.489947150731544e-07,
      "loss": 1.9418,
      "num_input_tokens_seen": 112512024,
      "step": 110500
    },
    {
      "epoch": 2.177551140950168,
      "grad_norm": 1.938106894493103,
      "learning_rate": 9.489039539783892e-07,
      "loss": 1.9478,
      "num_input_tokens_seen": 112613608,
      "step": 110600
    },
    {
      "epoch": 2.179519993699671,
      "grad_norm": 1.8697112798690796,
      "learning_rate": 9.488131165512928e-07,
      "loss": 1.9644,
      "num_input_tokens_seen": 112715192,
      "step": 110700
    },
    {
      "epoch": 2.181488846449174,
      "grad_norm": 2.2095909118652344,
      "learning_rate": 9.487222028073114e-07,
      "loss": 1.9624,
      "num_input_tokens_seen": 112817592,
      "step": 110800
    },
    {
      "epoch": 2.183457699198677,
      "grad_norm": 1.6604583263397217,
      "learning_rate": 9.486312127619043e-07,
      "loss": 1.9457,
      "num_input_tokens_seen": 112919992,
      "step": 110900
    },
    {
      "epoch": 2.1854265519481797,
      "grad_norm": 1.733359694480896,
      "learning_rate": 9.485401464305433e-07,
      "loss": 1.9655,
      "num_input_tokens_seen": 113022392,
      "step": 111000
    },
    {
      "epoch": 2.1873954046976825,
      "grad_norm": 1.8730971813201904,
      "learning_rate": 9.484490038287135e-07,
      "loss": 1.9608,
      "num_input_tokens_seen": 113124792,
      "step": 111100
    },
    {
      "epoch": 2.1893642574471857,
      "grad_norm": 2.102570056915283,
      "learning_rate": 9.483577849719131e-07,
      "loss": 1.9875,
      "num_input_tokens_seen": 113226256,
      "step": 111200
    },
    {
      "epoch": 2.1913331101966884,
      "grad_norm": 1.8639987707138062,
      "learning_rate": 9.482664898756529e-07,
      "loss": 2.0006,
      "num_input_tokens_seen": 113327992,
      "step": 111300
    },
    {
      "epoch": 2.193301962946191,
      "grad_norm": 1.9622057676315308,
      "learning_rate": 9.481751185554569e-07,
      "loss": 1.9812,
      "num_input_tokens_seen": 113429528,
      "step": 111400
    },
    {
      "epoch": 2.195270815695694,
      "grad_norm": 1.8260507583618164,
      "learning_rate": 9.480836710268623e-07,
      "loss": 2.0086,
      "num_input_tokens_seen": 113531232,
      "step": 111500
    },
    {
      "epoch": 2.197239668445197,
      "grad_norm": 1.7977197170257568,
      "learning_rate": 9.479921473054185e-07,
      "loss": 2.004,
      "num_input_tokens_seen": 113633632,
      "step": 111600
    },
    {
      "epoch": 2.1992085211947,
      "grad_norm": 1.9349136352539062,
      "learning_rate": 9.479005474066888e-07,
      "loss": 1.9878,
      "num_input_tokens_seen": 113736032,
      "step": 111700
    },
    {
      "epoch": 2.2011773739442027,
      "grad_norm": 1.8296550512313843,
      "learning_rate": 9.478088713462487e-07,
      "loss": 1.9464,
      "num_input_tokens_seen": 113838432,
      "step": 111800
    },
    {
      "epoch": 2.2031462266937054,
      "grad_norm": 2.0176846981048584,
      "learning_rate": 9.477171191396868e-07,
      "loss": 1.941,
      "num_input_tokens_seen": 113940256,
      "step": 111900
    },
    {
      "epoch": 2.2051150794432086,
      "grad_norm": 1.746730089187622,
      "learning_rate": 9.476252908026054e-07,
      "loss": 2.0294,
      "num_input_tokens_seen": 114042344,
      "step": 112000
    },
    {
      "epoch": 2.2070839321927114,
      "grad_norm": 2.1340911388397217,
      "learning_rate": 9.475333863506185e-07,
      "loss": 1.9986,
      "num_input_tokens_seen": 114144304,
      "step": 112100
    },
    {
      "epoch": 2.209052784942214,
      "grad_norm": 1.737505316734314,
      "learning_rate": 9.474414057993541e-07,
      "loss": 1.9797,
      "num_input_tokens_seen": 114246184,
      "step": 112200
    },
    {
      "epoch": 2.211021637691717,
      "grad_norm": 1.732858657836914,
      "learning_rate": 9.473493491644523e-07,
      "loss": 1.9783,
      "num_input_tokens_seen": 114347368,
      "step": 112300
    },
    {
      "epoch": 2.21299049044122,
      "grad_norm": 1.7610417604446411,
      "learning_rate": 9.472572164615668e-07,
      "loss": 1.9963,
      "num_input_tokens_seen": 114449064,
      "step": 112400
    },
    {
      "epoch": 2.214959343190723,
      "grad_norm": 1.9439780712127686,
      "learning_rate": 9.471650077063642e-07,
      "loss": 2.0007,
      "num_input_tokens_seen": 114550840,
      "step": 112500
    },
    {
      "epoch": 2.2169281959402256,
      "grad_norm": 2.0145537853240967,
      "learning_rate": 9.470727229145235e-07,
      "loss": 2.0029,
      "num_input_tokens_seen": 114653240,
      "step": 112600
    },
    {
      "epoch": 2.2188970486897284,
      "grad_norm": 2.5199801921844482,
      "learning_rate": 9.469803621017371e-07,
      "loss": 1.9654,
      "num_input_tokens_seen": 114755120,
      "step": 112700
    },
    {
      "epoch": 2.2208659014392316,
      "grad_norm": 1.873188853263855,
      "learning_rate": 9.468879252837101e-07,
      "loss": 1.9523,
      "num_input_tokens_seen": 114856672,
      "step": 112800
    },
    {
      "epoch": 2.2228347541887343,
      "grad_norm": 1.9388569593429565,
      "learning_rate": 9.467954124761606e-07,
      "loss": 2.0349,
      "num_input_tokens_seen": 114958008,
      "step": 112900
    },
    {
      "epoch": 2.224803606938237,
      "grad_norm": 2.1790852546691895,
      "learning_rate": 9.467028236948197e-07,
      "loss": 1.9762,
      "num_input_tokens_seen": 115060408,
      "step": 113000
    },
    {
      "epoch": 2.22677245968774,
      "grad_norm": 2.057921886444092,
      "learning_rate": 9.466101589554312e-07,
      "loss": 1.9801,
      "num_input_tokens_seen": 115161952,
      "step": 113100
    },
    {
      "epoch": 2.2287413124372426,
      "grad_norm": 1.8343178033828735,
      "learning_rate": 9.465174182737521e-07,
      "loss": 1.9496,
      "num_input_tokens_seen": 115264352,
      "step": 113200
    },
    {
      "epoch": 2.230710165186746,
      "grad_norm": 1.61873459815979,
      "learning_rate": 9.464246016655522e-07,
      "loss": 1.9687,
      "num_input_tokens_seen": 115366752,
      "step": 113300
    },
    {
      "epoch": 2.2326790179362486,
      "grad_norm": 1.95689058303833,
      "learning_rate": 9.463317091466141e-07,
      "loss": 2.0141,
      "num_input_tokens_seen": 115469152,
      "step": 113400
    },
    {
      "epoch": 2.2346478706857513,
      "grad_norm": 1.8117475509643555,
      "learning_rate": 9.462387407327332e-07,
      "loss": 1.9562,
      "num_input_tokens_seen": 115571552,
      "step": 113500
    },
    {
      "epoch": 2.236616723435254,
      "grad_norm": 1.9498783349990845,
      "learning_rate": 9.461456964397184e-07,
      "loss": 1.9446,
      "num_input_tokens_seen": 115673952,
      "step": 113600
    },
    {
      "epoch": 2.2385855761847573,
      "grad_norm": 1.789787769317627,
      "learning_rate": 9.460525762833908e-07,
      "loss": 2.0464,
      "num_input_tokens_seen": 115775504,
      "step": 113700
    },
    {
      "epoch": 2.24055442893426,
      "grad_norm": 3.6420681476593018,
      "learning_rate": 9.459593802795849e-07,
      "loss": 1.9711,
      "num_input_tokens_seen": 115877296,
      "step": 113800
    },
    {
      "epoch": 2.242523281683763,
      "grad_norm": 1.8349014520645142,
      "learning_rate": 9.458661084441478e-07,
      "loss": 1.9621,
      "num_input_tokens_seen": 115979200,
      "step": 113900
    },
    {
      "epoch": 2.2444921344332656,
      "grad_norm": 1.928267478942871,
      "learning_rate": 9.457727607929393e-07,
      "loss": 2.015,
      "num_input_tokens_seen": 116081096,
      "step": 114000
    },
    {
      "epoch": 2.2464609871827688,
      "grad_norm": 2.3008081912994385,
      "learning_rate": 9.456793373418328e-07,
      "loss": 1.9905,
      "num_input_tokens_seen": 116182640,
      "step": 114100
    },
    {
      "epoch": 2.2484298399322715,
      "grad_norm": 2.4185616970062256,
      "learning_rate": 9.455858381067141e-07,
      "loss": 1.9803,
      "num_input_tokens_seen": 116285040,
      "step": 114200
    },
    {
      "epoch": 2.2503986926817743,
      "grad_norm": 2.0490100383758545,
      "learning_rate": 9.454922631034818e-07,
      "loss": 2.0164,
      "num_input_tokens_seen": 116387440,
      "step": 114300
    },
    {
      "epoch": 2.252367545431277,
      "grad_norm": 1.9811676740646362,
      "learning_rate": 9.453986123480476e-07,
      "loss": 2.0168,
      "num_input_tokens_seen": 116489184,
      "step": 114400
    },
    {
      "epoch": 2.2543363981807802,
      "grad_norm": 2.4179511070251465,
      "learning_rate": 9.453048858563359e-07,
      "loss": 2.0379,
      "num_input_tokens_seen": 116591584,
      "step": 114500
    },
    {
      "epoch": 2.256305250930283,
      "grad_norm": 1.692130208015442,
      "learning_rate": 9.452110836442844e-07,
      "loss": 1.9564,
      "num_input_tokens_seen": 116693984,
      "step": 114600
    },
    {
      "epoch": 2.2582741036797858,
      "grad_norm": 1.9258095026016235,
      "learning_rate": 9.451172057278432e-07,
      "loss": 1.9786,
      "num_input_tokens_seen": 116795520,
      "step": 114700
    },
    {
      "epoch": 2.2602429564292885,
      "grad_norm": 2.063319683074951,
      "learning_rate": 9.450232521229755e-07,
      "loss": 1.9775,
      "num_input_tokens_seen": 116897920,
      "step": 114800
    },
    {
      "epoch": 2.2622118091787913,
      "grad_norm": 1.850327730178833,
      "learning_rate": 9.449292228456572e-07,
      "loss": 1.957,
      "num_input_tokens_seen": 116999192,
      "step": 114900
    },
    {
      "epoch": 2.2641806619282945,
      "grad_norm": 2.2068047523498535,
      "learning_rate": 9.448351179118773e-07,
      "loss": 1.9599,
      "num_input_tokens_seen": 117100776,
      "step": 115000
    },
    {
      "epoch": 2.2661495146777972,
      "grad_norm": 1.7473108768463135,
      "learning_rate": 9.447409373376375e-07,
      "loss": 2.0059,
      "num_input_tokens_seen": 117202496,
      "step": 115100
    },
    {
      "epoch": 2.2681183674273,
      "grad_norm": 1.8753230571746826,
      "learning_rate": 9.446466811389524e-07,
      "loss": 1.9853,
      "num_input_tokens_seen": 117304136,
      "step": 115200
    },
    {
      "epoch": 2.270087220176803,
      "grad_norm": 1.7122375965118408,
      "learning_rate": 9.445523493318496e-07,
      "loss": 1.9855,
      "num_input_tokens_seen": 117406088,
      "step": 115300
    },
    {
      "epoch": 2.272056072926306,
      "grad_norm": 1.9982937574386597,
      "learning_rate": 9.444579419323692e-07,
      "loss": 1.9693,
      "num_input_tokens_seen": 117508488,
      "step": 115400
    },
    {
      "epoch": 2.2740249256758087,
      "grad_norm": 1.8192331790924072,
      "learning_rate": 9.443634589565646e-07,
      "loss": 2.0152,
      "num_input_tokens_seen": 117610112,
      "step": 115500
    },
    {
      "epoch": 2.2759937784253115,
      "grad_norm": 1.9701948165893555,
      "learning_rate": 9.442689004205015e-07,
      "loss": 1.9559,
      "num_input_tokens_seen": 117712512,
      "step": 115600
    },
    {
      "epoch": 2.2779626311748142,
      "grad_norm": 1.8190864324569702,
      "learning_rate": 9.441742663402593e-07,
      "loss": 1.9861,
      "num_input_tokens_seen": 117813136,
      "step": 115700
    },
    {
      "epoch": 2.2799314839243174,
      "grad_norm": 2.1761672496795654,
      "learning_rate": 9.440795567319294e-07,
      "loss": 1.966,
      "num_input_tokens_seen": 117913968,
      "step": 115800
    },
    {
      "epoch": 2.28190033667382,
      "grad_norm": 1.8368778228759766,
      "learning_rate": 9.439847716116162e-07,
      "loss": 1.9665,
      "num_input_tokens_seen": 118013648,
      "step": 115900
    },
    {
      "epoch": 2.283869189423323,
      "grad_norm": 2.028080940246582,
      "learning_rate": 9.438899109954377e-07,
      "loss": 1.99,
      "num_input_tokens_seen": 118115552,
      "step": 116000
    },
    {
      "epoch": 2.2858380421728257,
      "grad_norm": 2.0657784938812256,
      "learning_rate": 9.437949748995235e-07,
      "loss": 1.9954,
      "num_input_tokens_seen": 118215696,
      "step": 116100
    },
    {
      "epoch": 2.287806894922329,
      "grad_norm": 2.0885541439056396,
      "learning_rate": 9.436999633400171e-07,
      "loss": 2.0151,
      "num_input_tokens_seen": 118316944,
      "step": 116200
    },
    {
      "epoch": 2.2897757476718317,
      "grad_norm": 1.8826969861984253,
      "learning_rate": 9.436048763330742e-07,
      "loss": 1.9737,
      "num_input_tokens_seen": 118418792,
      "step": 116300
    },
    {
      "epoch": 2.2917446004213344,
      "grad_norm": 1.8971164226531982,
      "learning_rate": 9.435097138948635e-07,
      "loss": 2.0081,
      "num_input_tokens_seen": 118520376,
      "step": 116400
    },
    {
      "epoch": 2.293713453170837,
      "grad_norm": 1.8199630975723267,
      "learning_rate": 9.434144760415669e-07,
      "loss": 1.9929,
      "num_input_tokens_seen": 118622776,
      "step": 116500
    },
    {
      "epoch": 2.2956823059203404,
      "grad_norm": 2.096458911895752,
      "learning_rate": 9.433191627893784e-07,
      "loss": 1.9575,
      "num_input_tokens_seen": 118723912,
      "step": 116600
    },
    {
      "epoch": 2.297651158669843,
      "grad_norm": 1.834140419960022,
      "learning_rate": 9.432237741545055e-07,
      "loss": 2.0431,
      "num_input_tokens_seen": 118825456,
      "step": 116700
    },
    {
      "epoch": 2.299620011419346,
      "grad_norm": 1.9655934572219849,
      "learning_rate": 9.431283101531681e-07,
      "loss": 1.9622,
      "num_input_tokens_seen": 118927856,
      "step": 116800
    },
    {
      "epoch": 2.3015888641688487,
      "grad_norm": 1.634841799736023,
      "learning_rate": 9.43032770801599e-07,
      "loss": 1.9598,
      "num_input_tokens_seen": 119030256,
      "step": 116900
    },
    {
      "epoch": 2.303557716918352,
      "grad_norm": 1.7568492889404297,
      "learning_rate": 9.429371561160442e-07,
      "loss": 2.0286,
      "num_input_tokens_seen": 119131464,
      "step": 117000
    },
    {
      "epoch": 2.3055265696678546,
      "grad_norm": 1.7758232355117798,
      "learning_rate": 9.428414661127618e-07,
      "loss": 1.9796,
      "num_input_tokens_seen": 119232936,
      "step": 117100
    },
    {
      "epoch": 2.3074954224173574,
      "grad_norm": 1.8784618377685547,
      "learning_rate": 9.427457008080229e-07,
      "loss": 1.9914,
      "num_input_tokens_seen": 119335040,
      "step": 117200
    },
    {
      "epoch": 2.30946427516686,
      "grad_norm": 1.991125464439392,
      "learning_rate": 9.42649860218112e-07,
      "loss": 2.0175,
      "num_input_tokens_seen": 119436864,
      "step": 117300
    },
    {
      "epoch": 2.311433127916363,
      "grad_norm": 1.7689714431762695,
      "learning_rate": 9.425539443593261e-07,
      "loss": 1.985,
      "num_input_tokens_seen": 119539264,
      "step": 117400
    },
    {
      "epoch": 2.313401980665866,
      "grad_norm": 1.9021596908569336,
      "learning_rate": 9.424579532479744e-07,
      "loss": 2.0146,
      "num_input_tokens_seen": 119641024,
      "step": 117500
    },
    {
      "epoch": 2.315370833415369,
      "grad_norm": 1.9273254871368408,
      "learning_rate": 9.423618869003798e-07,
      "loss": 2.0019,
      "num_input_tokens_seen": 119742472,
      "step": 117600
    },
    {
      "epoch": 2.3173396861648716,
      "grad_norm": 2.1978797912597656,
      "learning_rate": 9.422657453328772e-07,
      "loss": 1.9974,
      "num_input_tokens_seen": 119844872,
      "step": 117700
    },
    {
      "epoch": 2.319308538914375,
      "grad_norm": 1.8331992626190186,
      "learning_rate": 9.42169528561815e-07,
      "loss": 2.017,
      "num_input_tokens_seen": 119946560,
      "step": 117800
    },
    {
      "epoch": 2.3212773916638776,
      "grad_norm": 1.8678513765335083,
      "learning_rate": 9.420732366035538e-07,
      "loss": 1.9745,
      "num_input_tokens_seen": 120048392,
      "step": 117900
    },
    {
      "epoch": 2.3232462444133803,
      "grad_norm": 1.968079924583435,
      "learning_rate": 9.419768694744672e-07,
      "loss": 1.9955,
      "num_input_tokens_seen": 120149472,
      "step": 118000
    },
    {
      "epoch": 2.325215097162883,
      "grad_norm": 1.772836446762085,
      "learning_rate": 9.41880427190942e-07,
      "loss": 1.9925,
      "num_input_tokens_seen": 120251872,
      "step": 118100
    },
    {
      "epoch": 2.327183949912386,
      "grad_norm": 1.8192387819290161,
      "learning_rate": 9.41783909769377e-07,
      "loss": 2.022,
      "num_input_tokens_seen": 120354272,
      "step": 118200
    },
    {
      "epoch": 2.329152802661889,
      "grad_norm": 2.0357606410980225,
      "learning_rate": 9.416873172261843e-07,
      "loss": 2.015,
      "num_input_tokens_seen": 120456672,
      "step": 118300
    },
    {
      "epoch": 2.331121655411392,
      "grad_norm": 1.712955355644226,
      "learning_rate": 9.415906495777888e-07,
      "loss": 1.9741,
      "num_input_tokens_seen": 120558152,
      "step": 118400
    },
    {
      "epoch": 2.3330905081608946,
      "grad_norm": 1.790560007095337,
      "learning_rate": 9.414939068406278e-07,
      "loss": 1.9844,
      "num_input_tokens_seen": 120659776,
      "step": 118500
    },
    {
      "epoch": 2.3350593609103973,
      "grad_norm": 1.6297513246536255,
      "learning_rate": 9.413970890311517e-07,
      "loss": 1.9849,
      "num_input_tokens_seen": 120761304,
      "step": 118600
    },
    {
      "epoch": 2.3370282136599005,
      "grad_norm": 3.2493929862976074,
      "learning_rate": 9.413001961658235e-07,
      "loss": 2.046,
      "num_input_tokens_seen": 120863704,
      "step": 118700
    },
    {
      "epoch": 2.3389970664094033,
      "grad_norm": 1.8684008121490479,
      "learning_rate": 9.412032282611191e-07,
      "loss": 1.9968,
      "num_input_tokens_seen": 120965288,
      "step": 118800
    },
    {
      "epoch": 2.340965919158906,
      "grad_norm": 1.9268240928649902,
      "learning_rate": 9.411061853335269e-07,
      "loss": 2.0215,
      "num_input_tokens_seen": 121067240,
      "step": 118900
    },
    {
      "epoch": 2.342934771908409,
      "grad_norm": 2.0446953773498535,
      "learning_rate": 9.410090673995483e-07,
      "loss": 1.9718,
      "num_input_tokens_seen": 121167992,
      "step": 119000
    },
    {
      "epoch": 2.344903624657912,
      "grad_norm": 2.1195380687713623,
      "learning_rate": 9.409118744756977e-07,
      "loss": 1.9698,
      "num_input_tokens_seen": 121268824,
      "step": 119100
    },
    {
      "epoch": 2.346872477407415,
      "grad_norm": 2.0330278873443604,
      "learning_rate": 9.408146065785014e-07,
      "loss": 2.0111,
      "num_input_tokens_seen": 121370912,
      "step": 119200
    },
    {
      "epoch": 2.3488413301569175,
      "grad_norm": 2.2201755046844482,
      "learning_rate": 9.407172637244994e-07,
      "loss": 1.9716,
      "num_input_tokens_seen": 121472288,
      "step": 119300
    },
    {
      "epoch": 2.3508101829064203,
      "grad_norm": 1.8585072755813599,
      "learning_rate": 9.406198459302439e-07,
      "loss": 1.9807,
      "num_input_tokens_seen": 121574304,
      "step": 119400
    },
    {
      "epoch": 2.3527790356559235,
      "grad_norm": 1.8911980390548706,
      "learning_rate": 9.405223532122999e-07,
      "loss": 1.9585,
      "num_input_tokens_seen": 121676240,
      "step": 119500
    },
    {
      "epoch": 2.3547478884054263,
      "grad_norm": 2.043912172317505,
      "learning_rate": 9.404247855872452e-07,
      "loss": 1.9701,
      "num_input_tokens_seen": 121778016,
      "step": 119600
    },
    {
      "epoch": 2.356716741154929,
      "grad_norm": 1.8020490407943726,
      "learning_rate": 9.403271430716706e-07,
      "loss": 1.9727,
      "num_input_tokens_seen": 121880416,
      "step": 119700
    },
    {
      "epoch": 2.3586855939044318,
      "grad_norm": 2.0996904373168945,
      "learning_rate": 9.40229425682179e-07,
      "loss": 1.9843,
      "num_input_tokens_seen": 121982816,
      "step": 119800
    },
    {
      "epoch": 2.3606544466539345,
      "grad_norm": 1.9785363674163818,
      "learning_rate": 9.401316334353868e-07,
      "loss": 1.9667,
      "num_input_tokens_seen": 122084400,
      "step": 119900
    },
    {
      "epoch": 2.3626232994034377,
      "grad_norm": 1.6954538822174072,
      "learning_rate": 9.400337663479226e-07,
      "loss": 1.9978,
      "num_input_tokens_seen": 122186800,
      "step": 120000
    },
    {
      "epoch": 2.3645921521529405,
      "grad_norm": 2.209807872772217,
      "learning_rate": 9.399358244364277e-07,
      "loss": 2.0276,
      "num_input_tokens_seen": 122287616,
      "step": 120100
    },
    {
      "epoch": 2.3665610049024433,
      "grad_norm": 1.764782190322876,
      "learning_rate": 9.398378077175566e-07,
      "loss": 1.9761,
      "num_input_tokens_seen": 122388864,
      "step": 120200
    },
    {
      "epoch": 2.3685298576519465,
      "grad_norm": 1.9492393732070923,
      "learning_rate": 9.397397162079759e-07,
      "loss": 2.0153,
      "num_input_tokens_seen": 122491264,
      "step": 120300
    },
    {
      "epoch": 2.370498710401449,
      "grad_norm": 1.8193806409835815,
      "learning_rate": 9.396415499243655e-07,
      "loss": 1.9858,
      "num_input_tokens_seen": 122592256,
      "step": 120400
    },
    {
      "epoch": 2.372467563150952,
      "grad_norm": 1.8664618730545044,
      "learning_rate": 9.395433088834176e-07,
      "loss": 1.9605,
      "num_input_tokens_seen": 122693184,
      "step": 120500
    },
    {
      "epoch": 2.3744364159004547,
      "grad_norm": 1.7667503356933594,
      "learning_rate": 9.394449931018373e-07,
      "loss": 1.9666,
      "num_input_tokens_seen": 122795584,
      "step": 120600
    },
    {
      "epoch": 2.3764052686499575,
      "grad_norm": 1.727834939956665,
      "learning_rate": 9.393466025963423e-07,
      "loss": 1.9711,
      "num_input_tokens_seen": 122897248,
      "step": 120700
    },
    {
      "epoch": 2.3783741213994607,
      "grad_norm": 1.9298264980316162,
      "learning_rate": 9.392481373836633e-07,
      "loss": 1.959,
      "num_input_tokens_seen": 122999648,
      "step": 120800
    },
    {
      "epoch": 2.3803429741489635,
      "grad_norm": 2.3186757564544678,
      "learning_rate": 9.391495974805432e-07,
      "loss": 1.9725,
      "num_input_tokens_seen": 123101528,
      "step": 120900
    },
    {
      "epoch": 2.382311826898466,
      "grad_norm": 2.5058555603027344,
      "learning_rate": 9.390509829037381e-07,
      "loss": 1.9841,
      "num_input_tokens_seen": 123203056,
      "step": 121000
    },
    {
      "epoch": 2.384280679647969,
      "grad_norm": 6.203256607055664,
      "learning_rate": 9.389522936700165e-07,
      "loss": 1.9885,
      "num_input_tokens_seen": 123304688,
      "step": 121100
    },
    {
      "epoch": 2.386249532397472,
      "grad_norm": 1.8381099700927734,
      "learning_rate": 9.388535297961597e-07,
      "loss": 1.9858,
      "num_input_tokens_seen": 123407088,
      "step": 121200
    },
    {
      "epoch": 2.388218385146975,
      "grad_norm": 1.7783349752426147,
      "learning_rate": 9.387546912989614e-07,
      "loss": 2.0128,
      "num_input_tokens_seen": 123509488,
      "step": 121300
    },
    {
      "epoch": 2.3901872378964777,
      "grad_norm": 2.1013903617858887,
      "learning_rate": 9.386557781952287e-07,
      "loss": 1.9363,
      "num_input_tokens_seen": 123611888,
      "step": 121400
    },
    {
      "epoch": 2.3921560906459804,
      "grad_norm": 1.786528468132019,
      "learning_rate": 9.385567905017806e-07,
      "loss": 1.9847,
      "num_input_tokens_seen": 123713712,
      "step": 121500
    },
    {
      "epoch": 2.394124943395483,
      "grad_norm": 1.7002952098846436,
      "learning_rate": 9.384577282354493e-07,
      "loss": 2.011,
      "num_input_tokens_seen": 123814688,
      "step": 121600
    },
    {
      "epoch": 2.3960937961449864,
      "grad_norm": 1.940887451171875,
      "learning_rate": 9.383585914130794e-07,
      "loss": 1.946,
      "num_input_tokens_seen": 123916328,
      "step": 121700
    },
    {
      "epoch": 2.398062648894489,
      "grad_norm": 2.1516923904418945,
      "learning_rate": 9.382593800515283e-07,
      "loss": 1.9333,
      "num_input_tokens_seen": 124017312,
      "step": 121800
    },
    {
      "epoch": 2.400031501643992,
      "grad_norm": 1.819543480873108,
      "learning_rate": 9.381600941676661e-07,
      "loss": 2.0135,
      "num_input_tokens_seen": 124118480,
      "step": 121900
    },
    {
      "epoch": 2.402000354393495,
      "grad_norm": 1.9588466882705688,
      "learning_rate": 9.380607337783754e-07,
      "loss": 1.9691,
      "num_input_tokens_seen": 124220128,
      "step": 122000
    },
    {
      "epoch": 2.403969207142998,
      "grad_norm": 1.8389747142791748,
      "learning_rate": 9.379612989005517e-07,
      "loss": 2.0552,
      "num_input_tokens_seen": 124322528,
      "step": 122100
    },
    {
      "epoch": 2.4059380598925006,
      "grad_norm": 2.093642473220825,
      "learning_rate": 9.378617895511032e-07,
      "loss": 1.983,
      "num_input_tokens_seen": 124424560,
      "step": 122200
    },
    {
      "epoch": 2.4079069126420034,
      "grad_norm": 2.068624973297119,
      "learning_rate": 9.377622057469505e-07,
      "loss": 2.0576,
      "num_input_tokens_seen": 124525328,
      "step": 122300
    },
    {
      "epoch": 2.409875765391506,
      "grad_norm": 6.074371814727783,
      "learning_rate": 9.376625475050268e-07,
      "loss": 1.9936,
      "num_input_tokens_seen": 124627080,
      "step": 122400
    },
    {
      "epoch": 2.4118446181410094,
      "grad_norm": 1.7730644941329956,
      "learning_rate": 9.375628148422785e-07,
      "loss": 1.9775,
      "num_input_tokens_seen": 124729480,
      "step": 122500
    },
    {
      "epoch": 2.413813470890512,
      "grad_norm": 2.0941343307495117,
      "learning_rate": 9.37463007775664e-07,
      "loss": 1.9818,
      "num_input_tokens_seen": 124831880,
      "step": 122600
    },
    {
      "epoch": 2.415782323640015,
      "grad_norm": 2.197303056716919,
      "learning_rate": 9.373631263221547e-07,
      "loss": 1.9536,
      "num_input_tokens_seen": 124933616,
      "step": 122700
    },
    {
      "epoch": 2.4177511763895176,
      "grad_norm": 1.9617540836334229,
      "learning_rate": 9.372631704987347e-07,
      "loss": 2.0507,
      "num_input_tokens_seen": 125036016,
      "step": 122800
    },
    {
      "epoch": 2.419720029139021,
      "grad_norm": 1.848714828491211,
      "learning_rate": 9.371631403224008e-07,
      "loss": 1.9965,
      "num_input_tokens_seen": 125137824,
      "step": 122900
    },
    {
      "epoch": 2.4216888818885236,
      "grad_norm": 1.8612762689590454,
      "learning_rate": 9.370630358101618e-07,
      "loss": 1.9897,
      "num_input_tokens_seen": 125238720,
      "step": 123000
    },
    {
      "epoch": 2.4236577346380264,
      "grad_norm": 2.054504871368408,
      "learning_rate": 9.3696285697904e-07,
      "loss": 1.9864,
      "num_input_tokens_seen": 125340568,
      "step": 123100
    },
    {
      "epoch": 2.425626587387529,
      "grad_norm": 1.8542107343673706,
      "learning_rate": 9.368626038460699e-07,
      "loss": 1.9357,
      "num_input_tokens_seen": 125442968,
      "step": 123200
    },
    {
      "epoch": 2.4275954401370323,
      "grad_norm": 1.8898608684539795,
      "learning_rate": 9.367622764282987e-07,
      "loss": 1.9892,
      "num_input_tokens_seen": 125544640,
      "step": 123300
    },
    {
      "epoch": 2.429564292886535,
      "grad_norm": 1.8329038619995117,
      "learning_rate": 9.366618747427863e-07,
      "loss": 1.9773,
      "num_input_tokens_seen": 125647040,
      "step": 123400
    },
    {
      "epoch": 2.431533145636038,
      "grad_norm": 1.8550149202346802,
      "learning_rate": 9.36561398806605e-07,
      "loss": 2.033,
      "num_input_tokens_seen": 125748944,
      "step": 123500
    },
    {
      "epoch": 2.4335019983855406,
      "grad_norm": 1.8037301301956177,
      "learning_rate": 9.364608486368399e-07,
      "loss": 1.9951,
      "num_input_tokens_seen": 125851256,
      "step": 123600
    },
    {
      "epoch": 2.435470851135044,
      "grad_norm": 1.756620168685913,
      "learning_rate": 9.363602242505891e-07,
      "loss": 1.9914,
      "num_input_tokens_seen": 125953072,
      "step": 123700
    },
    {
      "epoch": 2.4374397038845466,
      "grad_norm": 2.3889143466949463,
      "learning_rate": 9.362595256649621e-07,
      "loss": 1.9774,
      "num_input_tokens_seen": 126054664,
      "step": 123800
    },
    {
      "epoch": 2.4394085566340493,
      "grad_norm": 1.8695127964019775,
      "learning_rate": 9.361587528970827e-07,
      "loss": 2.0321,
      "num_input_tokens_seen": 126155992,
      "step": 123900
    },
    {
      "epoch": 2.441377409383552,
      "grad_norm": 1.987545132637024,
      "learning_rate": 9.360579059640859e-07,
      "loss": 1.9582,
      "num_input_tokens_seen": 126257528,
      "step": 124000
    },
    {
      "epoch": 2.443346262133055,
      "grad_norm": 1.7185331583023071,
      "learning_rate": 9.359569848831203e-07,
      "loss": 1.9825,
      "num_input_tokens_seen": 126359416,
      "step": 124100
    },
    {
      "epoch": 2.445315114882558,
      "grad_norm": 2.0995638370513916,
      "learning_rate": 9.358559896713463e-07,
      "loss": 1.9601,
      "num_input_tokens_seen": 126461816,
      "step": 124200
    },
    {
      "epoch": 2.447283967632061,
      "grad_norm": 2.0495667457580566,
      "learning_rate": 9.357549203459374e-07,
      "loss": 1.966,
      "num_input_tokens_seen": 126562848,
      "step": 124300
    },
    {
      "epoch": 2.4492528203815636,
      "grad_norm": 1.9624818563461304,
      "learning_rate": 9.356537769240797e-07,
      "loss": 1.9607,
      "num_input_tokens_seen": 126664672,
      "step": 124400
    },
    {
      "epoch": 2.4512216731310668,
      "grad_norm": 1.7971683740615845,
      "learning_rate": 9.355525594229717e-07,
      "loss": 2.0049,
      "num_input_tokens_seen": 126765992,
      "step": 124500
    },
    {
      "epoch": 2.4531905258805695,
      "grad_norm": 1.8750773668289185,
      "learning_rate": 9.354512678598245e-07,
      "loss": 2.0275,
      "num_input_tokens_seen": 126868392,
      "step": 124600
    },
    {
      "epoch": 2.4551593786300723,
      "grad_norm": 1.8480435609817505,
      "learning_rate": 9.353499022518621e-07,
      "loss": 1.9962,
      "num_input_tokens_seen": 126969944,
      "step": 124700
    },
    {
      "epoch": 2.457128231379575,
      "grad_norm": 1.8100879192352295,
      "learning_rate": 9.352484626163206e-07,
      "loss": 1.9911,
      "num_input_tokens_seen": 127071792,
      "step": 124800
    },
    {
      "epoch": 2.459097084129078,
      "grad_norm": 1.913601279258728,
      "learning_rate": 9.351469489704491e-07,
      "loss": 1.9825,
      "num_input_tokens_seen": 127174192,
      "step": 124900
    },
    {
      "epoch": 2.461065936878581,
      "grad_norm": 1.8896150588989258,
      "learning_rate": 9.350453613315093e-07,
      "loss": 1.9921,
      "num_input_tokens_seen": 127273704,
      "step": 125000
    },
    {
      "epoch": 2.4630347896280838,
      "grad_norm": 2.3673813343048096,
      "learning_rate": 9.34943699716775e-07,
      "loss": 2.0182,
      "num_input_tokens_seen": 127375568,
      "step": 125100
    },
    {
      "epoch": 2.4650036423775865,
      "grad_norm": 2.0962560176849365,
      "learning_rate": 9.34841964143533e-07,
      "loss": 2.0025,
      "num_input_tokens_seen": 127477968,
      "step": 125200
    },
    {
      "epoch": 2.4669724951270893,
      "grad_norm": 1.9968198537826538,
      "learning_rate": 9.347401546290827e-07,
      "loss": 1.9551,
      "num_input_tokens_seen": 127580368,
      "step": 125300
    },
    {
      "epoch": 2.4689413478765925,
      "grad_norm": 1.636101245880127,
      "learning_rate": 9.346382711907359e-07,
      "loss": 1.9726,
      "num_input_tokens_seen": 127682768,
      "step": 125400
    },
    {
      "epoch": 2.4709102006260952,
      "grad_norm": 2.0196454524993896,
      "learning_rate": 9.34536313845817e-07,
      "loss": 1.9582,
      "num_input_tokens_seen": 127785168,
      "step": 125500
    },
    {
      "epoch": 2.472879053375598,
      "grad_norm": 2.038447380065918,
      "learning_rate": 9.344342826116629e-07,
      "loss": 1.9848,
      "num_input_tokens_seen": 127887568,
      "step": 125600
    },
    {
      "epoch": 2.4748479061251007,
      "grad_norm": 1.792228102684021,
      "learning_rate": 9.343321775056233e-07,
      "loss": 1.9615,
      "num_input_tokens_seen": 127989768,
      "step": 125700
    },
    {
      "epoch": 2.476816758874604,
      "grad_norm": 2.20328688621521,
      "learning_rate": 9.342299985450603e-07,
      "loss": 2.0324,
      "num_input_tokens_seen": 128090480,
      "step": 125800
    },
    {
      "epoch": 2.4787856116241067,
      "grad_norm": 1.8574694395065308,
      "learning_rate": 9.341277457473484e-07,
      "loss": 1.9777,
      "num_input_tokens_seen": 128192112,
      "step": 125900
    },
    {
      "epoch": 2.4807544643736095,
      "grad_norm": 1.7385035753250122,
      "learning_rate": 9.340254191298749e-07,
      "loss": 1.9614,
      "num_input_tokens_seen": 128294512,
      "step": 126000
    },
    {
      "epoch": 2.4827233171231122,
      "grad_norm": 1.932966709136963,
      "learning_rate": 9.339230187100398e-07,
      "loss": 1.9353,
      "num_input_tokens_seen": 128396912,
      "step": 126100
    },
    {
      "epoch": 2.4846921698726154,
      "grad_norm": 1.8148736953735352,
      "learning_rate": 9.338205445052551e-07,
      "loss": 1.963,
      "num_input_tokens_seen": 128498392,
      "step": 126200
    },
    {
      "epoch": 2.486661022622118,
      "grad_norm": 2.132577419281006,
      "learning_rate": 9.337179965329458e-07,
      "loss": 2.0064,
      "num_input_tokens_seen": 128600792,
      "step": 126300
    },
    {
      "epoch": 2.488629875371621,
      "grad_norm": 2.0273873805999756,
      "learning_rate": 9.336153748105494e-07,
      "loss": 1.9702,
      "num_input_tokens_seen": 128703192,
      "step": 126400
    },
    {
      "epoch": 2.4905987281211237,
      "grad_norm": 2.0853137969970703,
      "learning_rate": 9.335126793555157e-07,
      "loss": 1.9968,
      "num_input_tokens_seen": 128805592,
      "step": 126500
    },
    {
      "epoch": 2.4925675808706265,
      "grad_norm": 2.0068466663360596,
      "learning_rate": 9.334099101853075e-07,
      "loss": 2.0016,
      "num_input_tokens_seen": 128907112,
      "step": 126600
    },
    {
      "epoch": 2.4945364336201297,
      "grad_norm": 2.1706347465515137,
      "learning_rate": 9.333070673173994e-07,
      "loss": 1.9742,
      "num_input_tokens_seen": 129008776,
      "step": 126700
    },
    {
      "epoch": 2.4965052863696324,
      "grad_norm": 1.9625388383865356,
      "learning_rate": 9.332041507692793e-07,
      "loss": 1.9775,
      "num_input_tokens_seen": 129110656,
      "step": 126800
    },
    {
      "epoch": 2.498474139119135,
      "grad_norm": 1.8103257417678833,
      "learning_rate": 9.331011605584471e-07,
      "loss": 2.0023,
      "num_input_tokens_seen": 129212296,
      "step": 126900
    },
    {
      "epoch": 2.5004429918686384,
      "grad_norm": 1.702073097229004,
      "learning_rate": 9.329980967024156e-07,
      "loss": 1.9738,
      "num_input_tokens_seen": 129313952,
      "step": 127000
    },
    {
      "epoch": 2.502411844618141,
      "grad_norm": 1.8357709646224976,
      "learning_rate": 9.328949592187097e-07,
      "loss": 2.013,
      "num_input_tokens_seen": 129416352,
      "step": 127100
    },
    {
      "epoch": 2.504380697367644,
      "grad_norm": 2.4339730739593506,
      "learning_rate": 9.327917481248673e-07,
      "loss": 2.0074,
      "num_input_tokens_seen": 129517416,
      "step": 127200
    },
    {
      "epoch": 2.5063495501171467,
      "grad_norm": 1.863301396369934,
      "learning_rate": 9.326884634384384e-07,
      "loss": 1.935,
      "num_input_tokens_seen": 129619816,
      "step": 127300
    },
    {
      "epoch": 2.5083184028666494,
      "grad_norm": 2.1330418586730957,
      "learning_rate": 9.325851051769858e-07,
      "loss": 1.9947,
      "num_input_tokens_seen": 129721680,
      "step": 127400
    },
    {
      "epoch": 2.5102872556161526,
      "grad_norm": 1.7169445753097534,
      "learning_rate": 9.324816733580846e-07,
      "loss": 1.9705,
      "num_input_tokens_seen": 129823256,
      "step": 127500
    },
    {
      "epoch": 2.5122561083656554,
      "grad_norm": 2.6570725440979004,
      "learning_rate": 9.323781679993225e-07,
      "loss": 1.9716,
      "num_input_tokens_seen": 129924656,
      "step": 127600
    },
    {
      "epoch": 2.514224961115158,
      "grad_norm": 1.8668783903121948,
      "learning_rate": 9.322745891182999e-07,
      "loss": 2.0005,
      "num_input_tokens_seen": 130027056,
      "step": 127700
    },
    {
      "epoch": 2.516193813864661,
      "grad_norm": 1.8810783624649048,
      "learning_rate": 9.321709367326294e-07,
      "loss": 1.9997,
      "num_input_tokens_seen": 130129456,
      "step": 127800
    },
    {
      "epoch": 2.518162666614164,
      "grad_norm": 2.2021889686584473,
      "learning_rate": 9.320672108599361e-07,
      "loss": 2.0264,
      "num_input_tokens_seen": 130231016,
      "step": 127900
    },
    {
      "epoch": 2.520131519363667,
      "grad_norm": 2.0468647480010986,
      "learning_rate": 9.31963411517858e-07,
      "loss": 1.9705,
      "num_input_tokens_seen": 130333416,
      "step": 128000
    },
    {
      "epoch": 2.5221003721131696,
      "grad_norm": 1.8266481161117554,
      "learning_rate": 9.31859538724045e-07,
      "loss": 2.0478,
      "num_input_tokens_seen": 130434344,
      "step": 128100
    },
    {
      "epoch": 2.5240692248626724,
      "grad_norm": 1.7767115831375122,
      "learning_rate": 9.317555924961599e-07,
      "loss": 1.9972,
      "num_input_tokens_seen": 130536744,
      "step": 128200
    },
    {
      "epoch": 2.526038077612175,
      "grad_norm": 1.9214739799499512,
      "learning_rate": 9.316515728518779e-07,
      "loss": 2.027,
      "num_input_tokens_seen": 130639144,
      "step": 128300
    },
    {
      "epoch": 2.5280069303616783,
      "grad_norm": 1.867311954498291,
      "learning_rate": 9.315474798088867e-07,
      "loss": 1.9905,
      "num_input_tokens_seen": 130740896,
      "step": 128400
    },
    {
      "epoch": 2.529975783111181,
      "grad_norm": 2.0441529750823975,
      "learning_rate": 9.314433133848864e-07,
      "loss": 1.9569,
      "num_input_tokens_seen": 130842528,
      "step": 128500
    },
    {
      "epoch": 2.531944635860684,
      "grad_norm": 1.762548804283142,
      "learning_rate": 9.313390735975894e-07,
      "loss": 2.0254,
      "num_input_tokens_seen": 130944928,
      "step": 128600
    },
    {
      "epoch": 2.533913488610187,
      "grad_norm": 1.7779361009597778,
      "learning_rate": 9.312347604647213e-07,
      "loss": 1.989,
      "num_input_tokens_seen": 131046696,
      "step": 128700
    },
    {
      "epoch": 2.53588234135969,
      "grad_norm": 1.7223703861236572,
      "learning_rate": 9.311303740040193e-07,
      "loss": 1.9994,
      "num_input_tokens_seen": 131149096,
      "step": 128800
    },
    {
      "epoch": 2.5378511941091926,
      "grad_norm": 1.8027669191360474,
      "learning_rate": 9.310259142332335e-07,
      "loss": 2.004,
      "num_input_tokens_seen": 131250664,
      "step": 128900
    },
    {
      "epoch": 2.5398200468586953,
      "grad_norm": 1.9260095357894897,
      "learning_rate": 9.309213811701265e-07,
      "loss": 1.9928,
      "num_input_tokens_seen": 131353064,
      "step": 129000
    },
    {
      "epoch": 2.541788899608198,
      "grad_norm": 1.779481291770935,
      "learning_rate": 9.308167748324731e-07,
      "loss": 1.967,
      "num_input_tokens_seen": 131454872,
      "step": 129100
    },
    {
      "epoch": 2.5437577523577013,
      "grad_norm": 1.8147728443145752,
      "learning_rate": 9.307120952380607e-07,
      "loss": 1.9518,
      "num_input_tokens_seen": 131557272,
      "step": 129200
    },
    {
      "epoch": 2.545726605107204,
      "grad_norm": 1.7653882503509521,
      "learning_rate": 9.306073424046896e-07,
      "loss": 2.0163,
      "num_input_tokens_seen": 131659128,
      "step": 129300
    },
    {
      "epoch": 2.547695457856707,
      "grad_norm": 2.157727003097534,
      "learning_rate": 9.305025163501716e-07,
      "loss": 2.0305,
      "num_input_tokens_seen": 131760680,
      "step": 129400
    },
    {
      "epoch": 2.54966431060621,
      "grad_norm": 1.893654704093933,
      "learning_rate": 9.303976170923317e-07,
      "loss": 2.0263,
      "num_input_tokens_seen": 131863080,
      "step": 129500
    },
    {
      "epoch": 2.5516331633557128,
      "grad_norm": 1.816401720046997,
      "learning_rate": 9.302926446490073e-07,
      "loss": 1.9723,
      "num_input_tokens_seen": 131963232,
      "step": 129600
    },
    {
      "epoch": 2.5536020161052155,
      "grad_norm": 3.2825052738189697,
      "learning_rate": 9.301875990380479e-07,
      "loss": 2.0053,
      "num_input_tokens_seen": 132065096,
      "step": 129700
    },
    {
      "epoch": 2.5555708688547183,
      "grad_norm": 2.1008756160736084,
      "learning_rate": 9.300824802773156e-07,
      "loss": 2.0095,
      "num_input_tokens_seen": 132167496,
      "step": 129800
    },
    {
      "epoch": 2.557539721604221,
      "grad_norm": 1.709738850593567,
      "learning_rate": 9.299772883846851e-07,
      "loss": 2.0523,
      "num_input_tokens_seen": 132267848,
      "step": 129900
    },
    {
      "epoch": 2.559508574353724,
      "grad_norm": 1.909054160118103,
      "learning_rate": 9.298720233780432e-07,
      "loss": 1.9812,
      "num_input_tokens_seen": 132369320,
      "step": 130000
    },
    {
      "epoch": 2.561477427103227,
      "grad_norm": 1.8835586309432983,
      "learning_rate": 9.297666852752895e-07,
      "loss": 1.9506,
      "num_input_tokens_seen": 132471720,
      "step": 130100
    },
    {
      "epoch": 2.5634462798527298,
      "grad_norm": 1.8786972761154175,
      "learning_rate": 9.296612740943357e-07,
      "loss": 1.9678,
      "num_input_tokens_seen": 132573360,
      "step": 130200
    },
    {
      "epoch": 2.5654151326022325,
      "grad_norm": 2.090474843978882,
      "learning_rate": 9.295557898531064e-07,
      "loss": 1.9588,
      "num_input_tokens_seen": 132674944,
      "step": 130300
    },
    {
      "epoch": 2.5673839853517357,
      "grad_norm": 2.0154058933258057,
      "learning_rate": 9.294502325695378e-07,
      "loss": 1.9925,
      "num_input_tokens_seen": 132776768,
      "step": 130400
    },
    {
      "epoch": 2.5693528381012385,
      "grad_norm": 1.826133131980896,
      "learning_rate": 9.293446022615794e-07,
      "loss": 1.9585,
      "num_input_tokens_seen": 132878672,
      "step": 130500
    },
    {
      "epoch": 2.5713216908507412,
      "grad_norm": 1.9566258192062378,
      "learning_rate": 9.292388989471927e-07,
      "loss": 1.9453,
      "num_input_tokens_seen": 132980480,
      "step": 130600
    },
    {
      "epoch": 2.573290543600244,
      "grad_norm": 1.9297854900360107,
      "learning_rate": 9.291331226443516e-07,
      "loss": 1.9479,
      "num_input_tokens_seen": 133081544,
      "step": 130700
    },
    {
      "epoch": 2.5752593963497468,
      "grad_norm": 1.7936279773712158,
      "learning_rate": 9.290272733710425e-07,
      "loss": 1.9936,
      "num_input_tokens_seen": 133182768,
      "step": 130800
    },
    {
      "epoch": 2.57722824909925,
      "grad_norm": 1.8617814779281616,
      "learning_rate": 9.28921351145264e-07,
      "loss": 2.0311,
      "num_input_tokens_seen": 133284528,
      "step": 130900
    },
    {
      "epoch": 2.5791971018487527,
      "grad_norm": 1.7724997997283936,
      "learning_rate": 9.288153559850274e-07,
      "loss": 2.0212,
      "num_input_tokens_seen": 133386016,
      "step": 131000
    },
    {
      "epoch": 2.5811659545982555,
      "grad_norm": 1.9706727266311646,
      "learning_rate": 9.287092879083565e-07,
      "loss": 2.0261,
      "num_input_tokens_seen": 133487032,
      "step": 131100
    },
    {
      "epoch": 2.5831348073477587,
      "grad_norm": 6.432122707366943,
      "learning_rate": 9.286031469332871e-07,
      "loss": 1.9701,
      "num_input_tokens_seen": 133588608,
      "step": 131200
    },
    {
      "epoch": 2.5851036600972614,
      "grad_norm": 1.7948259115219116,
      "learning_rate": 9.284969330778674e-07,
      "loss": 1.9814,
      "num_input_tokens_seen": 133691008,
      "step": 131300
    },
    {
      "epoch": 2.587072512846764,
      "grad_norm": 1.7874032258987427,
      "learning_rate": 9.283906463601585e-07,
      "loss": 2.004,
      "num_input_tokens_seen": 133792328,
      "step": 131400
    },
    {
      "epoch": 2.589041365596267,
      "grad_norm": 1.7927825450897217,
      "learning_rate": 9.282842867982332e-07,
      "loss": 1.9606,
      "num_input_tokens_seen": 133894208,
      "step": 131500
    },
    {
      "epoch": 2.5910102183457697,
      "grad_norm": 2.154865026473999,
      "learning_rate": 9.281778544101775e-07,
      "loss": 1.9693,
      "num_input_tokens_seen": 133996608,
      "step": 131600
    },
    {
      "epoch": 2.592979071095273,
      "grad_norm": 1.800648808479309,
      "learning_rate": 9.28071349214089e-07,
      "loss": 1.9697,
      "num_input_tokens_seen": 134099008,
      "step": 131700
    },
    {
      "epoch": 2.5949479238447757,
      "grad_norm": 2.003859281539917,
      "learning_rate": 9.27964771228078e-07,
      "loss": 1.9429,
      "num_input_tokens_seen": 134201408,
      "step": 131800
    },
    {
      "epoch": 2.5969167765942784,
      "grad_norm": 1.988529920578003,
      "learning_rate": 9.278581204702672e-07,
      "loss": 1.9445,
      "num_input_tokens_seen": 134303808,
      "step": 131900
    },
    {
      "epoch": 2.5988856293437816,
      "grad_norm": 2.2726986408233643,
      "learning_rate": 9.277513969587919e-07,
      "loss": 1.9789,
      "num_input_tokens_seen": 134405456,
      "step": 132000
    },
    {
      "epoch": 2.6008544820932844,
      "grad_norm": 1.8805309534072876,
      "learning_rate": 9.276446007117993e-07,
      "loss": 1.985,
      "num_input_tokens_seen": 134507416,
      "step": 132100
    },
    {
      "epoch": 2.602823334842787,
      "grad_norm": 1.753617525100708,
      "learning_rate": 9.275377317474493e-07,
      "loss": 1.9638,
      "num_input_tokens_seen": 134609816,
      "step": 132200
    },
    {
      "epoch": 2.60479218759229,
      "grad_norm": 2.085961103439331,
      "learning_rate": 9.27430790083914e-07,
      "loss": 1.9709,
      "num_input_tokens_seen": 134711168,
      "step": 132300
    },
    {
      "epoch": 2.6067610403417927,
      "grad_norm": 1.8576356172561646,
      "learning_rate": 9.27323775739378e-07,
      "loss": 2.0193,
      "num_input_tokens_seen": 134813568,
      "step": 132400
    },
    {
      "epoch": 2.6087298930912954,
      "grad_norm": 1.9671186208724976,
      "learning_rate": 9.27216688732038e-07,
      "loss": 1.975,
      "num_input_tokens_seen": 134915104,
      "step": 132500
    },
    {
      "epoch": 2.6106987458407986,
      "grad_norm": 2.198086977005005,
      "learning_rate": 9.271095290801036e-07,
      "loss": 1.9641,
      "num_input_tokens_seen": 135017504,
      "step": 132600
    },
    {
      "epoch": 2.6126675985903014,
      "grad_norm": 1.8338600397109985,
      "learning_rate": 9.270022968017961e-07,
      "loss": 1.9719,
      "num_input_tokens_seen": 135117920,
      "step": 132700
    },
    {
      "epoch": 2.614636451339804,
      "grad_norm": 4.461668968200684,
      "learning_rate": 9.268949919153496e-07,
      "loss": 1.9957,
      "num_input_tokens_seen": 135219680,
      "step": 132800
    },
    {
      "epoch": 2.6166053040893074,
      "grad_norm": 2.001342535018921,
      "learning_rate": 9.267876144390102e-07,
      "loss": 1.9878,
      "num_input_tokens_seen": 135321280,
      "step": 132900
    },
    {
      "epoch": 2.61857415683881,
      "grad_norm": 1.8028596639633179,
      "learning_rate": 9.266801643910368e-07,
      "loss": 1.9808,
      "num_input_tokens_seen": 135422760,
      "step": 133000
    },
    {
      "epoch": 2.620543009588313,
      "grad_norm": 1.9720797538757324,
      "learning_rate": 9.265726417897002e-07,
      "loss": 2.0061,
      "num_input_tokens_seen": 135523760,
      "step": 133100
    },
    {
      "epoch": 2.6225118623378156,
      "grad_norm": 1.8140394687652588,
      "learning_rate": 9.264650466532836e-07,
      "loss": 1.9839,
      "num_input_tokens_seen": 135626160,
      "step": 133200
    },
    {
      "epoch": 2.6244807150873184,
      "grad_norm": 2.1675796508789062,
      "learning_rate": 9.263573790000829e-07,
      "loss": 1.974,
      "num_input_tokens_seen": 135728560,
      "step": 133300
    },
    {
      "epoch": 2.6264495678368216,
      "grad_norm": 2.0989134311676025,
      "learning_rate": 9.26249638848406e-07,
      "loss": 2.0111,
      "num_input_tokens_seen": 135830960,
      "step": 133400
    },
    {
      "epoch": 2.6284184205863244,
      "grad_norm": 2.260593891143799,
      "learning_rate": 9.261418262165731e-07,
      "loss": 2.0108,
      "num_input_tokens_seen": 135933360,
      "step": 133500
    },
    {
      "epoch": 2.630387273335827,
      "grad_norm": 2.1290645599365234,
      "learning_rate": 9.26033941122917e-07,
      "loss": 2.0208,
      "num_input_tokens_seen": 136035672,
      "step": 133600
    },
    {
      "epoch": 2.6323561260853303,
      "grad_norm": 2.066185235977173,
      "learning_rate": 9.259259835857825e-07,
      "loss": 2.0209,
      "num_input_tokens_seen": 136137256,
      "step": 133700
    },
    {
      "epoch": 2.634324978834833,
      "grad_norm": 1.9091438055038452,
      "learning_rate": 9.258179536235268e-07,
      "loss": 1.9751,
      "num_input_tokens_seen": 136239656,
      "step": 133800
    },
    {
      "epoch": 2.636293831584336,
      "grad_norm": 1.862436056137085,
      "learning_rate": 9.257098512545196e-07,
      "loss": 1.9894,
      "num_input_tokens_seen": 136342056,
      "step": 133900
    },
    {
      "epoch": 2.6382626843338386,
      "grad_norm": 1.8303775787353516,
      "learning_rate": 9.256016764971429e-07,
      "loss": 1.9832,
      "num_input_tokens_seen": 136442880,
      "step": 134000
    },
    {
      "epoch": 2.6402315370833414,
      "grad_norm": 2.034001350402832,
      "learning_rate": 9.254934293697909e-07,
      "loss": 1.9743,
      "num_input_tokens_seen": 136544760,
      "step": 134100
    },
    {
      "epoch": 2.6422003898328446,
      "grad_norm": 1.912581205368042,
      "learning_rate": 9.253851098908698e-07,
      "loss": 2.0062,
      "num_input_tokens_seen": 136647160,
      "step": 134200
    },
    {
      "epoch": 2.6441692425823473,
      "grad_norm": 2.018368721008301,
      "learning_rate": 9.252767180787988e-07,
      "loss": 2.0146,
      "num_input_tokens_seen": 136749560,
      "step": 134300
    },
    {
      "epoch": 2.64613809533185,
      "grad_norm": 1.8418313264846802,
      "learning_rate": 9.251682539520087e-07,
      "loss": 1.997,
      "num_input_tokens_seen": 136850640,
      "step": 134400
    },
    {
      "epoch": 2.648106948081353,
      "grad_norm": 1.9146231412887573,
      "learning_rate": 9.250597175289432e-07,
      "loss": 1.9582,
      "num_input_tokens_seen": 136953040,
      "step": 134500
    },
    {
      "epoch": 2.650075800830856,
      "grad_norm": 1.7362430095672607,
      "learning_rate": 9.249511088280577e-07,
      "loss": 2.0079,
      "num_input_tokens_seen": 137054728,
      "step": 134600
    },
    {
      "epoch": 2.652044653580359,
      "grad_norm": 1.7822071313858032,
      "learning_rate": 9.248424278678204e-07,
      "loss": 2.0114,
      "num_input_tokens_seen": 137157128,
      "step": 134700
    },
    {
      "epoch": 2.6540135063298615,
      "grad_norm": 1.8319240808486938,
      "learning_rate": 9.247336746667116e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 137258224,
      "step": 134800
    },
    {
      "epoch": 2.6559823590793643,
      "grad_norm": 2.119603395462036,
      "learning_rate": 9.246248492432239e-07,
      "loss": 1.9493,
      "num_input_tokens_seen": 137359768,
      "step": 134900
    },
    {
      "epoch": 2.657951211828867,
      "grad_norm": 1.8390417098999023,
      "learning_rate": 9.245159516158617e-07,
      "loss": 1.9964,
      "num_input_tokens_seen": 137461728,
      "step": 135000
    },
    {
      "epoch": 2.6599200645783703,
      "grad_norm": 1.992293119430542,
      "learning_rate": 9.244069818031426e-07,
      "loss": 1.9381,
      "num_input_tokens_seen": 137564128,
      "step": 135100
    },
    {
      "epoch": 2.661888917327873,
      "grad_norm": 1.7668195962905884,
      "learning_rate": 9.242979398235959e-07,
      "loss": 1.9997,
      "num_input_tokens_seen": 137665824,
      "step": 135200
    },
    {
      "epoch": 2.663857770077376,
      "grad_norm": 15.580219268798828,
      "learning_rate": 9.241888256957633e-07,
      "loss": 2.0132,
      "num_input_tokens_seen": 137768224,
      "step": 135300
    },
    {
      "epoch": 2.665826622826879,
      "grad_norm": 1.9911000728607178,
      "learning_rate": 9.240796394381985e-07,
      "loss": 1.9664,
      "num_input_tokens_seen": 137869808,
      "step": 135400
    },
    {
      "epoch": 2.6677954755763817,
      "grad_norm": 1.9079303741455078,
      "learning_rate": 9.23970381069468e-07,
      "loss": 1.9713,
      "num_input_tokens_seen": 137972208,
      "step": 135500
    },
    {
      "epoch": 2.6697643283258845,
      "grad_norm": 2.3192949295043945,
      "learning_rate": 9.238610506081501e-07,
      "loss": 1.9664,
      "num_input_tokens_seen": 138073880,
      "step": 135600
    },
    {
      "epoch": 2.6717331810753873,
      "grad_norm": 1.758529782295227,
      "learning_rate": 9.237516480728356e-07,
      "loss": 2.032,
      "num_input_tokens_seen": 138176280,
      "step": 135700
    },
    {
      "epoch": 2.67370203382489,
      "grad_norm": 2.1973562240600586,
      "learning_rate": 9.236421734821273e-07,
      "loss": 2.0422,
      "num_input_tokens_seen": 138276816,
      "step": 135800
    },
    {
      "epoch": 2.6756708865743932,
      "grad_norm": 2.0563364028930664,
      "learning_rate": 9.235326268546406e-07,
      "loss": 1.9659,
      "num_input_tokens_seen": 138378616,
      "step": 135900
    },
    {
      "epoch": 2.677639739323896,
      "grad_norm": 2.002542018890381,
      "learning_rate": 9.23423008209003e-07,
      "loss": 1.963,
      "num_input_tokens_seen": 138481016,
      "step": 136000
    },
    {
      "epoch": 2.6796085920733987,
      "grad_norm": 2.061286211013794,
      "learning_rate": 9.233133175638541e-07,
      "loss": 2.0181,
      "num_input_tokens_seen": 138582584,
      "step": 136100
    },
    {
      "epoch": 2.681577444822902,
      "grad_norm": 1.6337310075759888,
      "learning_rate": 9.23203554937846e-07,
      "loss": 1.9488,
      "num_input_tokens_seen": 138684168,
      "step": 136200
    },
    {
      "epoch": 2.6835462975724047,
      "grad_norm": 1.8827173709869385,
      "learning_rate": 9.230937203496428e-07,
      "loss": 1.9828,
      "num_input_tokens_seen": 138786568,
      "step": 136300
    },
    {
      "epoch": 2.6855151503219075,
      "grad_norm": 2.247293472290039,
      "learning_rate": 9.22983813817921e-07,
      "loss": 1.9879,
      "num_input_tokens_seen": 138888968,
      "step": 136400
    },
    {
      "epoch": 2.68748400307141,
      "grad_norm": 2.4556334018707275,
      "learning_rate": 9.228738353613693e-07,
      "loss": 1.9555,
      "num_input_tokens_seen": 138990600,
      "step": 136500
    },
    {
      "epoch": 2.689452855820913,
      "grad_norm": 3.499016523361206,
      "learning_rate": 9.227637849986886e-07,
      "loss": 1.9704,
      "num_input_tokens_seen": 139093000,
      "step": 136600
    },
    {
      "epoch": 2.691421708570416,
      "grad_norm": 2.174551248550415,
      "learning_rate": 9.22653662748592e-07,
      "loss": 1.9854,
      "num_input_tokens_seen": 139193960,
      "step": 136700
    },
    {
      "epoch": 2.693390561319919,
      "grad_norm": 1.9812849760055542,
      "learning_rate": 9.225434686298048e-07,
      "loss": 2.0031,
      "num_input_tokens_seen": 139293784,
      "step": 136800
    },
    {
      "epoch": 2.6953594140694217,
      "grad_norm": 1.7438524961471558,
      "learning_rate": 9.224332026610647e-07,
      "loss": 1.9305,
      "num_input_tokens_seen": 139396184,
      "step": 136900
    },
    {
      "epoch": 2.6973282668189245,
      "grad_norm": 1.6979296207427979,
      "learning_rate": 9.223228648611214e-07,
      "loss": 1.9994,
      "num_input_tokens_seen": 139498104,
      "step": 137000
    },
    {
      "epoch": 2.6992971195684277,
      "grad_norm": 1.828908085823059,
      "learning_rate": 9.222124552487372e-07,
      "loss": 1.9572,
      "num_input_tokens_seen": 139600504,
      "step": 137100
    },
    {
      "epoch": 2.7012659723179304,
      "grad_norm": 1.8999994993209839,
      "learning_rate": 9.221019738426859e-07,
      "loss": 1.9585,
      "num_input_tokens_seen": 139702904,
      "step": 137200
    },
    {
      "epoch": 2.703234825067433,
      "grad_norm": 2.0553765296936035,
      "learning_rate": 9.21991420661754e-07,
      "loss": 1.9695,
      "num_input_tokens_seen": 139803896,
      "step": 137300
    },
    {
      "epoch": 2.705203677816936,
      "grad_norm": 1.7925704717636108,
      "learning_rate": 9.218807957247406e-07,
      "loss": 1.9644,
      "num_input_tokens_seen": 139904408,
      "step": 137400
    },
    {
      "epoch": 2.7071725305664387,
      "grad_norm": 1.7893779277801514,
      "learning_rate": 9.217700990504559e-07,
      "loss": 2.0034,
      "num_input_tokens_seen": 140006168,
      "step": 137500
    },
    {
      "epoch": 2.709141383315942,
      "grad_norm": 1.9949560165405273,
      "learning_rate": 9.216593306577234e-07,
      "loss": 1.9777,
      "num_input_tokens_seen": 140108568,
      "step": 137600
    },
    {
      "epoch": 2.7111102360654447,
      "grad_norm": 2.1235673427581787,
      "learning_rate": 9.21548490565378e-07,
      "loss": 2.0064,
      "num_input_tokens_seen": 140210664,
      "step": 137700
    },
    {
      "epoch": 2.7130790888149474,
      "grad_norm": 2.0905706882476807,
      "learning_rate": 9.214375787922673e-07,
      "loss": 1.9855,
      "num_input_tokens_seen": 140312432,
      "step": 137800
    },
    {
      "epoch": 2.7150479415644506,
      "grad_norm": 1.8522093296051025,
      "learning_rate": 9.21326595357251e-07,
      "loss": 1.9697,
      "num_input_tokens_seen": 140413056,
      "step": 137900
    },
    {
      "epoch": 2.7170167943139534,
      "grad_norm": 1.9855432510375977,
      "learning_rate": 9.212155402792008e-07,
      "loss": 2.0105,
      "num_input_tokens_seen": 140513744,
      "step": 138000
    },
    {
      "epoch": 2.718985647063456,
      "grad_norm": 1.9174432754516602,
      "learning_rate": 9.211044135770005e-07,
      "loss": 2.0049,
      "num_input_tokens_seen": 140616040,
      "step": 138100
    },
    {
      "epoch": 2.720954499812959,
      "grad_norm": 1.7407281398773193,
      "learning_rate": 9.209932152695466e-07,
      "loss": 2.0164,
      "num_input_tokens_seen": 140717936,
      "step": 138200
    },
    {
      "epoch": 2.7229233525624617,
      "grad_norm": 1.7572743892669678,
      "learning_rate": 9.208819453757473e-07,
      "loss": 2.0132,
      "num_input_tokens_seen": 140820144,
      "step": 138300
    },
    {
      "epoch": 2.724892205311965,
      "grad_norm": 1.9282379150390625,
      "learning_rate": 9.207706039145229e-07,
      "loss": 1.9429,
      "num_input_tokens_seen": 140921808,
      "step": 138400
    },
    {
      "epoch": 2.7268610580614676,
      "grad_norm": 1.9206793308258057,
      "learning_rate": 9.206591909048063e-07,
      "loss": 2.027,
      "num_input_tokens_seen": 141023760,
      "step": 138500
    },
    {
      "epoch": 2.7288299108109704,
      "grad_norm": 2.001131296157837,
      "learning_rate": 9.205477063655424e-07,
      "loss": 2.0271,
      "num_input_tokens_seen": 141123640,
      "step": 138600
    },
    {
      "epoch": 2.7307987635604736,
      "grad_norm": 1.8215125799179077,
      "learning_rate": 9.204361503156881e-07,
      "loss": 2.0414,
      "num_input_tokens_seen": 141225992,
      "step": 138700
    },
    {
      "epoch": 2.7327676163099763,
      "grad_norm": 2.420888662338257,
      "learning_rate": 9.203245227742125e-07,
      "loss": 1.9799,
      "num_input_tokens_seen": 141328392,
      "step": 138800
    },
    {
      "epoch": 2.734736469059479,
      "grad_norm": 1.938777208328247,
      "learning_rate": 9.202128237600969e-07,
      "loss": 1.9957,
      "num_input_tokens_seen": 141429096,
      "step": 138900
    },
    {
      "epoch": 2.736705321808982,
      "grad_norm": 4.348512649536133,
      "learning_rate": 9.201010532923349e-07,
      "loss": 2.0387,
      "num_input_tokens_seen": 141530920,
      "step": 139000
    },
    {
      "epoch": 2.7386741745584846,
      "grad_norm": 1.7346715927124023,
      "learning_rate": 9.199892113899322e-07,
      "loss": 1.9955,
      "num_input_tokens_seen": 141631704,
      "step": 139100
    },
    {
      "epoch": 2.7406430273079874,
      "grad_norm": 1.8914837837219238,
      "learning_rate": 9.198772980719063e-07,
      "loss": 1.9765,
      "num_input_tokens_seen": 141733880,
      "step": 139200
    },
    {
      "epoch": 2.7426118800574906,
      "grad_norm": 1.7075005769729614,
      "learning_rate": 9.197653133572873e-07,
      "loss": 1.9777,
      "num_input_tokens_seen": 141834960,
      "step": 139300
    },
    {
      "epoch": 2.7445807328069933,
      "grad_norm": 2.315007448196411,
      "learning_rate": 9.196532572651172e-07,
      "loss": 2.0166,
      "num_input_tokens_seen": 141936176,
      "step": 139400
    },
    {
      "epoch": 2.746549585556496,
      "grad_norm": 1.8903642892837524,
      "learning_rate": 9.195411298144504e-07,
      "loss": 1.9442,
      "num_input_tokens_seen": 142038576,
      "step": 139500
    },
    {
      "epoch": 2.7485184383059993,
      "grad_norm": 1.965807557106018,
      "learning_rate": 9.194289310243528e-07,
      "loss": 1.9608,
      "num_input_tokens_seen": 142139456,
      "step": 139600
    },
    {
      "epoch": 2.750487291055502,
      "grad_norm": 1.9385151863098145,
      "learning_rate": 9.193166609139033e-07,
      "loss": 1.9783,
      "num_input_tokens_seen": 142241856,
      "step": 139700
    },
    {
      "epoch": 2.752456143805005,
      "grad_norm": 2.076444387435913,
      "learning_rate": 9.192043195021922e-07,
      "loss": 2.0202,
      "num_input_tokens_seen": 142344256,
      "step": 139800
    },
    {
      "epoch": 2.7544249965545076,
      "grad_norm": 4.172542095184326,
      "learning_rate": 9.190919068083224e-07,
      "loss": 2.0064,
      "num_input_tokens_seen": 142446032,
      "step": 139900
    },
    {
      "epoch": 2.7563938493040103,
      "grad_norm": 1.884360671043396,
      "learning_rate": 9.189794228514085e-07,
      "loss": 2.0104,
      "num_input_tokens_seen": 142548432,
      "step": 140000
    },
    {
      "epoch": 2.7583627020535135,
      "grad_norm": 1.7790875434875488,
      "learning_rate": 9.188668676505777e-07,
      "loss": 1.969,
      "num_input_tokens_seen": 142650280,
      "step": 140100
    },
    {
      "epoch": 2.7603315548030163,
      "grad_norm": 1.8419520854949951,
      "learning_rate": 9.187542412249689e-07,
      "loss": 1.9908,
      "num_input_tokens_seen": 142751704,
      "step": 140200
    },
    {
      "epoch": 2.762300407552519,
      "grad_norm": 1.9317359924316406,
      "learning_rate": 9.186415435937333e-07,
      "loss": 1.9929,
      "num_input_tokens_seen": 142854104,
      "step": 140300
    },
    {
      "epoch": 2.7642692603020222,
      "grad_norm": 2.1541104316711426,
      "learning_rate": 9.185287747760342e-07,
      "loss": 1.9594,
      "num_input_tokens_seen": 142955344,
      "step": 140400
    },
    {
      "epoch": 2.766238113051525,
      "grad_norm": 1.8215970993041992,
      "learning_rate": 9.18415934791047e-07,
      "loss": 2.0056,
      "num_input_tokens_seen": 143057744,
      "step": 140500
    },
    {
      "epoch": 2.7682069658010278,
      "grad_norm": 2.3591055870056152,
      "learning_rate": 9.183030236579591e-07,
      "loss": 1.9826,
      "num_input_tokens_seen": 143160144,
      "step": 140600
    },
    {
      "epoch": 2.7701758185505305,
      "grad_norm": 1.8065130710601807,
      "learning_rate": 9.181900413959703e-07,
      "loss": 2.0148,
      "num_input_tokens_seen": 143260632,
      "step": 140700
    },
    {
      "epoch": 2.7721446713000333,
      "grad_norm": 1.9266314506530762,
      "learning_rate": 9.18076988024292e-07,
      "loss": 1.9509,
      "num_input_tokens_seen": 143362720,
      "step": 140800
    },
    {
      "epoch": 2.7741135240495365,
      "grad_norm": 2.158468008041382,
      "learning_rate": 9.179638635621481e-07,
      "loss": 1.971,
      "num_input_tokens_seen": 143463128,
      "step": 140900
    },
    {
      "epoch": 2.7760823767990392,
      "grad_norm": 2.2344703674316406,
      "learning_rate": 9.178506680287746e-07,
      "loss": 1.9975,
      "num_input_tokens_seen": 143564736,
      "step": 141000
    },
    {
      "epoch": 2.778051229548542,
      "grad_norm": 1.9476065635681152,
      "learning_rate": 9.177374014434192e-07,
      "loss": 1.9718,
      "num_input_tokens_seen": 143664912,
      "step": 141100
    },
    {
      "epoch": 2.780020082298045,
      "grad_norm": 1.8100794553756714,
      "learning_rate": 9.176240638253421e-07,
      "loss": 1.9747,
      "num_input_tokens_seen": 143767312,
      "step": 141200
    },
    {
      "epoch": 2.781988935047548,
      "grad_norm": 1.922595739364624,
      "learning_rate": 9.175106551938153e-07,
      "loss": 1.9852,
      "num_input_tokens_seen": 143868904,
      "step": 141300
    },
    {
      "epoch": 2.7839577877970507,
      "grad_norm": 2.0649492740631104,
      "learning_rate": 9.173971755681231e-07,
      "loss": 1.9941,
      "num_input_tokens_seen": 143971304,
      "step": 141400
    },
    {
      "epoch": 2.7859266405465535,
      "grad_norm": 2.0110702514648438,
      "learning_rate": 9.172836249675617e-07,
      "loss": 1.9563,
      "num_input_tokens_seen": 144073704,
      "step": 141500
    },
    {
      "epoch": 2.7878954932960562,
      "grad_norm": 1.9530744552612305,
      "learning_rate": 9.171700034114394e-07,
      "loss": 2.0128,
      "num_input_tokens_seen": 144175584,
      "step": 141600
    },
    {
      "epoch": 2.789864346045559,
      "grad_norm": 2.0976758003234863,
      "learning_rate": 9.170563109190766e-07,
      "loss": 2.004,
      "num_input_tokens_seen": 144277984,
      "step": 141700
    },
    {
      "epoch": 2.791833198795062,
      "grad_norm": 2.0704092979431152,
      "learning_rate": 9.169425475098056e-07,
      "loss": 1.9728,
      "num_input_tokens_seen": 144379536,
      "step": 141800
    },
    {
      "epoch": 2.793802051544565,
      "grad_norm": 1.815583348274231,
      "learning_rate": 9.16828713202971e-07,
      "loss": 1.9135,
      "num_input_tokens_seen": 144481936,
      "step": 141900
    },
    {
      "epoch": 2.7957709042940677,
      "grad_norm": 1.7426722049713135,
      "learning_rate": 9.167148080179297e-07,
      "loss": 1.9656,
      "num_input_tokens_seen": 144584336,
      "step": 142000
    },
    {
      "epoch": 2.797739757043571,
      "grad_norm": 2.062602996826172,
      "learning_rate": 9.166008319740499e-07,
      "loss": 1.993,
      "num_input_tokens_seen": 144685840,
      "step": 142100
    },
    {
      "epoch": 2.7997086097930737,
      "grad_norm": 2.510030746459961,
      "learning_rate": 9.164867850907124e-07,
      "loss": 1.9507,
      "num_input_tokens_seen": 144788240,
      "step": 142200
    },
    {
      "epoch": 2.8016774625425764,
      "grad_norm": 1.8817150592803955,
      "learning_rate": 9.1637266738731e-07,
      "loss": 1.9942,
      "num_input_tokens_seen": 144889976,
      "step": 142300
    },
    {
      "epoch": 2.803646315292079,
      "grad_norm": 1.7231566905975342,
      "learning_rate": 9.162584788832473e-07,
      "loss": 2.0297,
      "num_input_tokens_seen": 144992376,
      "step": 142400
    },
    {
      "epoch": 2.805615168041582,
      "grad_norm": 1.8418123722076416,
      "learning_rate": 9.16144219597941e-07,
      "loss": 2.0164,
      "num_input_tokens_seen": 145094776,
      "step": 142500
    },
    {
      "epoch": 2.807584020791085,
      "grad_norm": 1.9155409336090088,
      "learning_rate": 9.160298895508202e-07,
      "loss": 1.948,
      "num_input_tokens_seen": 145195616,
      "step": 142600
    },
    {
      "epoch": 2.809552873540588,
      "grad_norm": 1.774415373802185,
      "learning_rate": 9.159154887613257e-07,
      "loss": 1.9702,
      "num_input_tokens_seen": 145298016,
      "step": 142700
    },
    {
      "epoch": 2.8115217262900907,
      "grad_norm": 1.7046304941177368,
      "learning_rate": 9.158010172489102e-07,
      "loss": 1.9656,
      "num_input_tokens_seen": 145400416,
      "step": 142800
    },
    {
      "epoch": 2.813490579039594,
      "grad_norm": 2.1160473823547363,
      "learning_rate": 9.156864750330389e-07,
      "loss": 1.9631,
      "num_input_tokens_seen": 145502816,
      "step": 142900
    },
    {
      "epoch": 2.8154594317890966,
      "grad_norm": 1.7345004081726074,
      "learning_rate": 9.155718621331885e-07,
      "loss": 1.9819,
      "num_input_tokens_seen": 145605216,
      "step": 143000
    },
    {
      "epoch": 2.8174282845385994,
      "grad_norm": 2.054319143295288,
      "learning_rate": 9.15457178568848e-07,
      "loss": 2.0453,
      "num_input_tokens_seen": 145707616,
      "step": 143100
    },
    {
      "epoch": 2.819397137288102,
      "grad_norm": 1.8499298095703125,
      "learning_rate": 9.153424243595185e-07,
      "loss": 1.9449,
      "num_input_tokens_seen": 145810016,
      "step": 143200
    },
    {
      "epoch": 2.821365990037605,
      "grad_norm": 1.8021115064620972,
      "learning_rate": 9.152275995247128e-07,
      "loss": 1.9548,
      "num_input_tokens_seen": 145912416,
      "step": 143300
    },
    {
      "epoch": 2.823334842787108,
      "grad_norm": 1.7581770420074463,
      "learning_rate": 9.15112704083956e-07,
      "loss": 1.9784,
      "num_input_tokens_seen": 146013976,
      "step": 143400
    },
    {
      "epoch": 2.825303695536611,
      "grad_norm": 2.0402488708496094,
      "learning_rate": 9.149977380567852e-07,
      "loss": 2.0034,
      "num_input_tokens_seen": 146115888,
      "step": 143500
    },
    {
      "epoch": 2.8272725482861136,
      "grad_norm": 1.9908995628356934,
      "learning_rate": 9.148827014627492e-07,
      "loss": 1.9587,
      "num_input_tokens_seen": 146218288,
      "step": 143600
    },
    {
      "epoch": 2.8292414010356164,
      "grad_norm": 1.9173650741577148,
      "learning_rate": 9.147675943214091e-07,
      "loss": 1.9649,
      "num_input_tokens_seen": 146320688,
      "step": 143700
    },
    {
      "epoch": 2.8312102537851196,
      "grad_norm": 1.8674744367599487,
      "learning_rate": 9.14652416652338e-07,
      "loss": 2.0627,
      "num_input_tokens_seen": 146422056,
      "step": 143800
    },
    {
      "epoch": 2.8331791065346223,
      "grad_norm": 1.8405635356903076,
      "learning_rate": 9.145371684751206e-07,
      "loss": 1.9678,
      "num_input_tokens_seen": 146523976,
      "step": 143900
    },
    {
      "epoch": 2.835147959284125,
      "grad_norm": 1.8517303466796875,
      "learning_rate": 9.144218498093541e-07,
      "loss": 1.9528,
      "num_input_tokens_seen": 146626376,
      "step": 144000
    },
    {
      "epoch": 2.837116812033628,
      "grad_norm": 2.000028610229492,
      "learning_rate": 9.143064606746476e-07,
      "loss": 1.9807,
      "num_input_tokens_seen": 146728776,
      "step": 144100
    },
    {
      "epoch": 2.8390856647831306,
      "grad_norm": 1.7356005907058716,
      "learning_rate": 9.141910010906218e-07,
      "loss": 2.0446,
      "num_input_tokens_seen": 146830576,
      "step": 144200
    },
    {
      "epoch": 2.841054517532634,
      "grad_norm": 1.9533261060714722,
      "learning_rate": 9.140754710769097e-07,
      "loss": 1.9862,
      "num_input_tokens_seen": 146932976,
      "step": 144300
    },
    {
      "epoch": 2.8430233702821366,
      "grad_norm": 1.9368846416473389,
      "learning_rate": 9.139598706531562e-07,
      "loss": 1.9735,
      "num_input_tokens_seen": 147034608,
      "step": 144400
    },
    {
      "epoch": 2.8449922230316393,
      "grad_norm": 2.0260725021362305,
      "learning_rate": 9.138441998390183e-07,
      "loss": 2.0029,
      "num_input_tokens_seen": 147135624,
      "step": 144500
    },
    {
      "epoch": 2.8469610757811425,
      "grad_norm": 4.3465352058410645,
      "learning_rate": 9.137284586541646e-07,
      "loss": 1.9741,
      "num_input_tokens_seen": 147238024,
      "step": 144600
    },
    {
      "epoch": 2.8489299285306453,
      "grad_norm": 1.8397555351257324,
      "learning_rate": 9.136126471182761e-07,
      "loss": 1.9228,
      "num_input_tokens_seen": 147338480,
      "step": 144700
    },
    {
      "epoch": 2.850898781280148,
      "grad_norm": 2.2060608863830566,
      "learning_rate": 9.134967652510456e-07,
      "loss": 1.9574,
      "num_input_tokens_seen": 147440360,
      "step": 144800
    },
    {
      "epoch": 2.852867634029651,
      "grad_norm": 1.9449845552444458,
      "learning_rate": 9.133808130721777e-07,
      "loss": 2.0137,
      "num_input_tokens_seen": 147542072,
      "step": 144900
    },
    {
      "epoch": 2.8548364867791536,
      "grad_norm": 1.9381047487258911,
      "learning_rate": 9.13264790601389e-07,
      "loss": 1.952,
      "num_input_tokens_seen": 147644472,
      "step": 145000
    },
    {
      "epoch": 2.856805339528657,
      "grad_norm": 1.975183129310608,
      "learning_rate": 9.131486978584085e-07,
      "loss": 2.0197,
      "num_input_tokens_seen": 147746872,
      "step": 145100
    },
    {
      "epoch": 2.8587741922781595,
      "grad_norm": 2.124565839767456,
      "learning_rate": 9.130325348629766e-07,
      "loss": 1.9374,
      "num_input_tokens_seen": 147849272,
      "step": 145200
    },
    {
      "epoch": 2.8607430450276623,
      "grad_norm": 2.788569927215576,
      "learning_rate": 9.129163016348458e-07,
      "loss": 1.9459,
      "num_input_tokens_seen": 147951672,
      "step": 145300
    },
    {
      "epoch": 2.8627118977771655,
      "grad_norm": 1.9773931503295898,
      "learning_rate": 9.127999981937806e-07,
      "loss": 1.9606,
      "num_input_tokens_seen": 148054072,
      "step": 145400
    },
    {
      "epoch": 2.8646807505266683,
      "grad_norm": 1.9272105693817139,
      "learning_rate": 9.126836245595572e-07,
      "loss": 2.0163,
      "num_input_tokens_seen": 148156472,
      "step": 145500
    },
    {
      "epoch": 2.866649603276171,
      "grad_norm": 1.7956663370132446,
      "learning_rate": 9.125671807519644e-07,
      "loss": 1.9884,
      "num_input_tokens_seen": 148257416,
      "step": 145600
    },
    {
      "epoch": 2.868618456025674,
      "grad_norm": 1.8678843975067139,
      "learning_rate": 9.124506667908023e-07,
      "loss": 2.0125,
      "num_input_tokens_seen": 148359816,
      "step": 145700
    },
    {
      "epoch": 2.8705873087751765,
      "grad_norm": 1.9273408651351929,
      "learning_rate": 9.12334082695883e-07,
      "loss": 1.9562,
      "num_input_tokens_seen": 148462216,
      "step": 145800
    },
    {
      "epoch": 2.8725561615246793,
      "grad_norm": 2.330432653427124,
      "learning_rate": 9.122174284870307e-07,
      "loss": 1.9803,
      "num_input_tokens_seen": 148563888,
      "step": 145900
    },
    {
      "epoch": 2.8745250142741825,
      "grad_norm": 1.7445673942565918,
      "learning_rate": 9.121007041840816e-07,
      "loss": 1.9816,
      "num_input_tokens_seen": 148665624,
      "step": 146000
    },
    {
      "epoch": 2.8764938670236853,
      "grad_norm": 2.0610992908477783,
      "learning_rate": 9.119839098068835e-07,
      "loss": 2.0486,
      "num_input_tokens_seen": 148766752,
      "step": 146100
    },
    {
      "epoch": 2.878462719773188,
      "grad_norm": 1.9714587926864624,
      "learning_rate": 9.118670453752964e-07,
      "loss": 1.9396,
      "num_input_tokens_seen": 148868736,
      "step": 146200
    },
    {
      "epoch": 2.880431572522691,
      "grad_norm": 1.8859783411026,
      "learning_rate": 9.11750110909192e-07,
      "loss": 1.9396,
      "num_input_tokens_seen": 148971136,
      "step": 146300
    },
    {
      "epoch": 2.882400425272194,
      "grad_norm": 2.111337423324585,
      "learning_rate": 9.116331064284542e-07,
      "loss": 2.0412,
      "num_input_tokens_seen": 149071728,
      "step": 146400
    },
    {
      "epoch": 2.8843692780216967,
      "grad_norm": 2.2499639987945557,
      "learning_rate": 9.115160319529783e-07,
      "loss": 2.0356,
      "num_input_tokens_seen": 149173560,
      "step": 146500
    },
    {
      "epoch": 2.8863381307711995,
      "grad_norm": 1.7903567552566528,
      "learning_rate": 9.113988875026724e-07,
      "loss": 2.0333,
      "num_input_tokens_seen": 149275328,
      "step": 146600
    },
    {
      "epoch": 2.8883069835207023,
      "grad_norm": 1.816192388534546,
      "learning_rate": 9.112816730974552e-07,
      "loss": 1.964,
      "num_input_tokens_seen": 149376408,
      "step": 146700
    },
    {
      "epoch": 2.8902758362702055,
      "grad_norm": 2.165644884109497,
      "learning_rate": 9.111643887572586e-07,
      "loss": 1.9503,
      "num_input_tokens_seen": 149478104,
      "step": 146800
    },
    {
      "epoch": 2.892244689019708,
      "grad_norm": 1.7579931020736694,
      "learning_rate": 9.110470345020257e-07,
      "loss": 1.9369,
      "num_input_tokens_seen": 149579984,
      "step": 146900
    },
    {
      "epoch": 2.894213541769211,
      "grad_norm": 2.125364303588867,
      "learning_rate": 9.109296103517115e-07,
      "loss": 2.013,
      "num_input_tokens_seen": 149682384,
      "step": 147000
    },
    {
      "epoch": 2.896182394518714,
      "grad_norm": 1.868359923362732,
      "learning_rate": 9.10812116326283e-07,
      "loss": 1.9916,
      "num_input_tokens_seen": 149784784,
      "step": 147100
    },
    {
      "epoch": 2.898151247268217,
      "grad_norm": 2.1574387550354004,
      "learning_rate": 9.10694552445719e-07,
      "loss": 1.9398,
      "num_input_tokens_seen": 149887184,
      "step": 147200
    },
    {
      "epoch": 2.9001201000177197,
      "grad_norm": 1.8687632083892822,
      "learning_rate": 9.105769187300102e-07,
      "loss": 2.0205,
      "num_input_tokens_seen": 149989584,
      "step": 147300
    },
    {
      "epoch": 2.9020889527672225,
      "grad_norm": 1.9556429386138916,
      "learning_rate": 9.104592151991594e-07,
      "loss": 2.0266,
      "num_input_tokens_seen": 150090624,
      "step": 147400
    },
    {
      "epoch": 2.904057805516725,
      "grad_norm": 1.843937873840332,
      "learning_rate": 9.103414418731811e-07,
      "loss": 1.9608,
      "num_input_tokens_seen": 150193024,
      "step": 147500
    },
    {
      "epoch": 2.9060266582662284,
      "grad_norm": 1.852134346961975,
      "learning_rate": 9.102235987721016e-07,
      "loss": 1.9864,
      "num_input_tokens_seen": 150294872,
      "step": 147600
    },
    {
      "epoch": 2.907995511015731,
      "grad_norm": 1.8409780263900757,
      "learning_rate": 9.10105685915959e-07,
      "loss": 1.9823,
      "num_input_tokens_seen": 150396128,
      "step": 147700
    },
    {
      "epoch": 2.909964363765234,
      "grad_norm": 4.855228424072266,
      "learning_rate": 9.099877033248036e-07,
      "loss": 1.9702,
      "num_input_tokens_seen": 150497856,
      "step": 147800
    },
    {
      "epoch": 2.911933216514737,
      "grad_norm": 1.8119330406188965,
      "learning_rate": 9.098696510186973e-07,
      "loss": 1.9961,
      "num_input_tokens_seen": 150599704,
      "step": 147900
    },
    {
      "epoch": 2.91390206926424,
      "grad_norm": 2.1123673915863037,
      "learning_rate": 9.097515290177138e-07,
      "loss": 1.9562,
      "num_input_tokens_seen": 150701416,
      "step": 148000
    },
    {
      "epoch": 2.9158709220137426,
      "grad_norm": 2.0797832012176514,
      "learning_rate": 9.096333373419387e-07,
      "loss": 1.9961,
      "num_input_tokens_seen": 150803816,
      "step": 148100
    },
    {
      "epoch": 2.9178397747632454,
      "grad_norm": 1.924194097518921,
      "learning_rate": 9.095150760114696e-07,
      "loss": 2.0,
      "num_input_tokens_seen": 150906216,
      "step": 148200
    },
    {
      "epoch": 2.919808627512748,
      "grad_norm": 2.2523162364959717,
      "learning_rate": 9.093967450464159e-07,
      "loss": 1.9633,
      "num_input_tokens_seen": 151007952,
      "step": 148300
    },
    {
      "epoch": 2.921777480262251,
      "grad_norm": 2.0440046787261963,
      "learning_rate": 9.092783444668988e-07,
      "loss": 1.9963,
      "num_input_tokens_seen": 151110080,
      "step": 148400
    },
    {
      "epoch": 2.923746333011754,
      "grad_norm": 1.921358585357666,
      "learning_rate": 9.091598742930511e-07,
      "loss": 1.9978,
      "num_input_tokens_seen": 151212480,
      "step": 148500
    },
    {
      "epoch": 2.925715185761257,
      "grad_norm": 2.0041520595550537,
      "learning_rate": 9.090413345450179e-07,
      "loss": 1.9963,
      "num_input_tokens_seen": 151314568,
      "step": 148600
    },
    {
      "epoch": 2.9276840385107596,
      "grad_norm": 2.030916213989258,
      "learning_rate": 9.089227252429557e-07,
      "loss": 2.0069,
      "num_input_tokens_seen": 151416344,
      "step": 148700
    },
    {
      "epoch": 2.929652891260263,
      "grad_norm": 1.925169587135315,
      "learning_rate": 9.08804046407033e-07,
      "loss": 2.0032,
      "num_input_tokens_seen": 151518016,
      "step": 148800
    },
    {
      "epoch": 2.9316217440097656,
      "grad_norm": 1.727295994758606,
      "learning_rate": 9.086852980574303e-07,
      "loss": 1.9748,
      "num_input_tokens_seen": 151619704,
      "step": 148900
    },
    {
      "epoch": 2.9335905967592684,
      "grad_norm": 2.8354527950286865,
      "learning_rate": 9.085664802143397e-07,
      "loss": 1.9605,
      "num_input_tokens_seen": 151722104,
      "step": 149000
    },
    {
      "epoch": 2.935559449508771,
      "grad_norm": 1.894550085067749,
      "learning_rate": 9.084475928979652e-07,
      "loss": 1.9814,
      "num_input_tokens_seen": 151823880,
      "step": 149100
    },
    {
      "epoch": 2.937528302258274,
      "grad_norm": 1.8652126789093018,
      "learning_rate": 9.083286361285225e-07,
      "loss": 1.9943,
      "num_input_tokens_seen": 151925528,
      "step": 149200
    },
    {
      "epoch": 2.939497155007777,
      "grad_norm": 3.667811393737793,
      "learning_rate": 9.082096099262391e-07,
      "loss": 1.952,
      "num_input_tokens_seen": 152027928,
      "step": 149300
    },
    {
      "epoch": 2.94146600775728,
      "grad_norm": 1.7258117198944092,
      "learning_rate": 9.080905143113546e-07,
      "loss": 1.9856,
      "num_input_tokens_seen": 152128712,
      "step": 149400
    },
    {
      "epoch": 2.9434348605067826,
      "grad_norm": 2.0203278064727783,
      "learning_rate": 9.079713493041203e-07,
      "loss": 1.9416,
      "num_input_tokens_seen": 152231112,
      "step": 149500
    },
    {
      "epoch": 2.945403713256286,
      "grad_norm": 1.8034933805465698,
      "learning_rate": 9.078521149247989e-07,
      "loss": 1.925,
      "num_input_tokens_seen": 152332144,
      "step": 149600
    },
    {
      "epoch": 2.9473725660057886,
      "grad_norm": 2.9003779888153076,
      "learning_rate": 9.077328111936653e-07,
      "loss": 1.9684,
      "num_input_tokens_seen": 152433712,
      "step": 149700
    },
    {
      "epoch": 2.9493414187552913,
      "grad_norm": 1.8845500946044922,
      "learning_rate": 9.076134381310061e-07,
      "loss": 1.9728,
      "num_input_tokens_seen": 152536112,
      "step": 149800
    },
    {
      "epoch": 2.951310271504794,
      "grad_norm": 1.877603530883789,
      "learning_rate": 9.074939957571199e-07,
      "loss": 1.963,
      "num_input_tokens_seen": 152637648,
      "step": 149900
    },
    {
      "epoch": 2.953279124254297,
      "grad_norm": 2.0671799182891846,
      "learning_rate": 9.073744840923165e-07,
      "loss": 1.9787,
      "num_input_tokens_seen": 152740048,
      "step": 150000
    },
    {
      "epoch": 2.9552479770038,
      "grad_norm": 1.756220817565918,
      "learning_rate": 9.072549031569181e-07,
      "loss": 1.9894,
      "num_input_tokens_seen": 152841808,
      "step": 150100
    },
    {
      "epoch": 2.957216829753303,
      "grad_norm": 1.8161519765853882,
      "learning_rate": 9.071352529712584e-07,
      "loss": 1.9597,
      "num_input_tokens_seen": 152944208,
      "step": 150200
    },
    {
      "epoch": 2.9591856825028056,
      "grad_norm": 1.9361835718154907,
      "learning_rate": 9.070155335556828e-07,
      "loss": 1.9777,
      "num_input_tokens_seen": 153045376,
      "step": 150300
    },
    {
      "epoch": 2.9611545352523083,
      "grad_norm": 1.94983971118927,
      "learning_rate": 9.068957449305488e-07,
      "loss": 1.9811,
      "num_input_tokens_seen": 153147776,
      "step": 150400
    },
    {
      "epoch": 2.9631233880018115,
      "grad_norm": 2.0812742710113525,
      "learning_rate": 9.067758871162252e-07,
      "loss": 2.0175,
      "num_input_tokens_seen": 153248736,
      "step": 150500
    },
    {
      "epoch": 2.9650922407513143,
      "grad_norm": 1.9933220148086548,
      "learning_rate": 9.066559601330929e-07,
      "loss": 1.9907,
      "num_input_tokens_seen": 153351136,
      "step": 150600
    },
    {
      "epoch": 2.967061093500817,
      "grad_norm": 1.9672973155975342,
      "learning_rate": 9.065359640015444e-07,
      "loss": 2.0143,
      "num_input_tokens_seen": 153453464,
      "step": 150700
    },
    {
      "epoch": 2.96902994625032,
      "grad_norm": 1.8607351779937744,
      "learning_rate": 9.064158987419842e-07,
      "loss": 1.9579,
      "num_input_tokens_seen": 153555360,
      "step": 150800
    },
    {
      "epoch": 2.9709987989998226,
      "grad_norm": 2.0569875240325928,
      "learning_rate": 9.062957643748284e-07,
      "loss": 2.006,
      "num_input_tokens_seen": 153657760,
      "step": 150900
    },
    {
      "epoch": 2.9729676517493258,
      "grad_norm": 1.9158655405044556,
      "learning_rate": 9.061755609205044e-07,
      "loss": 1.999,
      "num_input_tokens_seen": 153757472,
      "step": 151000
    },
    {
      "epoch": 2.9749365044988285,
      "grad_norm": 1.9854087829589844,
      "learning_rate": 9.060552883994524e-07,
      "loss": 1.9399,
      "num_input_tokens_seen": 153858400,
      "step": 151100
    },
    {
      "epoch": 2.9769053572483313,
      "grad_norm": 2.4062297344207764,
      "learning_rate": 9.059349468321234e-07,
      "loss": 1.9781,
      "num_input_tokens_seen": 153960240,
      "step": 151200
    },
    {
      "epoch": 2.9788742099978345,
      "grad_norm": 1.714268445968628,
      "learning_rate": 9.058145362389804e-07,
      "loss": 1.9746,
      "num_input_tokens_seen": 154061192,
      "step": 151300
    },
    {
      "epoch": 2.9808430627473372,
      "grad_norm": 1.889648199081421,
      "learning_rate": 9.056940566404983e-07,
      "loss": 1.9691,
      "num_input_tokens_seen": 154162960,
      "step": 151400
    },
    {
      "epoch": 2.98281191549684,
      "grad_norm": 1.926221251487732,
      "learning_rate": 9.055735080571636e-07,
      "loss": 2.0362,
      "num_input_tokens_seen": 154263232,
      "step": 151500
    },
    {
      "epoch": 2.9847807682463428,
      "grad_norm": 1.9325400590896606,
      "learning_rate": 9.054528905094747e-07,
      "loss": 1.9895,
      "num_input_tokens_seen": 154363224,
      "step": 151600
    },
    {
      "epoch": 2.9867496209958455,
      "grad_norm": 1.9317102432250977,
      "learning_rate": 9.053322040179416e-07,
      "loss": 1.9955,
      "num_input_tokens_seen": 154464776,
      "step": 151700
    },
    {
      "epoch": 2.9887184737453487,
      "grad_norm": 2.054033041000366,
      "learning_rate": 9.052114486030859e-07,
      "loss": 1.966,
      "num_input_tokens_seen": 154566408,
      "step": 151800
    },
    {
      "epoch": 2.9906873264948515,
      "grad_norm": 1.9863126277923584,
      "learning_rate": 9.05090624285441e-07,
      "loss": 1.9901,
      "num_input_tokens_seen": 154667800,
      "step": 151900
    },
    {
      "epoch": 2.9926561792443542,
      "grad_norm": 1.8175255060195923,
      "learning_rate": 9.049697310855521e-07,
      "loss": 1.9483,
      "num_input_tokens_seen": 154770200,
      "step": 152000
    },
    {
      "epoch": 2.9946250319938574,
      "grad_norm": 2.620454788208008,
      "learning_rate": 9.048487690239763e-07,
      "loss": 1.9753,
      "num_input_tokens_seen": 154872600,
      "step": 152100
    },
    {
      "epoch": 2.99659388474336,
      "grad_norm": 2.014535665512085,
      "learning_rate": 9.047277381212819e-07,
      "loss": 2.0067,
      "num_input_tokens_seen": 154974128,
      "step": 152200
    },
    {
      "epoch": 2.998562737492863,
      "grad_norm": 1.8836442232131958,
      "learning_rate": 9.046066383980491e-07,
      "loss": 1.9883,
      "num_input_tokens_seen": 155074968,
      "step": 152300
    },
    {
      "epoch": 3.0005315902423657,
      "grad_norm": 1.7077946662902832,
      "learning_rate": 9.044854698748703e-07,
      "loss": 1.9731,
      "num_input_tokens_seen": 155175984,
      "step": 152400
    },
    {
      "epoch": 3.0025004429918685,
      "grad_norm": 1.916077733039856,
      "learning_rate": 9.043642325723488e-07,
      "loss": 1.9718,
      "num_input_tokens_seen": 155277776,
      "step": 152500
    },
    {
      "epoch": 3.0044692957413717,
      "grad_norm": 1.922825813293457,
      "learning_rate": 9.042429265111e-07,
      "loss": 1.9881,
      "num_input_tokens_seen": 155380176,
      "step": 152600
    },
    {
      "epoch": 3.0064381484908744,
      "grad_norm": 1.6958569288253784,
      "learning_rate": 9.041215517117513e-07,
      "loss": 1.9728,
      "num_input_tokens_seen": 155482576,
      "step": 152700
    },
    {
      "epoch": 3.008407001240377,
      "grad_norm": 2.1971304416656494,
      "learning_rate": 9.040001081949412e-07,
      "loss": 1.9748,
      "num_input_tokens_seen": 155584976,
      "step": 152800
    },
    {
      "epoch": 3.01037585398988,
      "grad_norm": 1.9326709508895874,
      "learning_rate": 9.038785959813201e-07,
      "loss": 1.9952,
      "num_input_tokens_seen": 155687376,
      "step": 152900
    },
    {
      "epoch": 3.012344706739383,
      "grad_norm": 1.6472011804580688,
      "learning_rate": 9.037570150915502e-07,
      "loss": 1.9616,
      "num_input_tokens_seen": 155789776,
      "step": 153000
    },
    {
      "epoch": 3.014313559488886,
      "grad_norm": 2.083277940750122,
      "learning_rate": 9.036353655463054e-07,
      "loss": 2.0036,
      "num_input_tokens_seen": 155892176,
      "step": 153100
    },
    {
      "epoch": 3.0162824122383887,
      "grad_norm": 1.840825080871582,
      "learning_rate": 9.035136473662712e-07,
      "loss": 1.9263,
      "num_input_tokens_seen": 155994096,
      "step": 153200
    },
    {
      "epoch": 3.0182512649878914,
      "grad_norm": 2.437803030014038,
      "learning_rate": 9.033918605721445e-07,
      "loss": 1.9592,
      "num_input_tokens_seen": 156096496,
      "step": 153300
    },
    {
      "epoch": 3.0202201177373946,
      "grad_norm": 1.7386488914489746,
      "learning_rate": 9.032700051846345e-07,
      "loss": 1.9663,
      "num_input_tokens_seen": 156198120,
      "step": 153400
    },
    {
      "epoch": 3.0221889704868974,
      "grad_norm": 2.4257898330688477,
      "learning_rate": 9.031480812244612e-07,
      "loss": 2.0058,
      "num_input_tokens_seen": 156298320,
      "step": 153500
    },
    {
      "epoch": 3.0241578232364,
      "grad_norm": 1.9683313369750977,
      "learning_rate": 9.030260887123571e-07,
      "loss": 1.943,
      "num_input_tokens_seen": 156399448,
      "step": 153600
    },
    {
      "epoch": 3.026126675985903,
      "grad_norm": 2.007336139678955,
      "learning_rate": 9.029040276690658e-07,
      "loss": 2.0275,
      "num_input_tokens_seen": 156500760,
      "step": 153700
    },
    {
      "epoch": 3.0280955287354057,
      "grad_norm": 1.9971493482589722,
      "learning_rate": 9.02781898115343e-07,
      "loss": 1.95,
      "num_input_tokens_seen": 156603160,
      "step": 153800
    },
    {
      "epoch": 3.030064381484909,
      "grad_norm": 1.8072985410690308,
      "learning_rate": 9.026597000719555e-07,
      "loss": 1.9866,
      "num_input_tokens_seen": 156705560,
      "step": 153900
    },
    {
      "epoch": 3.0320332342344116,
      "grad_norm": 2.3550922870635986,
      "learning_rate": 9.025374335596824e-07,
      "loss": 1.9878,
      "num_input_tokens_seen": 156807512,
      "step": 154000
    },
    {
      "epoch": 3.0340020869839144,
      "grad_norm": 1.8513967990875244,
      "learning_rate": 9.024150985993138e-07,
      "loss": 2.004,
      "num_input_tokens_seen": 156907672,
      "step": 154100
    },
    {
      "epoch": 3.035970939733417,
      "grad_norm": 1.9500596523284912,
      "learning_rate": 9.022926952116516e-07,
      "loss": 1.9728,
      "num_input_tokens_seen": 157010072,
      "step": 154200
    },
    {
      "epoch": 3.0379397924829203,
      "grad_norm": 1.7841788530349731,
      "learning_rate": 9.0217022341751e-07,
      "loss": 1.9871,
      "num_input_tokens_seen": 157112472,
      "step": 154300
    },
    {
      "epoch": 3.039908645232423,
      "grad_norm": 1.9122768640518188,
      "learning_rate": 9.020476832377137e-07,
      "loss": 1.9695,
      "num_input_tokens_seen": 157214872,
      "step": 154400
    },
    {
      "epoch": 3.041877497981926,
      "grad_norm": 1.908725619316101,
      "learning_rate": 9.019250746931e-07,
      "loss": 1.9509,
      "num_input_tokens_seen": 157316408,
      "step": 154500
    },
    {
      "epoch": 3.0438463507314286,
      "grad_norm": 2.1317882537841797,
      "learning_rate": 9.018023978045173e-07,
      "loss": 1.9725,
      "num_input_tokens_seen": 157418208,
      "step": 154600
    },
    {
      "epoch": 3.045815203480932,
      "grad_norm": 2.0081725120544434,
      "learning_rate": 9.016796525928258e-07,
      "loss": 1.989,
      "num_input_tokens_seen": 157519448,
      "step": 154700
    },
    {
      "epoch": 3.0477840562304346,
      "grad_norm": 1.9159324169158936,
      "learning_rate": 9.015568390788974e-07,
      "loss": 2.0155,
      "num_input_tokens_seen": 157620976,
      "step": 154800
    },
    {
      "epoch": 3.0497529089799373,
      "grad_norm": 3.625356435775757,
      "learning_rate": 9.014339572836153e-07,
      "loss": 2.0194,
      "num_input_tokens_seen": 157722760,
      "step": 154900
    },
    {
      "epoch": 3.05172176172944,
      "grad_norm": 2.53078293800354,
      "learning_rate": 9.013110072278747e-07,
      "loss": 1.9585,
      "num_input_tokens_seen": 157824016,
      "step": 155000
    },
    {
      "epoch": 3.0536906144789433,
      "grad_norm": 2.018456220626831,
      "learning_rate": 9.011879889325821e-07,
      "loss": 1.9959,
      "num_input_tokens_seen": 157925568,
      "step": 155100
    },
    {
      "epoch": 3.055659467228446,
      "grad_norm": 1.8994927406311035,
      "learning_rate": 9.010649024186558e-07,
      "loss": 2.0125,
      "num_input_tokens_seen": 158027968,
      "step": 155200
    },
    {
      "epoch": 3.057628319977949,
      "grad_norm": 1.9639719724655151,
      "learning_rate": 9.009417477070254e-07,
      "loss": 1.9846,
      "num_input_tokens_seen": 158129152,
      "step": 155300
    },
    {
      "epoch": 3.0595971727274516,
      "grad_norm": 2.702502727508545,
      "learning_rate": 9.008185248186328e-07,
      "loss": 1.9872,
      "num_input_tokens_seen": 158231552,
      "step": 155400
    },
    {
      "epoch": 3.0615660254769548,
      "grad_norm": 2.3195114135742188,
      "learning_rate": 9.006952337744304e-07,
      "loss": 1.9969,
      "num_input_tokens_seen": 158333304,
      "step": 155500
    },
    {
      "epoch": 3.0635348782264575,
      "grad_norm": 1.6101160049438477,
      "learning_rate": 9.005718745953833e-07,
      "loss": 1.9884,
      "num_input_tokens_seen": 158434848,
      "step": 155600
    },
    {
      "epoch": 3.0655037309759603,
      "grad_norm": 1.9197417497634888,
      "learning_rate": 9.004484473024675e-07,
      "loss": 1.9332,
      "num_input_tokens_seen": 158537248,
      "step": 155700
    },
    {
      "epoch": 3.067472583725463,
      "grad_norm": 1.8220990896224976,
      "learning_rate": 9.003249519166707e-07,
      "loss": 1.9378,
      "num_input_tokens_seen": 158639648,
      "step": 155800
    },
    {
      "epoch": 3.069441436474966,
      "grad_norm": 2.0467464923858643,
      "learning_rate": 9.002013884589924e-07,
      "loss": 2.0053,
      "num_input_tokens_seen": 158741496,
      "step": 155900
    },
    {
      "epoch": 3.071410289224469,
      "grad_norm": 1.7307876348495483,
      "learning_rate": 9.000777569504435e-07,
      "loss": 1.9448,
      "num_input_tokens_seen": 158843424,
      "step": 156000
    },
    {
      "epoch": 3.0733791419739718,
      "grad_norm": 1.9886528253555298,
      "learning_rate": 8.999540574120462e-07,
      "loss": 2.0004,
      "num_input_tokens_seen": 158945048,
      "step": 156100
    },
    {
      "epoch": 3.0753479947234745,
      "grad_norm": 1.8458962440490723,
      "learning_rate": 8.99830289864835e-07,
      "loss": 2.0059,
      "num_input_tokens_seen": 159046584,
      "step": 156200
    },
    {
      "epoch": 3.0773168474729773,
      "grad_norm": 2.0900425910949707,
      "learning_rate": 8.997064543298553e-07,
      "loss": 1.9767,
      "num_input_tokens_seen": 159148504,
      "step": 156300
    },
    {
      "epoch": 3.0792857002224805,
      "grad_norm": 2.185340404510498,
      "learning_rate": 8.995825508281643e-07,
      "loss": 1.9557,
      "num_input_tokens_seen": 159249696,
      "step": 156400
    },
    {
      "epoch": 3.0812545529719833,
      "grad_norm": 1.8409805297851562,
      "learning_rate": 8.994585793808308e-07,
      "loss": 1.9945,
      "num_input_tokens_seen": 159352096,
      "step": 156500
    },
    {
      "epoch": 3.083223405721486,
      "grad_norm": 1.7478899955749512,
      "learning_rate": 8.993345400089351e-07,
      "loss": 1.9673,
      "num_input_tokens_seen": 159454496,
      "step": 156600
    },
    {
      "epoch": 3.0851922584709888,
      "grad_norm": 1.7332470417022705,
      "learning_rate": 8.992104327335689e-07,
      "loss": 1.9692,
      "num_input_tokens_seen": 159556896,
      "step": 156700
    },
    {
      "epoch": 3.087161111220492,
      "grad_norm": 1.8158246278762817,
      "learning_rate": 8.990862575758358e-07,
      "loss": 2.0218,
      "num_input_tokens_seen": 159659296,
      "step": 156800
    },
    {
      "epoch": 3.0891299639699947,
      "grad_norm": 1.8258429765701294,
      "learning_rate": 8.989620145568506e-07,
      "loss": 1.9707,
      "num_input_tokens_seen": 159761696,
      "step": 156900
    },
    {
      "epoch": 3.0910988167194975,
      "grad_norm": 1.9211344718933105,
      "learning_rate": 8.988377036977398e-07,
      "loss": 2.0424,
      "num_input_tokens_seen": 159864096,
      "step": 157000
    },
    {
      "epoch": 3.0930676694690002,
      "grad_norm": 2.6851234436035156,
      "learning_rate": 8.987133250196414e-07,
      "loss": 1.9635,
      "num_input_tokens_seen": 159966496,
      "step": 157100
    },
    {
      "epoch": 3.0950365222185034,
      "grad_norm": 1.910126805305481,
      "learning_rate": 8.985888785437052e-07,
      "loss": 1.9844,
      "num_input_tokens_seen": 160067936,
      "step": 157200
    },
    {
      "epoch": 3.097005374968006,
      "grad_norm": 1.9952406883239746,
      "learning_rate": 8.984643642910918e-07,
      "loss": 1.9908,
      "num_input_tokens_seen": 160170336,
      "step": 157300
    },
    {
      "epoch": 3.098974227717509,
      "grad_norm": 1.881360411643982,
      "learning_rate": 8.983397822829744e-07,
      "loss": 1.9762,
      "num_input_tokens_seen": 160272736,
      "step": 157400
    },
    {
      "epoch": 3.1009430804670117,
      "grad_norm": 1.9617085456848145,
      "learning_rate": 8.982151325405365e-07,
      "loss": 1.984,
      "num_input_tokens_seen": 160375136,
      "step": 157500
    },
    {
      "epoch": 3.102911933216515,
      "grad_norm": 1.795054316520691,
      "learning_rate": 8.98090415084974e-07,
      "loss": 1.937,
      "num_input_tokens_seen": 160476704,
      "step": 157600
    },
    {
      "epoch": 3.1048807859660177,
      "grad_norm": 1.9739768505096436,
      "learning_rate": 8.979656299374941e-07,
      "loss": 2.0462,
      "num_input_tokens_seen": 160579104,
      "step": 157700
    },
    {
      "epoch": 3.1068496387155204,
      "grad_norm": 1.8118436336517334,
      "learning_rate": 8.978407771193155e-07,
      "loss": 1.9849,
      "num_input_tokens_seen": 160680800,
      "step": 157800
    },
    {
      "epoch": 3.108818491465023,
      "grad_norm": 2.1058084964752197,
      "learning_rate": 8.977158566516683e-07,
      "loss": 2.0085,
      "num_input_tokens_seen": 160782568,
      "step": 157900
    },
    {
      "epoch": 3.1107873442145264,
      "grad_norm": 1.942762017250061,
      "learning_rate": 8.97590868555794e-07,
      "loss": 2.0127,
      "num_input_tokens_seen": 160884968,
      "step": 158000
    },
    {
      "epoch": 3.112756196964029,
      "grad_norm": 2.611391305923462,
      "learning_rate": 8.974658128529461e-07,
      "loss": 1.9945,
      "num_input_tokens_seen": 160987368,
      "step": 158100
    },
    {
      "epoch": 3.114725049713532,
      "grad_norm": 1.9599195718765259,
      "learning_rate": 8.973406895643891e-07,
      "loss": 1.9623,
      "num_input_tokens_seen": 161089768,
      "step": 158200
    },
    {
      "epoch": 3.1166939024630347,
      "grad_norm": 1.860360026359558,
      "learning_rate": 8.972154987113991e-07,
      "loss": 1.9532,
      "num_input_tokens_seen": 161191304,
      "step": 158300
    },
    {
      "epoch": 3.1186627552125374,
      "grad_norm": 1.8410513401031494,
      "learning_rate": 8.970902403152638e-07,
      "loss": 1.9718,
      "num_input_tokens_seen": 161292304,
      "step": 158400
    },
    {
      "epoch": 3.1206316079620406,
      "grad_norm": 2.022204637527466,
      "learning_rate": 8.969649143972824e-07,
      "loss": 2.0167,
      "num_input_tokens_seen": 161394160,
      "step": 158500
    },
    {
      "epoch": 3.1226004607115434,
      "grad_norm": 1.9852044582366943,
      "learning_rate": 8.968395209787656e-07,
      "loss": 2.0541,
      "num_input_tokens_seen": 161496560,
      "step": 158600
    },
    {
      "epoch": 3.124569313461046,
      "grad_norm": 41.12730407714844,
      "learning_rate": 8.967140600810351e-07,
      "loss": 2.0817,
      "num_input_tokens_seen": 161598008,
      "step": 158700
    },
    {
      "epoch": 3.126538166210549,
      "grad_norm": 2.2703306674957275,
      "learning_rate": 8.965885317254249e-07,
      "loss": 2.0078,
      "num_input_tokens_seen": 161699800,
      "step": 158800
    },
    {
      "epoch": 3.128507018960052,
      "grad_norm": 2.369110345840454,
      "learning_rate": 8.964629359332797e-07,
      "loss": 1.9669,
      "num_input_tokens_seen": 161801704,
      "step": 158900
    },
    {
      "epoch": 3.130475871709555,
      "grad_norm": 2.0651791095733643,
      "learning_rate": 8.963372727259564e-07,
      "loss": 1.9613,
      "num_input_tokens_seen": 161903336,
      "step": 159000
    },
    {
      "epoch": 3.1324447244590576,
      "grad_norm": 1.8558191061019897,
      "learning_rate": 8.962115421248225e-07,
      "loss": 1.9656,
      "num_input_tokens_seen": 162005736,
      "step": 159100
    },
    {
      "epoch": 3.1344135772085604,
      "grad_norm": 2.227570056915283,
      "learning_rate": 8.960857441512578e-07,
      "loss": 1.9927,
      "num_input_tokens_seen": 162106912,
      "step": 159200
    },
    {
      "epoch": 3.1363824299580636,
      "grad_norm": 1.9104275703430176,
      "learning_rate": 8.95959878826653e-07,
      "loss": 2.0015,
      "num_input_tokens_seen": 162208520,
      "step": 159300
    },
    {
      "epoch": 3.1383512827075664,
      "grad_norm": 2.1558897495269775,
      "learning_rate": 8.958339461724104e-07,
      "loss": 1.9662,
      "num_input_tokens_seen": 162310920,
      "step": 159400
    },
    {
      "epoch": 3.140320135457069,
      "grad_norm": 1.891343593597412,
      "learning_rate": 8.957079462099442e-07,
      "loss": 1.9294,
      "num_input_tokens_seen": 162412504,
      "step": 159500
    },
    {
      "epoch": 3.142288988206572,
      "grad_norm": 1.9379512071609497,
      "learning_rate": 8.95581878960679e-07,
      "loss": 1.9676,
      "num_input_tokens_seen": 162514304,
      "step": 159600
    },
    {
      "epoch": 3.144257840956075,
      "grad_norm": 2.2635998725891113,
      "learning_rate": 8.954557444460518e-07,
      "loss": 2.025,
      "num_input_tokens_seen": 162615960,
      "step": 159700
    },
    {
      "epoch": 3.146226693705578,
      "grad_norm": 1.8451220989227295,
      "learning_rate": 8.953295426875107e-07,
      "loss": 1.9757,
      "num_input_tokens_seen": 162717824,
      "step": 159800
    },
    {
      "epoch": 3.1481955464550806,
      "grad_norm": 1.8659178018569946,
      "learning_rate": 8.952032737065153e-07,
      "loss": 1.9565,
      "num_input_tokens_seen": 162819704,
      "step": 159900
    },
    {
      "epoch": 3.1501643992045834,
      "grad_norm": 1.7440221309661865,
      "learning_rate": 8.950769375245364e-07,
      "loss": 1.9858,
      "num_input_tokens_seen": 162922104,
      "step": 160000
    },
    {
      "epoch": 3.1521332519540866,
      "grad_norm": 2.0307304859161377,
      "learning_rate": 8.949505341630566e-07,
      "loss": 2.0013,
      "num_input_tokens_seen": 163022320,
      "step": 160100
    },
    {
      "epoch": 3.1541021047035893,
      "grad_norm": 1.9746315479278564,
      "learning_rate": 8.948240636435695e-07,
      "loss": 1.9869,
      "num_input_tokens_seen": 163123584,
      "step": 160200
    },
    {
      "epoch": 3.156070957453092,
      "grad_norm": 1.9546113014221191,
      "learning_rate": 8.946975259875806e-07,
      "loss": 1.9716,
      "num_input_tokens_seen": 163225344,
      "step": 160300
    },
    {
      "epoch": 3.158039810202595,
      "grad_norm": 2.3259716033935547,
      "learning_rate": 8.945709212166066e-07,
      "loss": 1.9689,
      "num_input_tokens_seen": 163327008,
      "step": 160400
    },
    {
      "epoch": 3.160008662952098,
      "grad_norm": 1.8341962099075317,
      "learning_rate": 8.944442493521751e-07,
      "loss": 1.9864,
      "num_input_tokens_seen": 163429408,
      "step": 160500
    },
    {
      "epoch": 3.161977515701601,
      "grad_norm": 1.8844000101089478,
      "learning_rate": 8.94317510415826e-07,
      "loss": 2.0079,
      "num_input_tokens_seen": 163531808,
      "step": 160600
    },
    {
      "epoch": 3.1639463684511036,
      "grad_norm": 2.1464009284973145,
      "learning_rate": 8.941907044291102e-07,
      "loss": 2.0031,
      "num_input_tokens_seen": 163634208,
      "step": 160700
    },
    {
      "epoch": 3.1659152212006063,
      "grad_norm": 1.9537999629974365,
      "learning_rate": 8.940638314135898e-07,
      "loss": 2.0465,
      "num_input_tokens_seen": 163735848,
      "step": 160800
    },
    {
      "epoch": 3.167884073950109,
      "grad_norm": 1.8776942491531372,
      "learning_rate": 8.939368913908385e-07,
      "loss": 2.007,
      "num_input_tokens_seen": 163837680,
      "step": 160900
    },
    {
      "epoch": 3.1698529266996123,
      "grad_norm": 1.7506014108657837,
      "learning_rate": 8.938098843824415e-07,
      "loss": 1.9985,
      "num_input_tokens_seen": 163939272,
      "step": 161000
    },
    {
      "epoch": 3.171821779449115,
      "grad_norm": 2.0206222534179688,
      "learning_rate": 8.936828104099952e-07,
      "loss": 1.9931,
      "num_input_tokens_seen": 164040760,
      "step": 161100
    },
    {
      "epoch": 3.173790632198618,
      "grad_norm": 1.89096999168396,
      "learning_rate": 8.935556694951074e-07,
      "loss": 2.0344,
      "num_input_tokens_seen": 164142384,
      "step": 161200
    },
    {
      "epoch": 3.1757594849481205,
      "grad_norm": 1.959899663925171,
      "learning_rate": 8.934284616593975e-07,
      "loss": 1.9667,
      "num_input_tokens_seen": 164244784,
      "step": 161300
    },
    {
      "epoch": 3.1777283376976238,
      "grad_norm": 1.9966684579849243,
      "learning_rate": 8.933011869244959e-07,
      "loss": 1.9644,
      "num_input_tokens_seen": 164346480,
      "step": 161400
    },
    {
      "epoch": 3.1796971904471265,
      "grad_norm": 2.167038917541504,
      "learning_rate": 8.931738453120448e-07,
      "loss": 1.9888,
      "num_input_tokens_seen": 164448264,
      "step": 161500
    },
    {
      "epoch": 3.1816660431966293,
      "grad_norm": 1.9101742506027222,
      "learning_rate": 8.930464368436974e-07,
      "loss": 2.0127,
      "num_input_tokens_seen": 164550664,
      "step": 161600
    },
    {
      "epoch": 3.183634895946132,
      "grad_norm": 1.678812026977539,
      "learning_rate": 8.929189615411185e-07,
      "loss": 1.9547,
      "num_input_tokens_seen": 164652440,
      "step": 161700
    },
    {
      "epoch": 3.1856037486956352,
      "grad_norm": 1.9156702756881714,
      "learning_rate": 8.927914194259845e-07,
      "loss": 1.9994,
      "num_input_tokens_seen": 164754008,
      "step": 161800
    },
    {
      "epoch": 3.187572601445138,
      "grad_norm": 2.3595893383026123,
      "learning_rate": 8.926638105199824e-07,
      "loss": 1.9736,
      "num_input_tokens_seen": 164854520,
      "step": 161900
    },
    {
      "epoch": 3.1895414541946407,
      "grad_norm": 2.0386016368865967,
      "learning_rate": 8.925361348448112e-07,
      "loss": 1.9764,
      "num_input_tokens_seen": 164956552,
      "step": 162000
    },
    {
      "epoch": 3.1915103069441435,
      "grad_norm": 2.2614169120788574,
      "learning_rate": 8.92408392422181e-07,
      "loss": 1.9695,
      "num_input_tokens_seen": 165058504,
      "step": 162100
    },
    {
      "epoch": 3.1934791596936467,
      "grad_norm": 1.8114501237869263,
      "learning_rate": 8.922805832738133e-07,
      "loss": 1.9296,
      "num_input_tokens_seen": 165160904,
      "step": 162200
    },
    {
      "epoch": 3.1954480124431495,
      "grad_norm": 2.0567026138305664,
      "learning_rate": 8.921527074214414e-07,
      "loss": 1.9708,
      "num_input_tokens_seen": 165262840,
      "step": 162300
    },
    {
      "epoch": 3.1974168651926522,
      "grad_norm": 1.8833788633346558,
      "learning_rate": 8.92024764886809e-07,
      "loss": 1.9825,
      "num_input_tokens_seen": 165364128,
      "step": 162400
    },
    {
      "epoch": 3.199385717942155,
      "grad_norm": 2.330120801925659,
      "learning_rate": 8.918967556916719e-07,
      "loss": 1.9494,
      "num_input_tokens_seen": 165465896,
      "step": 162500
    },
    {
      "epoch": 3.2013545706916577,
      "grad_norm": 2.012997627258301,
      "learning_rate": 8.917686798577968e-07,
      "loss": 1.9949,
      "num_input_tokens_seen": 165568296,
      "step": 162600
    },
    {
      "epoch": 3.203323423441161,
      "grad_norm": 1.964740514755249,
      "learning_rate": 8.91640537406962e-07,
      "loss": 1.9576,
      "num_input_tokens_seen": 165669936,
      "step": 162700
    },
    {
      "epoch": 3.2052922761906637,
      "grad_norm": 1.7610605955123901,
      "learning_rate": 8.915123283609572e-07,
      "loss": 2.018,
      "num_input_tokens_seen": 165771440,
      "step": 162800
    },
    {
      "epoch": 3.2072611289401665,
      "grad_norm": 2.0105230808258057,
      "learning_rate": 8.91384052741583e-07,
      "loss": 1.9705,
      "num_input_tokens_seen": 165873840,
      "step": 162900
    },
    {
      "epoch": 3.2092299816896697,
      "grad_norm": 1.9277650117874146,
      "learning_rate": 8.912557105706517e-07,
      "loss": 1.9828,
      "num_input_tokens_seen": 165975416,
      "step": 163000
    },
    {
      "epoch": 3.2111988344391724,
      "grad_norm": 1.7110425233840942,
      "learning_rate": 8.911273018699869e-07,
      "loss": 1.999,
      "num_input_tokens_seen": 166075720,
      "step": 163100
    },
    {
      "epoch": 3.213167687188675,
      "grad_norm": 1.8858184814453125,
      "learning_rate": 8.909988266614233e-07,
      "loss": 1.9918,
      "num_input_tokens_seen": 166176728,
      "step": 163200
    },
    {
      "epoch": 3.215136539938178,
      "grad_norm": 2.62264084815979,
      "learning_rate": 8.90870284966807e-07,
      "loss": 2.0197,
      "num_input_tokens_seen": 166279128,
      "step": 163300
    },
    {
      "epoch": 3.2171053926876807,
      "grad_norm": 1.7170372009277344,
      "learning_rate": 8.907416768079956e-07,
      "loss": 1.9746,
      "num_input_tokens_seen": 166379728,
      "step": 163400
    },
    {
      "epoch": 3.219074245437184,
      "grad_norm": 1.7519105672836304,
      "learning_rate": 8.906130022068577e-07,
      "loss": 2.0108,
      "num_input_tokens_seen": 166482128,
      "step": 163500
    },
    {
      "epoch": 3.2210430981866867,
      "grad_norm": 1.9250478744506836,
      "learning_rate": 8.904842611852734e-07,
      "loss": 2.0074,
      "num_input_tokens_seen": 166583688,
      "step": 163600
    },
    {
      "epoch": 3.2230119509361894,
      "grad_norm": 2.1288650035858154,
      "learning_rate": 8.903554537651339e-07,
      "loss": 1.9904,
      "num_input_tokens_seen": 166686088,
      "step": 163700
    },
    {
      "epoch": 3.224980803685692,
      "grad_norm": 2.430946111679077,
      "learning_rate": 8.90226579968342e-07,
      "loss": 1.951,
      "num_input_tokens_seen": 166788488,
      "step": 163800
    },
    {
      "epoch": 3.2269496564351954,
      "grad_norm": 2.035384178161621,
      "learning_rate": 8.900976398168113e-07,
      "loss": 1.9685,
      "num_input_tokens_seen": 166889480,
      "step": 163900
    },
    {
      "epoch": 3.228918509184698,
      "grad_norm": 1.9620369672775269,
      "learning_rate": 8.899686333324674e-07,
      "loss": 2.015,
      "num_input_tokens_seen": 166991232,
      "step": 164000
    },
    {
      "epoch": 3.230887361934201,
      "grad_norm": 1.7356258630752563,
      "learning_rate": 8.898395605372464e-07,
      "loss": 2.0182,
      "num_input_tokens_seen": 167092816,
      "step": 164100
    },
    {
      "epoch": 3.2328562146837037,
      "grad_norm": 1.9568620920181274,
      "learning_rate": 8.897104214530962e-07,
      "loss": 1.9717,
      "num_input_tokens_seen": 167193408,
      "step": 164200
    },
    {
      "epoch": 3.234825067433207,
      "grad_norm": 2.3432846069335938,
      "learning_rate": 8.895812161019759e-07,
      "loss": 2.0157,
      "num_input_tokens_seen": 167295312,
      "step": 164300
    },
    {
      "epoch": 3.2367939201827096,
      "grad_norm": 1.7052772045135498,
      "learning_rate": 8.894519445058556e-07,
      "loss": 1.922,
      "num_input_tokens_seen": 167397088,
      "step": 164400
    },
    {
      "epoch": 3.2387627729322124,
      "grad_norm": 1.9668554067611694,
      "learning_rate": 8.89322606686717e-07,
      "loss": 1.9658,
      "num_input_tokens_seen": 167498976,
      "step": 164500
    },
    {
      "epoch": 3.240731625681715,
      "grad_norm": 1.9469531774520874,
      "learning_rate": 8.891932026665528e-07,
      "loss": 1.9703,
      "num_input_tokens_seen": 167601376,
      "step": 164600
    },
    {
      "epoch": 3.2427004784312183,
      "grad_norm": 1.912940263748169,
      "learning_rate": 8.890637324673671e-07,
      "loss": 1.9732,
      "num_input_tokens_seen": 167703768,
      "step": 164700
    },
    {
      "epoch": 3.244669331180721,
      "grad_norm": 1.7971786260604858,
      "learning_rate": 8.889341961111751e-07,
      "loss": 2.0166,
      "num_input_tokens_seen": 167806168,
      "step": 164800
    },
    {
      "epoch": 3.246638183930224,
      "grad_norm": 1.7801483869552612,
      "learning_rate": 8.888045936200036e-07,
      "loss": 1.9954,
      "num_input_tokens_seen": 167908568,
      "step": 164900
    },
    {
      "epoch": 3.2486070366797266,
      "grad_norm": 2.1926114559173584,
      "learning_rate": 8.886749250158903e-07,
      "loss": 2.0289,
      "num_input_tokens_seen": 168010376,
      "step": 165000
    },
    {
      "epoch": 3.2505758894292294,
      "grad_norm": 1.8983944654464722,
      "learning_rate": 8.885451903208841e-07,
      "loss": 1.9615,
      "num_input_tokens_seen": 168111576,
      "step": 165100
    },
    {
      "epoch": 3.2525447421787326,
      "grad_norm": 2.0520403385162354,
      "learning_rate": 8.884153895570456e-07,
      "loss": 1.9826,
      "num_input_tokens_seen": 168213104,
      "step": 165200
    },
    {
      "epoch": 3.2545135949282353,
      "grad_norm": 1.9233990907669067,
      "learning_rate": 8.882855227464461e-07,
      "loss": 1.9622,
      "num_input_tokens_seen": 168314768,
      "step": 165300
    },
    {
      "epoch": 3.256482447677738,
      "grad_norm": 1.8786237239837646,
      "learning_rate": 8.881555899111683e-07,
      "loss": 1.9824,
      "num_input_tokens_seen": 168415584,
      "step": 165400
    },
    {
      "epoch": 3.2584513004272413,
      "grad_norm": 1.89914870262146,
      "learning_rate": 8.880255910733063e-07,
      "loss": 1.9423,
      "num_input_tokens_seen": 168517064,
      "step": 165500
    },
    {
      "epoch": 3.260420153176744,
      "grad_norm": 1.8395462036132812,
      "learning_rate": 8.878955262549655e-07,
      "loss": 1.9966,
      "num_input_tokens_seen": 168618720,
      "step": 165600
    },
    {
      "epoch": 3.262389005926247,
      "grad_norm": 2.196533679962158,
      "learning_rate": 8.877653954782619e-07,
      "loss": 1.9929,
      "num_input_tokens_seen": 168721120,
      "step": 165700
    },
    {
      "epoch": 3.2643578586757496,
      "grad_norm": 1.9165867567062378,
      "learning_rate": 8.876351987653235e-07,
      "loss": 1.9942,
      "num_input_tokens_seen": 168822248,
      "step": 165800
    },
    {
      "epoch": 3.2663267114252523,
      "grad_norm": 1.9275126457214355,
      "learning_rate": 8.875049361382889e-07,
      "loss": 1.9777,
      "num_input_tokens_seen": 168923696,
      "step": 165900
    },
    {
      "epoch": 3.2682955641747555,
      "grad_norm": 1.802634596824646,
      "learning_rate": 8.873746076193084e-07,
      "loss": 1.9761,
      "num_input_tokens_seen": 169024496,
      "step": 166000
    },
    {
      "epoch": 3.2702644169242583,
      "grad_norm": 2.0053305625915527,
      "learning_rate": 8.872442132305431e-07,
      "loss": 2.0066,
      "num_input_tokens_seen": 169126304,
      "step": 166100
    },
    {
      "epoch": 3.272233269673761,
      "grad_norm": 1.9875487089157104,
      "learning_rate": 8.871137529941655e-07,
      "loss": 1.967,
      "num_input_tokens_seen": 169226296,
      "step": 166200
    },
    {
      "epoch": 3.274202122423264,
      "grad_norm": 1.7244879007339478,
      "learning_rate": 8.869832269323592e-07,
      "loss": 1.9969,
      "num_input_tokens_seen": 169327776,
      "step": 166300
    },
    {
      "epoch": 3.276170975172767,
      "grad_norm": 1.9510544538497925,
      "learning_rate": 8.868526350673191e-07,
      "loss": 1.9694,
      "num_input_tokens_seen": 169430176,
      "step": 166400
    },
    {
      "epoch": 3.2781398279222698,
      "grad_norm": 1.7385942935943604,
      "learning_rate": 8.867219774212514e-07,
      "loss": 1.959,
      "num_input_tokens_seen": 169532488,
      "step": 166500
    },
    {
      "epoch": 3.2801086806717725,
      "grad_norm": 1.9729104042053223,
      "learning_rate": 8.865912540163731e-07,
      "loss": 2.009,
      "num_input_tokens_seen": 169634888,
      "step": 166600
    },
    {
      "epoch": 3.2820775334212753,
      "grad_norm": 2.0269644260406494,
      "learning_rate": 8.864604648749129e-07,
      "loss": 1.9963,
      "num_input_tokens_seen": 169737288,
      "step": 166700
    },
    {
      "epoch": 3.284046386170778,
      "grad_norm": 2.124392032623291,
      "learning_rate": 8.863296100191102e-07,
      "loss": 1.9577,
      "num_input_tokens_seen": 169839688,
      "step": 166800
    },
    {
      "epoch": 3.2860152389202812,
      "grad_norm": 2.3548800945281982,
      "learning_rate": 8.861986894712155e-07,
      "loss": 2.025,
      "num_input_tokens_seen": 169942088,
      "step": 166900
    },
    {
      "epoch": 3.287984091669784,
      "grad_norm": 1.7009906768798828,
      "learning_rate": 8.860677032534913e-07,
      "loss": 1.9597,
      "num_input_tokens_seen": 170044488,
      "step": 167000
    },
    {
      "epoch": 3.2899529444192868,
      "grad_norm": 2.1477997303009033,
      "learning_rate": 8.859366513882103e-07,
      "loss": 1.9842,
      "num_input_tokens_seen": 170146368,
      "step": 167100
    },
    {
      "epoch": 3.29192179716879,
      "grad_norm": 2.09075665473938,
      "learning_rate": 8.858055338976566e-07,
      "loss": 1.9543,
      "num_input_tokens_seen": 170247912,
      "step": 167200
    },
    {
      "epoch": 3.2938906499182927,
      "grad_norm": 1.9375501871109009,
      "learning_rate": 8.856743508041262e-07,
      "loss": 2.0287,
      "num_input_tokens_seen": 170349456,
      "step": 167300
    },
    {
      "epoch": 3.2958595026677955,
      "grad_norm": 2.09952974319458,
      "learning_rate": 8.855431021299253e-07,
      "loss": 1.9911,
      "num_input_tokens_seen": 170451856,
      "step": 167400
    },
    {
      "epoch": 3.2978283554172982,
      "grad_norm": 1.9498170614242554,
      "learning_rate": 8.854117878973718e-07,
      "loss": 2.0775,
      "num_input_tokens_seen": 170553328,
      "step": 167500
    },
    {
      "epoch": 3.299797208166801,
      "grad_norm": 2.5795865058898926,
      "learning_rate": 8.852804081287943e-07,
      "loss": 2.0236,
      "num_input_tokens_seen": 170655680,
      "step": 167600
    },
    {
      "epoch": 3.301766060916304,
      "grad_norm": 1.757726788520813,
      "learning_rate": 8.851489628465331e-07,
      "loss": 2.0025,
      "num_input_tokens_seen": 170757408,
      "step": 167700
    },
    {
      "epoch": 3.303734913665807,
      "grad_norm": 1.8712294101715088,
      "learning_rate": 8.850174520729395e-07,
      "loss": 2.0344,
      "num_input_tokens_seen": 170858144,
      "step": 167800
    },
    {
      "epoch": 3.3057037664153097,
      "grad_norm": 2.155019521713257,
      "learning_rate": 8.848858758303752e-07,
      "loss": 1.9832,
      "num_input_tokens_seen": 170960232,
      "step": 167900
    },
    {
      "epoch": 3.3076726191648125,
      "grad_norm": 1.9364187717437744,
      "learning_rate": 8.84754234141214e-07,
      "loss": 1.9534,
      "num_input_tokens_seen": 171062632,
      "step": 168000
    },
    {
      "epoch": 3.3096414719143157,
      "grad_norm": 1.8489073514938354,
      "learning_rate": 8.846225270278406e-07,
      "loss": 1.9614,
      "num_input_tokens_seen": 171165032,
      "step": 168100
    },
    {
      "epoch": 3.3116103246638184,
      "grad_norm": 1.943538784980774,
      "learning_rate": 8.844907545126506e-07,
      "loss": 1.9851,
      "num_input_tokens_seen": 171267144,
      "step": 168200
    },
    {
      "epoch": 3.313579177413321,
      "grad_norm": 1.8676447868347168,
      "learning_rate": 8.843589166180506e-07,
      "loss": 1.9712,
      "num_input_tokens_seen": 171369544,
      "step": 168300
    },
    {
      "epoch": 3.315548030162824,
      "grad_norm": 1.9416911602020264,
      "learning_rate": 8.842270133664586e-07,
      "loss": 1.9966,
      "num_input_tokens_seen": 171470576,
      "step": 168400
    },
    {
      "epoch": 3.317516882912327,
      "grad_norm": 1.9778763055801392,
      "learning_rate": 8.84095044780304e-07,
      "loss": 1.9937,
      "num_input_tokens_seen": 171572976,
      "step": 168500
    },
    {
      "epoch": 3.31948573566183,
      "grad_norm": 2.098393678665161,
      "learning_rate": 8.839630108820264e-07,
      "loss": 2.0208,
      "num_input_tokens_seen": 171673944,
      "step": 168600
    },
    {
      "epoch": 3.3214545884113327,
      "grad_norm": 2.0689074993133545,
      "learning_rate": 8.838309116940773e-07,
      "loss": 1.9673,
      "num_input_tokens_seen": 171776344,
      "step": 168700
    },
    {
      "epoch": 3.3234234411608354,
      "grad_norm": 2.3261332511901855,
      "learning_rate": 8.83698747238919e-07,
      "loss": 1.9714,
      "num_input_tokens_seen": 171878744,
      "step": 168800
    },
    {
      "epoch": 3.3253922939103386,
      "grad_norm": 1.6794986724853516,
      "learning_rate": 8.835665175390252e-07,
      "loss": 2.0109,
      "num_input_tokens_seen": 171980688,
      "step": 168900
    },
    {
      "epoch": 3.3273611466598414,
      "grad_norm": 1.9332787990570068,
      "learning_rate": 8.8343422261688e-07,
      "loss": 1.9729,
      "num_input_tokens_seen": 172083088,
      "step": 169000
    },
    {
      "epoch": 3.329329999409344,
      "grad_norm": 1.6862245798110962,
      "learning_rate": 8.833018624949795e-07,
      "loss": 2.0067,
      "num_input_tokens_seen": 172185488,
      "step": 169100
    },
    {
      "epoch": 3.331298852158847,
      "grad_norm": 1.9138447046279907,
      "learning_rate": 8.831694371958302e-07,
      "loss": 2.0186,
      "num_input_tokens_seen": 172285832,
      "step": 169200
    },
    {
      "epoch": 3.3332677049083497,
      "grad_norm": 1.9838453531265259,
      "learning_rate": 8.830369467419497e-07,
      "loss": 1.9824,
      "num_input_tokens_seen": 172388232,
      "step": 169300
    },
    {
      "epoch": 3.335236557657853,
      "grad_norm": 2.972688674926758,
      "learning_rate": 8.829043911558671e-07,
      "loss": 2.0135,
      "num_input_tokens_seen": 172489536,
      "step": 169400
    },
    {
      "epoch": 3.3372054104073556,
      "grad_norm": 1.8738408088684082,
      "learning_rate": 8.827717704601225e-07,
      "loss": 1.9755,
      "num_input_tokens_seen": 172590632,
      "step": 169500
    },
    {
      "epoch": 3.3391742631568584,
      "grad_norm": 1.9847636222839355,
      "learning_rate": 8.826390846772668e-07,
      "loss": 1.938,
      "num_input_tokens_seen": 172691736,
      "step": 169600
    },
    {
      "epoch": 3.3411431159063616,
      "grad_norm": 1.9578179121017456,
      "learning_rate": 8.825063338298619e-07,
      "loss": 1.9784,
      "num_input_tokens_seen": 172793368,
      "step": 169700
    },
    {
      "epoch": 3.3431119686558644,
      "grad_norm": 1.7819938659667969,
      "learning_rate": 8.823735179404812e-07,
      "loss": 1.9772,
      "num_input_tokens_seen": 172894400,
      "step": 169800
    },
    {
      "epoch": 3.345080821405367,
      "grad_norm": 2.017322301864624,
      "learning_rate": 8.82240637031709e-07,
      "loss": 1.9745,
      "num_input_tokens_seen": 172996800,
      "step": 169900
    },
    {
      "epoch": 3.34704967415487,
      "grad_norm": 2.1448822021484375,
      "learning_rate": 8.821076911261403e-07,
      "loss": 1.9564,
      "num_input_tokens_seen": 173099200,
      "step": 170000
    },
    {
      "epoch": 3.3490185269043726,
      "grad_norm": 1.874789834022522,
      "learning_rate": 8.819746802463817e-07,
      "loss": 1.9886,
      "num_input_tokens_seen": 173200976,
      "step": 170100
    },
    {
      "epoch": 3.350987379653876,
      "grad_norm": 2.202587604522705,
      "learning_rate": 8.818416044150503e-07,
      "loss": 1.9684,
      "num_input_tokens_seen": 173303376,
      "step": 170200
    },
    {
      "epoch": 3.3529562324033786,
      "grad_norm": 2.053558826446533,
      "learning_rate": 8.817084636547747e-07,
      "loss": 1.9877,
      "num_input_tokens_seen": 173404800,
      "step": 170300
    },
    {
      "epoch": 3.3549250851528813,
      "grad_norm": 1.969000220298767,
      "learning_rate": 8.815752579881944e-07,
      "loss": 1.9827,
      "num_input_tokens_seen": 173507200,
      "step": 170400
    },
    {
      "epoch": 3.356893937902384,
      "grad_norm": 1.8853901624679565,
      "learning_rate": 8.814419874379598e-07,
      "loss": 1.9758,
      "num_input_tokens_seen": 173608488,
      "step": 170500
    },
    {
      "epoch": 3.3588627906518873,
      "grad_norm": 5.5333380699157715,
      "learning_rate": 8.813086520267324e-07,
      "loss": 1.9703,
      "num_input_tokens_seen": 173708024,
      "step": 170600
    },
    {
      "epoch": 3.36083164340139,
      "grad_norm": 2.043829917907715,
      "learning_rate": 8.811752517771848e-07,
      "loss": 1.9842,
      "num_input_tokens_seen": 173810064,
      "step": 170700
    },
    {
      "epoch": 3.362800496150893,
      "grad_norm": 1.8781522512435913,
      "learning_rate": 8.810417867120008e-07,
      "loss": 1.9513,
      "num_input_tokens_seen": 173912464,
      "step": 170800
    },
    {
      "epoch": 3.3647693489003956,
      "grad_norm": 3.270911693572998,
      "learning_rate": 8.809082568538748e-07,
      "loss": 2.0031,
      "num_input_tokens_seen": 174014216,
      "step": 170900
    },
    {
      "epoch": 3.366738201649899,
      "grad_norm": 2.1133625507354736,
      "learning_rate": 8.807746622255123e-07,
      "loss": 1.9944,
      "num_input_tokens_seen": 174116616,
      "step": 171000
    },
    {
      "epoch": 3.3687070543994015,
      "grad_norm": 1.9044172763824463,
      "learning_rate": 8.806410028496302e-07,
      "loss": 1.9605,
      "num_input_tokens_seen": 174219008,
      "step": 171100
    },
    {
      "epoch": 3.3706759071489043,
      "grad_norm": 1.9099133014678955,
      "learning_rate": 8.805072787489561e-07,
      "loss": 2.0292,
      "num_input_tokens_seen": 174320560,
      "step": 171200
    },
    {
      "epoch": 3.372644759898407,
      "grad_norm": 2.0432288646698,
      "learning_rate": 8.803734899462285e-07,
      "loss": 2.0272,
      "num_input_tokens_seen": 174421456,
      "step": 171300
    },
    {
      "epoch": 3.3746136126479103,
      "grad_norm": 1.81580650806427,
      "learning_rate": 8.802396364641971e-07,
      "loss": 2.0202,
      "num_input_tokens_seen": 174523856,
      "step": 171400
    },
    {
      "epoch": 3.376582465397413,
      "grad_norm": 2.0548923015594482,
      "learning_rate": 8.801057183256226e-07,
      "loss": 1.9989,
      "num_input_tokens_seen": 174626256,
      "step": 171500
    },
    {
      "epoch": 3.378551318146916,
      "grad_norm": 1.7348370552062988,
      "learning_rate": 8.799717355532767e-07,
      "loss": 1.9893,
      "num_input_tokens_seen": 174726680,
      "step": 171600
    },
    {
      "epoch": 3.3805201708964185,
      "grad_norm": 1.9614585638046265,
      "learning_rate": 8.798376881699421e-07,
      "loss": 1.9935,
      "num_input_tokens_seen": 174827712,
      "step": 171700
    },
    {
      "epoch": 3.3824890236459213,
      "grad_norm": 2.0643503665924072,
      "learning_rate": 8.797035761984123e-07,
      "loss": 1.9866,
      "num_input_tokens_seen": 174930112,
      "step": 171800
    },
    {
      "epoch": 3.3844578763954245,
      "grad_norm": 1.7682868242263794,
      "learning_rate": 8.795693996614919e-07,
      "loss": 1.9844,
      "num_input_tokens_seen": 175031400,
      "step": 171900
    },
    {
      "epoch": 3.3864267291449273,
      "grad_norm": 2.0654399394989014,
      "learning_rate": 8.794351585819965e-07,
      "loss": 2.0195,
      "num_input_tokens_seen": 175133800,
      "step": 172000
    },
    {
      "epoch": 3.38839558189443,
      "grad_norm": 1.854973316192627,
      "learning_rate": 8.793008529827527e-07,
      "loss": 1.9906,
      "num_input_tokens_seen": 175234848,
      "step": 172100
    },
    {
      "epoch": 3.3903644346439332,
      "grad_norm": 1.7924410104751587,
      "learning_rate": 8.79166482886598e-07,
      "loss": 1.9701,
      "num_input_tokens_seen": 175337248,
      "step": 172200
    },
    {
      "epoch": 3.392333287393436,
      "grad_norm": 1.903087854385376,
      "learning_rate": 8.790320483163809e-07,
      "loss": 1.987,
      "num_input_tokens_seen": 175438224,
      "step": 172300
    },
    {
      "epoch": 3.3943021401429387,
      "grad_norm": 1.9773424863815308,
      "learning_rate": 8.78897549294961e-07,
      "loss": 2.0261,
      "num_input_tokens_seen": 175539688,
      "step": 172400
    },
    {
      "epoch": 3.3962709928924415,
      "grad_norm": 2.204087972640991,
      "learning_rate": 8.787629858452085e-07,
      "loss": 2.0154,
      "num_input_tokens_seen": 175640208,
      "step": 172500
    },
    {
      "epoch": 3.3982398456419443,
      "grad_norm": 1.6858693361282349,
      "learning_rate": 8.786283579900049e-07,
      "loss": 1.982,
      "num_input_tokens_seen": 175742608,
      "step": 172600
    },
    {
      "epoch": 3.4002086983914475,
      "grad_norm": 2.1794304847717285,
      "learning_rate": 8.784936657522426e-07,
      "loss": 1.9456,
      "num_input_tokens_seen": 175844480,
      "step": 172700
    },
    {
      "epoch": 3.40217755114095,
      "grad_norm": 1.7944830656051636,
      "learning_rate": 8.783589091548247e-07,
      "loss": 1.9911,
      "num_input_tokens_seen": 175946008,
      "step": 172800
    },
    {
      "epoch": 3.404146403890453,
      "grad_norm": 1.7953822612762451,
      "learning_rate": 8.782240882206655e-07,
      "loss": 1.9776,
      "num_input_tokens_seen": 176048408,
      "step": 172900
    },
    {
      "epoch": 3.4061152566399557,
      "grad_norm": 2.0612521171569824,
      "learning_rate": 8.780892029726904e-07,
      "loss": 2.0142,
      "num_input_tokens_seen": 176149992,
      "step": 173000
    },
    {
      "epoch": 3.408084109389459,
      "grad_norm": 1.9655766487121582,
      "learning_rate": 8.779542534338351e-07,
      "loss": 1.959,
      "num_input_tokens_seen": 176251648,
      "step": 173100
    },
    {
      "epoch": 3.4100529621389617,
      "grad_norm": 1.9699556827545166,
      "learning_rate": 8.77819239627047e-07,
      "loss": 2.0098,
      "num_input_tokens_seen": 176351816,
      "step": 173200
    },
    {
      "epoch": 3.4120218148884645,
      "grad_norm": 1.851715087890625,
      "learning_rate": 8.776841615752837e-07,
      "loss": 1.9588,
      "num_input_tokens_seen": 176453520,
      "step": 173300
    },
    {
      "epoch": 3.413990667637967,
      "grad_norm": 2.1681926250457764,
      "learning_rate": 8.775490193015143e-07,
      "loss": 2.0074,
      "num_input_tokens_seen": 176555920,
      "step": 173400
    },
    {
      "epoch": 3.41595952038747,
      "grad_norm": 1.6440446376800537,
      "learning_rate": 8.774138128287185e-07,
      "loss": 1.9695,
      "num_input_tokens_seen": 176658320,
      "step": 173500
    },
    {
      "epoch": 3.417928373136973,
      "grad_norm": 2.0168633460998535,
      "learning_rate": 8.772785421798872e-07,
      "loss": 1.926,
      "num_input_tokens_seen": 176760720,
      "step": 173600
    },
    {
      "epoch": 3.419897225886476,
      "grad_norm": 1.994805932044983,
      "learning_rate": 8.771432073780219e-07,
      "loss": 1.9823,
      "num_input_tokens_seen": 176862416,
      "step": 173700
    },
    {
      "epoch": 3.4218660786359787,
      "grad_norm": 1.8778208494186401,
      "learning_rate": 8.770078084461351e-07,
      "loss": 1.9637,
      "num_input_tokens_seen": 176964816,
      "step": 173800
    },
    {
      "epoch": 3.423834931385482,
      "grad_norm": 2.268155574798584,
      "learning_rate": 8.768723454072502e-07,
      "loss": 1.9809,
      "num_input_tokens_seen": 177066648,
      "step": 173900
    },
    {
      "epoch": 3.4258037841349847,
      "grad_norm": 1.6476994752883911,
      "learning_rate": 8.767368182844018e-07,
      "loss": 2.001,
      "num_input_tokens_seen": 177169048,
      "step": 174000
    },
    {
      "epoch": 3.4277726368844874,
      "grad_norm": 1.9171645641326904,
      "learning_rate": 8.766012271006351e-07,
      "loss": 1.9709,
      "num_input_tokens_seen": 177270760,
      "step": 174100
    },
    {
      "epoch": 3.42974148963399,
      "grad_norm": 1.7686083316802979,
      "learning_rate": 8.764655718790059e-07,
      "loss": 2.039,
      "num_input_tokens_seen": 177373160,
      "step": 174200
    },
    {
      "epoch": 3.431710342383493,
      "grad_norm": 2.0800998210906982,
      "learning_rate": 8.763298526425816e-07,
      "loss": 2.0096,
      "num_input_tokens_seen": 177475128,
      "step": 174300
    },
    {
      "epoch": 3.433679195132996,
      "grad_norm": 1.8019814491271973,
      "learning_rate": 8.761940694144399e-07,
      "loss": 1.9807,
      "num_input_tokens_seen": 177577528,
      "step": 174400
    },
    {
      "epoch": 3.435648047882499,
      "grad_norm": 2.1396100521087646,
      "learning_rate": 8.760582222176698e-07,
      "loss": 1.9268,
      "num_input_tokens_seen": 177679928,
      "step": 174500
    },
    {
      "epoch": 3.4376169006320016,
      "grad_norm": 2.138073205947876,
      "learning_rate": 8.759223110753708e-07,
      "loss": 1.9931,
      "num_input_tokens_seen": 177782328,
      "step": 174600
    },
    {
      "epoch": 3.4395857533815044,
      "grad_norm": 2.0238165855407715,
      "learning_rate": 8.757863360106534e-07,
      "loss": 2.0324,
      "num_input_tokens_seen": 177884728,
      "step": 174700
    },
    {
      "epoch": 3.4415546061310076,
      "grad_norm": 1.9930150508880615,
      "learning_rate": 8.756502970466391e-07,
      "loss": 2.0009,
      "num_input_tokens_seen": 177987128,
      "step": 174800
    },
    {
      "epoch": 3.4435234588805104,
      "grad_norm": 1.9081519842147827,
      "learning_rate": 8.755141942064601e-07,
      "loss": 1.937,
      "num_input_tokens_seen": 178088752,
      "step": 174900
    },
    {
      "epoch": 3.445492311630013,
      "grad_norm": 1.9462133646011353,
      "learning_rate": 8.753780275132598e-07,
      "loss": 1.9945,
      "num_input_tokens_seen": 178191152,
      "step": 175000
    },
    {
      "epoch": 3.447461164379516,
      "grad_norm": 1.8115335702896118,
      "learning_rate": 8.752417969901918e-07,
      "loss": 2.0213,
      "num_input_tokens_seen": 178292944,
      "step": 175100
    },
    {
      "epoch": 3.449430017129019,
      "grad_norm": 1.675045132637024,
      "learning_rate": 8.751055026604213e-07,
      "loss": 2.0624,
      "num_input_tokens_seen": 178394768,
      "step": 175200
    },
    {
      "epoch": 3.451398869878522,
      "grad_norm": 2.788806915283203,
      "learning_rate": 8.749691445471236e-07,
      "loss": 2.0736,
      "num_input_tokens_seen": 178496768,
      "step": 175300
    },
    {
      "epoch": 3.4533677226280246,
      "grad_norm": 1.9393644332885742,
      "learning_rate": 8.748327226734858e-07,
      "loss": 1.9192,
      "num_input_tokens_seen": 178598128,
      "step": 175400
    },
    {
      "epoch": 3.4553365753775274,
      "grad_norm": 2.0350983142852783,
      "learning_rate": 8.746962370627049e-07,
      "loss": 1.9605,
      "num_input_tokens_seen": 178699840,
      "step": 175500
    },
    {
      "epoch": 3.4573054281270306,
      "grad_norm": 1.8694217205047607,
      "learning_rate": 8.745596877379893e-07,
      "loss": 2.0159,
      "num_input_tokens_seen": 178800144,
      "step": 175600
    },
    {
      "epoch": 3.4592742808765333,
      "grad_norm": 1.9558669328689575,
      "learning_rate": 8.744230747225579e-07,
      "loss": 1.9723,
      "num_input_tokens_seen": 178902544,
      "step": 175700
    },
    {
      "epoch": 3.461243133626036,
      "grad_norm": 1.9053242206573486,
      "learning_rate": 8.742863980396407e-07,
      "loss": 1.9882,
      "num_input_tokens_seen": 179004304,
      "step": 175800
    },
    {
      "epoch": 3.463211986375539,
      "grad_norm": 1.9949630498886108,
      "learning_rate": 8.741496577124783e-07,
      "loss": 2.0504,
      "num_input_tokens_seen": 179106704,
      "step": 175900
    },
    {
      "epoch": 3.4651808391250416,
      "grad_norm": 1.9565402269363403,
      "learning_rate": 8.740128537643225e-07,
      "loss": 2.0043,
      "num_input_tokens_seen": 179209104,
      "step": 176000
    },
    {
      "epoch": 3.467149691874545,
      "grad_norm": 1.7130528688430786,
      "learning_rate": 8.738759862184353e-07,
      "loss": 1.9861,
      "num_input_tokens_seen": 179310104,
      "step": 176100
    },
    {
      "epoch": 3.4691185446240476,
      "grad_norm": 1.7138453722000122,
      "learning_rate": 8.737390550980901e-07,
      "loss": 1.9759,
      "num_input_tokens_seen": 179411736,
      "step": 176200
    },
    {
      "epoch": 3.4710873973735503,
      "grad_norm": 1.835710883140564,
      "learning_rate": 8.73602060426571e-07,
      "loss": 1.9581,
      "num_input_tokens_seen": 179514136,
      "step": 176300
    },
    {
      "epoch": 3.4730562501230535,
      "grad_norm": 5.932350158691406,
      "learning_rate": 8.734650022271725e-07,
      "loss": 1.945,
      "num_input_tokens_seen": 179616536,
      "step": 176400
    },
    {
      "epoch": 3.4750251028725563,
      "grad_norm": 1.84043550491333,
      "learning_rate": 8.733278805232003e-07,
      "loss": 1.9158,
      "num_input_tokens_seen": 179718936,
      "step": 176500
    },
    {
      "epoch": 3.476993955622059,
      "grad_norm": 2.0021297931671143,
      "learning_rate": 8.731906953379709e-07,
      "loss": 1.9646,
      "num_input_tokens_seen": 179821336,
      "step": 176600
    },
    {
      "epoch": 3.478962808371562,
      "grad_norm": 2.124436378479004,
      "learning_rate": 8.730534466948113e-07,
      "loss": 1.9504,
      "num_input_tokens_seen": 179922088,
      "step": 176700
    },
    {
      "epoch": 3.4809316611210646,
      "grad_norm": 2.1138103008270264,
      "learning_rate": 8.729161346170596e-07,
      "loss": 1.9762,
      "num_input_tokens_seen": 180024488,
      "step": 176800
    },
    {
      "epoch": 3.4829005138705678,
      "grad_norm": 2.0620594024658203,
      "learning_rate": 8.727787591280646e-07,
      "loss": 2.0408,
      "num_input_tokens_seen": 180126888,
      "step": 176900
    },
    {
      "epoch": 3.4848693666200705,
      "grad_norm": 1.9161667823791504,
      "learning_rate": 8.726413202511856e-07,
      "loss": 1.9324,
      "num_input_tokens_seen": 180228744,
      "step": 177000
    },
    {
      "epoch": 3.4868382193695733,
      "grad_norm": 1.696247935295105,
      "learning_rate": 8.725038180097932e-07,
      "loss": 1.9831,
      "num_input_tokens_seen": 180330384,
      "step": 177100
    },
    {
      "epoch": 3.488807072119076,
      "grad_norm": 1.9214447736740112,
      "learning_rate": 8.723662524272684e-07,
      "loss": 1.9566,
      "num_input_tokens_seen": 180432784,
      "step": 177200
    },
    {
      "epoch": 3.4907759248685792,
      "grad_norm": 2.0514333248138428,
      "learning_rate": 8.722286235270031e-07,
      "loss": 2.0537,
      "num_input_tokens_seen": 180534240,
      "step": 177300
    },
    {
      "epoch": 3.492744777618082,
      "grad_norm": 2.177111864089966,
      "learning_rate": 8.720909313323999e-07,
      "loss": 2.0004,
      "num_input_tokens_seen": 180636640,
      "step": 177400
    },
    {
      "epoch": 3.4947136303675848,
      "grad_norm": 1.830773949623108,
      "learning_rate": 8.71953175866872e-07,
      "loss": 1.9714,
      "num_input_tokens_seen": 180737616,
      "step": 177500
    },
    {
      "epoch": 3.4966824831170875,
      "grad_norm": 1.8422319889068604,
      "learning_rate": 8.718153571538439e-07,
      "loss": 2.0145,
      "num_input_tokens_seen": 180838904,
      "step": 177600
    },
    {
      "epoch": 3.4986513358665907,
      "grad_norm": 1.8522621393203735,
      "learning_rate": 8.716774752167504e-07,
      "loss": 1.9813,
      "num_input_tokens_seen": 180941304,
      "step": 177700
    },
    {
      "epoch": 3.5006201886160935,
      "grad_norm": 1.8867874145507812,
      "learning_rate": 8.715395300790369e-07,
      "loss": 2.0262,
      "num_input_tokens_seen": 181042648,
      "step": 177800
    },
    {
      "epoch": 3.5025890413655962,
      "grad_norm": 2.1557350158691406,
      "learning_rate": 8.714015217641601e-07,
      "loss": 1.9373,
      "num_input_tokens_seen": 181145048,
      "step": 177900
    },
    {
      "epoch": 3.504557894115099,
      "grad_norm": 2.2549006938934326,
      "learning_rate": 8.712634502955872e-07,
      "loss": 2.0458,
      "num_input_tokens_seen": 181247448,
      "step": 178000
    },
    {
      "epoch": 3.506526746864602,
      "grad_norm": 19.7275333404541,
      "learning_rate": 8.711253156967959e-07,
      "loss": 2.0102,
      "num_input_tokens_seen": 181349704,
      "step": 178100
    },
    {
      "epoch": 3.508495599614105,
      "grad_norm": 1.8544641733169556,
      "learning_rate": 8.709871179912749e-07,
      "loss": 2.0012,
      "num_input_tokens_seen": 181450736,
      "step": 178200
    },
    {
      "epoch": 3.5104644523636077,
      "grad_norm": 2.025705575942993,
      "learning_rate": 8.708488572025234e-07,
      "loss": 1.9897,
      "num_input_tokens_seen": 181550728,
      "step": 178300
    },
    {
      "epoch": 3.5124333051131105,
      "grad_norm": 1.967421531677246,
      "learning_rate": 8.707105333540518e-07,
      "loss": 2.0217,
      "num_input_tokens_seen": 181653128,
      "step": 178400
    },
    {
      "epoch": 3.5144021578626132,
      "grad_norm": 1.8565772771835327,
      "learning_rate": 8.705721464693805e-07,
      "loss": 2.0151,
      "num_input_tokens_seen": 181755528,
      "step": 178500
    },
    {
      "epoch": 3.5163710106121164,
      "grad_norm": 1.5747811794281006,
      "learning_rate": 8.704336965720414e-07,
      "loss": 1.9748,
      "num_input_tokens_seen": 181857176,
      "step": 178600
    },
    {
      "epoch": 3.518339863361619,
      "grad_norm": 2.0144262313842773,
      "learning_rate": 8.702951836855766e-07,
      "loss": 1.9825,
      "num_input_tokens_seen": 181959576,
      "step": 178700
    },
    {
      "epoch": 3.520308716111122,
      "grad_norm": 2.0384912490844727,
      "learning_rate": 8.70156607833539e-07,
      "loss": 1.959,
      "num_input_tokens_seen": 182061536,
      "step": 178800
    },
    {
      "epoch": 3.522277568860625,
      "grad_norm": 1.6842730045318604,
      "learning_rate": 8.700179690394922e-07,
      "loss": 1.9947,
      "num_input_tokens_seen": 182163464,
      "step": 178900
    },
    {
      "epoch": 3.524246421610128,
      "grad_norm": 1.6675080060958862,
      "learning_rate": 8.698792673270108e-07,
      "loss": 1.9709,
      "num_input_tokens_seen": 182265160,
      "step": 179000
    },
    {
      "epoch": 3.5262152743596307,
      "grad_norm": 0.0,
      "learning_rate": 8.697405027196796e-07,
      "loss": 2.0081,
      "num_input_tokens_seen": 182366032,
      "step": 179100
    },
    {
      "epoch": 3.5281841271091334,
      "grad_norm": 2.0393857955932617,
      "learning_rate": 8.696016752410944e-07,
      "loss": 1.9512,
      "num_input_tokens_seen": 182468432,
      "step": 179200
    },
    {
      "epoch": 3.530152979858636,
      "grad_norm": 1.8243776559829712,
      "learning_rate": 8.694627849148616e-07,
      "loss": 2.0498,
      "num_input_tokens_seen": 182570832,
      "step": 179300
    },
    {
      "epoch": 3.5321218326081394,
      "grad_norm": 2.2270710468292236,
      "learning_rate": 8.693238317645988e-07,
      "loss": 1.9975,
      "num_input_tokens_seen": 182673232,
      "step": 179400
    },
    {
      "epoch": 3.534090685357642,
      "grad_norm": 1.792210340499878,
      "learning_rate": 8.691848158139331e-07,
      "loss": 1.9612,
      "num_input_tokens_seen": 182775632,
      "step": 179500
    },
    {
      "epoch": 3.536059538107145,
      "grad_norm": 2.04830002784729,
      "learning_rate": 8.690457370865034e-07,
      "loss": 1.9844,
      "num_input_tokens_seen": 182878032,
      "step": 179600
    },
    {
      "epoch": 3.538028390856648,
      "grad_norm": 2.1170568466186523,
      "learning_rate": 8.689065956059588e-07,
      "loss": 1.9488,
      "num_input_tokens_seen": 182979704,
      "step": 179700
    },
    {
      "epoch": 3.539997243606151,
      "grad_norm": 2.1218490600585938,
      "learning_rate": 8.687673913959591e-07,
      "loss": 1.9948,
      "num_input_tokens_seen": 183081480,
      "step": 179800
    },
    {
      "epoch": 3.5419660963556536,
      "grad_norm": 1.859352946281433,
      "learning_rate": 8.686281244801748e-07,
      "loss": 1.9588,
      "num_input_tokens_seen": 183183880,
      "step": 179900
    },
    {
      "epoch": 3.5439349491051564,
      "grad_norm": 1.9357470273971558,
      "learning_rate": 8.684887948822872e-07,
      "loss": 1.9905,
      "num_input_tokens_seen": 183286280,
      "step": 180000
    },
    {
      "epoch": 3.545903801854659,
      "grad_norm": 2.0041255950927734,
      "learning_rate": 8.683494026259879e-07,
      "loss": 2.01,
      "num_input_tokens_seen": 183388680,
      "step": 180100
    },
    {
      "epoch": 3.547872654604162,
      "grad_norm": 1.6589006185531616,
      "learning_rate": 8.682099477349795e-07,
      "loss": 1.9872,
      "num_input_tokens_seen": 183489744,
      "step": 180200
    },
    {
      "epoch": 3.549841507353665,
      "grad_norm": 2.102747917175293,
      "learning_rate": 8.680704302329751e-07,
      "loss": 2.0279,
      "num_input_tokens_seen": 183591624,
      "step": 180300
    },
    {
      "epoch": 3.551810360103168,
      "grad_norm": 1.8964042663574219,
      "learning_rate": 8.679308501436985e-07,
      "loss": 2.0346,
      "num_input_tokens_seen": 183694024,
      "step": 180400
    },
    {
      "epoch": 3.5537792128526706,
      "grad_norm": 2.020772933959961,
      "learning_rate": 8.677912074908842e-07,
      "loss": 2.0251,
      "num_input_tokens_seen": 183795032,
      "step": 180500
    },
    {
      "epoch": 3.555748065602174,
      "grad_norm": 5.046718597412109,
      "learning_rate": 8.67651502298277e-07,
      "loss": 1.9951,
      "num_input_tokens_seen": 183895856,
      "step": 180600
    },
    {
      "epoch": 3.5577169183516766,
      "grad_norm": 2.6011369228363037,
      "learning_rate": 8.675117345896328e-07,
      "loss": 1.9797,
      "num_input_tokens_seen": 183998160,
      "step": 180700
    },
    {
      "epoch": 3.5596857711011793,
      "grad_norm": 1.867495059967041,
      "learning_rate": 8.67371904388718e-07,
      "loss": 2.0258,
      "num_input_tokens_seen": 184100080,
      "step": 180800
    },
    {
      "epoch": 3.561654623850682,
      "grad_norm": 2.237550735473633,
      "learning_rate": 8.672320117193093e-07,
      "loss": 1.9952,
      "num_input_tokens_seen": 184200392,
      "step": 180900
    },
    {
      "epoch": 3.563623476600185,
      "grad_norm": 1.8391515016555786,
      "learning_rate": 8.670920566051943e-07,
      "loss": 1.9979,
      "num_input_tokens_seen": 184302000,
      "step": 181000
    },
    {
      "epoch": 3.565592329349688,
      "grad_norm": 1.8157563209533691,
      "learning_rate": 8.669520390701715e-07,
      "loss": 1.9619,
      "num_input_tokens_seen": 184404400,
      "step": 181100
    },
    {
      "epoch": 3.567561182099191,
      "grad_norm": 2.065551996231079,
      "learning_rate": 8.668119591380492e-07,
      "loss": 1.9436,
      "num_input_tokens_seen": 184506800,
      "step": 181200
    },
    {
      "epoch": 3.5695300348486936,
      "grad_norm": 1.8277415037155151,
      "learning_rate": 8.666718168326472e-07,
      "loss": 1.991,
      "num_input_tokens_seen": 184608424,
      "step": 181300
    },
    {
      "epoch": 3.571498887598197,
      "grad_norm": 1.8323853015899658,
      "learning_rate": 8.665316121777955e-07,
      "loss": 1.9495,
      "num_input_tokens_seen": 184710528,
      "step": 181400
    },
    {
      "epoch": 3.5734677403476995,
      "grad_norm": 1.8363794088363647,
      "learning_rate": 8.663913451973346e-07,
      "loss": 2.0341,
      "num_input_tokens_seen": 184810776,
      "step": 181500
    },
    {
      "epoch": 3.5754365930972023,
      "grad_norm": 1.8086886405944824,
      "learning_rate": 8.662510159151156e-07,
      "loss": 1.9911,
      "num_input_tokens_seen": 184913176,
      "step": 181600
    },
    {
      "epoch": 3.577405445846705,
      "grad_norm": 2.2289958000183105,
      "learning_rate": 8.661106243550005e-07,
      "loss": 2.0103,
      "num_input_tokens_seen": 185013848,
      "step": 181700
    },
    {
      "epoch": 3.579374298596208,
      "grad_norm": 1.8465604782104492,
      "learning_rate": 8.659701705408617e-07,
      "loss": 2.0237,
      "num_input_tokens_seen": 185115936,
      "step": 181800
    },
    {
      "epoch": 3.581343151345711,
      "grad_norm": 1.8184815645217896,
      "learning_rate": 8.658296544965822e-07,
      "loss": 1.962,
      "num_input_tokens_seen": 185217720,
      "step": 181900
    },
    {
      "epoch": 3.5833120040952138,
      "grad_norm": 2.013380527496338,
      "learning_rate": 8.656890762460553e-07,
      "loss": 1.9993,
      "num_input_tokens_seen": 185318928,
      "step": 182000
    },
    {
      "epoch": 3.5852808568447165,
      "grad_norm": 2.047400712966919,
      "learning_rate": 8.655484358131855e-07,
      "loss": 1.9998,
      "num_input_tokens_seen": 185419008,
      "step": 182100
    },
    {
      "epoch": 3.5872497095942193,
      "grad_norm": 1.7890983819961548,
      "learning_rate": 8.654077332218873e-07,
      "loss": 1.9903,
      "num_input_tokens_seen": 185520616,
      "step": 182200
    },
    {
      "epoch": 3.5892185623437225,
      "grad_norm": 2.072608709335327,
      "learning_rate": 8.65266968496086e-07,
      "loss": 1.9607,
      "num_input_tokens_seen": 185621336,
      "step": 182300
    },
    {
      "epoch": 3.5911874150932253,
      "grad_norm": 1.9646893739700317,
      "learning_rate": 8.651261416597178e-07,
      "loss": 2.0364,
      "num_input_tokens_seen": 185722904,
      "step": 182400
    },
    {
      "epoch": 3.593156267842728,
      "grad_norm": 1.9087058305740356,
      "learning_rate": 8.649852527367286e-07,
      "loss": 1.9962,
      "num_input_tokens_seen": 185825304,
      "step": 182500
    },
    {
      "epoch": 3.5951251205922308,
      "grad_norm": 1.9879136085510254,
      "learning_rate": 8.648443017510756e-07,
      "loss": 1.9647,
      "num_input_tokens_seen": 185926208,
      "step": 182600
    },
    {
      "epoch": 3.5970939733417335,
      "grad_norm": 1.9980089664459229,
      "learning_rate": 8.647032887267262e-07,
      "loss": 1.9971,
      "num_input_tokens_seen": 186027944,
      "step": 182700
    },
    {
      "epoch": 3.5990628260912367,
      "grad_norm": 1.8720948696136475,
      "learning_rate": 8.645622136876589e-07,
      "loss": 1.9784,
      "num_input_tokens_seen": 186129224,
      "step": 182800
    },
    {
      "epoch": 3.6010316788407395,
      "grad_norm": 2.2965409755706787,
      "learning_rate": 8.64421076657862e-07,
      "loss": 1.9903,
      "num_input_tokens_seen": 186230952,
      "step": 182900
    },
    {
      "epoch": 3.6030005315902423,
      "grad_norm": 2.0069820880889893,
      "learning_rate": 8.642798776613344e-07,
      "loss": 1.9691,
      "num_input_tokens_seen": 186333352,
      "step": 183000
    },
    {
      "epoch": 3.6049693843397455,
      "grad_norm": 1.9990017414093018,
      "learning_rate": 8.641386167220863e-07,
      "loss": 2.0193,
      "num_input_tokens_seen": 186435752,
      "step": 183100
    },
    {
      "epoch": 3.606938237089248,
      "grad_norm": 1.9313108921051025,
      "learning_rate": 8.639972938641375e-07,
      "loss": 1.96,
      "num_input_tokens_seen": 186536600,
      "step": 183200
    },
    {
      "epoch": 3.608907089838751,
      "grad_norm": 1.6989598274230957,
      "learning_rate": 8.638559091115191e-07,
      "loss": 1.9586,
      "num_input_tokens_seen": 186639000,
      "step": 183300
    },
    {
      "epoch": 3.6108759425882537,
      "grad_norm": 2.0604283809661865,
      "learning_rate": 8.637144624882724e-07,
      "loss": 2.0078,
      "num_input_tokens_seen": 186741400,
      "step": 183400
    },
    {
      "epoch": 3.6128447953377565,
      "grad_norm": 1.9702231884002686,
      "learning_rate": 8.63572954018449e-07,
      "loss": 1.9799,
      "num_input_tokens_seen": 186843048,
      "step": 183500
    },
    {
      "epoch": 3.6148136480872597,
      "grad_norm": 2.0850439071655273,
      "learning_rate": 8.634313837261111e-07,
      "loss": 1.9358,
      "num_input_tokens_seen": 186945448,
      "step": 183600
    },
    {
      "epoch": 3.6167825008367624,
      "grad_norm": 1.6503615379333496,
      "learning_rate": 8.632897516353319e-07,
      "loss": 1.9519,
      "num_input_tokens_seen": 187047848,
      "step": 183700
    },
    {
      "epoch": 3.618751353586265,
      "grad_norm": 1.742605209350586,
      "learning_rate": 8.631480577701947e-07,
      "loss": 1.9903,
      "num_input_tokens_seen": 187148944,
      "step": 183800
    },
    {
      "epoch": 3.6207202063357684,
      "grad_norm": 1.9477397203445435,
      "learning_rate": 8.630063021547929e-07,
      "loss": 1.9741,
      "num_input_tokens_seen": 187250504,
      "step": 183900
    },
    {
      "epoch": 3.622689059085271,
      "grad_norm": 1.8840001821517944,
      "learning_rate": 8.628644848132314e-07,
      "loss": 2.0098,
      "num_input_tokens_seen": 187352904,
      "step": 184000
    },
    {
      "epoch": 3.624657911834774,
      "grad_norm": 2.2380168437957764,
      "learning_rate": 8.627226057696245e-07,
      "loss": 2.0108,
      "num_input_tokens_seen": 187454432,
      "step": 184100
    },
    {
      "epoch": 3.6266267645842767,
      "grad_norm": 1.914587140083313,
      "learning_rate": 8.62580665048098e-07,
      "loss": 1.9512,
      "num_input_tokens_seen": 187556832,
      "step": 184200
    },
    {
      "epoch": 3.6285956173337794,
      "grad_norm": 1.8810619115829468,
      "learning_rate": 8.624386626727875e-07,
      "loss": 1.9921,
      "num_input_tokens_seen": 187658920,
      "step": 184300
    },
    {
      "epoch": 3.630564470083282,
      "grad_norm": 2.317134380340576,
      "learning_rate": 8.622965986678396e-07,
      "loss": 1.9784,
      "num_input_tokens_seen": 187760560,
      "step": 184400
    },
    {
      "epoch": 3.6325333228327854,
      "grad_norm": 1.864627718925476,
      "learning_rate": 8.621544730574106e-07,
      "loss": 1.9742,
      "num_input_tokens_seen": 187862960,
      "step": 184500
    },
    {
      "epoch": 3.634502175582288,
      "grad_norm": 1.7810827493667603,
      "learning_rate": 8.62012285865668e-07,
      "loss": 1.9929,
      "num_input_tokens_seen": 187963976,
      "step": 184600
    },
    {
      "epoch": 3.636471028331791,
      "grad_norm": 3.099287748336792,
      "learning_rate": 8.618700371167898e-07,
      "loss": 2.0315,
      "num_input_tokens_seen": 188065560,
      "step": 184700
    },
    {
      "epoch": 3.638439881081294,
      "grad_norm": 1.766345500946045,
      "learning_rate": 8.617277268349636e-07,
      "loss": 1.9878,
      "num_input_tokens_seen": 188167736,
      "step": 184800
    },
    {
      "epoch": 3.640408733830797,
      "grad_norm": 1.857459306716919,
      "learning_rate": 8.615853550443887e-07,
      "loss": 1.9637,
      "num_input_tokens_seen": 188269344,
      "step": 184900
    },
    {
      "epoch": 3.6423775865802996,
      "grad_norm": 1.7559006214141846,
      "learning_rate": 8.614429217692738e-07,
      "loss": 2.0003,
      "num_input_tokens_seen": 188371744,
      "step": 185000
    },
    {
      "epoch": 3.6443464393298024,
      "grad_norm": 1.821175217628479,
      "learning_rate": 8.613004270338384e-07,
      "loss": 2.0161,
      "num_input_tokens_seen": 188472352,
      "step": 185100
    },
    {
      "epoch": 3.646315292079305,
      "grad_norm": 1.9865384101867676,
      "learning_rate": 8.611578708623131e-07,
      "loss": 1.9891,
      "num_input_tokens_seen": 188574752,
      "step": 185200
    },
    {
      "epoch": 3.6482841448288084,
      "grad_norm": 2.212289571762085,
      "learning_rate": 8.610152532789377e-07,
      "loss": 1.9667,
      "num_input_tokens_seen": 188676600,
      "step": 185300
    },
    {
      "epoch": 3.650252997578311,
      "grad_norm": 1.9270789623260498,
      "learning_rate": 8.608725743079636e-07,
      "loss": 1.9937,
      "num_input_tokens_seen": 188779000,
      "step": 185400
    },
    {
      "epoch": 3.652221850327814,
      "grad_norm": 2.1115171909332275,
      "learning_rate": 8.60729833973652e-07,
      "loss": 1.9928,
      "num_input_tokens_seen": 188881400,
      "step": 185500
    },
    {
      "epoch": 3.654190703077317,
      "grad_norm": 1.9058347940444946,
      "learning_rate": 8.605870323002745e-07,
      "loss": 1.9409,
      "num_input_tokens_seen": 188983800,
      "step": 185600
    },
    {
      "epoch": 3.65615955582682,
      "grad_norm": 1.9823269844055176,
      "learning_rate": 8.604441693121136e-07,
      "loss": 1.9801,
      "num_input_tokens_seen": 189085472,
      "step": 185700
    },
    {
      "epoch": 3.6581284085763226,
      "grad_norm": 1.990706205368042,
      "learning_rate": 8.603012450334619e-07,
      "loss": 1.9457,
      "num_input_tokens_seen": 189187872,
      "step": 185800
    },
    {
      "epoch": 3.6600972613258254,
      "grad_norm": 1.8885613679885864,
      "learning_rate": 8.601582594886223e-07,
      "loss": 1.9638,
      "num_input_tokens_seen": 189289560,
      "step": 185900
    },
    {
      "epoch": 3.662066114075328,
      "grad_norm": 1.9282426834106445,
      "learning_rate": 8.600152127019085e-07,
      "loss": 1.9725,
      "num_input_tokens_seen": 189391960,
      "step": 186000
    },
    {
      "epoch": 3.6640349668248313,
      "grad_norm": 1.8946210145950317,
      "learning_rate": 8.598721046976441e-07,
      "loss": 1.9753,
      "num_input_tokens_seen": 189494360,
      "step": 186100
    },
    {
      "epoch": 3.666003819574334,
      "grad_norm": 1.8834301233291626,
      "learning_rate": 8.597289355001638e-07,
      "loss": 1.9869,
      "num_input_tokens_seen": 189596760,
      "step": 186200
    },
    {
      "epoch": 3.667972672323837,
      "grad_norm": 1.8436224460601807,
      "learning_rate": 8.595857051338121e-07,
      "loss": 1.9507,
      "num_input_tokens_seen": 189698688,
      "step": 186300
    },
    {
      "epoch": 3.66994152507334,
      "grad_norm": 1.9974068403244019,
      "learning_rate": 8.594424136229439e-07,
      "loss": 1.9651,
      "num_input_tokens_seen": 189800464,
      "step": 186400
    },
    {
      "epoch": 3.671910377822843,
      "grad_norm": 2.129647970199585,
      "learning_rate": 8.592990609919251e-07,
      "loss": 1.9953,
      "num_input_tokens_seen": 189901968,
      "step": 186500
    },
    {
      "epoch": 3.6738792305723456,
      "grad_norm": 1.7398163080215454,
      "learning_rate": 8.591556472651315e-07,
      "loss": 1.9789,
      "num_input_tokens_seen": 190004368,
      "step": 186600
    },
    {
      "epoch": 3.6758480833218483,
      "grad_norm": 2.255389451980591,
      "learning_rate": 8.590121724669492e-07,
      "loss": 1.9799,
      "num_input_tokens_seen": 190104880,
      "step": 186700
    },
    {
      "epoch": 3.677816936071351,
      "grad_norm": 6.116527080535889,
      "learning_rate": 8.58868636621775e-07,
      "loss": 1.9964,
      "num_input_tokens_seen": 190205592,
      "step": 186800
    },
    {
      "epoch": 3.679785788820854,
      "grad_norm": 2.1795454025268555,
      "learning_rate": 8.587250397540163e-07,
      "loss": 2.011,
      "num_input_tokens_seen": 190307888,
      "step": 186900
    },
    {
      "epoch": 3.681754641570357,
      "grad_norm": 2.2472429275512695,
      "learning_rate": 8.585813818880898e-07,
      "loss": 1.9983,
      "num_input_tokens_seen": 190409176,
      "step": 187000
    },
    {
      "epoch": 3.68372349431986,
      "grad_norm": 1.9176465272903442,
      "learning_rate": 8.58437663048424e-07,
      "loss": 1.9523,
      "num_input_tokens_seen": 190511576,
      "step": 187100
    },
    {
      "epoch": 3.6856923470693626,
      "grad_norm": 1.9553813934326172,
      "learning_rate": 8.582938832594566e-07,
      "loss": 1.9646,
      "num_input_tokens_seen": 190612832,
      "step": 187200
    },
    {
      "epoch": 3.6876611998188658,
      "grad_norm": 1.890999436378479,
      "learning_rate": 8.581500425456365e-07,
      "loss": 1.9459,
      "num_input_tokens_seen": 190715232,
      "step": 187300
    },
    {
      "epoch": 3.6896300525683685,
      "grad_norm": 2.7544963359832764,
      "learning_rate": 8.580061409314225e-07,
      "loss": 2.0061,
      "num_input_tokens_seen": 190816392,
      "step": 187400
    },
    {
      "epoch": 3.6915989053178713,
      "grad_norm": 1.9823936223983765,
      "learning_rate": 8.578621784412836e-07,
      "loss": 1.9828,
      "num_input_tokens_seen": 190918792,
      "step": 187500
    },
    {
      "epoch": 3.693567758067374,
      "grad_norm": 1.8486562967300415,
      "learning_rate": 8.577181550996998e-07,
      "loss": 1.9355,
      "num_input_tokens_seen": 191021192,
      "step": 187600
    },
    {
      "epoch": 3.695536610816877,
      "grad_norm": 1.8696311712265015,
      "learning_rate": 8.575740709311607e-07,
      "loss": 1.9764,
      "num_input_tokens_seen": 191123592,
      "step": 187700
    },
    {
      "epoch": 3.69750546356638,
      "grad_norm": 2.0588529109954834,
      "learning_rate": 8.57429925960167e-07,
      "loss": 1.9945,
      "num_input_tokens_seen": 191224936,
      "step": 187800
    },
    {
      "epoch": 3.6994743163158827,
      "grad_norm": 1.979798436164856,
      "learning_rate": 8.572857202112288e-07,
      "loss": 1.9826,
      "num_input_tokens_seen": 191327336,
      "step": 187900
    },
    {
      "epoch": 3.7014431690653855,
      "grad_norm": 2.138587713241577,
      "learning_rate": 8.571414537088675e-07,
      "loss": 1.9534,
      "num_input_tokens_seen": 191429736,
      "step": 188000
    },
    {
      "epoch": 3.7034120218148887,
      "grad_norm": 1.895857572555542,
      "learning_rate": 8.569971264776141e-07,
      "loss": 1.9745,
      "num_input_tokens_seen": 191531208,
      "step": 188100
    },
    {
      "epoch": 3.7053808745643915,
      "grad_norm": 1.8352304697036743,
      "learning_rate": 8.568527385420105e-07,
      "loss": 1.9534,
      "num_input_tokens_seen": 191633608,
      "step": 188200
    },
    {
      "epoch": 3.7073497273138942,
      "grad_norm": 2.2083489894866943,
      "learning_rate": 8.567082899266087e-07,
      "loss": 1.9698,
      "num_input_tokens_seen": 191735480,
      "step": 188300
    },
    {
      "epoch": 3.709318580063397,
      "grad_norm": 2.0960662364959717,
      "learning_rate": 8.565637806559707e-07,
      "loss": 1.9902,
      "num_input_tokens_seen": 191837880,
      "step": 188400
    },
    {
      "epoch": 3.7112874328128997,
      "grad_norm": 1.8242642879486084,
      "learning_rate": 8.564192107546693e-07,
      "loss": 1.977,
      "num_input_tokens_seen": 191940280,
      "step": 188500
    },
    {
      "epoch": 3.713256285562403,
      "grad_norm": 1.9868113994598389,
      "learning_rate": 8.562745802472871e-07,
      "loss": 1.9903,
      "num_input_tokens_seen": 192042680,
      "step": 188600
    },
    {
      "epoch": 3.7152251383119057,
      "grad_norm": 1.7235865592956543,
      "learning_rate": 8.561298891584177e-07,
      "loss": 1.9602,
      "num_input_tokens_seen": 192144536,
      "step": 188700
    },
    {
      "epoch": 3.7171939910614085,
      "grad_norm": 2.305779218673706,
      "learning_rate": 8.559851375126641e-07,
      "loss": 1.9796,
      "num_input_tokens_seen": 192246656,
      "step": 188800
    },
    {
      "epoch": 3.7191628438109112,
      "grad_norm": 1.788556456565857,
      "learning_rate": 8.558403253346407e-07,
      "loss": 2.01,
      "num_input_tokens_seen": 192348552,
      "step": 188900
    },
    {
      "epoch": 3.7211316965604144,
      "grad_norm": 3.87287974357605,
      "learning_rate": 8.556954526489711e-07,
      "loss": 1.9889,
      "num_input_tokens_seen": 192449680,
      "step": 189000
    },
    {
      "epoch": 3.723100549309917,
      "grad_norm": 2.1033787727355957,
      "learning_rate": 8.555505194802899e-07,
      "loss": 2.0566,
      "num_input_tokens_seen": 192551208,
      "step": 189100
    },
    {
      "epoch": 3.72506940205942,
      "grad_norm": 1.7448862791061401,
      "learning_rate": 8.554055258532415e-07,
      "loss": 1.9601,
      "num_input_tokens_seen": 192651576,
      "step": 189200
    },
    {
      "epoch": 3.7270382548089227,
      "grad_norm": 1.8359603881835938,
      "learning_rate": 8.552604717924813e-07,
      "loss": 1.9817,
      "num_input_tokens_seen": 192753976,
      "step": 189300
    },
    {
      "epoch": 3.7290071075584255,
      "grad_norm": 1.9870225191116333,
      "learning_rate": 8.551153573226742e-07,
      "loss": 1.9943,
      "num_input_tokens_seen": 192855344,
      "step": 189400
    },
    {
      "epoch": 3.7309759603079287,
      "grad_norm": 1.8825901746749878,
      "learning_rate": 8.549701824684958e-07,
      "loss": 1.9913,
      "num_input_tokens_seen": 192957104,
      "step": 189500
    },
    {
      "epoch": 3.7329448130574314,
      "grad_norm": 1.7999411821365356,
      "learning_rate": 8.548249472546318e-07,
      "loss": 1.9509,
      "num_input_tokens_seen": 193058936,
      "step": 189600
    },
    {
      "epoch": 3.734913665806934,
      "grad_norm": 1.7239336967468262,
      "learning_rate": 8.546796517057783e-07,
      "loss": 2.0122,
      "num_input_tokens_seen": 193159800,
      "step": 189700
    },
    {
      "epoch": 3.7368825185564374,
      "grad_norm": 2.1975951194763184,
      "learning_rate": 8.545342958466415e-07,
      "loss": 2.0278,
      "num_input_tokens_seen": 193260264,
      "step": 189800
    },
    {
      "epoch": 3.73885137130594,
      "grad_norm": 2.1563830375671387,
      "learning_rate": 8.543888797019379e-07,
      "loss": 1.99,
      "num_input_tokens_seen": 193361032,
      "step": 189900
    },
    {
      "epoch": 3.740820224055443,
      "grad_norm": 1.9104723930358887,
      "learning_rate": 8.542434032963945e-07,
      "loss": 2.0445,
      "num_input_tokens_seen": 193462888,
      "step": 190000
    },
    {
      "epoch": 3.7427890768049457,
      "grad_norm": 1.787787914276123,
      "learning_rate": 8.540978666547481e-07,
      "loss": 2.0123,
      "num_input_tokens_seen": 193564496,
      "step": 190100
    },
    {
      "epoch": 3.7447579295544484,
      "grad_norm": 1.7190896272659302,
      "learning_rate": 8.539522698017461e-07,
      "loss": 1.9979,
      "num_input_tokens_seen": 193666456,
      "step": 190200
    },
    {
      "epoch": 3.7467267823039516,
      "grad_norm": 1.9271742105484009,
      "learning_rate": 8.53806612762146e-07,
      "loss": 2.0014,
      "num_input_tokens_seen": 193768856,
      "step": 190300
    },
    {
      "epoch": 3.7486956350534544,
      "grad_norm": 1.8336818218231201,
      "learning_rate": 8.536608955607156e-07,
      "loss": 2.0196,
      "num_input_tokens_seen": 193870632,
      "step": 190400
    },
    {
      "epoch": 3.750664487802957,
      "grad_norm": 1.9987962245941162,
      "learning_rate": 8.535151182222327e-07,
      "loss": 1.9436,
      "num_input_tokens_seen": 193973032,
      "step": 190500
    },
    {
      "epoch": 3.7526333405524603,
      "grad_norm": 1.9764909744262695,
      "learning_rate": 8.533692807714858e-07,
      "loss": 1.997,
      "num_input_tokens_seen": 194075016,
      "step": 190600
    },
    {
      "epoch": 3.754602193301963,
      "grad_norm": 1.976149559020996,
      "learning_rate": 8.532233832332729e-07,
      "loss": 1.9966,
      "num_input_tokens_seen": 194176776,
      "step": 190700
    },
    {
      "epoch": 3.756571046051466,
      "grad_norm": 2.051839590072632,
      "learning_rate": 8.530774256324032e-07,
      "loss": 1.9886,
      "num_input_tokens_seen": 194278600,
      "step": 190800
    },
    {
      "epoch": 3.7585398988009686,
      "grad_norm": 1.795717477798462,
      "learning_rate": 8.529314079936951e-07,
      "loss": 2.0213,
      "num_input_tokens_seen": 194380088,
      "step": 190900
    },
    {
      "epoch": 3.7605087515504714,
      "grad_norm": 2.2465925216674805,
      "learning_rate": 8.527853303419779e-07,
      "loss": 1.9507,
      "num_input_tokens_seen": 194481016,
      "step": 191000
    },
    {
      "epoch": 3.762477604299974,
      "grad_norm": 1.9409157037734985,
      "learning_rate": 8.526391927020907e-07,
      "loss": 2.0071,
      "num_input_tokens_seen": 194580744,
      "step": 191100
    },
    {
      "epoch": 3.7644464570494773,
      "grad_norm": 1.8164609670639038,
      "learning_rate": 8.52492995098883e-07,
      "loss": 1.9863,
      "num_input_tokens_seen": 194682592,
      "step": 191200
    },
    {
      "epoch": 3.76641530979898,
      "grad_norm": 1.8520736694335938,
      "learning_rate": 8.523467375572148e-07,
      "loss": 1.9554,
      "num_input_tokens_seen": 194784152,
      "step": 191300
    },
    {
      "epoch": 3.768384162548483,
      "grad_norm": 1.793675184249878,
      "learning_rate": 8.522004201019556e-07,
      "loss": 1.9964,
      "num_input_tokens_seen": 194884664,
      "step": 191400
    },
    {
      "epoch": 3.770353015297986,
      "grad_norm": 1.9452241659164429,
      "learning_rate": 8.520540427579854e-07,
      "loss": 1.9821,
      "num_input_tokens_seen": 194987064,
      "step": 191500
    },
    {
      "epoch": 3.772321868047489,
      "grad_norm": 1.653780221939087,
      "learning_rate": 8.519076055501948e-07,
      "loss": 1.9732,
      "num_input_tokens_seen": 195088424,
      "step": 191600
    },
    {
      "epoch": 3.7742907207969916,
      "grad_norm": 2.1882855892181396,
      "learning_rate": 8.517611085034839e-07,
      "loss": 1.973,
      "num_input_tokens_seen": 195190344,
      "step": 191700
    },
    {
      "epoch": 3.7762595735464943,
      "grad_norm": 1.9713499546051025,
      "learning_rate": 8.516145516427635e-07,
      "loss": 1.9973,
      "num_input_tokens_seen": 195290768,
      "step": 191800
    },
    {
      "epoch": 3.778228426295997,
      "grad_norm": 1.9395986795425415,
      "learning_rate": 8.514679349929541e-07,
      "loss": 1.9831,
      "num_input_tokens_seen": 195393168,
      "step": 191900
    },
    {
      "epoch": 3.7801972790455003,
      "grad_norm": 2.054429292678833,
      "learning_rate": 8.513212585789869e-07,
      "loss": 1.9653,
      "num_input_tokens_seen": 195495224,
      "step": 192000
    },
    {
      "epoch": 3.782166131795003,
      "grad_norm": 2.1292591094970703,
      "learning_rate": 8.511745224258028e-07,
      "loss": 1.9975,
      "num_input_tokens_seen": 195597624,
      "step": 192100
    },
    {
      "epoch": 3.784134984544506,
      "grad_norm": 2.0953760147094727,
      "learning_rate": 8.510277265583532e-07,
      "loss": 1.9709,
      "num_input_tokens_seen": 195700024,
      "step": 192200
    },
    {
      "epoch": 3.786103837294009,
      "grad_norm": 1.8644973039627075,
      "learning_rate": 8.508808710015992e-07,
      "loss": 1.9736,
      "num_input_tokens_seen": 195801880,
      "step": 192300
    },
    {
      "epoch": 3.7880726900435118,
      "grad_norm": 1.8061989545822144,
      "learning_rate": 8.507339557805127e-07,
      "loss": 1.9991,
      "num_input_tokens_seen": 195903720,
      "step": 192400
    },
    {
      "epoch": 3.7900415427930145,
      "grad_norm": 2.189606189727783,
      "learning_rate": 8.505869809200752e-07,
      "loss": 2.0377,
      "num_input_tokens_seen": 196006120,
      "step": 192500
    },
    {
      "epoch": 3.7920103955425173,
      "grad_norm": 1.904588222503662,
      "learning_rate": 8.504399464452785e-07,
      "loss": 1.9689,
      "num_input_tokens_seen": 196108520,
      "step": 192600
    },
    {
      "epoch": 3.79397924829202,
      "grad_norm": 2.192648410797119,
      "learning_rate": 8.502928523811249e-07,
      "loss": 1.9871,
      "num_input_tokens_seen": 196210176,
      "step": 192700
    },
    {
      "epoch": 3.7959481010415232,
      "grad_norm": 1.6559892892837524,
      "learning_rate": 8.50145698752626e-07,
      "loss": 1.98,
      "num_input_tokens_seen": 196312576,
      "step": 192800
    },
    {
      "epoch": 3.797916953791026,
      "grad_norm": 2.859254837036133,
      "learning_rate": 8.499984855848044e-07,
      "loss": 2.0401,
      "num_input_tokens_seen": 196414976,
      "step": 192900
    },
    {
      "epoch": 3.7998858065405288,
      "grad_norm": 2.1295316219329834,
      "learning_rate": 8.498512129026924e-07,
      "loss": 2.027,
      "num_input_tokens_seen": 196516168,
      "step": 193000
    },
    {
      "epoch": 3.801854659290032,
      "grad_norm": 2.002180814743042,
      "learning_rate": 8.497038807313324e-07,
      "loss": 2.0009,
      "num_input_tokens_seen": 196618568,
      "step": 193100
    },
    {
      "epoch": 3.8038235120395347,
      "grad_norm": 1.9041650295257568,
      "learning_rate": 8.495564890957769e-07,
      "loss": 2.003,
      "num_input_tokens_seen": 196719792,
      "step": 193200
    },
    {
      "epoch": 3.8057923647890375,
      "grad_norm": 2.1067962646484375,
      "learning_rate": 8.494090380210888e-07,
      "loss": 1.9892,
      "num_input_tokens_seen": 196821376,
      "step": 193300
    },
    {
      "epoch": 3.8077612175385402,
      "grad_norm": 2.053758382797241,
      "learning_rate": 8.492615275323409e-07,
      "loss": 1.9502,
      "num_input_tokens_seen": 196923776,
      "step": 193400
    },
    {
      "epoch": 3.809730070288043,
      "grad_norm": 2.005551815032959,
      "learning_rate": 8.49113957654616e-07,
      "loss": 1.9644,
      "num_input_tokens_seen": 197023584,
      "step": 193500
    },
    {
      "epoch": 3.8116989230375458,
      "grad_norm": 1.8815425634384155,
      "learning_rate": 8.489663284130071e-07,
      "loss": 1.9929,
      "num_input_tokens_seen": 197125984,
      "step": 193600
    },
    {
      "epoch": 3.813667775787049,
      "grad_norm": 1.7223732471466064,
      "learning_rate": 8.488186398326175e-07,
      "loss": 1.9873,
      "num_input_tokens_seen": 197227784,
      "step": 193700
    },
    {
      "epoch": 3.8156366285365517,
      "grad_norm": 1.9018996953964233,
      "learning_rate": 8.486708919385601e-07,
      "loss": 1.9699,
      "num_input_tokens_seen": 197330184,
      "step": 193800
    },
    {
      "epoch": 3.8176054812860545,
      "grad_norm": 1.9016982316970825,
      "learning_rate": 8.485230847559586e-07,
      "loss": 1.9702,
      "num_input_tokens_seen": 197432584,
      "step": 193900
    },
    {
      "epoch": 3.8195743340355577,
      "grad_norm": 2.1803274154663086,
      "learning_rate": 8.48375218309946e-07,
      "loss": 1.9834,
      "num_input_tokens_seen": 197534984,
      "step": 194000
    },
    {
      "epoch": 3.8215431867850604,
      "grad_norm": 1.8374851942062378,
      "learning_rate": 8.482272926256657e-07,
      "loss": 2.0257,
      "num_input_tokens_seen": 197636808,
      "step": 194100
    },
    {
      "epoch": 3.823512039534563,
      "grad_norm": 1.8388861417770386,
      "learning_rate": 8.480793077282715e-07,
      "loss": 1.9356,
      "num_input_tokens_seen": 197739208,
      "step": 194200
    },
    {
      "epoch": 3.825480892284066,
      "grad_norm": 3.2752463817596436,
      "learning_rate": 8.47931263642927e-07,
      "loss": 1.9702,
      "num_input_tokens_seen": 197840224,
      "step": 194300
    },
    {
      "epoch": 3.8274497450335687,
      "grad_norm": 1.958634614944458,
      "learning_rate": 8.477831603948056e-07,
      "loss": 1.9928,
      "num_input_tokens_seen": 197941960,
      "step": 194400
    },
    {
      "epoch": 3.829418597783072,
      "grad_norm": 1.793360948562622,
      "learning_rate": 8.47634998009091e-07,
      "loss": 1.9904,
      "num_input_tokens_seen": 198043864,
      "step": 194500
    },
    {
      "epoch": 3.8313874505325747,
      "grad_norm": 1.9204010963439941,
      "learning_rate": 8.474867765109772e-07,
      "loss": 2.0096,
      "num_input_tokens_seen": 198145640,
      "step": 194600
    },
    {
      "epoch": 3.8333563032820774,
      "grad_norm": 2.309009552001953,
      "learning_rate": 8.473384959256678e-07,
      "loss": 1.979,
      "num_input_tokens_seen": 198248040,
      "step": 194700
    },
    {
      "epoch": 3.8353251560315806,
      "grad_norm": 1.9729424715042114,
      "learning_rate": 8.471901562783767e-07,
      "loss": 2.0032,
      "num_input_tokens_seen": 198348840,
      "step": 194800
    },
    {
      "epoch": 3.8372940087810834,
      "grad_norm": 1.9262510538101196,
      "learning_rate": 8.47041757594328e-07,
      "loss": 1.9444,
      "num_input_tokens_seen": 198451240,
      "step": 194900
    },
    {
      "epoch": 3.839262861530586,
      "grad_norm": 2.2204275131225586,
      "learning_rate": 8.468932998987556e-07,
      "loss": 1.9965,
      "num_input_tokens_seen": 198552328,
      "step": 195000
    },
    {
      "epoch": 3.841231714280089,
      "grad_norm": 1.9872573614120483,
      "learning_rate": 8.467447832169032e-07,
      "loss": 2.0584,
      "num_input_tokens_seen": 198653104,
      "step": 195100
    },
    {
      "epoch": 3.8432005670295917,
      "grad_norm": 2.0090408325195312,
      "learning_rate": 8.465962075740252e-07,
      "loss": 1.9778,
      "num_input_tokens_seen": 198754936,
      "step": 195200
    },
    {
      "epoch": 3.845169419779095,
      "grad_norm": 2.006701946258545,
      "learning_rate": 8.464475729953853e-07,
      "loss": 1.969,
      "num_input_tokens_seen": 198857336,
      "step": 195300
    },
    {
      "epoch": 3.8471382725285976,
      "grad_norm": 1.786949634552002,
      "learning_rate": 8.462988795062575e-07,
      "loss": 1.9855,
      "num_input_tokens_seen": 198959736,
      "step": 195400
    },
    {
      "epoch": 3.8491071252781004,
      "grad_norm": 1.8043653964996338,
      "learning_rate": 8.461501271319262e-07,
      "loss": 1.9842,
      "num_input_tokens_seen": 199062136,
      "step": 195500
    },
    {
      "epoch": 3.851075978027603,
      "grad_norm": 1.873810887336731,
      "learning_rate": 8.460013158976856e-07,
      "loss": 2.0087,
      "num_input_tokens_seen": 199164536,
      "step": 195600
    },
    {
      "epoch": 3.8530448307771064,
      "grad_norm": 1.781235933303833,
      "learning_rate": 8.458524458288392e-07,
      "loss": 2.0274,
      "num_input_tokens_seen": 199266432,
      "step": 195700
    },
    {
      "epoch": 3.855013683526609,
      "grad_norm": 1.8602921962738037,
      "learning_rate": 8.457035169507017e-07,
      "loss": 1.9717,
      "num_input_tokens_seen": 199368832,
      "step": 195800
    },
    {
      "epoch": 3.856982536276112,
      "grad_norm": 11.825242042541504,
      "learning_rate": 8.455545292885966e-07,
      "loss": 2.0065,
      "num_input_tokens_seen": 199470312,
      "step": 195900
    },
    {
      "epoch": 3.8589513890256146,
      "grad_norm": 1.8490667343139648,
      "learning_rate": 8.454054828678586e-07,
      "loss": 1.9557,
      "num_input_tokens_seen": 199572712,
      "step": 196000
    },
    {
      "epoch": 3.8609202417751174,
      "grad_norm": 1.6781002283096313,
      "learning_rate": 8.452563777138315e-07,
      "loss": 2.0068,
      "num_input_tokens_seen": 199674312,
      "step": 196100
    },
    {
      "epoch": 3.8628890945246206,
      "grad_norm": 1.9158780574798584,
      "learning_rate": 8.451072138518694e-07,
      "loss": 1.9492,
      "num_input_tokens_seen": 199776712,
      "step": 196200
    },
    {
      "epoch": 3.8648579472741234,
      "grad_norm": 1.8332237005233765,
      "learning_rate": 8.449579913073362e-07,
      "loss": 1.9589,
      "num_input_tokens_seen": 199876728,
      "step": 196300
    },
    {
      "epoch": 3.866826800023626,
      "grad_norm": 1.9842472076416016,
      "learning_rate": 8.448087101056063e-07,
      "loss": 1.996,
      "num_input_tokens_seen": 199978424,
      "step": 196400
    },
    {
      "epoch": 3.8687956527731293,
      "grad_norm": 2.268526315689087,
      "learning_rate": 8.446593702720633e-07,
      "loss": 1.987,
      "num_input_tokens_seen": 200079264,
      "step": 196500
    },
    {
      "epoch": 3.870764505522632,
      "grad_norm": 1.9832903146743774,
      "learning_rate": 8.445099718321014e-07,
      "loss": 1.9845,
      "num_input_tokens_seen": 200180888,
      "step": 196600
    },
    {
      "epoch": 3.872733358272135,
      "grad_norm": 1.9900301694869995,
      "learning_rate": 8.443605148111247e-07,
      "loss": 1.9131,
      "num_input_tokens_seen": 200283288,
      "step": 196700
    },
    {
      "epoch": 3.8747022110216376,
      "grad_norm": 1.9500422477722168,
      "learning_rate": 8.442109992345468e-07,
      "loss": 1.9981,
      "num_input_tokens_seen": 200384456,
      "step": 196800
    },
    {
      "epoch": 3.8766710637711403,
      "grad_norm": 1.8457444906234741,
      "learning_rate": 8.440614251277918e-07,
      "loss": 1.9838,
      "num_input_tokens_seen": 200485656,
      "step": 196900
    },
    {
      "epoch": 3.8786399165206435,
      "grad_norm": 1.8744587898254395,
      "learning_rate": 8.439117925162933e-07,
      "loss": 2.0387,
      "num_input_tokens_seen": 200587456,
      "step": 197000
    },
    {
      "epoch": 3.8806087692701463,
      "grad_norm": 2.4993038177490234,
      "learning_rate": 8.437621014254952e-07,
      "loss": 2.0066,
      "num_input_tokens_seen": 200689856,
      "step": 197100
    },
    {
      "epoch": 3.882577622019649,
      "grad_norm": 2.0390734672546387,
      "learning_rate": 8.436123518808511e-07,
      "loss": 1.9351,
      "num_input_tokens_seen": 200792256,
      "step": 197200
    },
    {
      "epoch": 3.8845464747691523,
      "grad_norm": 1.6963281631469727,
      "learning_rate": 8.434625439078247e-07,
      "loss": 2.0019,
      "num_input_tokens_seen": 200893392,
      "step": 197300
    },
    {
      "epoch": 3.886515327518655,
      "grad_norm": 2.2399237155914307,
      "learning_rate": 8.433126775318899e-07,
      "loss": 1.9791,
      "num_input_tokens_seen": 200995272,
      "step": 197400
    },
    {
      "epoch": 3.888484180268158,
      "grad_norm": 1.9176384210586548,
      "learning_rate": 8.431627527785297e-07,
      "loss": 2.0071,
      "num_input_tokens_seen": 201097672,
      "step": 197500
    },
    {
      "epoch": 3.8904530330176605,
      "grad_norm": 1.8128925561904907,
      "learning_rate": 8.430127696732376e-07,
      "loss": 1.9774,
      "num_input_tokens_seen": 201200072,
      "step": 197600
    },
    {
      "epoch": 3.8924218857671633,
      "grad_norm": 2.0059993267059326,
      "learning_rate": 8.428627282415173e-07,
      "loss": 1.973,
      "num_input_tokens_seen": 201301280,
      "step": 197700
    },
    {
      "epoch": 3.894390738516666,
      "grad_norm": 2.077951669692993,
      "learning_rate": 8.427126285088819e-07,
      "loss": 1.9979,
      "num_input_tokens_seen": 201403680,
      "step": 197800
    },
    {
      "epoch": 3.8963595912661693,
      "grad_norm": 1.7671644687652588,
      "learning_rate": 8.425624705008546e-07,
      "loss": 1.9887,
      "num_input_tokens_seen": 201505152,
      "step": 197900
    },
    {
      "epoch": 3.898328444015672,
      "grad_norm": 1.8615261316299438,
      "learning_rate": 8.424122542429685e-07,
      "loss": 1.9901,
      "num_input_tokens_seen": 201607552,
      "step": 198000
    },
    {
      "epoch": 3.900297296765175,
      "grad_norm": 1.9583107233047485,
      "learning_rate": 8.422619797607667e-07,
      "loss": 1.9962,
      "num_input_tokens_seen": 201709952,
      "step": 198100
    },
    {
      "epoch": 3.902266149514678,
      "grad_norm": 1.7911033630371094,
      "learning_rate": 8.421116470798018e-07,
      "loss": 1.9556,
      "num_input_tokens_seen": 201810960,
      "step": 198200
    },
    {
      "epoch": 3.9042350022641807,
      "grad_norm": 1.959963321685791,
      "learning_rate": 8.419612562256369e-07,
      "loss": 2.028,
      "num_input_tokens_seen": 201913360,
      "step": 198300
    },
    {
      "epoch": 3.9062038550136835,
      "grad_norm": 1.7791067361831665,
      "learning_rate": 8.418108072238446e-07,
      "loss": 1.9523,
      "num_input_tokens_seen": 202015176,
      "step": 198400
    },
    {
      "epoch": 3.9081727077631863,
      "grad_norm": 2.198005437850952,
      "learning_rate": 8.416603001000075e-07,
      "loss": 2.0028,
      "num_input_tokens_seen": 202116712,
      "step": 198500
    },
    {
      "epoch": 3.910141560512689,
      "grad_norm": 2.0756611824035645,
      "learning_rate": 8.415097348797181e-07,
      "loss": 1.9562,
      "num_input_tokens_seen": 202219112,
      "step": 198600
    },
    {
      "epoch": 3.912110413262192,
      "grad_norm": 2.0650599002838135,
      "learning_rate": 8.413591115885788e-07,
      "loss": 1.9894,
      "num_input_tokens_seen": 202321512,
      "step": 198700
    },
    {
      "epoch": 3.914079266011695,
      "grad_norm": 2.096550226211548,
      "learning_rate": 8.412084302522016e-07,
      "loss": 1.996,
      "num_input_tokens_seen": 202423496,
      "step": 198800
    },
    {
      "epoch": 3.9160481187611977,
      "grad_norm": 1.5226213932037354,
      "learning_rate": 8.410576908962087e-07,
      "loss": 1.9765,
      "num_input_tokens_seen": 202525256,
      "step": 198900
    },
    {
      "epoch": 3.918016971510701,
      "grad_norm": 2.0522027015686035,
      "learning_rate": 8.409068935462322e-07,
      "loss": 1.9696,
      "num_input_tokens_seen": 202627656,
      "step": 199000
    },
    {
      "epoch": 3.9199858242602037,
      "grad_norm": 1.8121548891067505,
      "learning_rate": 8.407560382279136e-07,
      "loss": 1.9455,
      "num_input_tokens_seen": 202730056,
      "step": 199100
    },
    {
      "epoch": 3.9219546770097065,
      "grad_norm": 1.8804686069488525,
      "learning_rate": 8.406051249669049e-07,
      "loss": 1.9787,
      "num_input_tokens_seen": 202831760,
      "step": 199200
    },
    {
      "epoch": 3.923923529759209,
      "grad_norm": 1.9517990350723267,
      "learning_rate": 8.404541537888674e-07,
      "loss": 1.9847,
      "num_input_tokens_seen": 202934160,
      "step": 199300
    },
    {
      "epoch": 3.925892382508712,
      "grad_norm": 2.3334410190582275,
      "learning_rate": 8.403031247194728e-07,
      "loss": 1.9484,
      "num_input_tokens_seen": 203036560,
      "step": 199400
    },
    {
      "epoch": 3.927861235258215,
      "grad_norm": 1.965456247329712,
      "learning_rate": 8.401520377844018e-07,
      "loss": 1.959,
      "num_input_tokens_seen": 203138032,
      "step": 199500
    },
    {
      "epoch": 3.929830088007718,
      "grad_norm": 1.795345425605774,
      "learning_rate": 8.400008930093459e-07,
      "loss": 1.9658,
      "num_input_tokens_seen": 203240432,
      "step": 199600
    },
    {
      "epoch": 3.9317989407572207,
      "grad_norm": 2.119727611541748,
      "learning_rate": 8.398496904200059e-07,
      "loss": 1.9998,
      "num_input_tokens_seen": 203340840,
      "step": 199700
    },
    {
      "epoch": 3.933767793506724,
      "grad_norm": 1.9842429161071777,
      "learning_rate": 8.396984300420924e-07,
      "loss": 2.0053,
      "num_input_tokens_seen": 203440832,
      "step": 199800
    },
    {
      "epoch": 3.9357366462562267,
      "grad_norm": 2.071290969848633,
      "learning_rate": 8.39547111901326e-07,
      "loss": 2.041,
      "num_input_tokens_seen": 203542464,
      "step": 199900
    },
    {
      "epoch": 3.9377054990057294,
      "grad_norm": 1.716937780380249,
      "learning_rate": 8.393957360234371e-07,
      "loss": 1.9668,
      "num_input_tokens_seen": 203644096,
      "step": 200000
    },
    {
      "epoch": 3.939674351755232,
      "grad_norm": 1.91750168800354,
      "learning_rate": 8.39244302434166e-07,
      "loss": 2.036,
      "num_input_tokens_seen": 203746496,
      "step": 200100
    },
    {
      "epoch": 3.941643204504735,
      "grad_norm": 1.9597529172897339,
      "learning_rate": 8.390928111592624e-07,
      "loss": 1.998,
      "num_input_tokens_seen": 203848288,
      "step": 200200
    },
    {
      "epoch": 3.9436120572542377,
      "grad_norm": 2.6929409503936768,
      "learning_rate": 8.389412622244865e-07,
      "loss": 1.9637,
      "num_input_tokens_seen": 203950688,
      "step": 200300
    },
    {
      "epoch": 3.945580910003741,
      "grad_norm": 1.871503233909607,
      "learning_rate": 8.387896556556076e-07,
      "loss": 2.0262,
      "num_input_tokens_seen": 204053088,
      "step": 200400
    },
    {
      "epoch": 3.9475497627532437,
      "grad_norm": 1.8585203886032104,
      "learning_rate": 8.386379914784052e-07,
      "loss": 1.9907,
      "num_input_tokens_seen": 204155488,
      "step": 200500
    },
    {
      "epoch": 3.9495186155027464,
      "grad_norm": 2.4684081077575684,
      "learning_rate": 8.384862697186685e-07,
      "loss": 1.9869,
      "num_input_tokens_seen": 204256816,
      "step": 200600
    },
    {
      "epoch": 3.9514874682522496,
      "grad_norm": 1.9001214504241943,
      "learning_rate": 8.383344904021967e-07,
      "loss": 1.9518,
      "num_input_tokens_seen": 204358720,
      "step": 200700
    },
    {
      "epoch": 3.9534563210017524,
      "grad_norm": 2.191331624984741,
      "learning_rate": 8.381826535547985e-07,
      "loss": 2.001,
      "num_input_tokens_seen": 204461120,
      "step": 200800
    },
    {
      "epoch": 3.955425173751255,
      "grad_norm": 1.946645736694336,
      "learning_rate": 8.380307592022924e-07,
      "loss": 1.9508,
      "num_input_tokens_seen": 204563520,
      "step": 200900
    },
    {
      "epoch": 3.957394026500758,
      "grad_norm": 1.9473180770874023,
      "learning_rate": 8.378788073705068e-07,
      "loss": 1.9856,
      "num_input_tokens_seen": 204664960,
      "step": 201000
    },
    {
      "epoch": 3.9593628792502606,
      "grad_norm": 1.8633493185043335,
      "learning_rate": 8.377267980852795e-07,
      "loss": 2.0254,
      "num_input_tokens_seen": 204766272,
      "step": 201100
    },
    {
      "epoch": 3.961331731999764,
      "grad_norm": 1.9327882528305054,
      "learning_rate": 8.375747313724591e-07,
      "loss": 1.9597,
      "num_input_tokens_seen": 204868136,
      "step": 201200
    },
    {
      "epoch": 3.9633005847492666,
      "grad_norm": 2.6450247764587402,
      "learning_rate": 8.374226072579027e-07,
      "loss": 2.0266,
      "num_input_tokens_seen": 204970536,
      "step": 201300
    },
    {
      "epoch": 3.9652694374987694,
      "grad_norm": 1.8664968013763428,
      "learning_rate": 8.372704257674779e-07,
      "loss": 1.9837,
      "num_input_tokens_seen": 205072080,
      "step": 201400
    },
    {
      "epoch": 3.9672382902482726,
      "grad_norm": 1.7720860242843628,
      "learning_rate": 8.371181869270618e-07,
      "loss": 2.0002,
      "num_input_tokens_seen": 205173616,
      "step": 201500
    },
    {
      "epoch": 3.9692071429977753,
      "grad_norm": 2.5905239582061768,
      "learning_rate": 8.369658907625415e-07,
      "loss": 1.9925,
      "num_input_tokens_seen": 205275200,
      "step": 201600
    },
    {
      "epoch": 3.971175995747278,
      "grad_norm": 2.0033373832702637,
      "learning_rate": 8.368135372998135e-07,
      "loss": 2.0251,
      "num_input_tokens_seen": 205375624,
      "step": 201700
    },
    {
      "epoch": 3.973144848496781,
      "grad_norm": 2.1199939250946045,
      "learning_rate": 8.366611265647844e-07,
      "loss": 1.9952,
      "num_input_tokens_seen": 205478024,
      "step": 201800
    },
    {
      "epoch": 3.9751137012462836,
      "grad_norm": 2.075946092605591,
      "learning_rate": 8.365086585833701e-07,
      "loss": 2.0151,
      "num_input_tokens_seen": 205579664,
      "step": 201900
    },
    {
      "epoch": 3.977082553995787,
      "grad_norm": 1.9389138221740723,
      "learning_rate": 8.363561333814968e-07,
      "loss": 1.9899,
      "num_input_tokens_seen": 205681512,
      "step": 202000
    },
    {
      "epoch": 3.9790514067452896,
      "grad_norm": 1.8847119808197021,
      "learning_rate": 8.362035509851e-07,
      "loss": 2.0283,
      "num_input_tokens_seen": 205782336,
      "step": 202100
    },
    {
      "epoch": 3.9810202594947923,
      "grad_norm": 5.401323318481445,
      "learning_rate": 8.36050911420125e-07,
      "loss": 1.9538,
      "num_input_tokens_seen": 205883464,
      "step": 202200
    },
    {
      "epoch": 3.9829891122442955,
      "grad_norm": 2.1095244884490967,
      "learning_rate": 8.358982147125269e-07,
      "loss": 1.9767,
      "num_input_tokens_seen": 205985864,
      "step": 202300
    },
    {
      "epoch": 3.9849579649937983,
      "grad_norm": 2.071115255355835,
      "learning_rate": 8.357454608882704e-07,
      "loss": 1.9626,
      "num_input_tokens_seen": 206087512,
      "step": 202400
    },
    {
      "epoch": 3.986926817743301,
      "grad_norm": 1.9583262205123901,
      "learning_rate": 8.355926499733301e-07,
      "loss": 1.9848,
      "num_input_tokens_seen": 206188584,
      "step": 202500
    },
    {
      "epoch": 3.988895670492804,
      "grad_norm": 1.683656096458435,
      "learning_rate": 8.354397819936902e-07,
      "loss": 1.9682,
      "num_input_tokens_seen": 206289464,
      "step": 202600
    },
    {
      "epoch": 3.9908645232423066,
      "grad_norm": 2.043354034423828,
      "learning_rate": 8.352868569753446e-07,
      "loss": 1.9459,
      "num_input_tokens_seen": 206390680,
      "step": 202700
    },
    {
      "epoch": 3.9928333759918093,
      "grad_norm": 1.8235090970993042,
      "learning_rate": 8.351338749442969e-07,
      "loss": 1.9382,
      "num_input_tokens_seen": 206493080,
      "step": 202800
    },
    {
      "epoch": 3.9948022287413125,
      "grad_norm": 1.6625016927719116,
      "learning_rate": 8.349808359265604e-07,
      "loss": 2.0174,
      "num_input_tokens_seen": 206594872,
      "step": 202900
    },
    {
      "epoch": 3.9967710814908153,
      "grad_norm": 1.902225375175476,
      "learning_rate": 8.348277399481583e-07,
      "loss": 2.0029,
      "num_input_tokens_seen": 206696512,
      "step": 203000
    },
    {
      "epoch": 3.998739934240318,
      "grad_norm": 2.0428802967071533,
      "learning_rate": 8.346745870351228e-07,
      "loss": 1.9667,
      "num_input_tokens_seen": 206798912,
      "step": 203100
    },
    {
      "epoch": 4.000708786989821,
      "grad_norm": 1.9964284896850586,
      "learning_rate": 8.345213772134968e-07,
      "loss": 1.9123,
      "num_input_tokens_seen": 206901312,
      "step": 203200
    },
    {
      "epoch": 4.002677639739324,
      "grad_norm": 2.084804058074951,
      "learning_rate": 8.34368110509332e-07,
      "loss": 1.982,
      "num_input_tokens_seen": 207003112,
      "step": 203300
    },
    {
      "epoch": 4.004646492488827,
      "grad_norm": 1.9210829734802246,
      "learning_rate": 8.342147869486902e-07,
      "loss": 1.9427,
      "num_input_tokens_seen": 207105512,
      "step": 203400
    },
    {
      "epoch": 4.0066153452383295,
      "grad_norm": 1.6910932064056396,
      "learning_rate": 8.340614065576427e-07,
      "loss": 1.9943,
      "num_input_tokens_seen": 207206536,
      "step": 203500
    },
    {
      "epoch": 4.008584197987832,
      "grad_norm": 1.9586155414581299,
      "learning_rate": 8.339079693622706e-07,
      "loss": 1.9707,
      "num_input_tokens_seen": 207308400,
      "step": 203600
    },
    {
      "epoch": 4.010553050737335,
      "grad_norm": 1.8755433559417725,
      "learning_rate": 8.337544753886648e-07,
      "loss": 1.9741,
      "num_input_tokens_seen": 207410800,
      "step": 203700
    },
    {
      "epoch": 4.012521903486838,
      "grad_norm": 1.9017244577407837,
      "learning_rate": 8.336009246629253e-07,
      "loss": 1.9616,
      "num_input_tokens_seen": 207512616,
      "step": 203800
    },
    {
      "epoch": 4.014490756236341,
      "grad_norm": 1.9800962209701538,
      "learning_rate": 8.334473172111624e-07,
      "loss": 1.9774,
      "num_input_tokens_seen": 207614440,
      "step": 203900
    },
    {
      "epoch": 4.016459608985844,
      "grad_norm": 2.0316872596740723,
      "learning_rate": 8.332936530594956e-07,
      "loss": 1.9815,
      "num_input_tokens_seen": 207715888,
      "step": 204000
    },
    {
      "epoch": 4.018428461735347,
      "grad_norm": 1.707642674446106,
      "learning_rate": 8.331399322340542e-07,
      "loss": 1.9691,
      "num_input_tokens_seen": 207818288,
      "step": 204100
    },
    {
      "epoch": 4.02039731448485,
      "grad_norm": 1.788368582725525,
      "learning_rate": 8.329861547609771e-07,
      "loss": 1.9727,
      "num_input_tokens_seen": 207920688,
      "step": 204200
    },
    {
      "epoch": 4.0223661672343525,
      "grad_norm": 1.995225191116333,
      "learning_rate": 8.328323206664129e-07,
      "loss": 1.9449,
      "num_input_tokens_seen": 208022560,
      "step": 204300
    },
    {
      "epoch": 4.024335019983855,
      "grad_norm": 2.0321285724639893,
      "learning_rate": 8.3267842997652e-07,
      "loss": 2.0058,
      "num_input_tokens_seen": 208122328,
      "step": 204400
    },
    {
      "epoch": 4.026303872733358,
      "grad_norm": 1.9423996210098267,
      "learning_rate": 8.32524482717466e-07,
      "loss": 1.9511,
      "num_input_tokens_seen": 208224728,
      "step": 204500
    },
    {
      "epoch": 4.028272725482861,
      "grad_norm": 1.8356170654296875,
      "learning_rate": 8.323704789154282e-07,
      "loss": 1.962,
      "num_input_tokens_seen": 208327128,
      "step": 204600
    },
    {
      "epoch": 4.030241578232364,
      "grad_norm": 1.8499250411987305,
      "learning_rate": 8.322164185965939e-07,
      "loss": 2.0148,
      "num_input_tokens_seen": 208429528,
      "step": 204700
    },
    {
      "epoch": 4.032210430981867,
      "grad_norm": 1.967336893081665,
      "learning_rate": 8.320623017871596e-07,
      "loss": 1.956,
      "num_input_tokens_seen": 208531264,
      "step": 204800
    },
    {
      "epoch": 4.03417928373137,
      "grad_norm": 2.0016117095947266,
      "learning_rate": 8.319081285133316e-07,
      "loss": 1.96,
      "num_input_tokens_seen": 208632984,
      "step": 204900
    },
    {
      "epoch": 4.036148136480873,
      "grad_norm": 1.9814389944076538,
      "learning_rate": 8.317538988013259e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 208735384,
      "step": 205000
    },
    {
      "epoch": 4.038116989230375,
      "grad_norm": 2.057405710220337,
      "learning_rate": 8.315996126773678e-07,
      "loss": 1.9073,
      "num_input_tokens_seen": 208837336,
      "step": 205100
    },
    {
      "epoch": 4.040085841979878,
      "grad_norm": 1.9322025775909424,
      "learning_rate": 8.314452701676924e-07,
      "loss": 1.9671,
      "num_input_tokens_seen": 208939544,
      "step": 205200
    },
    {
      "epoch": 4.042054694729381,
      "grad_norm": 2.1665055751800537,
      "learning_rate": 8.312908712985443e-07,
      "loss": 1.9526,
      "num_input_tokens_seen": 209041944,
      "step": 205300
    },
    {
      "epoch": 4.044023547478884,
      "grad_norm": 2.12907075881958,
      "learning_rate": 8.311364160961778e-07,
      "loss": 2.0005,
      "num_input_tokens_seen": 209143344,
      "step": 205400
    },
    {
      "epoch": 4.045992400228387,
      "grad_norm": 2.1272454261779785,
      "learning_rate": 8.309819045868567e-07,
      "loss": 1.9922,
      "num_input_tokens_seen": 209244968,
      "step": 205500
    },
    {
      "epoch": 4.04796125297789,
      "grad_norm": 2.0681750774383545,
      "learning_rate": 8.308273367968543e-07,
      "loss": 2.0,
      "num_input_tokens_seen": 209347368,
      "step": 205600
    },
    {
      "epoch": 4.049930105727393,
      "grad_norm": 1.7734335660934448,
      "learning_rate": 8.306727127524536e-07,
      "loss": 1.9963,
      "num_input_tokens_seen": 209449768,
      "step": 205700
    },
    {
      "epoch": 4.051898958476896,
      "grad_norm": 2.0544943809509277,
      "learning_rate": 8.305180324799472e-07,
      "loss": 1.9861,
      "num_input_tokens_seen": 209551592,
      "step": 205800
    },
    {
      "epoch": 4.053867811226398,
      "grad_norm": 3.1180179119110107,
      "learning_rate": 8.303632960056372e-07,
      "loss": 1.9687,
      "num_input_tokens_seen": 209652912,
      "step": 205900
    },
    {
      "epoch": 4.055836663975901,
      "grad_norm": 1.825851321220398,
      "learning_rate": 8.302085033558349e-07,
      "loss": 1.9902,
      "num_input_tokens_seen": 209754880,
      "step": 206000
    },
    {
      "epoch": 4.057805516725404,
      "grad_norm": 1.6018146276474,
      "learning_rate": 8.300536545568618e-07,
      "loss": 1.9743,
      "num_input_tokens_seen": 209857280,
      "step": 206100
    },
    {
      "epoch": 4.059774369474907,
      "grad_norm": 2.071413993835449,
      "learning_rate": 8.298987496350486e-07,
      "loss": 2.0353,
      "num_input_tokens_seen": 209958416,
      "step": 206200
    },
    {
      "epoch": 4.061743222224409,
      "grad_norm": 1.945900321006775,
      "learning_rate": 8.297437886167356e-07,
      "loss": 1.9995,
      "num_input_tokens_seen": 210060256,
      "step": 206300
    },
    {
      "epoch": 4.063712074973913,
      "grad_norm": 2.2497408390045166,
      "learning_rate": 8.295887715282723e-07,
      "loss": 2.0072,
      "num_input_tokens_seen": 210162656,
      "step": 206400
    },
    {
      "epoch": 4.065680927723416,
      "grad_norm": 2.5082108974456787,
      "learning_rate": 8.294336983960184e-07,
      "loss": 2.0172,
      "num_input_tokens_seen": 210265056,
      "step": 206500
    },
    {
      "epoch": 4.067649780472919,
      "grad_norm": 1.8619681596755981,
      "learning_rate": 8.292785692463426e-07,
      "loss": 1.9583,
      "num_input_tokens_seen": 210367456,
      "step": 206600
    },
    {
      "epoch": 4.069618633222421,
      "grad_norm": 1.8918906450271606,
      "learning_rate": 8.291233841056235e-07,
      "loss": 1.9861,
      "num_input_tokens_seen": 210469856,
      "step": 206700
    },
    {
      "epoch": 4.071587485971924,
      "grad_norm": 1.8196094036102295,
      "learning_rate": 8.289681430002487e-07,
      "loss": 1.9968,
      "num_input_tokens_seen": 210572256,
      "step": 206800
    },
    {
      "epoch": 4.073556338721427,
      "grad_norm": 1.860552191734314,
      "learning_rate": 8.28812845956616e-07,
      "loss": 1.9992,
      "num_input_tokens_seen": 210674120,
      "step": 206900
    },
    {
      "epoch": 4.07552519147093,
      "grad_norm": 3.457857131958008,
      "learning_rate": 8.28657493001132e-07,
      "loss": 1.9544,
      "num_input_tokens_seen": 210776520,
      "step": 207000
    },
    {
      "epoch": 4.077494044220432,
      "grad_norm": 1.7468923330307007,
      "learning_rate": 8.285020841602135e-07,
      "loss": 2.0996,
      "num_input_tokens_seen": 210877536,
      "step": 207100
    },
    {
      "epoch": 4.079462896969936,
      "grad_norm": 1.8271055221557617,
      "learning_rate": 8.283466194602862e-07,
      "loss": 1.9664,
      "num_input_tokens_seen": 210979416,
      "step": 207200
    },
    {
      "epoch": 4.081431749719439,
      "grad_norm": 1.7764694690704346,
      "learning_rate": 8.281910989277858e-07,
      "loss": 2.0187,
      "num_input_tokens_seen": 211081816,
      "step": 207300
    },
    {
      "epoch": 4.0834006024689415,
      "grad_norm": 1.8398135900497437,
      "learning_rate": 8.28035522589157e-07,
      "loss": 1.9931,
      "num_input_tokens_seen": 211182888,
      "step": 207400
    },
    {
      "epoch": 4.085369455218444,
      "grad_norm": 1.9261059761047363,
      "learning_rate": 8.278798904708543e-07,
      "loss": 1.9812,
      "num_input_tokens_seen": 211285288,
      "step": 207500
    },
    {
      "epoch": 4.087338307967947,
      "grad_norm": 2.3591842651367188,
      "learning_rate": 8.277242025993418e-07,
      "loss": 1.9635,
      "num_input_tokens_seen": 211387040,
      "step": 207600
    },
    {
      "epoch": 4.08930716071745,
      "grad_norm": 1.8212112188339233,
      "learning_rate": 8.275684590010928e-07,
      "loss": 1.9566,
      "num_input_tokens_seen": 211488600,
      "step": 207700
    },
    {
      "epoch": 4.091276013466953,
      "grad_norm": 2.126899242401123,
      "learning_rate": 8.274126597025901e-07,
      "loss": 1.9676,
      "num_input_tokens_seen": 211591000,
      "step": 207800
    },
    {
      "epoch": 4.093244866216455,
      "grad_norm": 2.0518031120300293,
      "learning_rate": 8.272568047303263e-07,
      "loss": 1.9952,
      "num_input_tokens_seen": 211693224,
      "step": 207900
    },
    {
      "epoch": 4.095213718965958,
      "grad_norm": 1.884930968284607,
      "learning_rate": 8.27100894110803e-07,
      "loss": 1.9882,
      "num_input_tokens_seen": 211793832,
      "step": 208000
    },
    {
      "epoch": 4.097182571715462,
      "grad_norm": 2.251709222793579,
      "learning_rate": 8.269449278705315e-07,
      "loss": 1.9612,
      "num_input_tokens_seen": 211896232,
      "step": 208100
    },
    {
      "epoch": 4.0991514244649645,
      "grad_norm": 1.875095248222351,
      "learning_rate": 8.267889060360327e-07,
      "loss": 2.0129,
      "num_input_tokens_seen": 211996336,
      "step": 208200
    },
    {
      "epoch": 4.101120277214467,
      "grad_norm": 1.8317281007766724,
      "learning_rate": 8.266328286338367e-07,
      "loss": 1.9711,
      "num_input_tokens_seen": 212096544,
      "step": 208300
    },
    {
      "epoch": 4.10308912996397,
      "grad_norm": 1.7605141401290894,
      "learning_rate": 8.264766956904829e-07,
      "loss": 2.0407,
      "num_input_tokens_seen": 212197544,
      "step": 208400
    },
    {
      "epoch": 4.105057982713473,
      "grad_norm": 2.077226161956787,
      "learning_rate": 8.263205072325208e-07,
      "loss": 2.0048,
      "num_input_tokens_seen": 212298336,
      "step": 208500
    },
    {
      "epoch": 4.1070268354629755,
      "grad_norm": 2.268589973449707,
      "learning_rate": 8.261642632865088e-07,
      "loss": 1.9792,
      "num_input_tokens_seen": 212400512,
      "step": 208600
    },
    {
      "epoch": 4.108995688212478,
      "grad_norm": 1.9634552001953125,
      "learning_rate": 8.260079638790147e-07,
      "loss": 1.982,
      "num_input_tokens_seen": 212502912,
      "step": 208700
    },
    {
      "epoch": 4.110964540961981,
      "grad_norm": 1.8080847263336182,
      "learning_rate": 8.258516090366161e-07,
      "loss": 2.0355,
      "num_input_tokens_seen": 212605312,
      "step": 208800
    },
    {
      "epoch": 4.112933393711485,
      "grad_norm": 1.8589696884155273,
      "learning_rate": 8.256951987858998e-07,
      "loss": 2.001,
      "num_input_tokens_seen": 212706944,
      "step": 208900
    },
    {
      "epoch": 4.1149022464609875,
      "grad_norm": 2.4436771869659424,
      "learning_rate": 8.255387331534619e-07,
      "loss": 1.9415,
      "num_input_tokens_seen": 212809344,
      "step": 209000
    },
    {
      "epoch": 4.11687109921049,
      "grad_norm": 1.895723581314087,
      "learning_rate": 8.253822121659082e-07,
      "loss": 2.0208,
      "num_input_tokens_seen": 212910824,
      "step": 209100
    },
    {
      "epoch": 4.118839951959993,
      "grad_norm": 1.7055015563964844,
      "learning_rate": 8.252256358498538e-07,
      "loss": 1.942,
      "num_input_tokens_seen": 213013224,
      "step": 209200
    },
    {
      "epoch": 4.120808804709496,
      "grad_norm": 1.7317379713058472,
      "learning_rate": 8.250690042319229e-07,
      "loss": 1.9911,
      "num_input_tokens_seen": 213115624,
      "step": 209300
    },
    {
      "epoch": 4.1227776574589985,
      "grad_norm": 2.1245877742767334,
      "learning_rate": 8.249123173387497e-07,
      "loss": 2.0294,
      "num_input_tokens_seen": 213215712,
      "step": 209400
    },
    {
      "epoch": 4.124746510208501,
      "grad_norm": 2.0705044269561768,
      "learning_rate": 8.247555751969773e-07,
      "loss": 2.0028,
      "num_input_tokens_seen": 213317184,
      "step": 209500
    },
    {
      "epoch": 4.126715362958004,
      "grad_norm": 1.9951545000076294,
      "learning_rate": 8.245987778332586e-07,
      "loss": 1.9659,
      "num_input_tokens_seen": 213418608,
      "step": 209600
    },
    {
      "epoch": 4.128684215707508,
      "grad_norm": 1.8356311321258545,
      "learning_rate": 8.244419252742553e-07,
      "loss": 1.9893,
      "num_input_tokens_seen": 213521008,
      "step": 209700
    },
    {
      "epoch": 4.13065306845701,
      "grad_norm": 2.0035300254821777,
      "learning_rate": 8.242850175466392e-07,
      "loss": 2.003,
      "num_input_tokens_seen": 213622856,
      "step": 209800
    },
    {
      "epoch": 4.132621921206513,
      "grad_norm": 1.7533024549484253,
      "learning_rate": 8.24128054677091e-07,
      "loss": 1.9662,
      "num_input_tokens_seen": 213724440,
      "step": 209900
    },
    {
      "epoch": 4.134590773956016,
      "grad_norm": 1.7321463823318481,
      "learning_rate": 8.23971036692301e-07,
      "loss": 1.9721,
      "num_input_tokens_seen": 213826840,
      "step": 210000
    },
    {
      "epoch": 4.136559626705519,
      "grad_norm": 1.9174144268035889,
      "learning_rate": 8.238139636189687e-07,
      "loss": 1.9552,
      "num_input_tokens_seen": 213929240,
      "step": 210100
    },
    {
      "epoch": 4.1385284794550214,
      "grad_norm": 2.0311577320098877,
      "learning_rate": 8.23656835483803e-07,
      "loss": 1.9565,
      "num_input_tokens_seen": 214031640,
      "step": 210200
    },
    {
      "epoch": 4.140497332204524,
      "grad_norm": 1.895086646080017,
      "learning_rate": 8.234996523135224e-07,
      "loss": 1.9584,
      "num_input_tokens_seen": 214134040,
      "step": 210300
    },
    {
      "epoch": 4.142466184954027,
      "grad_norm": 1.6728761196136475,
      "learning_rate": 8.233424141348542e-07,
      "loss": 1.949,
      "num_input_tokens_seen": 214236440,
      "step": 210400
    },
    {
      "epoch": 4.14443503770353,
      "grad_norm": 2.218950033187866,
      "learning_rate": 8.23185120974536e-07,
      "loss": 1.9878,
      "num_input_tokens_seen": 214337696,
      "step": 210500
    },
    {
      "epoch": 4.146403890453033,
      "grad_norm": 1.7907037734985352,
      "learning_rate": 8.230277728593136e-07,
      "loss": 2.0009,
      "num_input_tokens_seen": 214439176,
      "step": 210600
    },
    {
      "epoch": 4.148372743202536,
      "grad_norm": 2.00350022315979,
      "learning_rate": 8.228703698159429e-07,
      "loss": 1.9421,
      "num_input_tokens_seen": 214540328,
      "step": 210700
    },
    {
      "epoch": 4.150341595952039,
      "grad_norm": 1.9227792024612427,
      "learning_rate": 8.227129118711892e-07,
      "loss": 1.9908,
      "num_input_tokens_seen": 214641656,
      "step": 210800
    },
    {
      "epoch": 4.152310448701542,
      "grad_norm": 1.8052361011505127,
      "learning_rate": 8.225553990518267e-07,
      "loss": 1.9454,
      "num_input_tokens_seen": 214744056,
      "step": 210900
    },
    {
      "epoch": 4.154279301451044,
      "grad_norm": 2.1096808910369873,
      "learning_rate": 8.223978313846392e-07,
      "loss": 2.0,
      "num_input_tokens_seen": 214845760,
      "step": 211000
    },
    {
      "epoch": 4.156248154200547,
      "grad_norm": 1.8590428829193115,
      "learning_rate": 8.222402088964195e-07,
      "loss": 2.0573,
      "num_input_tokens_seen": 214947368,
      "step": 211100
    },
    {
      "epoch": 4.15821700695005,
      "grad_norm": 1.908556342124939,
      "learning_rate": 8.220825316139703e-07,
      "loss": 1.929,
      "num_input_tokens_seen": 215049768,
      "step": 211200
    },
    {
      "epoch": 4.160185859699553,
      "grad_norm": 1.777230978012085,
      "learning_rate": 8.219247995641032e-07,
      "loss": 1.9528,
      "num_input_tokens_seen": 215151352,
      "step": 211300
    },
    {
      "epoch": 4.162154712449056,
      "grad_norm": 6.116434574127197,
      "learning_rate": 8.217670127736391e-07,
      "loss": 1.9722,
      "num_input_tokens_seen": 215252352,
      "step": 211400
    },
    {
      "epoch": 4.164123565198559,
      "grad_norm": 2.3004748821258545,
      "learning_rate": 8.216091712694086e-07,
      "loss": 2.0707,
      "num_input_tokens_seen": 215353984,
      "step": 211500
    },
    {
      "epoch": 4.166092417948062,
      "grad_norm": 1.8602970838546753,
      "learning_rate": 8.214512750782509e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 215455768,
      "step": 211600
    },
    {
      "epoch": 4.168061270697565,
      "grad_norm": 1.7709438800811768,
      "learning_rate": 8.212933242270151e-07,
      "loss": 1.9846,
      "num_input_tokens_seen": 215557464,
      "step": 211700
    },
    {
      "epoch": 4.170030123447067,
      "grad_norm": 1.8944815397262573,
      "learning_rate": 8.211353187425593e-07,
      "loss": 1.9754,
      "num_input_tokens_seen": 215658584,
      "step": 211800
    },
    {
      "epoch": 4.17199897619657,
      "grad_norm": 2.141850471496582,
      "learning_rate": 8.209772586517513e-07,
      "loss": 1.9906,
      "num_input_tokens_seen": 215760984,
      "step": 211900
    },
    {
      "epoch": 4.173967828946073,
      "grad_norm": 2.012176036834717,
      "learning_rate": 8.208191439814679e-07,
      "loss": 2.0133,
      "num_input_tokens_seen": 215862536,
      "step": 212000
    },
    {
      "epoch": 4.175936681695576,
      "grad_norm": 2.2813560962677,
      "learning_rate": 8.206609747585949e-07,
      "loss": 2.0069,
      "num_input_tokens_seen": 215963792,
      "step": 212100
    },
    {
      "epoch": 4.177905534445079,
      "grad_norm": 2.2592687606811523,
      "learning_rate": 8.205027510100275e-07,
      "loss": 1.9822,
      "num_input_tokens_seen": 216066192,
      "step": 212200
    },
    {
      "epoch": 4.179874387194582,
      "grad_norm": 2.3675546646118164,
      "learning_rate": 8.203444727626708e-07,
      "loss": 1.9829,
      "num_input_tokens_seen": 216168000,
      "step": 212300
    },
    {
      "epoch": 4.181843239944085,
      "grad_norm": 1.8685001134872437,
      "learning_rate": 8.201861400434382e-07,
      "loss": 2.0066,
      "num_input_tokens_seen": 216268720,
      "step": 212400
    },
    {
      "epoch": 4.183812092693588,
      "grad_norm": 2.0438554286956787,
      "learning_rate": 8.200277528792531e-07,
      "loss": 1.9843,
      "num_input_tokens_seen": 216371120,
      "step": 212500
    },
    {
      "epoch": 4.18578094544309,
      "grad_norm": 2.06326961517334,
      "learning_rate": 8.198693112970478e-07,
      "loss": 1.9634,
      "num_input_tokens_seen": 216473520,
      "step": 212600
    },
    {
      "epoch": 4.187749798192593,
      "grad_norm": 1.8550713062286377,
      "learning_rate": 8.197108153237642e-07,
      "loss": 1.9755,
      "num_input_tokens_seen": 216575920,
      "step": 212700
    },
    {
      "epoch": 4.189718650942096,
      "grad_norm": 1.854211688041687,
      "learning_rate": 8.19552264986353e-07,
      "loss": 1.9761,
      "num_input_tokens_seen": 216678320,
      "step": 212800
    },
    {
      "epoch": 4.191687503691599,
      "grad_norm": 2.236013889312744,
      "learning_rate": 8.193936603117741e-07,
      "loss": 1.9775,
      "num_input_tokens_seen": 216780072,
      "step": 212900
    },
    {
      "epoch": 4.193656356441101,
      "grad_norm": 1.5717908143997192,
      "learning_rate": 8.192350013269975e-07,
      "loss": 2.0049,
      "num_input_tokens_seen": 216882472,
      "step": 213000
    },
    {
      "epoch": 4.195625209190605,
      "grad_norm": 1.8222144842147827,
      "learning_rate": 8.190762880590013e-07,
      "loss": 2.0121,
      "num_input_tokens_seen": 216984072,
      "step": 213100
    },
    {
      "epoch": 4.197594061940108,
      "grad_norm": 2.036623477935791,
      "learning_rate": 8.189175205347735e-07,
      "loss": 2.0088,
      "num_input_tokens_seen": 217085240,
      "step": 213200
    },
    {
      "epoch": 4.1995629146896105,
      "grad_norm": 1.8548247814178467,
      "learning_rate": 8.187586987813113e-07,
      "loss": 1.9197,
      "num_input_tokens_seen": 217186816,
      "step": 213300
    },
    {
      "epoch": 4.201531767439113,
      "grad_norm": 1.9620369672775269,
      "learning_rate": 8.185998228256209e-07,
      "loss": 1.9741,
      "num_input_tokens_seen": 217288736,
      "step": 213400
    },
    {
      "epoch": 4.203500620188616,
      "grad_norm": 2.084591865539551,
      "learning_rate": 8.184408926947179e-07,
      "loss": 1.9746,
      "num_input_tokens_seen": 217391136,
      "step": 213500
    },
    {
      "epoch": 4.205469472938119,
      "grad_norm": 1.709738850593567,
      "learning_rate": 8.182819084156268e-07,
      "loss": 1.955,
      "num_input_tokens_seen": 217493536,
      "step": 213600
    },
    {
      "epoch": 4.2074383256876215,
      "grad_norm": 2.3697879314422607,
      "learning_rate": 8.181228700153817e-07,
      "loss": 1.9938,
      "num_input_tokens_seen": 217595096,
      "step": 213700
    },
    {
      "epoch": 4.209407178437124,
      "grad_norm": 1.8364559412002563,
      "learning_rate": 8.179637775210258e-07,
      "loss": 1.9411,
      "num_input_tokens_seen": 217697496,
      "step": 213800
    },
    {
      "epoch": 4.211376031186628,
      "grad_norm": 2.0698494911193848,
      "learning_rate": 8.178046309596111e-07,
      "loss": 2.0071,
      "num_input_tokens_seen": 217799896,
      "step": 213900
    },
    {
      "epoch": 4.213344883936131,
      "grad_norm": 2.2861826419830322,
      "learning_rate": 8.176454303581998e-07,
      "loss": 2.0019,
      "num_input_tokens_seen": 217902296,
      "step": 214000
    },
    {
      "epoch": 4.2153137366856335,
      "grad_norm": 2.09855055809021,
      "learning_rate": 8.17486175743862e-07,
      "loss": 1.9549,
      "num_input_tokens_seen": 218004696,
      "step": 214100
    },
    {
      "epoch": 4.217282589435136,
      "grad_norm": 1.7645584344863892,
      "learning_rate": 8.173268671436779e-07,
      "loss": 1.9808,
      "num_input_tokens_seen": 218105096,
      "step": 214200
    },
    {
      "epoch": 4.219251442184639,
      "grad_norm": 2.2229950428009033,
      "learning_rate": 8.171675045847363e-07,
      "loss": 2.0014,
      "num_input_tokens_seen": 218207496,
      "step": 214300
    },
    {
      "epoch": 4.221220294934142,
      "grad_norm": 2.069821834564209,
      "learning_rate": 8.170080880941359e-07,
      "loss": 1.9452,
      "num_input_tokens_seen": 218309896,
      "step": 214400
    },
    {
      "epoch": 4.2231891476836445,
      "grad_norm": 2.1184885501861572,
      "learning_rate": 8.168486176989838e-07,
      "loss": 2.0071,
      "num_input_tokens_seen": 218411864,
      "step": 214500
    },
    {
      "epoch": 4.225158000433147,
      "grad_norm": 2.1211719512939453,
      "learning_rate": 8.166890934263968e-07,
      "loss": 1.9598,
      "num_input_tokens_seen": 218513744,
      "step": 214600
    },
    {
      "epoch": 4.22712685318265,
      "grad_norm": 1.9384825229644775,
      "learning_rate": 8.165295153035003e-07,
      "loss": 1.922,
      "num_input_tokens_seen": 218616144,
      "step": 214700
    },
    {
      "epoch": 4.229095705932154,
      "grad_norm": 2.017470598220825,
      "learning_rate": 8.163698833574297e-07,
      "loss": 2.0144,
      "num_input_tokens_seen": 218718544,
      "step": 214800
    },
    {
      "epoch": 4.231064558681656,
      "grad_norm": 2.218912124633789,
      "learning_rate": 8.162101976153286e-07,
      "loss": 1.997,
      "num_input_tokens_seen": 218820944,
      "step": 214900
    },
    {
      "epoch": 4.233033411431159,
      "grad_norm": 2.1257431507110596,
      "learning_rate": 8.160504581043506e-07,
      "loss": 1.987,
      "num_input_tokens_seen": 218923128,
      "step": 215000
    },
    {
      "epoch": 4.235002264180662,
      "grad_norm": 1.788659691810608,
      "learning_rate": 8.158906648516578e-07,
      "loss": 1.928,
      "num_input_tokens_seen": 219024752,
      "step": 215100
    },
    {
      "epoch": 4.236971116930165,
      "grad_norm": 1.8117871284484863,
      "learning_rate": 8.157308178844218e-07,
      "loss": 2.0293,
      "num_input_tokens_seen": 219126840,
      "step": 215200
    },
    {
      "epoch": 4.2389399696796675,
      "grad_norm": 1.7583394050598145,
      "learning_rate": 8.155709172298233e-07,
      "loss": 2.0154,
      "num_input_tokens_seen": 219229240,
      "step": 215300
    },
    {
      "epoch": 4.24090882242917,
      "grad_norm": 1.9496679306030273,
      "learning_rate": 8.154109629150518e-07,
      "loss": 1.9819,
      "num_input_tokens_seen": 219331224,
      "step": 215400
    },
    {
      "epoch": 4.242877675178673,
      "grad_norm": 1.9294391870498657,
      "learning_rate": 8.152509549673066e-07,
      "loss": 1.9872,
      "num_input_tokens_seen": 219432704,
      "step": 215500
    },
    {
      "epoch": 4.244846527928177,
      "grad_norm": 1.7344541549682617,
      "learning_rate": 8.150908934137952e-07,
      "loss": 1.9877,
      "num_input_tokens_seen": 219535104,
      "step": 215600
    },
    {
      "epoch": 4.246815380677679,
      "grad_norm": 3.2139742374420166,
      "learning_rate": 8.149307782817352e-07,
      "loss": 1.9592,
      "num_input_tokens_seen": 219636744,
      "step": 215700
    },
    {
      "epoch": 4.248784233427182,
      "grad_norm": 1.9002854824066162,
      "learning_rate": 8.147706095983525e-07,
      "loss": 1.9351,
      "num_input_tokens_seen": 219739144,
      "step": 215800
    },
    {
      "epoch": 4.250753086176685,
      "grad_norm": 1.993449091911316,
      "learning_rate": 8.146103873908825e-07,
      "loss": 1.9986,
      "num_input_tokens_seen": 219841544,
      "step": 215900
    },
    {
      "epoch": 4.252721938926188,
      "grad_norm": 2.0600881576538086,
      "learning_rate": 8.144501116865696e-07,
      "loss": 1.9911,
      "num_input_tokens_seen": 219943944,
      "step": 216000
    },
    {
      "epoch": 4.25469079167569,
      "grad_norm": 1.9502677917480469,
      "learning_rate": 8.142897825126675e-07,
      "loss": 2.0185,
      "num_input_tokens_seen": 220046248,
      "step": 216100
    },
    {
      "epoch": 4.256659644425193,
      "grad_norm": 2.3295066356658936,
      "learning_rate": 8.141293998964388e-07,
      "loss": 1.9602,
      "num_input_tokens_seen": 220148648,
      "step": 216200
    },
    {
      "epoch": 4.258628497174696,
      "grad_norm": 1.9984863996505737,
      "learning_rate": 8.13968963865155e-07,
      "loss": 1.9947,
      "num_input_tokens_seen": 220250280,
      "step": 216300
    },
    {
      "epoch": 4.2605973499242,
      "grad_norm": 1.7733039855957031,
      "learning_rate": 8.138084744460971e-07,
      "loss": 2.0138,
      "num_input_tokens_seen": 220351728,
      "step": 216400
    },
    {
      "epoch": 4.262566202673702,
      "grad_norm": 1.9590157270431519,
      "learning_rate": 8.136479316665549e-07,
      "loss": 1.9595,
      "num_input_tokens_seen": 220453528,
      "step": 216500
    },
    {
      "epoch": 4.264535055423205,
      "grad_norm": 1.9967883825302124,
      "learning_rate": 8.134873355538274e-07,
      "loss": 1.9962,
      "num_input_tokens_seen": 220554864,
      "step": 216600
    },
    {
      "epoch": 4.266503908172708,
      "grad_norm": 1.8763107061386108,
      "learning_rate": 8.133266861352226e-07,
      "loss": 2.0066,
      "num_input_tokens_seen": 220656696,
      "step": 216700
    },
    {
      "epoch": 4.268472760922211,
      "grad_norm": 2.052912712097168,
      "learning_rate": 8.131659834380574e-07,
      "loss": 1.9398,
      "num_input_tokens_seen": 220759096,
      "step": 216800
    },
    {
      "epoch": 4.270441613671713,
      "grad_norm": 1.9175455570220947,
      "learning_rate": 8.130052274896581e-07,
      "loss": 1.9424,
      "num_input_tokens_seen": 220860648,
      "step": 216900
    },
    {
      "epoch": 4.272410466421216,
      "grad_norm": 1.7989146709442139,
      "learning_rate": 8.128444183173599e-07,
      "loss": 2.0372,
      "num_input_tokens_seen": 220963048,
      "step": 217000
    },
    {
      "epoch": 4.274379319170719,
      "grad_norm": 1.8976370096206665,
      "learning_rate": 8.12683555948507e-07,
      "loss": 1.9601,
      "num_input_tokens_seen": 221065448,
      "step": 217100
    },
    {
      "epoch": 4.2763481719202225,
      "grad_norm": 1.6844745874404907,
      "learning_rate": 8.125226404104528e-07,
      "loss": 1.9583,
      "num_input_tokens_seen": 221167848,
      "step": 217200
    },
    {
      "epoch": 4.278317024669725,
      "grad_norm": 1.6370481252670288,
      "learning_rate": 8.123616717305595e-07,
      "loss": 1.996,
      "num_input_tokens_seen": 221269760,
      "step": 217300
    },
    {
      "epoch": 4.280285877419228,
      "grad_norm": 2.272353410720825,
      "learning_rate": 8.122006499361984e-07,
      "loss": 1.9959,
      "num_input_tokens_seen": 221371328,
      "step": 217400
    },
    {
      "epoch": 4.282254730168731,
      "grad_norm": 2.06241774559021,
      "learning_rate": 8.120395750547501e-07,
      "loss": 1.9787,
      "num_input_tokens_seen": 221473728,
      "step": 217500
    },
    {
      "epoch": 4.284223582918234,
      "grad_norm": 1.9038654565811157,
      "learning_rate": 8.11878447113604e-07,
      "loss": 1.9408,
      "num_input_tokens_seen": 221575632,
      "step": 217600
    },
    {
      "epoch": 4.286192435667736,
      "grad_norm": 1.8895894289016724,
      "learning_rate": 8.117172661401584e-07,
      "loss": 1.9984,
      "num_input_tokens_seen": 221677184,
      "step": 217700
    },
    {
      "epoch": 4.288161288417239,
      "grad_norm": 2.1330461502075195,
      "learning_rate": 8.115560321618206e-07,
      "loss": 1.9746,
      "num_input_tokens_seen": 221779584,
      "step": 217800
    },
    {
      "epoch": 4.290130141166742,
      "grad_norm": 2.0095295906066895,
      "learning_rate": 8.113947452060076e-07,
      "loss": 2.002,
      "num_input_tokens_seen": 221881984,
      "step": 217900
    },
    {
      "epoch": 4.292098993916245,
      "grad_norm": 1.8864161968231201,
      "learning_rate": 8.112334053001444e-07,
      "loss": 1.985,
      "num_input_tokens_seen": 221983856,
      "step": 218000
    },
    {
      "epoch": 4.294067846665748,
      "grad_norm": 2.039607286453247,
      "learning_rate": 8.110720124716659e-07,
      "loss": 1.985,
      "num_input_tokens_seen": 222085744,
      "step": 218100
    },
    {
      "epoch": 4.296036699415251,
      "grad_norm": 1.9328733682632446,
      "learning_rate": 8.109105667480153e-07,
      "loss": 1.9594,
      "num_input_tokens_seen": 222187536,
      "step": 218200
    },
    {
      "epoch": 4.298005552164754,
      "grad_norm": 2.0978965759277344,
      "learning_rate": 8.107490681566451e-07,
      "loss": 1.982,
      "num_input_tokens_seen": 222289352,
      "step": 218300
    },
    {
      "epoch": 4.2999744049142565,
      "grad_norm": 1.7404478788375854,
      "learning_rate": 8.10587516725017e-07,
      "loss": 1.9871,
      "num_input_tokens_seen": 222391752,
      "step": 218400
    },
    {
      "epoch": 4.301943257663759,
      "grad_norm": 1.785993218421936,
      "learning_rate": 8.104259124806012e-07,
      "loss": 1.9659,
      "num_input_tokens_seen": 222494152,
      "step": 218500
    },
    {
      "epoch": 4.303912110413262,
      "grad_norm": 2.1978917121887207,
      "learning_rate": 8.102642554508772e-07,
      "loss": 1.9512,
      "num_input_tokens_seen": 222595672,
      "step": 218600
    },
    {
      "epoch": 4.305880963162765,
      "grad_norm": 1.7965723276138306,
      "learning_rate": 8.101025456633334e-07,
      "loss": 1.9729,
      "num_input_tokens_seen": 222698072,
      "step": 218700
    },
    {
      "epoch": 4.307849815912268,
      "grad_norm": 2.0766849517822266,
      "learning_rate": 8.099407831454674e-07,
      "loss": 2.0122,
      "num_input_tokens_seen": 222798968,
      "step": 218800
    },
    {
      "epoch": 4.309818668661771,
      "grad_norm": 2.1054444313049316,
      "learning_rate": 8.097789679247853e-07,
      "loss": 1.959,
      "num_input_tokens_seen": 222901368,
      "step": 218900
    },
    {
      "epoch": 4.311787521411274,
      "grad_norm": 1.8245131969451904,
      "learning_rate": 8.096171000288025e-07,
      "loss": 2.0052,
      "num_input_tokens_seen": 223003768,
      "step": 219000
    },
    {
      "epoch": 4.313756374160777,
      "grad_norm": 1.8209213018417358,
      "learning_rate": 8.094551794850432e-07,
      "loss": 2.0242,
      "num_input_tokens_seen": 223105800,
      "step": 219100
    },
    {
      "epoch": 4.3157252269102795,
      "grad_norm": 1.971667766571045,
      "learning_rate": 8.092932063210409e-07,
      "loss": 1.9974,
      "num_input_tokens_seen": 223208200,
      "step": 219200
    },
    {
      "epoch": 4.317694079659782,
      "grad_norm": 2.0280184745788574,
      "learning_rate": 8.091311805643375e-07,
      "loss": 2.0031,
      "num_input_tokens_seen": 223310600,
      "step": 219300
    },
    {
      "epoch": 4.319662932409285,
      "grad_norm": 2.1408958435058594,
      "learning_rate": 8.089691022424841e-07,
      "loss": 1.9802,
      "num_input_tokens_seen": 223413000,
      "step": 219400
    },
    {
      "epoch": 4.321631785158788,
      "grad_norm": 2.1364798545837402,
      "learning_rate": 8.088069713830408e-07,
      "loss": 1.9982,
      "num_input_tokens_seen": 223514864,
      "step": 219500
    },
    {
      "epoch": 4.3236006379082905,
      "grad_norm": 2.9700281620025635,
      "learning_rate": 8.086447880135767e-07,
      "loss": 1.9846,
      "num_input_tokens_seen": 223616984,
      "step": 219600
    },
    {
      "epoch": 4.325569490657793,
      "grad_norm": 2.3307693004608154,
      "learning_rate": 8.084825521616696e-07,
      "loss": 1.9966,
      "num_input_tokens_seen": 223717896,
      "step": 219700
    },
    {
      "epoch": 4.327538343407297,
      "grad_norm": 5.6779255867004395,
      "learning_rate": 8.083202638549063e-07,
      "loss": 2.0122,
      "num_input_tokens_seen": 223819536,
      "step": 219800
    },
    {
      "epoch": 4.3295071961568,
      "grad_norm": 1.9002426862716675,
      "learning_rate": 8.081579231208827e-07,
      "loss": 1.9533,
      "num_input_tokens_seen": 223921936,
      "step": 219900
    },
    {
      "epoch": 4.3314760489063024,
      "grad_norm": 1.9026561975479126,
      "learning_rate": 8.079955299872034e-07,
      "loss": 1.956,
      "num_input_tokens_seen": 224023568,
      "step": 220000
    },
    {
      "epoch": 4.333444901655805,
      "grad_norm": 1.7311137914657593,
      "learning_rate": 8.078330844814819e-07,
      "loss": 1.9851,
      "num_input_tokens_seen": 224125272,
      "step": 220100
    },
    {
      "epoch": 4.335413754405308,
      "grad_norm": 1.9070080518722534,
      "learning_rate": 8.076705866313408e-07,
      "loss": 2.0065,
      "num_input_tokens_seen": 224227672,
      "step": 220200
    },
    {
      "epoch": 4.337382607154811,
      "grad_norm": 1.7994564771652222,
      "learning_rate": 8.075080364644116e-07,
      "loss": 2.0414,
      "num_input_tokens_seen": 224329800,
      "step": 220300
    },
    {
      "epoch": 4.3393514599043135,
      "grad_norm": 1.876141905784607,
      "learning_rate": 8.073454340083342e-07,
      "loss": 2.0027,
      "num_input_tokens_seen": 224431352,
      "step": 220400
    },
    {
      "epoch": 4.341320312653816,
      "grad_norm": 1.9209034442901611,
      "learning_rate": 8.071827792907582e-07,
      "loss": 1.9898,
      "num_input_tokens_seen": 224532992,
      "step": 220500
    },
    {
      "epoch": 4.34328916540332,
      "grad_norm": 2.1137211322784424,
      "learning_rate": 8.070200723393414e-07,
      "loss": 2.0278,
      "num_input_tokens_seen": 224634448,
      "step": 220600
    },
    {
      "epoch": 4.345258018152823,
      "grad_norm": 1.9331674575805664,
      "learning_rate": 8.068573131817508e-07,
      "loss": 2.0014,
      "num_input_tokens_seen": 224735272,
      "step": 220700
    },
    {
      "epoch": 4.347226870902325,
      "grad_norm": 1.8493545055389404,
      "learning_rate": 8.066945018456623e-07,
      "loss": 1.9861,
      "num_input_tokens_seen": 224836856,
      "step": 220800
    },
    {
      "epoch": 4.349195723651828,
      "grad_norm": 1.6915216445922852,
      "learning_rate": 8.065316383587606e-07,
      "loss": 1.991,
      "num_input_tokens_seen": 224938760,
      "step": 220900
    },
    {
      "epoch": 4.351164576401331,
      "grad_norm": 2.0094316005706787,
      "learning_rate": 8.06368722748739e-07,
      "loss": 1.9606,
      "num_input_tokens_seen": 225040864,
      "step": 221000
    },
    {
      "epoch": 4.353133429150834,
      "grad_norm": 2.79548978805542,
      "learning_rate": 8.062057550433003e-07,
      "loss": 1.9728,
      "num_input_tokens_seen": 225143264,
      "step": 221100
    },
    {
      "epoch": 4.355102281900336,
      "grad_norm": 1.8829996585845947,
      "learning_rate": 8.060427352701557e-07,
      "loss": 2.0219,
      "num_input_tokens_seen": 225245080,
      "step": 221200
    },
    {
      "epoch": 4.357071134649839,
      "grad_norm": 2.938256025314331,
      "learning_rate": 8.058796634570251e-07,
      "loss": 1.9436,
      "num_input_tokens_seen": 225346856,
      "step": 221300
    },
    {
      "epoch": 4.359039987399342,
      "grad_norm": 1.8306713104248047,
      "learning_rate": 8.057165396316377e-07,
      "loss": 2.0174,
      "num_input_tokens_seen": 225448656,
      "step": 221400
    },
    {
      "epoch": 4.361008840148846,
      "grad_norm": 2.3685410022735596,
      "learning_rate": 8.055533638217312e-07,
      "loss": 1.9718,
      "num_input_tokens_seen": 225551056,
      "step": 221500
    },
    {
      "epoch": 4.362977692898348,
      "grad_norm": 1.8687312602996826,
      "learning_rate": 8.053901360550523e-07,
      "loss": 1.9771,
      "num_input_tokens_seen": 225652584,
      "step": 221600
    },
    {
      "epoch": 4.364946545647851,
      "grad_norm": 2.394420862197876,
      "learning_rate": 8.052268563593567e-07,
      "loss": 2.0222,
      "num_input_tokens_seen": 225754984,
      "step": 221700
    },
    {
      "epoch": 4.366915398397354,
      "grad_norm": 2.370257616043091,
      "learning_rate": 8.050635247624086e-07,
      "loss": 1.9827,
      "num_input_tokens_seen": 225856552,
      "step": 221800
    },
    {
      "epoch": 4.368884251146857,
      "grad_norm": 1.8775486946105957,
      "learning_rate": 8.04900141291981e-07,
      "loss": 1.9879,
      "num_input_tokens_seen": 225958496,
      "step": 221900
    },
    {
      "epoch": 4.370853103896359,
      "grad_norm": 1.8185542821884155,
      "learning_rate": 8.047367059758559e-07,
      "loss": 2.0074,
      "num_input_tokens_seen": 226060896,
      "step": 222000
    },
    {
      "epoch": 4.372821956645862,
      "grad_norm": 2.056126117706299,
      "learning_rate": 8.045732188418244e-07,
      "loss": 1.9692,
      "num_input_tokens_seen": 226161880,
      "step": 222100
    },
    {
      "epoch": 4.374790809395365,
      "grad_norm": 1.9742474555969238,
      "learning_rate": 8.04409679917686e-07,
      "loss": 1.9683,
      "num_input_tokens_seen": 226263008,
      "step": 222200
    },
    {
      "epoch": 4.376759662144869,
      "grad_norm": 1.935840129852295,
      "learning_rate": 8.04246089231249e-07,
      "loss": 1.9961,
      "num_input_tokens_seen": 226364176,
      "step": 222300
    },
    {
      "epoch": 4.378728514894371,
      "grad_norm": 1.9568092823028564,
      "learning_rate": 8.040824468103306e-07,
      "loss": 1.982,
      "num_input_tokens_seen": 226466576,
      "step": 222400
    },
    {
      "epoch": 4.380697367643874,
      "grad_norm": 1.8954073190689087,
      "learning_rate": 8.039187526827568e-07,
      "loss": 1.996,
      "num_input_tokens_seen": 226568976,
      "step": 222500
    },
    {
      "epoch": 4.382666220393377,
      "grad_norm": 2.0917599201202393,
      "learning_rate": 8.037550068763625e-07,
      "loss": 1.9948,
      "num_input_tokens_seen": 226671376,
      "step": 222600
    },
    {
      "epoch": 4.38463507314288,
      "grad_norm": 1.7765631675720215,
      "learning_rate": 8.035912094189914e-07,
      "loss": 1.9881,
      "num_input_tokens_seen": 226773776,
      "step": 222700
    },
    {
      "epoch": 4.386603925892382,
      "grad_norm": 5.36970853805542,
      "learning_rate": 8.034273603384956e-07,
      "loss": 1.9644,
      "num_input_tokens_seen": 226876176,
      "step": 222800
    },
    {
      "epoch": 4.388572778641885,
      "grad_norm": 1.9368995428085327,
      "learning_rate": 8.032634596627366e-07,
      "loss": 1.934,
      "num_input_tokens_seen": 226976896,
      "step": 222900
    },
    {
      "epoch": 4.390541631391388,
      "grad_norm": 2.9717519283294678,
      "learning_rate": 8.030995074195839e-07,
      "loss": 2.0059,
      "num_input_tokens_seen": 227078272,
      "step": 223000
    },
    {
      "epoch": 4.3925104841408915,
      "grad_norm": 1.9175190925598145,
      "learning_rate": 8.029355036369164e-07,
      "loss": 1.9639,
      "num_input_tokens_seen": 227180088,
      "step": 223100
    },
    {
      "epoch": 4.394479336890394,
      "grad_norm": 1.7901525497436523,
      "learning_rate": 8.027714483426218e-07,
      "loss": 2.0324,
      "num_input_tokens_seen": 227281896,
      "step": 223200
    },
    {
      "epoch": 4.396448189639897,
      "grad_norm": 2.0427134037017822,
      "learning_rate": 8.02607341564596e-07,
      "loss": 2.0029,
      "num_input_tokens_seen": 227383400,
      "step": 223300
    },
    {
      "epoch": 4.3984170423894,
      "grad_norm": 1.894559383392334,
      "learning_rate": 8.02443183330744e-07,
      "loss": 1.9632,
      "num_input_tokens_seen": 227485800,
      "step": 223400
    },
    {
      "epoch": 4.4003858951389025,
      "grad_norm": 1.9449197053909302,
      "learning_rate": 8.022789736689796e-07,
      "loss": 1.97,
      "num_input_tokens_seen": 227587272,
      "step": 223500
    },
    {
      "epoch": 4.402354747888405,
      "grad_norm": 1.855339765548706,
      "learning_rate": 8.021147126072254e-07,
      "loss": 2.0101,
      "num_input_tokens_seen": 227688816,
      "step": 223600
    },
    {
      "epoch": 4.404323600637908,
      "grad_norm": 3.689824104309082,
      "learning_rate": 8.019504001734124e-07,
      "loss": 2.0184,
      "num_input_tokens_seen": 227789008,
      "step": 223700
    },
    {
      "epoch": 4.406292453387411,
      "grad_norm": 1.8142951726913452,
      "learning_rate": 8.017860363954807e-07,
      "loss": 1.9591,
      "num_input_tokens_seen": 227890552,
      "step": 223800
    },
    {
      "epoch": 4.4082613061369145,
      "grad_norm": 1.777241587638855,
      "learning_rate": 8.016216213013788e-07,
      "loss": 1.9961,
      "num_input_tokens_seen": 227992664,
      "step": 223900
    },
    {
      "epoch": 4.410230158886417,
      "grad_norm": 2.256772518157959,
      "learning_rate": 8.014571549190642e-07,
      "loss": 2.0172,
      "num_input_tokens_seen": 228094568,
      "step": 224000
    },
    {
      "epoch": 4.41219901163592,
      "grad_norm": 2.0180375576019287,
      "learning_rate": 8.012926372765029e-07,
      "loss": 2.0225,
      "num_input_tokens_seen": 228196104,
      "step": 224100
    },
    {
      "epoch": 4.414167864385423,
      "grad_norm": 2.085426092147827,
      "learning_rate": 8.011280684016699e-07,
      "loss": 1.9956,
      "num_input_tokens_seen": 228298504,
      "step": 224200
    },
    {
      "epoch": 4.4161367171349255,
      "grad_norm": 1.8967337608337402,
      "learning_rate": 8.009634483225489e-07,
      "loss": 1.9948,
      "num_input_tokens_seen": 228399888,
      "step": 224300
    },
    {
      "epoch": 4.418105569884428,
      "grad_norm": 2.135732650756836,
      "learning_rate": 8.007987770671318e-07,
      "loss": 2.0094,
      "num_input_tokens_seen": 228500976,
      "step": 224400
    },
    {
      "epoch": 4.420074422633931,
      "grad_norm": 1.991890788078308,
      "learning_rate": 8.006340546634197e-07,
      "loss": 1.9365,
      "num_input_tokens_seen": 228603376,
      "step": 224500
    },
    {
      "epoch": 4.422043275383434,
      "grad_norm": 2.9969050884246826,
      "learning_rate": 8.004692811394223e-07,
      "loss": 2.0129,
      "num_input_tokens_seen": 228705776,
      "step": 224600
    },
    {
      "epoch": 4.4240121281329365,
      "grad_norm": 1.8463053703308105,
      "learning_rate": 8.00304456523158e-07,
      "loss": 1.9556,
      "num_input_tokens_seen": 228808176,
      "step": 224700
    },
    {
      "epoch": 4.42598098088244,
      "grad_norm": 1.9226841926574707,
      "learning_rate": 8.001395808426537e-07,
      "loss": 1.9627,
      "num_input_tokens_seen": 228909720,
      "step": 224800
    },
    {
      "epoch": 4.427949833631943,
      "grad_norm": 1.9595040082931519,
      "learning_rate": 7.999746541259451e-07,
      "loss": 2.0559,
      "num_input_tokens_seen": 229012120,
      "step": 224900
    },
    {
      "epoch": 4.429918686381446,
      "grad_norm": 1.708908200263977,
      "learning_rate": 7.998096764010767e-07,
      "loss": 1.9864,
      "num_input_tokens_seen": 229113904,
      "step": 225000
    },
    {
      "epoch": 4.4318875391309485,
      "grad_norm": 1.9762471914291382,
      "learning_rate": 7.996446476961015e-07,
      "loss": 1.9841,
      "num_input_tokens_seen": 229216120,
      "step": 225100
    },
    {
      "epoch": 4.433856391880451,
      "grad_norm": 1.8990074396133423,
      "learning_rate": 7.994795680390814e-07,
      "loss": 1.9934,
      "num_input_tokens_seen": 229317288,
      "step": 225200
    },
    {
      "epoch": 4.435825244629954,
      "grad_norm": 2.5449118614196777,
      "learning_rate": 7.993144374580866e-07,
      "loss": 1.9653,
      "num_input_tokens_seen": 229418184,
      "step": 225300
    },
    {
      "epoch": 4.437794097379457,
      "grad_norm": 3.2167248725891113,
      "learning_rate": 7.991492559811964e-07,
      "loss": 1.9907,
      "num_input_tokens_seen": 229520584,
      "step": 225400
    },
    {
      "epoch": 4.4397629501289595,
      "grad_norm": 1.797454595565796,
      "learning_rate": 7.989840236364981e-07,
      "loss": 1.9663,
      "num_input_tokens_seen": 229622984,
      "step": 225500
    },
    {
      "epoch": 4.441731802878463,
      "grad_norm": 1.874466896057129,
      "learning_rate": 7.988187404520884e-07,
      "loss": 2.0149,
      "num_input_tokens_seen": 229724216,
      "step": 225600
    },
    {
      "epoch": 4.443700655627966,
      "grad_norm": 2.0009689331054688,
      "learning_rate": 7.986534064560723e-07,
      "loss": 2.0108,
      "num_input_tokens_seen": 229825784,
      "step": 225700
    },
    {
      "epoch": 4.445669508377469,
      "grad_norm": 1.888047695159912,
      "learning_rate": 7.984880216765635e-07,
      "loss": 2.0118,
      "num_input_tokens_seen": 229926592,
      "step": 225800
    },
    {
      "epoch": 4.447638361126971,
      "grad_norm": 1.9812793731689453,
      "learning_rate": 7.98322586141684e-07,
      "loss": 1.9749,
      "num_input_tokens_seen": 230027664,
      "step": 225900
    },
    {
      "epoch": 4.449607213876474,
      "grad_norm": 1.8331326246261597,
      "learning_rate": 7.98157099879565e-07,
      "loss": 2.0502,
      "num_input_tokens_seen": 230129584,
      "step": 226000
    },
    {
      "epoch": 4.451576066625977,
      "grad_norm": 1.9073522090911865,
      "learning_rate": 7.979915629183458e-07,
      "loss": 1.9923,
      "num_input_tokens_seen": 230230648,
      "step": 226100
    },
    {
      "epoch": 4.45354491937548,
      "grad_norm": 1.8403912782669067,
      "learning_rate": 7.978259752861747e-07,
      "loss": 1.9725,
      "num_input_tokens_seen": 230332176,
      "step": 226200
    },
    {
      "epoch": 4.4555137721249825,
      "grad_norm": 1.805899977684021,
      "learning_rate": 7.976603370112087e-07,
      "loss": 2.0041,
      "num_input_tokens_seen": 230433304,
      "step": 226300
    },
    {
      "epoch": 4.457482624874485,
      "grad_norm": 3.5743937492370605,
      "learning_rate": 7.974946481216128e-07,
      "loss": 2.0362,
      "num_input_tokens_seen": 230532976,
      "step": 226400
    },
    {
      "epoch": 4.459451477623989,
      "grad_norm": 2.0350921154022217,
      "learning_rate": 7.973289086455612e-07,
      "loss": 1.987,
      "num_input_tokens_seen": 230634808,
      "step": 226500
    },
    {
      "epoch": 4.461420330373492,
      "grad_norm": 1.9272820949554443,
      "learning_rate": 7.971631186112367e-07,
      "loss": 1.9851,
      "num_input_tokens_seen": 230736392,
      "step": 226600
    },
    {
      "epoch": 4.463389183122994,
      "grad_norm": 2.0048458576202393,
      "learning_rate": 7.969972780468299e-07,
      "loss": 1.9943,
      "num_input_tokens_seen": 230838792,
      "step": 226700
    },
    {
      "epoch": 4.465358035872497,
      "grad_norm": 1.8114912509918213,
      "learning_rate": 7.968313869805412e-07,
      "loss": 1.9732,
      "num_input_tokens_seen": 230941192,
      "step": 226800
    },
    {
      "epoch": 4.467326888622,
      "grad_norm": 2.2808680534362793,
      "learning_rate": 7.966654454405785e-07,
      "loss": 2.0134,
      "num_input_tokens_seen": 231042680,
      "step": 226900
    },
    {
      "epoch": 4.469295741371503,
      "grad_norm": 1.9703865051269531,
      "learning_rate": 7.964994534551592e-07,
      "loss": 1.9556,
      "num_input_tokens_seen": 231145080,
      "step": 227000
    },
    {
      "epoch": 4.471264594121005,
      "grad_norm": 2.1411333084106445,
      "learning_rate": 7.963334110525086e-07,
      "loss": 1.9598,
      "num_input_tokens_seen": 231247480,
      "step": 227100
    },
    {
      "epoch": 4.473233446870508,
      "grad_norm": 1.9498172998428345,
      "learning_rate": 7.961673182608609e-07,
      "loss": 2.0634,
      "num_input_tokens_seen": 231348544,
      "step": 227200
    },
    {
      "epoch": 4.475202299620012,
      "grad_norm": 1.9369093179702759,
      "learning_rate": 7.960011751084587e-07,
      "loss": 2.003,
      "num_input_tokens_seen": 231450944,
      "step": 227300
    },
    {
      "epoch": 4.477171152369515,
      "grad_norm": 1.945993423461914,
      "learning_rate": 7.958349816235533e-07,
      "loss": 1.9968,
      "num_input_tokens_seen": 231551472,
      "step": 227400
    },
    {
      "epoch": 4.479140005119017,
      "grad_norm": 1.9946876764297485,
      "learning_rate": 7.956687378344043e-07,
      "loss": 1.9972,
      "num_input_tokens_seen": 231652656,
      "step": 227500
    },
    {
      "epoch": 4.48110885786852,
      "grad_norm": 2.0146830081939697,
      "learning_rate": 7.955024437692803e-07,
      "loss": 1.9865,
      "num_input_tokens_seen": 231754496,
      "step": 227600
    },
    {
      "epoch": 4.483077710618023,
      "grad_norm": 1.8295978307724,
      "learning_rate": 7.953360994564581e-07,
      "loss": 1.9826,
      "num_input_tokens_seen": 231856896,
      "step": 227700
    },
    {
      "epoch": 4.485046563367526,
      "grad_norm": 1.7086530923843384,
      "learning_rate": 7.951697049242232e-07,
      "loss": 1.9761,
      "num_input_tokens_seen": 231958848,
      "step": 227800
    },
    {
      "epoch": 4.487015416117028,
      "grad_norm": 1.8575228452682495,
      "learning_rate": 7.950032602008695e-07,
      "loss": 1.9852,
      "num_input_tokens_seen": 232061248,
      "step": 227900
    },
    {
      "epoch": 4.488984268866531,
      "grad_norm": 2.264408826828003,
      "learning_rate": 7.948367653146997e-07,
      "loss": 1.9611,
      "num_input_tokens_seen": 232163648,
      "step": 228000
    },
    {
      "epoch": 4.490953121616034,
      "grad_norm": 1.838910698890686,
      "learning_rate": 7.946702202940246e-07,
      "loss": 1.9767,
      "num_input_tokens_seen": 232265032,
      "step": 228100
    },
    {
      "epoch": 4.4929219743655375,
      "grad_norm": 2.100069522857666,
      "learning_rate": 7.945036251671639e-07,
      "loss": 1.9756,
      "num_input_tokens_seen": 232367432,
      "step": 228200
    },
    {
      "epoch": 4.49489082711504,
      "grad_norm": 2.0678296089172363,
      "learning_rate": 7.943369799624458e-07,
      "loss": 1.9786,
      "num_input_tokens_seen": 232469728,
      "step": 228300
    },
    {
      "epoch": 4.496859679864543,
      "grad_norm": 2.136457920074463,
      "learning_rate": 7.941702847082066e-07,
      "loss": 1.9877,
      "num_input_tokens_seen": 232569256,
      "step": 228400
    },
    {
      "epoch": 4.498828532614046,
      "grad_norm": 1.9161975383758545,
      "learning_rate": 7.940035394327918e-07,
      "loss": 1.9814,
      "num_input_tokens_seen": 232670504,
      "step": 228500
    },
    {
      "epoch": 4.500797385363549,
      "grad_norm": 1.863601803779602,
      "learning_rate": 7.938367441645548e-07,
      "loss": 1.9505,
      "num_input_tokens_seen": 232772904,
      "step": 228600
    },
    {
      "epoch": 4.502766238113051,
      "grad_norm": 1.7487550973892212,
      "learning_rate": 7.936698989318579e-07,
      "loss": 1.9631,
      "num_input_tokens_seen": 232875304,
      "step": 228700
    },
    {
      "epoch": 4.504735090862554,
      "grad_norm": 1.996494174003601,
      "learning_rate": 7.935030037630715e-07,
      "loss": 2.028,
      "num_input_tokens_seen": 232977704,
      "step": 228800
    },
    {
      "epoch": 4.506703943612058,
      "grad_norm": 1.7284367084503174,
      "learning_rate": 7.93336058686575e-07,
      "loss": 2.0266,
      "num_input_tokens_seen": 233079184,
      "step": 228900
    },
    {
      "epoch": 4.5086727963615605,
      "grad_norm": 1.7391117811203003,
      "learning_rate": 7.931690637307556e-07,
      "loss": 2.0386,
      "num_input_tokens_seen": 233181168,
      "step": 229000
    },
    {
      "epoch": 4.510641649111063,
      "grad_norm": 1.9778952598571777,
      "learning_rate": 7.930020189240098e-07,
      "loss": 2.0128,
      "num_input_tokens_seen": 233283568,
      "step": 229100
    },
    {
      "epoch": 4.512610501860566,
      "grad_norm": 1.9555339813232422,
      "learning_rate": 7.928349242947417e-07,
      "loss": 1.9569,
      "num_input_tokens_seen": 233385224,
      "step": 229200
    },
    {
      "epoch": 4.514579354610069,
      "grad_norm": 1.83670973777771,
      "learning_rate": 7.926677798713648e-07,
      "loss": 1.9724,
      "num_input_tokens_seen": 233487624,
      "step": 229300
    },
    {
      "epoch": 4.5165482073595715,
      "grad_norm": 1.8802381753921509,
      "learning_rate": 7.925005856823005e-07,
      "loss": 1.9751,
      "num_input_tokens_seen": 233590024,
      "step": 229400
    },
    {
      "epoch": 4.518517060109074,
      "grad_norm": 2.3416171073913574,
      "learning_rate": 7.923333417559785e-07,
      "loss": 1.9788,
      "num_input_tokens_seen": 233690840,
      "step": 229500
    },
    {
      "epoch": 4.520485912858577,
      "grad_norm": 1.822119951248169,
      "learning_rate": 7.921660481208375e-07,
      "loss": 2.0116,
      "num_input_tokens_seen": 233793240,
      "step": 229600
    },
    {
      "epoch": 4.52245476560808,
      "grad_norm": 2.103149890899658,
      "learning_rate": 7.919987048053243e-07,
      "loss": 2.0102,
      "num_input_tokens_seen": 233894824,
      "step": 229700
    },
    {
      "epoch": 4.5244236183575826,
      "grad_norm": 1.893359899520874,
      "learning_rate": 7.91831311837894e-07,
      "loss": 2.0355,
      "num_input_tokens_seen": 233996416,
      "step": 229800
    },
    {
      "epoch": 4.526392471107086,
      "grad_norm": 1.7088468074798584,
      "learning_rate": 7.916638692470107e-07,
      "loss": 2.0399,
      "num_input_tokens_seen": 234098728,
      "step": 229900
    },
    {
      "epoch": 4.528361323856589,
      "grad_norm": 1.9777233600616455,
      "learning_rate": 7.914963770611464e-07,
      "loss": 1.9528,
      "num_input_tokens_seen": 234200336,
      "step": 230000
    },
    {
      "epoch": 4.530330176606092,
      "grad_norm": 1.9436982870101929,
      "learning_rate": 7.913288353087817e-07,
      "loss": 2.0033,
      "num_input_tokens_seen": 234302736,
      "step": 230100
    },
    {
      "epoch": 4.5322990293555945,
      "grad_norm": 2.070924997329712,
      "learning_rate": 7.911612440184057e-07,
      "loss": 1.9786,
      "num_input_tokens_seen": 234403176,
      "step": 230200
    },
    {
      "epoch": 4.534267882105097,
      "grad_norm": 2.2279651165008545,
      "learning_rate": 7.909936032185161e-07,
      "loss": 2.056,
      "num_input_tokens_seen": 234505000,
      "step": 230300
    },
    {
      "epoch": 4.5362367348546,
      "grad_norm": 2.046231269836426,
      "learning_rate": 7.908259129376185e-07,
      "loss": 1.985,
      "num_input_tokens_seen": 234605928,
      "step": 230400
    },
    {
      "epoch": 4.538205587604103,
      "grad_norm": 1.8107647895812988,
      "learning_rate": 7.906581732042275e-07,
      "loss": 1.9942,
      "num_input_tokens_seen": 234707848,
      "step": 230500
    },
    {
      "epoch": 4.540174440353606,
      "grad_norm": 1.6536964178085327,
      "learning_rate": 7.904903840468655e-07,
      "loss": 1.9839,
      "num_input_tokens_seen": 234809720,
      "step": 230600
    },
    {
      "epoch": 4.542143293103109,
      "grad_norm": 1.911716103553772,
      "learning_rate": 7.903225454940639e-07,
      "loss": 1.955,
      "num_input_tokens_seen": 234911384,
      "step": 230700
    },
    {
      "epoch": 4.544112145852612,
      "grad_norm": 1.9053138494491577,
      "learning_rate": 7.901546575743621e-07,
      "loss": 2.0247,
      "num_input_tokens_seen": 235013416,
      "step": 230800
    },
    {
      "epoch": 4.546080998602115,
      "grad_norm": 1.8903558254241943,
      "learning_rate": 7.89986720316308e-07,
      "loss": 1.9646,
      "num_input_tokens_seen": 235115008,
      "step": 230900
    },
    {
      "epoch": 4.548049851351617,
      "grad_norm": 1.9929543733596802,
      "learning_rate": 7.89818733748458e-07,
      "loss": 1.961,
      "num_input_tokens_seen": 235217408,
      "step": 231000
    },
    {
      "epoch": 4.55001870410112,
      "grad_norm": 1.9232136011123657,
      "learning_rate": 7.896506978993767e-07,
      "loss": 2.0015,
      "num_input_tokens_seen": 235319256,
      "step": 231100
    },
    {
      "epoch": 4.551987556850623,
      "grad_norm": 1.882171392440796,
      "learning_rate": 7.894826127976374e-07,
      "loss": 2.015,
      "num_input_tokens_seen": 235420928,
      "step": 231200
    },
    {
      "epoch": 4.553956409600126,
      "grad_norm": 2.100560426712036,
      "learning_rate": 7.893144784718213e-07,
      "loss": 1.997,
      "num_input_tokens_seen": 235522856,
      "step": 231300
    },
    {
      "epoch": 4.5559252623496285,
      "grad_norm": 1.7758921384811401,
      "learning_rate": 7.891462949505185e-07,
      "loss": 1.9465,
      "num_input_tokens_seen": 235625256,
      "step": 231400
    },
    {
      "epoch": 4.557894115099132,
      "grad_norm": 1.914080023765564,
      "learning_rate": 7.889780622623268e-07,
      "loss": 2.0049,
      "num_input_tokens_seen": 235725664,
      "step": 231500
    },
    {
      "epoch": 4.559862967848635,
      "grad_norm": 1.8459874391555786,
      "learning_rate": 7.888097804358531e-07,
      "loss": 2.0241,
      "num_input_tokens_seen": 235826840,
      "step": 231600
    },
    {
      "epoch": 4.561831820598138,
      "grad_norm": 1.9404900074005127,
      "learning_rate": 7.88641449499712e-07,
      "loss": 1.9947,
      "num_input_tokens_seen": 235927976,
      "step": 231700
    },
    {
      "epoch": 4.56380067334764,
      "grad_norm": 2.0413923263549805,
      "learning_rate": 7.884730694825271e-07,
      "loss": 2.003,
      "num_input_tokens_seen": 236029696,
      "step": 231800
    },
    {
      "epoch": 4.565769526097143,
      "grad_norm": 1.9381991624832153,
      "learning_rate": 7.883046404129299e-07,
      "loss": 1.9189,
      "num_input_tokens_seen": 236132096,
      "step": 231900
    },
    {
      "epoch": 4.567738378846646,
      "grad_norm": 1.8498059511184692,
      "learning_rate": 7.8813616231956e-07,
      "loss": 1.9752,
      "num_input_tokens_seen": 236234496,
      "step": 232000
    },
    {
      "epoch": 4.569707231596149,
      "grad_norm": 1.8928567171096802,
      "learning_rate": 7.879676352310662e-07,
      "loss": 1.9923,
      "num_input_tokens_seen": 236336752,
      "step": 232100
    },
    {
      "epoch": 4.571676084345651,
      "grad_norm": 2.515831470489502,
      "learning_rate": 7.877990591761046e-07,
      "loss": 1.9548,
      "num_input_tokens_seen": 236439152,
      "step": 232200
    },
    {
      "epoch": 4.573644937095155,
      "grad_norm": 1.9485191106796265,
      "learning_rate": 7.876304341833407e-07,
      "loss": 1.9822,
      "num_input_tokens_seen": 236540784,
      "step": 232300
    },
    {
      "epoch": 4.575613789844658,
      "grad_norm": 1.871766448020935,
      "learning_rate": 7.874617602814472e-07,
      "loss": 1.9805,
      "num_input_tokens_seen": 236642752,
      "step": 232400
    },
    {
      "epoch": 4.577582642594161,
      "grad_norm": 1.8358638286590576,
      "learning_rate": 7.87293037499106e-07,
      "loss": 2.002,
      "num_input_tokens_seen": 236745152,
      "step": 232500
    },
    {
      "epoch": 4.579551495343663,
      "grad_norm": 2.0074095726013184,
      "learning_rate": 7.87124265865007e-07,
      "loss": 1.9208,
      "num_input_tokens_seen": 236846792,
      "step": 232600
    },
    {
      "epoch": 4.581520348093166,
      "grad_norm": 1.966914176940918,
      "learning_rate": 7.869554454078482e-07,
      "loss": 1.9783,
      "num_input_tokens_seen": 236949192,
      "step": 232700
    },
    {
      "epoch": 4.583489200842669,
      "grad_norm": 1.9215971231460571,
      "learning_rate": 7.867865761563364e-07,
      "loss": 1.9966,
      "num_input_tokens_seen": 237051000,
      "step": 232800
    },
    {
      "epoch": 4.585458053592172,
      "grad_norm": 2.2058589458465576,
      "learning_rate": 7.866176581391861e-07,
      "loss": 1.9756,
      "num_input_tokens_seen": 237152552,
      "step": 232900
    },
    {
      "epoch": 4.587426906341674,
      "grad_norm": 1.9091542959213257,
      "learning_rate": 7.864486913851203e-07,
      "loss": 2.0251,
      "num_input_tokens_seen": 237253944,
      "step": 233000
    },
    {
      "epoch": 4.589395759091177,
      "grad_norm": 2.0750956535339355,
      "learning_rate": 7.862796759228707e-07,
      "loss": 1.9849,
      "num_input_tokens_seen": 237355384,
      "step": 233100
    },
    {
      "epoch": 4.591364611840681,
      "grad_norm": 1.8790674209594727,
      "learning_rate": 7.861106117811767e-07,
      "loss": 1.9724,
      "num_input_tokens_seen": 237456600,
      "step": 233200
    },
    {
      "epoch": 4.5933334645901835,
      "grad_norm": 1.7590235471725464,
      "learning_rate": 7.859414989887865e-07,
      "loss": 1.9423,
      "num_input_tokens_seen": 237558360,
      "step": 233300
    },
    {
      "epoch": 4.595302317339686,
      "grad_norm": 2.190117835998535,
      "learning_rate": 7.85772337574456e-07,
      "loss": 2.0129,
      "num_input_tokens_seen": 237660760,
      "step": 233400
    },
    {
      "epoch": 4.597271170089189,
      "grad_norm": 1.9098066091537476,
      "learning_rate": 7.856031275669499e-07,
      "loss": 1.9746,
      "num_input_tokens_seen": 237761864,
      "step": 233500
    },
    {
      "epoch": 4.599240022838692,
      "grad_norm": 2.2540605068206787,
      "learning_rate": 7.854338689950408e-07,
      "loss": 1.9703,
      "num_input_tokens_seen": 237862968,
      "step": 233600
    },
    {
      "epoch": 4.601208875588195,
      "grad_norm": 2.0733883380889893,
      "learning_rate": 7.852645618875098e-07,
      "loss": 1.9833,
      "num_input_tokens_seen": 237964040,
      "step": 233700
    },
    {
      "epoch": 4.603177728337697,
      "grad_norm": 1.6968687772750854,
      "learning_rate": 7.85095206273146e-07,
      "loss": 2.0093,
      "num_input_tokens_seen": 238064352,
      "step": 233800
    },
    {
      "epoch": 4.605146581087201,
      "grad_norm": 1.6747245788574219,
      "learning_rate": 7.84925802180747e-07,
      "loss": 1.9902,
      "num_input_tokens_seen": 238165888,
      "step": 233900
    },
    {
      "epoch": 4.607115433836704,
      "grad_norm": 2.083350658416748,
      "learning_rate": 7.847563496391185e-07,
      "loss": 1.9852,
      "num_input_tokens_seen": 238268288,
      "step": 234000
    },
    {
      "epoch": 4.6090842865862065,
      "grad_norm": 1.9042161703109741,
      "learning_rate": 7.845868486770746e-07,
      "loss": 2.0179,
      "num_input_tokens_seen": 238370688,
      "step": 234100
    },
    {
      "epoch": 4.611053139335709,
      "grad_norm": 1.9082952737808228,
      "learning_rate": 7.844172993234372e-07,
      "loss": 1.9391,
      "num_input_tokens_seen": 238472456,
      "step": 234200
    },
    {
      "epoch": 4.613021992085212,
      "grad_norm": 1.8290693759918213,
      "learning_rate": 7.84247701607037e-07,
      "loss": 1.9625,
      "num_input_tokens_seen": 238574856,
      "step": 234300
    },
    {
      "epoch": 4.614990844834715,
      "grad_norm": 1.919039011001587,
      "learning_rate": 7.840780555567128e-07,
      "loss": 1.9387,
      "num_input_tokens_seen": 238676608,
      "step": 234400
    },
    {
      "epoch": 4.6169596975842175,
      "grad_norm": 2.044429302215576,
      "learning_rate": 7.839083612013111e-07,
      "loss": 1.9293,
      "num_input_tokens_seen": 238779008,
      "step": 234500
    },
    {
      "epoch": 4.61892855033372,
      "grad_norm": 1.6604371070861816,
      "learning_rate": 7.837386185696872e-07,
      "loss": 1.9873,
      "num_input_tokens_seen": 238880936,
      "step": 234600
    },
    {
      "epoch": 4.620897403083223,
      "grad_norm": 1.8480556011199951,
      "learning_rate": 7.835688276907044e-07,
      "loss": 1.9705,
      "num_input_tokens_seen": 238983336,
      "step": 234700
    },
    {
      "epoch": 4.622866255832726,
      "grad_norm": 1.7892968654632568,
      "learning_rate": 7.833989885932342e-07,
      "loss": 2.0523,
      "num_input_tokens_seen": 239084888,
      "step": 234800
    },
    {
      "epoch": 4.6248351085822295,
      "grad_norm": 1.9779351949691772,
      "learning_rate": 7.832291013061561e-07,
      "loss": 1.9507,
      "num_input_tokens_seen": 239186672,
      "step": 234900
    },
    {
      "epoch": 4.626803961331732,
      "grad_norm": 2.0236270427703857,
      "learning_rate": 7.830591658583584e-07,
      "loss": 1.9871,
      "num_input_tokens_seen": 239288384,
      "step": 235000
    },
    {
      "epoch": 4.628772814081235,
      "grad_norm": 1.9210270643234253,
      "learning_rate": 7.828891822787369e-07,
      "loss": 2.0223,
      "num_input_tokens_seen": 239389464,
      "step": 235100
    },
    {
      "epoch": 4.630741666830738,
      "grad_norm": 1.7915340662002563,
      "learning_rate": 7.827191505961959e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 239491864,
      "step": 235200
    },
    {
      "epoch": 4.6327105195802405,
      "grad_norm": 1.9686694145202637,
      "learning_rate": 7.82549070839648e-07,
      "loss": 2.0114,
      "num_input_tokens_seen": 239594264,
      "step": 235300
    },
    {
      "epoch": 4.634679372329743,
      "grad_norm": 2.0891788005828857,
      "learning_rate": 7.823789430380136e-07,
      "loss": 1.9559,
      "num_input_tokens_seen": 239696664,
      "step": 235400
    },
    {
      "epoch": 4.636648225079246,
      "grad_norm": 1.9873641729354858,
      "learning_rate": 7.822087672202217e-07,
      "loss": 1.9618,
      "num_input_tokens_seen": 239798208,
      "step": 235500
    },
    {
      "epoch": 4.63861707782875,
      "grad_norm": 1.9363912343978882,
      "learning_rate": 7.820385434152093e-07,
      "loss": 1.9937,
      "num_input_tokens_seen": 239899792,
      "step": 235600
    },
    {
      "epoch": 4.640585930578252,
      "grad_norm": 2.0016725063323975,
      "learning_rate": 7.818682716519214e-07,
      "loss": 2.0103,
      "num_input_tokens_seen": 240000832,
      "step": 235700
    },
    {
      "epoch": 4.642554783327755,
      "grad_norm": 1.9424642324447632,
      "learning_rate": 7.816979519593112e-07,
      "loss": 1.9981,
      "num_input_tokens_seen": 240103232,
      "step": 235800
    },
    {
      "epoch": 4.644523636077258,
      "grad_norm": 1.9264036417007446,
      "learning_rate": 7.815275843663402e-07,
      "loss": 2.0369,
      "num_input_tokens_seen": 240204352,
      "step": 235900
    },
    {
      "epoch": 4.646492488826761,
      "grad_norm": 1.7386269569396973,
      "learning_rate": 7.813571689019781e-07,
      "loss": 2.0238,
      "num_input_tokens_seen": 240306112,
      "step": 236000
    },
    {
      "epoch": 4.6484613415762635,
      "grad_norm": 1.8147231340408325,
      "learning_rate": 7.811867055952024e-07,
      "loss": 2.0585,
      "num_input_tokens_seen": 240407656,
      "step": 236100
    },
    {
      "epoch": 4.650430194325766,
      "grad_norm": 1.8757692575454712,
      "learning_rate": 7.810161944749991e-07,
      "loss": 1.9717,
      "num_input_tokens_seen": 240510056,
      "step": 236200
    },
    {
      "epoch": 4.652399047075269,
      "grad_norm": 1.6022590398788452,
      "learning_rate": 7.808456355703621e-07,
      "loss": 2.0203,
      "num_input_tokens_seen": 240610992,
      "step": 236300
    },
    {
      "epoch": 4.654367899824772,
      "grad_norm": 4.077651023864746,
      "learning_rate": 7.806750289102935e-07,
      "loss": 1.9609,
      "num_input_tokens_seen": 240712640,
      "step": 236400
    },
    {
      "epoch": 4.6563367525742745,
      "grad_norm": 2.067807197570801,
      "learning_rate": 7.805043745238037e-07,
      "loss": 1.9732,
      "num_input_tokens_seen": 240815040,
      "step": 236500
    },
    {
      "epoch": 4.658305605323778,
      "grad_norm": 2.07737135887146,
      "learning_rate": 7.803336724399107e-07,
      "loss": 2.0,
      "num_input_tokens_seen": 240916976,
      "step": 236600
    },
    {
      "epoch": 4.660274458073281,
      "grad_norm": 2.0691514015197754,
      "learning_rate": 7.801629226876414e-07,
      "loss": 1.9896,
      "num_input_tokens_seen": 241018552,
      "step": 236700
    },
    {
      "epoch": 4.662243310822784,
      "grad_norm": 2.2837865352630615,
      "learning_rate": 7.799921252960299e-07,
      "loss": 1.9879,
      "num_input_tokens_seen": 241120952,
      "step": 236800
    },
    {
      "epoch": 4.664212163572286,
      "grad_norm": 1.9081482887268066,
      "learning_rate": 7.798212802941191e-07,
      "loss": 2.0108,
      "num_input_tokens_seen": 241223280,
      "step": 236900
    },
    {
      "epoch": 4.666181016321789,
      "grad_norm": 2.322063446044922,
      "learning_rate": 7.796503877109597e-07,
      "loss": 1.9737,
      "num_input_tokens_seen": 241325680,
      "step": 237000
    },
    {
      "epoch": 4.668149869071292,
      "grad_norm": 1.907496452331543,
      "learning_rate": 7.794794475756106e-07,
      "loss": 1.9647,
      "num_input_tokens_seen": 241427512,
      "step": 237100
    },
    {
      "epoch": 4.670118721820795,
      "grad_norm": 1.9791024923324585,
      "learning_rate": 7.793084599171387e-07,
      "loss": 1.9795,
      "num_input_tokens_seen": 241529912,
      "step": 237200
    },
    {
      "epoch": 4.672087574570298,
      "grad_norm": 1.9470134973526,
      "learning_rate": 7.791374247646188e-07,
      "loss": 1.933,
      "num_input_tokens_seen": 241631400,
      "step": 237300
    },
    {
      "epoch": 4.674056427319801,
      "grad_norm": 1.9613935947418213,
      "learning_rate": 7.789663421471343e-07,
      "loss": 1.9493,
      "num_input_tokens_seen": 241733800,
      "step": 237400
    },
    {
      "epoch": 4.676025280069304,
      "grad_norm": 2.3004748821258545,
      "learning_rate": 7.787952120937761e-07,
      "loss": 1.9725,
      "num_input_tokens_seen": 241835624,
      "step": 237500
    },
    {
      "epoch": 4.677994132818807,
      "grad_norm": 1.9286905527114868,
      "learning_rate": 7.78624034633644e-07,
      "loss": 2.0052,
      "num_input_tokens_seen": 241937448,
      "step": 237600
    },
    {
      "epoch": 4.679962985568309,
      "grad_norm": 1.8778142929077148,
      "learning_rate": 7.784528097958444e-07,
      "loss": 1.9878,
      "num_input_tokens_seen": 242038424,
      "step": 237700
    },
    {
      "epoch": 4.681931838317812,
      "grad_norm": 1.7701457738876343,
      "learning_rate": 7.782815376094931e-07,
      "loss": 1.9773,
      "num_input_tokens_seen": 242139376,
      "step": 237800
    },
    {
      "epoch": 4.683900691067315,
      "grad_norm": 1.7408747673034668,
      "learning_rate": 7.781102181037136e-07,
      "loss": 2.0059,
      "num_input_tokens_seen": 242241136,
      "step": 237900
    },
    {
      "epoch": 4.685869543816818,
      "grad_norm": 1.8353618383407593,
      "learning_rate": 7.779388513076373e-07,
      "loss": 1.9885,
      "num_input_tokens_seen": 242342904,
      "step": 238000
    },
    {
      "epoch": 4.68783839656632,
      "grad_norm": 1.8905603885650635,
      "learning_rate": 7.777674372504033e-07,
      "loss": 1.9919,
      "num_input_tokens_seen": 242443632,
      "step": 238100
    },
    {
      "epoch": 4.689807249315824,
      "grad_norm": 1.9271143674850464,
      "learning_rate": 7.775959759611595e-07,
      "loss": 1.9468,
      "num_input_tokens_seen": 242544712,
      "step": 238200
    },
    {
      "epoch": 4.691776102065327,
      "grad_norm": 2.4840543270111084,
      "learning_rate": 7.774244674690613e-07,
      "loss": 1.9898,
      "num_input_tokens_seen": 242646256,
      "step": 238300
    },
    {
      "epoch": 4.69374495481483,
      "grad_norm": 1.6891206502914429,
      "learning_rate": 7.772529118032725e-07,
      "loss": 1.9578,
      "num_input_tokens_seen": 242748128,
      "step": 238400
    },
    {
      "epoch": 4.695713807564332,
      "grad_norm": 1.9536232948303223,
      "learning_rate": 7.770813089929643e-07,
      "loss": 1.9804,
      "num_input_tokens_seen": 242850528,
      "step": 238500
    },
    {
      "epoch": 4.697682660313835,
      "grad_norm": 2.0244576930999756,
      "learning_rate": 7.769096590673167e-07,
      "loss": 1.9641,
      "num_input_tokens_seen": 242951256,
      "step": 238600
    },
    {
      "epoch": 4.699651513063338,
      "grad_norm": 1.9010989665985107,
      "learning_rate": 7.76737962055517e-07,
      "loss": 1.9761,
      "num_input_tokens_seen": 243053656,
      "step": 238700
    },
    {
      "epoch": 4.701620365812841,
      "grad_norm": 2.0288681983947754,
      "learning_rate": 7.76566217986761e-07,
      "loss": 1.9956,
      "num_input_tokens_seen": 243156056,
      "step": 238800
    },
    {
      "epoch": 4.703589218562343,
      "grad_norm": 2.0047388076782227,
      "learning_rate": 7.763944268902524e-07,
      "loss": 1.955,
      "num_input_tokens_seen": 243258456,
      "step": 238900
    },
    {
      "epoch": 4.705558071311847,
      "grad_norm": 7.127472877502441,
      "learning_rate": 7.762225887952025e-07,
      "loss": 2.0129,
      "num_input_tokens_seen": 243360856,
      "step": 239000
    },
    {
      "epoch": 4.70752692406135,
      "grad_norm": 2.2163915634155273,
      "learning_rate": 7.760507037308313e-07,
      "loss": 2.0004,
      "num_input_tokens_seen": 243463256,
      "step": 239100
    },
    {
      "epoch": 4.7094957768108525,
      "grad_norm": 1.742059350013733,
      "learning_rate": 7.758787717263662e-07,
      "loss": 2.0329,
      "num_input_tokens_seen": 243564056,
      "step": 239200
    },
    {
      "epoch": 4.711464629560355,
      "grad_norm": 1.989970088005066,
      "learning_rate": 7.757067928110431e-07,
      "loss": 1.9915,
      "num_input_tokens_seen": 243665560,
      "step": 239300
    },
    {
      "epoch": 4.713433482309858,
      "grad_norm": 1.9462517499923706,
      "learning_rate": 7.755347670141049e-07,
      "loss": 1.9581,
      "num_input_tokens_seen": 243767960,
      "step": 239400
    },
    {
      "epoch": 4.715402335059361,
      "grad_norm": 1.9584484100341797,
      "learning_rate": 7.753626943648038e-07,
      "loss": 1.9533,
      "num_input_tokens_seen": 243869880,
      "step": 239500
    },
    {
      "epoch": 4.7173711878088636,
      "grad_norm": 1.6583120822906494,
      "learning_rate": 7.751905748923989e-07,
      "loss": 2.0057,
      "num_input_tokens_seen": 243971688,
      "step": 239600
    },
    {
      "epoch": 4.719340040558366,
      "grad_norm": 2.032684087753296,
      "learning_rate": 7.750184086261578e-07,
      "loss": 1.99,
      "num_input_tokens_seen": 244074088,
      "step": 239700
    },
    {
      "epoch": 4.721308893307869,
      "grad_norm": 1.8053139448165894,
      "learning_rate": 7.74846195595356e-07,
      "loss": 1.979,
      "num_input_tokens_seen": 244176488,
      "step": 239800
    },
    {
      "epoch": 4.723277746057373,
      "grad_norm": 2.0863256454467773,
      "learning_rate": 7.746739358292769e-07,
      "loss": 1.9828,
      "num_input_tokens_seen": 244277368,
      "step": 239900
    },
    {
      "epoch": 4.7252465988068755,
      "grad_norm": 1.8091161251068115,
      "learning_rate": 7.745016293572115e-07,
      "loss": 1.965,
      "num_input_tokens_seen": 244379768,
      "step": 240000
    },
    {
      "epoch": 4.727215451556378,
      "grad_norm": 1.9357178211212158,
      "learning_rate": 7.743292762084594e-07,
      "loss": 1.9679,
      "num_input_tokens_seen": 244482168,
      "step": 240100
    },
    {
      "epoch": 4.729184304305881,
      "grad_norm": 5.501689434051514,
      "learning_rate": 7.741568764123276e-07,
      "loss": 2.0063,
      "num_input_tokens_seen": 244583776,
      "step": 240200
    },
    {
      "epoch": 4.731153157055384,
      "grad_norm": 1.9443832635879517,
      "learning_rate": 7.739844299981315e-07,
      "loss": 1.9971,
      "num_input_tokens_seen": 244684488,
      "step": 240300
    },
    {
      "epoch": 4.7331220098048865,
      "grad_norm": 1.8885704278945923,
      "learning_rate": 7.738119369951936e-07,
      "loss": 1.9554,
      "num_input_tokens_seen": 244784608,
      "step": 240400
    },
    {
      "epoch": 4.735090862554389,
      "grad_norm": 2.113469362258911,
      "learning_rate": 7.736393974328456e-07,
      "loss": 1.9819,
      "num_input_tokens_seen": 244887008,
      "step": 240500
    },
    {
      "epoch": 4.737059715303893,
      "grad_norm": 1.9098161458969116,
      "learning_rate": 7.734668113404259e-07,
      "loss": 2.0177,
      "num_input_tokens_seen": 244988952,
      "step": 240600
    },
    {
      "epoch": 4.739028568053396,
      "grad_norm": 1.8703964948654175,
      "learning_rate": 7.732941787472815e-07,
      "loss": 1.983,
      "num_input_tokens_seen": 245090336,
      "step": 240700
    },
    {
      "epoch": 4.740997420802898,
      "grad_norm": 1.9677131175994873,
      "learning_rate": 7.731214996827671e-07,
      "loss": 1.993,
      "num_input_tokens_seen": 245191096,
      "step": 240800
    },
    {
      "epoch": 4.742966273552401,
      "grad_norm": 1.7599327564239502,
      "learning_rate": 7.729487741762453e-07,
      "loss": 2.0268,
      "num_input_tokens_seen": 245293496,
      "step": 240900
    },
    {
      "epoch": 4.744935126301904,
      "grad_norm": 1.8329025506973267,
      "learning_rate": 7.727760022570866e-07,
      "loss": 1.9771,
      "num_input_tokens_seen": 245394472,
      "step": 241000
    },
    {
      "epoch": 4.746903979051407,
      "grad_norm": 1.8511635065078735,
      "learning_rate": 7.726031839546694e-07,
      "loss": 1.9415,
      "num_input_tokens_seen": 245496872,
      "step": 241100
    },
    {
      "epoch": 4.7488728318009095,
      "grad_norm": 1.6652101278305054,
      "learning_rate": 7.724303192983798e-07,
      "loss": 1.9613,
      "num_input_tokens_seen": 245598696,
      "step": 241200
    },
    {
      "epoch": 4.750841684550412,
      "grad_norm": 2.142704725265503,
      "learning_rate": 7.722574083176126e-07,
      "loss": 1.9493,
      "num_input_tokens_seen": 245700520,
      "step": 241300
    },
    {
      "epoch": 4.752810537299915,
      "grad_norm": 1.9635742902755737,
      "learning_rate": 7.72084451041769e-07,
      "loss": 1.9343,
      "num_input_tokens_seen": 245802376,
      "step": 241400
    },
    {
      "epoch": 4.754779390049418,
      "grad_norm": 1.9217370748519897,
      "learning_rate": 7.719114475002598e-07,
      "loss": 1.9815,
      "num_input_tokens_seen": 245903904,
      "step": 241500
    },
    {
      "epoch": 4.756748242798921,
      "grad_norm": 1.9156200885772705,
      "learning_rate": 7.717383977225021e-07,
      "loss": 1.959,
      "num_input_tokens_seen": 246006304,
      "step": 241600
    },
    {
      "epoch": 4.758717095548424,
      "grad_norm": 2.073235273361206,
      "learning_rate": 7.715653017379217e-07,
      "loss": 1.9583,
      "num_input_tokens_seen": 246108040,
      "step": 241700
    },
    {
      "epoch": 4.760685948297927,
      "grad_norm": 1.8034470081329346,
      "learning_rate": 7.713921595759525e-07,
      "loss": 2.0054,
      "num_input_tokens_seen": 246209776,
      "step": 241800
    },
    {
      "epoch": 4.76265480104743,
      "grad_norm": 1.9783844947814941,
      "learning_rate": 7.712189712660355e-07,
      "loss": 2.0034,
      "num_input_tokens_seen": 246311672,
      "step": 241900
    },
    {
      "epoch": 4.764623653796932,
      "grad_norm": 1.8855631351470947,
      "learning_rate": 7.7104573683762e-07,
      "loss": 1.9775,
      "num_input_tokens_seen": 246413528,
      "step": 242000
    },
    {
      "epoch": 4.766592506546435,
      "grad_norm": 1.9352155923843384,
      "learning_rate": 7.708724563201629e-07,
      "loss": 2.0206,
      "num_input_tokens_seen": 246515928,
      "step": 242100
    },
    {
      "epoch": 4.768561359295938,
      "grad_norm": 1.8944944143295288,
      "learning_rate": 7.706991297431294e-07,
      "loss": 2.0112,
      "num_input_tokens_seen": 246616336,
      "step": 242200
    },
    {
      "epoch": 4.770530212045442,
      "grad_norm": 1.7942332029342651,
      "learning_rate": 7.705257571359921e-07,
      "loss": 1.9552,
      "num_input_tokens_seen": 246718736,
      "step": 242300
    },
    {
      "epoch": 4.772499064794944,
      "grad_norm": 2.022024631500244,
      "learning_rate": 7.703523385282314e-07,
      "loss": 1.9995,
      "num_input_tokens_seen": 246820328,
      "step": 242400
    },
    {
      "epoch": 4.774467917544447,
      "grad_norm": 1.8557579517364502,
      "learning_rate": 7.70178873949336e-07,
      "loss": 1.9152,
      "num_input_tokens_seen": 246921128,
      "step": 242500
    },
    {
      "epoch": 4.77643677029395,
      "grad_norm": 1.7127655744552612,
      "learning_rate": 7.700053634288018e-07,
      "loss": 2.0022,
      "num_input_tokens_seen": 247022976,
      "step": 242600
    },
    {
      "epoch": 4.778405623043453,
      "grad_norm": 2.2102367877960205,
      "learning_rate": 7.698318069961327e-07,
      "loss": 1.9786,
      "num_input_tokens_seen": 247125376,
      "step": 242700
    },
    {
      "epoch": 4.780374475792955,
      "grad_norm": 1.8471670150756836,
      "learning_rate": 7.696582046808409e-07,
      "loss": 2.0114,
      "num_input_tokens_seen": 247227224,
      "step": 242800
    },
    {
      "epoch": 4.782343328542458,
      "grad_norm": 2.032442808151245,
      "learning_rate": 7.694845565124456e-07,
      "loss": 1.9674,
      "num_input_tokens_seen": 247328480,
      "step": 242900
    },
    {
      "epoch": 4.784312181291961,
      "grad_norm": 1.9215418100357056,
      "learning_rate": 7.693108625204745e-07,
      "loss": 1.9807,
      "num_input_tokens_seen": 247430120,
      "step": 243000
    },
    {
      "epoch": 4.786281034041464,
      "grad_norm": 1.954582929611206,
      "learning_rate": 7.691371227344628e-07,
      "loss": 2.0155,
      "num_input_tokens_seen": 247530976,
      "step": 243100
    },
    {
      "epoch": 4.788249886790966,
      "grad_norm": 1.8124114274978638,
      "learning_rate": 7.689633371839533e-07,
      "loss": 1.9671,
      "num_input_tokens_seen": 247631784,
      "step": 243200
    },
    {
      "epoch": 4.79021873954047,
      "grad_norm": 1.6148899793624878,
      "learning_rate": 7.687895058984968e-07,
      "loss": 1.9918,
      "num_input_tokens_seen": 247732560,
      "step": 243300
    },
    {
      "epoch": 4.792187592289973,
      "grad_norm": 1.8852741718292236,
      "learning_rate": 7.686156289076519e-07,
      "loss": 2.0244,
      "num_input_tokens_seen": 247834960,
      "step": 243400
    },
    {
      "epoch": 4.794156445039476,
      "grad_norm": 2.124105930328369,
      "learning_rate": 7.684417062409849e-07,
      "loss": 1.9407,
      "num_input_tokens_seen": 247936560,
      "step": 243500
    },
    {
      "epoch": 4.796125297788978,
      "grad_norm": 2.2060279846191406,
      "learning_rate": 7.682677379280699e-07,
      "loss": 1.9684,
      "num_input_tokens_seen": 248038960,
      "step": 243600
    },
    {
      "epoch": 4.798094150538481,
      "grad_norm": 1.9046642780303955,
      "learning_rate": 7.680937239984888e-07,
      "loss": 1.9862,
      "num_input_tokens_seen": 248139920,
      "step": 243700
    },
    {
      "epoch": 4.800063003287984,
      "grad_norm": 1.8380968570709229,
      "learning_rate": 7.679196644818313e-07,
      "loss": 2.0188,
      "num_input_tokens_seen": 248242320,
      "step": 243800
    },
    {
      "epoch": 4.802031856037487,
      "grad_norm": 1.8512117862701416,
      "learning_rate": 7.677455594076943e-07,
      "loss": 1.9799,
      "num_input_tokens_seen": 248344720,
      "step": 243900
    },
    {
      "epoch": 4.80400070878699,
      "grad_norm": 1.8946309089660645,
      "learning_rate": 7.675714088056833e-07,
      "loss": 1.9191,
      "num_input_tokens_seen": 248446640,
      "step": 244000
    },
    {
      "epoch": 4.805969561536493,
      "grad_norm": 1.989884376525879,
      "learning_rate": 7.673972127054112e-07,
      "loss": 1.976,
      "num_input_tokens_seen": 248546576,
      "step": 244100
    },
    {
      "epoch": 4.807938414285996,
      "grad_norm": 2.0823934078216553,
      "learning_rate": 7.672229711364981e-07,
      "loss": 2.0169,
      "num_input_tokens_seen": 248648840,
      "step": 244200
    },
    {
      "epoch": 4.8099072670354985,
      "grad_norm": 2.1578969955444336,
      "learning_rate": 7.670486841285728e-07,
      "loss": 2.0082,
      "num_input_tokens_seen": 248751240,
      "step": 244300
    },
    {
      "epoch": 4.811876119785001,
      "grad_norm": 1.9351460933685303,
      "learning_rate": 7.668743517112713e-07,
      "loss": 2.003,
      "num_input_tokens_seen": 248852264,
      "step": 244400
    },
    {
      "epoch": 4.813844972534504,
      "grad_norm": 3.4517440795898438,
      "learning_rate": 7.66699973914237e-07,
      "loss": 1.9917,
      "num_input_tokens_seen": 248954360,
      "step": 244500
    },
    {
      "epoch": 4.815813825284007,
      "grad_norm": 1.9812885522842407,
      "learning_rate": 7.665255507671216e-07,
      "loss": 1.9517,
      "num_input_tokens_seen": 249056760,
      "step": 244600
    },
    {
      "epoch": 4.81778267803351,
      "grad_norm": 1.8989157676696777,
      "learning_rate": 7.663510822995842e-07,
      "loss": 1.9795,
      "num_input_tokens_seen": 249159160,
      "step": 244700
    },
    {
      "epoch": 4.819751530783012,
      "grad_norm": 2.909083843231201,
      "learning_rate": 7.661765685412919e-07,
      "loss": 2.0136,
      "num_input_tokens_seen": 249260792,
      "step": 244800
    },
    {
      "epoch": 4.821720383532516,
      "grad_norm": 2.0052597522735596,
      "learning_rate": 7.660020095219192e-07,
      "loss": 1.9627,
      "num_input_tokens_seen": 249362848,
      "step": 244900
    },
    {
      "epoch": 4.823689236282019,
      "grad_norm": 1.970702052116394,
      "learning_rate": 7.658274052711483e-07,
      "loss": 1.9917,
      "num_input_tokens_seen": 249464624,
      "step": 245000
    },
    {
      "epoch": 4.8256580890315215,
      "grad_norm": 1.9335062503814697,
      "learning_rate": 7.656527558186692e-07,
      "loss": 2.0013,
      "num_input_tokens_seen": 249566456,
      "step": 245100
    },
    {
      "epoch": 4.827626941781024,
      "grad_norm": 2.063624620437622,
      "learning_rate": 7.654780611941796e-07,
      "loss": 2.0469,
      "num_input_tokens_seen": 249667304,
      "step": 245200
    },
    {
      "epoch": 4.829595794530527,
      "grad_norm": 1.7945432662963867,
      "learning_rate": 7.653033214273849e-07,
      "loss": 1.9682,
      "num_input_tokens_seen": 249769704,
      "step": 245300
    },
    {
      "epoch": 4.83156464728003,
      "grad_norm": 3.2311344146728516,
      "learning_rate": 7.651285365479979e-07,
      "loss": 1.945,
      "num_input_tokens_seen": 249872104,
      "step": 245400
    },
    {
      "epoch": 4.8335335000295325,
      "grad_norm": 1.8495867252349854,
      "learning_rate": 7.649537065857397e-07,
      "loss": 1.9813,
      "num_input_tokens_seen": 249974504,
      "step": 245500
    },
    {
      "epoch": 4.835502352779035,
      "grad_norm": 1.802419900894165,
      "learning_rate": 7.647788315703381e-07,
      "loss": 1.9672,
      "num_input_tokens_seen": 250076056,
      "step": 245600
    },
    {
      "epoch": 4.837471205528539,
      "grad_norm": 1.8500380516052246,
      "learning_rate": 7.646039115315297e-07,
      "loss": 2.0243,
      "num_input_tokens_seen": 250178000,
      "step": 245700
    },
    {
      "epoch": 4.839440058278042,
      "grad_norm": 1.6966948509216309,
      "learning_rate": 7.644289464990578e-07,
      "loss": 1.9632,
      "num_input_tokens_seen": 250279832,
      "step": 245800
    },
    {
      "epoch": 4.8414089110275444,
      "grad_norm": 1.8331348896026611,
      "learning_rate": 7.642539365026739e-07,
      "loss": 1.9973,
      "num_input_tokens_seen": 250382232,
      "step": 245900
    },
    {
      "epoch": 4.843377763777047,
      "grad_norm": 1.5486363172531128,
      "learning_rate": 7.640788815721368e-07,
      "loss": 1.9749,
      "num_input_tokens_seen": 250484632,
      "step": 246000
    },
    {
      "epoch": 4.84534661652655,
      "grad_norm": 2.655693531036377,
      "learning_rate": 7.639037817372135e-07,
      "loss": 2.0112,
      "num_input_tokens_seen": 250586272,
      "step": 246100
    },
    {
      "epoch": 4.847315469276053,
      "grad_norm": 2.0464234352111816,
      "learning_rate": 7.637286370276778e-07,
      "loss": 1.9555,
      "num_input_tokens_seen": 250688672,
      "step": 246200
    },
    {
      "epoch": 4.8492843220255555,
      "grad_norm": 1.990045428276062,
      "learning_rate": 7.635534474733118e-07,
      "loss": 2.0138,
      "num_input_tokens_seen": 250790488,
      "step": 246300
    },
    {
      "epoch": 4.851253174775058,
      "grad_norm": 1.757544755935669,
      "learning_rate": 7.633782131039048e-07,
      "loss": 1.9635,
      "num_input_tokens_seen": 250892368,
      "step": 246400
    },
    {
      "epoch": 4.853222027524561,
      "grad_norm": 2.010892152786255,
      "learning_rate": 7.632029339492543e-07,
      "loss": 1.9928,
      "num_input_tokens_seen": 250993536,
      "step": 246500
    },
    {
      "epoch": 4.855190880274065,
      "grad_norm": 2.1679794788360596,
      "learning_rate": 7.630276100391647e-07,
      "loss": 2.0663,
      "num_input_tokens_seen": 251093456,
      "step": 246600
    },
    {
      "epoch": 4.857159733023567,
      "grad_norm": 2.170764923095703,
      "learning_rate": 7.628522414034486e-07,
      "loss": 1.9865,
      "num_input_tokens_seen": 251195032,
      "step": 246700
    },
    {
      "epoch": 4.85912858577307,
      "grad_norm": 1.8072079420089722,
      "learning_rate": 7.626768280719258e-07,
      "loss": 1.9895,
      "num_input_tokens_seen": 251297432,
      "step": 246800
    },
    {
      "epoch": 4.861097438522573,
      "grad_norm": 2.074186325073242,
      "learning_rate": 7.62501370074424e-07,
      "loss": 2.0129,
      "num_input_tokens_seen": 251398832,
      "step": 246900
    },
    {
      "epoch": 4.863066291272076,
      "grad_norm": 1.8249096870422363,
      "learning_rate": 7.623258674407782e-07,
      "loss": 1.9659,
      "num_input_tokens_seen": 251501032,
      "step": 247000
    },
    {
      "epoch": 4.865035144021578,
      "grad_norm": 4.696745872497559,
      "learning_rate": 7.621503202008312e-07,
      "loss": 2.0045,
      "num_input_tokens_seen": 251602816,
      "step": 247100
    },
    {
      "epoch": 4.867003996771081,
      "grad_norm": 2.030505418777466,
      "learning_rate": 7.619747283844332e-07,
      "loss": 1.9664,
      "num_input_tokens_seen": 251705216,
      "step": 247200
    },
    {
      "epoch": 4.868972849520585,
      "grad_norm": 2.0291748046875,
      "learning_rate": 7.617990920214425e-07,
      "loss": 1.9633,
      "num_input_tokens_seen": 251807616,
      "step": 247300
    },
    {
      "epoch": 4.870941702270088,
      "grad_norm": 2.032618999481201,
      "learning_rate": 7.61623411141724e-07,
      "loss": 1.9673,
      "num_input_tokens_seen": 251910016,
      "step": 247400
    },
    {
      "epoch": 4.87291055501959,
      "grad_norm": 1.7918150424957275,
      "learning_rate": 7.614476857751513e-07,
      "loss": 1.9885,
      "num_input_tokens_seen": 252012416,
      "step": 247500
    },
    {
      "epoch": 4.874879407769093,
      "grad_norm": 1.6047450304031372,
      "learning_rate": 7.612719159516046e-07,
      "loss": 1.9819,
      "num_input_tokens_seen": 252114816,
      "step": 247600
    },
    {
      "epoch": 4.876848260518596,
      "grad_norm": 2.1791837215423584,
      "learning_rate": 7.610961017009725e-07,
      "loss": 2.0085,
      "num_input_tokens_seen": 252216488,
      "step": 247700
    },
    {
      "epoch": 4.878817113268099,
      "grad_norm": 2.147240161895752,
      "learning_rate": 7.609202430531503e-07,
      "loss": 1.997,
      "num_input_tokens_seen": 252318344,
      "step": 247800
    },
    {
      "epoch": 4.880785966017601,
      "grad_norm": 3.31123685836792,
      "learning_rate": 7.607443400380415e-07,
      "loss": 2.068,
      "num_input_tokens_seen": 252420184,
      "step": 247900
    },
    {
      "epoch": 4.882754818767104,
      "grad_norm": 2.0299837589263916,
      "learning_rate": 7.60568392685557e-07,
      "loss": 2.034,
      "num_input_tokens_seen": 252521064,
      "step": 248000
    },
    {
      "epoch": 4.884723671516607,
      "grad_norm": 1.8934005498886108,
      "learning_rate": 7.603924010256149e-07,
      "loss": 1.9944,
      "num_input_tokens_seen": 252622640,
      "step": 248100
    },
    {
      "epoch": 4.88669252426611,
      "grad_norm": 1.7607197761535645,
      "learning_rate": 7.602163650881415e-07,
      "loss": 1.9715,
      "num_input_tokens_seen": 252723392,
      "step": 248200
    },
    {
      "epoch": 4.888661377015613,
      "grad_norm": 1.6598211526870728,
      "learning_rate": 7.600402849030698e-07,
      "loss": 2.0128,
      "num_input_tokens_seen": 252825080,
      "step": 248300
    },
    {
      "epoch": 4.890630229765116,
      "grad_norm": 1.8595517873764038,
      "learning_rate": 7.598641605003408e-07,
      "loss": 1.994,
      "num_input_tokens_seen": 252926560,
      "step": 248400
    },
    {
      "epoch": 4.892599082514619,
      "grad_norm": 1.9030126333236694,
      "learning_rate": 7.596879919099033e-07,
      "loss": 1.9687,
      "num_input_tokens_seen": 253028960,
      "step": 248500
    },
    {
      "epoch": 4.894567935264122,
      "grad_norm": 2.1952364444732666,
      "learning_rate": 7.595117791617131e-07,
      "loss": 2.0137,
      "num_input_tokens_seen": 253130320,
      "step": 248600
    },
    {
      "epoch": 4.896536788013624,
      "grad_norm": 1.8906571865081787,
      "learning_rate": 7.593355222857336e-07,
      "loss": 1.9851,
      "num_input_tokens_seen": 253232720,
      "step": 248700
    },
    {
      "epoch": 4.898505640763127,
      "grad_norm": 2.167135000228882,
      "learning_rate": 7.591592213119359e-07,
      "loss": 1.9977,
      "num_input_tokens_seen": 253334480,
      "step": 248800
    },
    {
      "epoch": 4.90047449351263,
      "grad_norm": 1.8287575244903564,
      "learning_rate": 7.589828762702986e-07,
      "loss": 1.9797,
      "num_input_tokens_seen": 253436360,
      "step": 248900
    },
    {
      "epoch": 4.9024433462621335,
      "grad_norm": 1.7591739892959595,
      "learning_rate": 7.588064871908073e-07,
      "loss": 2.0528,
      "num_input_tokens_seen": 253535936,
      "step": 249000
    },
    {
      "epoch": 4.904412199011636,
      "grad_norm": 1.7703303098678589,
      "learning_rate": 7.586300541034559e-07,
      "loss": 1.9703,
      "num_input_tokens_seen": 253637656,
      "step": 249100
    },
    {
      "epoch": 4.906381051761139,
      "grad_norm": 2.017404317855835,
      "learning_rate": 7.584535770382451e-07,
      "loss": 1.9691,
      "num_input_tokens_seen": 253740056,
      "step": 249200
    },
    {
      "epoch": 4.908349904510642,
      "grad_norm": 2.018389940261841,
      "learning_rate": 7.582770560251835e-07,
      "loss": 1.9676,
      "num_input_tokens_seen": 253842456,
      "step": 249300
    },
    {
      "epoch": 4.9103187572601446,
      "grad_norm": 1.9108238220214844,
      "learning_rate": 7.581004910942869e-07,
      "loss": 1.9926,
      "num_input_tokens_seen": 253944440,
      "step": 249400
    },
    {
      "epoch": 4.912287610009647,
      "grad_norm": 1.774223804473877,
      "learning_rate": 7.579238822755787e-07,
      "loss": 1.9619,
      "num_input_tokens_seen": 254046840,
      "step": 249500
    },
    {
      "epoch": 4.91425646275915,
      "grad_norm": 1.94884192943573,
      "learning_rate": 7.577472295990896e-07,
      "loss": 1.9519,
      "num_input_tokens_seen": 254148752,
      "step": 249600
    },
    {
      "epoch": 4.916225315508653,
      "grad_norm": 1.7490347623825073,
      "learning_rate": 7.575705330948583e-07,
      "loss": 2.0006,
      "num_input_tokens_seen": 254250520,
      "step": 249700
    },
    {
      "epoch": 4.918194168258156,
      "grad_norm": 1.9168856143951416,
      "learning_rate": 7.573937927929302e-07,
      "loss": 1.9962,
      "num_input_tokens_seen": 254352920,
      "step": 249800
    },
    {
      "epoch": 4.920163021007658,
      "grad_norm": 1.9100327491760254,
      "learning_rate": 7.572170087233586e-07,
      "loss": 2.0098,
      "num_input_tokens_seen": 254454720,
      "step": 249900
    },
    {
      "epoch": 4.922131873757162,
      "grad_norm": 2.0106396675109863,
      "learning_rate": 7.570401809162042e-07,
      "loss": 1.9462,
      "num_input_tokens_seen": 254557120,
      "step": 250000
    },
    {
      "epoch": 4.924100726506665,
      "grad_norm": 1.8019355535507202,
      "learning_rate": 7.568633094015349e-07,
      "loss": 2.0181,
      "num_input_tokens_seen": 254658400,
      "step": 250100
    },
    {
      "epoch": 4.9260695792561675,
      "grad_norm": 2.2052829265594482,
      "learning_rate": 7.566863942094263e-07,
      "loss": 1.9828,
      "num_input_tokens_seen": 254760800,
      "step": 250200
    },
    {
      "epoch": 4.92803843200567,
      "grad_norm": 2.1917173862457275,
      "learning_rate": 7.565094353699614e-07,
      "loss": 1.9979,
      "num_input_tokens_seen": 254862552,
      "step": 250300
    },
    {
      "epoch": 4.930007284755173,
      "grad_norm": 1.8937517404556274,
      "learning_rate": 7.563324329132306e-07,
      "loss": 1.9988,
      "num_input_tokens_seen": 254964096,
      "step": 250400
    },
    {
      "epoch": 4.931976137504676,
      "grad_norm": 1.8392560482025146,
      "learning_rate": 7.561553868693313e-07,
      "loss": 1.984,
      "num_input_tokens_seen": 255066496,
      "step": 250500
    },
    {
      "epoch": 4.9339449902541785,
      "grad_norm": 1.7257837057113647,
      "learning_rate": 7.559782972683694e-07,
      "loss": 1.9893,
      "num_input_tokens_seen": 255168896,
      "step": 250600
    },
    {
      "epoch": 4.935913843003682,
      "grad_norm": 1.9757217168807983,
      "learning_rate": 7.558011641404568e-07,
      "loss": 2.0202,
      "num_input_tokens_seen": 255271000,
      "step": 250700
    },
    {
      "epoch": 4.937882695753185,
      "grad_norm": 2.108100652694702,
      "learning_rate": 7.556239875157139e-07,
      "loss": 2.0018,
      "num_input_tokens_seen": 255371784,
      "step": 250800
    },
    {
      "epoch": 4.939851548502688,
      "grad_norm": 2.2486793994903564,
      "learning_rate": 7.554467674242678e-07,
      "loss": 1.9993,
      "num_input_tokens_seen": 255474184,
      "step": 250900
    },
    {
      "epoch": 4.9418204012521905,
      "grad_norm": 2.750699520111084,
      "learning_rate": 7.552695038962535e-07,
      "loss": 1.9614,
      "num_input_tokens_seen": 255576584,
      "step": 251000
    },
    {
      "epoch": 4.943789254001693,
      "grad_norm": 1.7878025770187378,
      "learning_rate": 7.550921969618132e-07,
      "loss": 2.0229,
      "num_input_tokens_seen": 255678280,
      "step": 251100
    },
    {
      "epoch": 4.945758106751196,
      "grad_norm": 2.116180181503296,
      "learning_rate": 7.549148466510964e-07,
      "loss": 1.9928,
      "num_input_tokens_seen": 255780680,
      "step": 251200
    },
    {
      "epoch": 4.947726959500699,
      "grad_norm": 1.764121413230896,
      "learning_rate": 7.547374529942597e-07,
      "loss": 1.9626,
      "num_input_tokens_seen": 255883080,
      "step": 251300
    },
    {
      "epoch": 4.9496958122502015,
      "grad_norm": 2.0100488662719727,
      "learning_rate": 7.545600160214679e-07,
      "loss": 1.9804,
      "num_input_tokens_seen": 255984064,
      "step": 251400
    },
    {
      "epoch": 4.951664664999704,
      "grad_norm": 1.9737484455108643,
      "learning_rate": 7.543825357628924e-07,
      "loss": 1.9367,
      "num_input_tokens_seen": 256086464,
      "step": 251500
    },
    {
      "epoch": 4.953633517749208,
      "grad_norm": 1.7976274490356445,
      "learning_rate": 7.542050122487122e-07,
      "loss": 1.9902,
      "num_input_tokens_seen": 256188104,
      "step": 251600
    },
    {
      "epoch": 4.955602370498711,
      "grad_norm": 1.7301064729690552,
      "learning_rate": 7.540274455091136e-07,
      "loss": 1.9746,
      "num_input_tokens_seen": 256289736,
      "step": 251700
    },
    {
      "epoch": 4.957571223248213,
      "grad_norm": 1.71588933467865,
      "learning_rate": 7.538498355742907e-07,
      "loss": 1.9361,
      "num_input_tokens_seen": 256392136,
      "step": 251800
    },
    {
      "epoch": 4.959540075997716,
      "grad_norm": 7.013174533843994,
      "learning_rate": 7.536721824744443e-07,
      "loss": 2.0139,
      "num_input_tokens_seen": 256494176,
      "step": 251900
    },
    {
      "epoch": 4.961508928747219,
      "grad_norm": 2.029755115509033,
      "learning_rate": 7.534944862397827e-07,
      "loss": 2.0195,
      "num_input_tokens_seen": 256596032,
      "step": 252000
    },
    {
      "epoch": 4.963477781496722,
      "grad_norm": 1.9294846057891846,
      "learning_rate": 7.533167469005219e-07,
      "loss": 2.0266,
      "num_input_tokens_seen": 256694728,
      "step": 252100
    },
    {
      "epoch": 4.9654466342462245,
      "grad_norm": 2.0432777404785156,
      "learning_rate": 7.531389644868848e-07,
      "loss": 1.9878,
      "num_input_tokens_seen": 256796448,
      "step": 252200
    },
    {
      "epoch": 4.967415486995727,
      "grad_norm": 1.7484453916549683,
      "learning_rate": 7.529611390291019e-07,
      "loss": 1.9393,
      "num_input_tokens_seen": 256896400,
      "step": 252300
    },
    {
      "epoch": 4.969384339745231,
      "grad_norm": 1.9300607442855835,
      "learning_rate": 7.52783270557411e-07,
      "loss": 1.979,
      "num_input_tokens_seen": 256998800,
      "step": 252400
    },
    {
      "epoch": 4.971353192494734,
      "grad_norm": 2.1486616134643555,
      "learning_rate": 7.526053591020568e-07,
      "loss": 2.0132,
      "num_input_tokens_seen": 257100432,
      "step": 252500
    },
    {
      "epoch": 4.973322045244236,
      "grad_norm": 1.813297986984253,
      "learning_rate": 7.524274046932919e-07,
      "loss": 1.9673,
      "num_input_tokens_seen": 257202832,
      "step": 252600
    },
    {
      "epoch": 4.975290897993739,
      "grad_norm": 1.8517218828201294,
      "learning_rate": 7.52249407361376e-07,
      "loss": 2.0336,
      "num_input_tokens_seen": 257304424,
      "step": 252700
    },
    {
      "epoch": 4.977259750743242,
      "grad_norm": 3.19309139251709,
      "learning_rate": 7.520713671365759e-07,
      "loss": 2.0102,
      "num_input_tokens_seen": 257405552,
      "step": 252800
    },
    {
      "epoch": 4.979228603492745,
      "grad_norm": 2.1946089267730713,
      "learning_rate": 7.518932840491659e-07,
      "loss": 1.9739,
      "num_input_tokens_seen": 257507224,
      "step": 252900
    },
    {
      "epoch": 4.981197456242247,
      "grad_norm": 1.8575719594955444,
      "learning_rate": 7.517151581294275e-07,
      "loss": 1.9371,
      "num_input_tokens_seen": 257609624,
      "step": 253000
    },
    {
      "epoch": 4.98316630899175,
      "grad_norm": 1.8853888511657715,
      "learning_rate": 7.515369894076494e-07,
      "loss": 1.9715,
      "num_input_tokens_seen": 257709776,
      "step": 253100
    },
    {
      "epoch": 4.985135161741253,
      "grad_norm": 2.0849671363830566,
      "learning_rate": 7.513587779141279e-07,
      "loss": 1.9851,
      "num_input_tokens_seen": 257811656,
      "step": 253200
    },
    {
      "epoch": 4.987104014490757,
      "grad_norm": 1.8125005960464478,
      "learning_rate": 7.511805236791664e-07,
      "loss": 1.9891,
      "num_input_tokens_seen": 257912328,
      "step": 253300
    },
    {
      "epoch": 4.989072867240259,
      "grad_norm": 1.8635026216506958,
      "learning_rate": 7.510022267330754e-07,
      "loss": 1.9633,
      "num_input_tokens_seen": 258014728,
      "step": 253400
    },
    {
      "epoch": 4.991041719989762,
      "grad_norm": 1.9553800821304321,
      "learning_rate": 7.508238871061725e-07,
      "loss": 1.9748,
      "num_input_tokens_seen": 258115848,
      "step": 253500
    },
    {
      "epoch": 4.993010572739265,
      "grad_norm": 2.6224100589752197,
      "learning_rate": 7.506455048287834e-07,
      "loss": 2.0179,
      "num_input_tokens_seen": 258217816,
      "step": 253600
    },
    {
      "epoch": 4.994979425488768,
      "grad_norm": 1.7693684101104736,
      "learning_rate": 7.504670799312403e-07,
      "loss": 1.9841,
      "num_input_tokens_seen": 258319440,
      "step": 253700
    },
    {
      "epoch": 4.99694827823827,
      "grad_norm": 1.981896162033081,
      "learning_rate": 7.502886124438826e-07,
      "loss": 2.0438,
      "num_input_tokens_seen": 258421840,
      "step": 253800
    },
    {
      "epoch": 4.998917130987773,
      "grad_norm": 1.9829676151275635,
      "learning_rate": 7.501101023970575e-07,
      "loss": 1.9882,
      "num_input_tokens_seen": 258524240,
      "step": 253900
    },
    {
      "epoch": 5.000885983737276,
      "grad_norm": 2.067901134490967,
      "learning_rate": 7.499315498211189e-07,
      "loss": 1.9443,
      "num_input_tokens_seen": 258626640,
      "step": 254000
    },
    {
      "epoch": 5.0028548364867795,
      "grad_norm": 2.22636342048645,
      "learning_rate": 7.497529547464284e-07,
      "loss": 1.9898,
      "num_input_tokens_seen": 258728416,
      "step": 254100
    },
    {
      "epoch": 5.004823689236282,
      "grad_norm": 1.7298223972320557,
      "learning_rate": 7.495743172033544e-07,
      "loss": 2.0092,
      "num_input_tokens_seen": 258830384,
      "step": 254200
    },
    {
      "epoch": 5.006792541985785,
      "grad_norm": 2.0167651176452637,
      "learning_rate": 7.493956372222728e-07,
      "loss": 2.0209,
      "num_input_tokens_seen": 258931456,
      "step": 254300
    },
    {
      "epoch": 5.008761394735288,
      "grad_norm": 1.8186672925949097,
      "learning_rate": 7.492169148335665e-07,
      "loss": 1.9968,
      "num_input_tokens_seen": 259032920,
      "step": 254400
    },
    {
      "epoch": 5.010730247484791,
      "grad_norm": 2.0461297035217285,
      "learning_rate": 7.49038150067626e-07,
      "loss": 2.0187,
      "num_input_tokens_seen": 259133080,
      "step": 254500
    },
    {
      "epoch": 5.012699100234293,
      "grad_norm": 2.007765769958496,
      "learning_rate": 7.488593429548484e-07,
      "loss": 1.9514,
      "num_input_tokens_seen": 259235480,
      "step": 254600
    },
    {
      "epoch": 5.014667952983796,
      "grad_norm": 1.9249708652496338,
      "learning_rate": 7.486804935256386e-07,
      "loss": 1.9713,
      "num_input_tokens_seen": 259337880,
      "step": 254700
    },
    {
      "epoch": 5.016636805733299,
      "grad_norm": 2.003084421157837,
      "learning_rate": 7.485016018104082e-07,
      "loss": 2.0099,
      "num_input_tokens_seen": 259440280,
      "step": 254800
    },
    {
      "epoch": 5.0186056584828025,
      "grad_norm": 1.9281221628189087,
      "learning_rate": 7.483226678395767e-07,
      "loss": 1.986,
      "num_input_tokens_seen": 259542000,
      "step": 254900
    },
    {
      "epoch": 5.020574511232305,
      "grad_norm": 9.370895385742188,
      "learning_rate": 7.481436916435696e-07,
      "loss": 1.9969,
      "num_input_tokens_seen": 259642976,
      "step": 255000
    },
    {
      "epoch": 5.022543363981808,
      "grad_norm": 1.9999701976776123,
      "learning_rate": 7.47964673252821e-07,
      "loss": 1.9783,
      "num_input_tokens_seen": 259743568,
      "step": 255100
    },
    {
      "epoch": 5.024512216731311,
      "grad_norm": 1.772251009941101,
      "learning_rate": 7.477856126977709e-07,
      "loss": 1.9714,
      "num_input_tokens_seen": 259845968,
      "step": 255200
    },
    {
      "epoch": 5.0264810694808135,
      "grad_norm": 1.958411455154419,
      "learning_rate": 7.476065100088674e-07,
      "loss": 1.9458,
      "num_input_tokens_seen": 259947256,
      "step": 255300
    },
    {
      "epoch": 5.028449922230316,
      "grad_norm": 1.9385576248168945,
      "learning_rate": 7.474273652165652e-07,
      "loss": 1.9767,
      "num_input_tokens_seen": 260049344,
      "step": 255400
    },
    {
      "epoch": 5.030418774979819,
      "grad_norm": 1.998084545135498,
      "learning_rate": 7.472481783513266e-07,
      "loss": 1.9445,
      "num_input_tokens_seen": 260151744,
      "step": 255500
    },
    {
      "epoch": 5.032387627729322,
      "grad_norm": 1.7117881774902344,
      "learning_rate": 7.470689494436204e-07,
      "loss": 1.9259,
      "num_input_tokens_seen": 260252792,
      "step": 255600
    },
    {
      "epoch": 5.034356480478825,
      "grad_norm": 1.8511531352996826,
      "learning_rate": 7.468896785239234e-07,
      "loss": 2.0109,
      "num_input_tokens_seen": 260354624,
      "step": 255700
    },
    {
      "epoch": 5.036325333228328,
      "grad_norm": 2.6892521381378174,
      "learning_rate": 7.467103656227189e-07,
      "loss": 1.9802,
      "num_input_tokens_seen": 260455456,
      "step": 255800
    },
    {
      "epoch": 5.038294185977831,
      "grad_norm": 2.0311710834503174,
      "learning_rate": 7.465310107704976e-07,
      "loss": 1.9659,
      "num_input_tokens_seen": 260557856,
      "step": 255900
    },
    {
      "epoch": 5.040263038727334,
      "grad_norm": 1.8495655059814453,
      "learning_rate": 7.463516139977573e-07,
      "loss": 1.9694,
      "num_input_tokens_seen": 260660256,
      "step": 256000
    },
    {
      "epoch": 5.0422318914768365,
      "grad_norm": 1.7853916883468628,
      "learning_rate": 7.461721753350027e-07,
      "loss": 2.0493,
      "num_input_tokens_seen": 260762656,
      "step": 256100
    },
    {
      "epoch": 5.044200744226339,
      "grad_norm": 2.032388687133789,
      "learning_rate": 7.459926948127462e-07,
      "loss": 1.9538,
      "num_input_tokens_seen": 260864416,
      "step": 256200
    },
    {
      "epoch": 5.046169596975842,
      "grad_norm": 2.157252550125122,
      "learning_rate": 7.458131724615067e-07,
      "loss": 1.9323,
      "num_input_tokens_seen": 260966816,
      "step": 256300
    },
    {
      "epoch": 5.048138449725345,
      "grad_norm": 2.6756553649902344,
      "learning_rate": 7.456336083118105e-07,
      "loss": 1.9893,
      "num_input_tokens_seen": 261067816,
      "step": 256400
    },
    {
      "epoch": 5.0501073024748475,
      "grad_norm": 1.8276116847991943,
      "learning_rate": 7.454540023941908e-07,
      "loss": 1.9601,
      "num_input_tokens_seen": 261170216,
      "step": 256500
    },
    {
      "epoch": 5.052076155224351,
      "grad_norm": 2.02026629447937,
      "learning_rate": 7.452743547391884e-07,
      "loss": 2.0006,
      "num_input_tokens_seen": 261271224,
      "step": 256600
    },
    {
      "epoch": 5.054045007973854,
      "grad_norm": 2.038952350616455,
      "learning_rate": 7.450946653773506e-07,
      "loss": 2.0365,
      "num_input_tokens_seen": 261372472,
      "step": 256700
    },
    {
      "epoch": 5.056013860723357,
      "grad_norm": 9.953147888183594,
      "learning_rate": 7.449149343392325e-07,
      "loss": 1.9775,
      "num_input_tokens_seen": 261472112,
      "step": 256800
    },
    {
      "epoch": 5.057982713472859,
      "grad_norm": 1.6050832271575928,
      "learning_rate": 7.447351616553951e-07,
      "loss": 1.9589,
      "num_input_tokens_seen": 261574512,
      "step": 256900
    },
    {
      "epoch": 5.059951566222362,
      "grad_norm": 1.9243571758270264,
      "learning_rate": 7.445553473564079e-07,
      "loss": 1.954,
      "num_input_tokens_seen": 261676912,
      "step": 257000
    },
    {
      "epoch": 5.061920418971865,
      "grad_norm": 1.702769160270691,
      "learning_rate": 7.443754914728465e-07,
      "loss": 2.0053,
      "num_input_tokens_seen": 261778688,
      "step": 257100
    },
    {
      "epoch": 5.063889271721368,
      "grad_norm": 5.400176048278809,
      "learning_rate": 7.44195594035294e-07,
      "loss": 1.9565,
      "num_input_tokens_seen": 261881088,
      "step": 257200
    },
    {
      "epoch": 5.0658581244708705,
      "grad_norm": 5.50039529800415,
      "learning_rate": 7.440156550743403e-07,
      "loss": 1.97,
      "num_input_tokens_seen": 261983488,
      "step": 257300
    },
    {
      "epoch": 5.067826977220374,
      "grad_norm": 1.8719820976257324,
      "learning_rate": 7.438356746205825e-07,
      "loss": 1.9726,
      "num_input_tokens_seen": 262085384,
      "step": 257400
    },
    {
      "epoch": 5.069795829969877,
      "grad_norm": 2.159721851348877,
      "learning_rate": 7.43655652704625e-07,
      "loss": 2.0127,
      "num_input_tokens_seen": 262187104,
      "step": 257500
    },
    {
      "epoch": 5.07176468271938,
      "grad_norm": 1.9829753637313843,
      "learning_rate": 7.434755893570787e-07,
      "loss": 1.9911,
      "num_input_tokens_seen": 262289400,
      "step": 257600
    },
    {
      "epoch": 5.073733535468882,
      "grad_norm": 1.8157498836517334,
      "learning_rate": 7.43295484608562e-07,
      "loss": 1.9775,
      "num_input_tokens_seen": 262391800,
      "step": 257700
    },
    {
      "epoch": 5.075702388218385,
      "grad_norm": 1.903546690940857,
      "learning_rate": 7.431153384897004e-07,
      "loss": 1.9711,
      "num_input_tokens_seen": 262493344,
      "step": 257800
    },
    {
      "epoch": 5.077671240967888,
      "grad_norm": 2.0336084365844727,
      "learning_rate": 7.42935151031126e-07,
      "loss": 1.9974,
      "num_input_tokens_seen": 262594912,
      "step": 257900
    },
    {
      "epoch": 5.079640093717391,
      "grad_norm": 2.0679564476013184,
      "learning_rate": 7.427549222634779e-07,
      "loss": 1.9706,
      "num_input_tokens_seen": 262697048,
      "step": 258000
    },
    {
      "epoch": 5.081608946466893,
      "grad_norm": 2.1089797019958496,
      "learning_rate": 7.42574652217403e-07,
      "loss": 1.9856,
      "num_input_tokens_seen": 262799448,
      "step": 258100
    },
    {
      "epoch": 5.083577799216396,
      "grad_norm": 2.1195034980773926,
      "learning_rate": 7.423943409235543e-07,
      "loss": 2.0093,
      "num_input_tokens_seen": 262901072,
      "step": 258200
    },
    {
      "epoch": 5.0855466519659,
      "grad_norm": 1.9945552349090576,
      "learning_rate": 7.422139884125924e-07,
      "loss": 2.0036,
      "num_input_tokens_seen": 263003472,
      "step": 258300
    },
    {
      "epoch": 5.087515504715403,
      "grad_norm": 1.980982780456543,
      "learning_rate": 7.420335947151849e-07,
      "loss": 2.0192,
      "num_input_tokens_seen": 263105160,
      "step": 258400
    },
    {
      "epoch": 5.089484357464905,
      "grad_norm": 1.9468029737472534,
      "learning_rate": 7.418531598620058e-07,
      "loss": 1.9804,
      "num_input_tokens_seen": 263205152,
      "step": 258500
    },
    {
      "epoch": 5.091453210214408,
      "grad_norm": 2.0993528366088867,
      "learning_rate": 7.416726838837368e-07,
      "loss": 1.9799,
      "num_input_tokens_seen": 263306312,
      "step": 258600
    },
    {
      "epoch": 5.093422062963911,
      "grad_norm": 2.0031890869140625,
      "learning_rate": 7.414921668110663e-07,
      "loss": 1.9747,
      "num_input_tokens_seen": 263408096,
      "step": 258700
    },
    {
      "epoch": 5.095390915713414,
      "grad_norm": 1.8896909952163696,
      "learning_rate": 7.413116086746897e-07,
      "loss": 2.0072,
      "num_input_tokens_seen": 263509784,
      "step": 258800
    },
    {
      "epoch": 5.097359768462916,
      "grad_norm": 1.9193813800811768,
      "learning_rate": 7.411310095053093e-07,
      "loss": 1.9938,
      "num_input_tokens_seen": 263611888,
      "step": 258900
    },
    {
      "epoch": 5.099328621212419,
      "grad_norm": 2.0843522548675537,
      "learning_rate": 7.409503693336346e-07,
      "loss": 1.9905,
      "num_input_tokens_seen": 263713360,
      "step": 259000
    },
    {
      "epoch": 5.101297473961923,
      "grad_norm": 2.25429630279541,
      "learning_rate": 7.407696881903818e-07,
      "loss": 2.0082,
      "num_input_tokens_seen": 263814176,
      "step": 259100
    },
    {
      "epoch": 5.1032663267114255,
      "grad_norm": 2.1886987686157227,
      "learning_rate": 7.405889661062743e-07,
      "loss": 1.999,
      "num_input_tokens_seen": 263915088,
      "step": 259200
    },
    {
      "epoch": 5.105235179460928,
      "grad_norm": 1.743133544921875,
      "learning_rate": 7.404082031120422e-07,
      "loss": 1.9333,
      "num_input_tokens_seen": 264016672,
      "step": 259300
    },
    {
      "epoch": 5.107204032210431,
      "grad_norm": 1.8552533388137817,
      "learning_rate": 7.40227399238423e-07,
      "loss": 1.994,
      "num_input_tokens_seen": 264118272,
      "step": 259400
    },
    {
      "epoch": 5.109172884959934,
      "grad_norm": 1.9470716714859009,
      "learning_rate": 7.400465545161607e-07,
      "loss": 1.9545,
      "num_input_tokens_seen": 264220672,
      "step": 259500
    },
    {
      "epoch": 5.111141737709437,
      "grad_norm": 1.7200884819030762,
      "learning_rate": 7.398656689760065e-07,
      "loss": 1.964,
      "num_input_tokens_seen": 264323072,
      "step": 259600
    },
    {
      "epoch": 5.113110590458939,
      "grad_norm": 1.9774038791656494,
      "learning_rate": 7.396847426487183e-07,
      "loss": 1.9751,
      "num_input_tokens_seen": 264424128,
      "step": 259700
    },
    {
      "epoch": 5.115079443208442,
      "grad_norm": 1.903700828552246,
      "learning_rate": 7.395037755650613e-07,
      "loss": 2.0092,
      "num_input_tokens_seen": 264525896,
      "step": 259800
    },
    {
      "epoch": 5.117048295957945,
      "grad_norm": 1.8798542022705078,
      "learning_rate": 7.393227677558072e-07,
      "loss": 1.9909,
      "num_input_tokens_seen": 264628296,
      "step": 259900
    },
    {
      "epoch": 5.1190171487074485,
      "grad_norm": 2.2121682167053223,
      "learning_rate": 7.391417192517352e-07,
      "loss": 1.9974,
      "num_input_tokens_seen": 264728216,
      "step": 260000
    },
    {
      "epoch": 5.120986001456951,
      "grad_norm": 1.7667813301086426,
      "learning_rate": 7.389606300836306e-07,
      "loss": 1.9967,
      "num_input_tokens_seen": 264830616,
      "step": 260100
    },
    {
      "epoch": 5.122954854206454,
      "grad_norm": 1.9347078800201416,
      "learning_rate": 7.387795002822866e-07,
      "loss": 1.9633,
      "num_input_tokens_seen": 264932528,
      "step": 260200
    },
    {
      "epoch": 5.124923706955957,
      "grad_norm": 11.775089263916016,
      "learning_rate": 7.385983298785023e-07,
      "loss": 1.9677,
      "num_input_tokens_seen": 265034360,
      "step": 260300
    },
    {
      "epoch": 5.1268925597054595,
      "grad_norm": 1.8936445713043213,
      "learning_rate": 7.384171189030848e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 265136048,
      "step": 260400
    },
    {
      "epoch": 5.128861412454962,
      "grad_norm": 1.8390278816223145,
      "learning_rate": 7.382358673868467e-07,
      "loss": 1.9801,
      "num_input_tokens_seen": 265237584,
      "step": 260500
    },
    {
      "epoch": 5.130830265204465,
      "grad_norm": 2.2224788665771484,
      "learning_rate": 7.380545753606091e-07,
      "loss": 2.0165,
      "num_input_tokens_seen": 265339432,
      "step": 260600
    },
    {
      "epoch": 5.132799117953968,
      "grad_norm": 2.1867525577545166,
      "learning_rate": 7.378732428551985e-07,
      "loss": 1.9829,
      "num_input_tokens_seen": 265440536,
      "step": 260700
    },
    {
      "epoch": 5.1347679707034715,
      "grad_norm": 1.8687938451766968,
      "learning_rate": 7.376918699014494e-07,
      "loss": 1.9595,
      "num_input_tokens_seen": 265542360,
      "step": 260800
    },
    {
      "epoch": 5.136736823452974,
      "grad_norm": 1.8443448543548584,
      "learning_rate": 7.375104565302027e-07,
      "loss": 1.9618,
      "num_input_tokens_seen": 265644760,
      "step": 260900
    },
    {
      "epoch": 5.138705676202477,
      "grad_norm": 2.1384170055389404,
      "learning_rate": 7.373290027723059e-07,
      "loss": 2.0161,
      "num_input_tokens_seen": 265747160,
      "step": 261000
    },
    {
      "epoch": 5.14067452895198,
      "grad_norm": 2.0410032272338867,
      "learning_rate": 7.37147508658614e-07,
      "loss": 2.0163,
      "num_input_tokens_seen": 265846320,
      "step": 261100
    },
    {
      "epoch": 5.1426433817014825,
      "grad_norm": 2.0527236461639404,
      "learning_rate": 7.369659742199886e-07,
      "loss": 1.9463,
      "num_input_tokens_seen": 265948720,
      "step": 261200
    },
    {
      "epoch": 5.144612234450985,
      "grad_norm": 2.003117322921753,
      "learning_rate": 7.367843994872978e-07,
      "loss": 2.0115,
      "num_input_tokens_seen": 266049864,
      "step": 261300
    },
    {
      "epoch": 5.146581087200488,
      "grad_norm": 1.7844789028167725,
      "learning_rate": 7.36602784491417e-07,
      "loss": 1.9823,
      "num_input_tokens_seen": 266152264,
      "step": 261400
    },
    {
      "epoch": 5.148549939949991,
      "grad_norm": 2.2948644161224365,
      "learning_rate": 7.364211292632283e-07,
      "loss": 1.9628,
      "num_input_tokens_seen": 266254664,
      "step": 261500
    },
    {
      "epoch": 5.150518792699494,
      "grad_norm": 11.594809532165527,
      "learning_rate": 7.362394338336206e-07,
      "loss": 1.9456,
      "num_input_tokens_seen": 266355376,
      "step": 261600
    },
    {
      "epoch": 5.152487645448997,
      "grad_norm": 1.8288902044296265,
      "learning_rate": 7.360576982334898e-07,
      "loss": 1.9937,
      "num_input_tokens_seen": 266457776,
      "step": 261700
    },
    {
      "epoch": 5.1544564981985,
      "grad_norm": 2.08296799659729,
      "learning_rate": 7.358759224937383e-07,
      "loss": 1.9522,
      "num_input_tokens_seen": 266560176,
      "step": 261800
    },
    {
      "epoch": 5.156425350948003,
      "grad_norm": 1.9673665761947632,
      "learning_rate": 7.356941066452757e-07,
      "loss": 1.9974,
      "num_input_tokens_seen": 266661616,
      "step": 261900
    },
    {
      "epoch": 5.1583942036975055,
      "grad_norm": 2.078920364379883,
      "learning_rate": 7.355122507190184e-07,
      "loss": 2.0107,
      "num_input_tokens_seen": 266762344,
      "step": 262000
    },
    {
      "epoch": 5.160363056447008,
      "grad_norm": 1.879564881324768,
      "learning_rate": 7.353303547458891e-07,
      "loss": 1.9975,
      "num_input_tokens_seen": 266864744,
      "step": 262100
    },
    {
      "epoch": 5.162331909196511,
      "grad_norm": 1.6097413301467896,
      "learning_rate": 7.351484187568177e-07,
      "loss": 1.9982,
      "num_input_tokens_seen": 266966320,
      "step": 262200
    },
    {
      "epoch": 5.164300761946014,
      "grad_norm": 1.8999217748641968,
      "learning_rate": 7.349664427827413e-07,
      "loss": 1.9637,
      "num_input_tokens_seen": 267068720,
      "step": 262300
    },
    {
      "epoch": 5.166269614695517,
      "grad_norm": 2.17246675491333,
      "learning_rate": 7.347844268546029e-07,
      "loss": 2.0391,
      "num_input_tokens_seen": 267170592,
      "step": 262400
    },
    {
      "epoch": 5.16823846744502,
      "grad_norm": 1.8745571374893188,
      "learning_rate": 7.346023710033531e-07,
      "loss": 2.0162,
      "num_input_tokens_seen": 267272944,
      "step": 262500
    },
    {
      "epoch": 5.170207320194523,
      "grad_norm": 1.9142630100250244,
      "learning_rate": 7.344202752599488e-07,
      "loss": 2.0217,
      "num_input_tokens_seen": 267375344,
      "step": 262600
    },
    {
      "epoch": 5.172176172944026,
      "grad_norm": 1.863242745399475,
      "learning_rate": 7.342381396553539e-07,
      "loss": 1.9593,
      "num_input_tokens_seen": 267477744,
      "step": 262700
    },
    {
      "epoch": 5.174145025693528,
      "grad_norm": 1.8339025974273682,
      "learning_rate": 7.340559642205388e-07,
      "loss": 2.0127,
      "num_input_tokens_seen": 267578648,
      "step": 262800
    },
    {
      "epoch": 5.176113878443031,
      "grad_norm": 1.8508989810943604,
      "learning_rate": 7.338737489864814e-07,
      "loss": 1.9917,
      "num_input_tokens_seen": 267681048,
      "step": 262900
    },
    {
      "epoch": 5.178082731192534,
      "grad_norm": 1.758082389831543,
      "learning_rate": 7.336914939841654e-07,
      "loss": 1.9616,
      "num_input_tokens_seen": 267783448,
      "step": 263000
    },
    {
      "epoch": 5.180051583942037,
      "grad_norm": 1.7795236110687256,
      "learning_rate": 7.335091992445819e-07,
      "loss": 1.9658,
      "num_input_tokens_seen": 267885848,
      "step": 263100
    },
    {
      "epoch": 5.1820204366915394,
      "grad_norm": 1.7803370952606201,
      "learning_rate": 7.333268647987286e-07,
      "loss": 1.995,
      "num_input_tokens_seen": 267987696,
      "step": 263200
    },
    {
      "epoch": 5.183989289441043,
      "grad_norm": 1.8964024782180786,
      "learning_rate": 7.3314449067761e-07,
      "loss": 1.961,
      "num_input_tokens_seen": 268090096,
      "step": 263300
    },
    {
      "epoch": 5.185958142190546,
      "grad_norm": 2.2240588665008545,
      "learning_rate": 7.329620769122371e-07,
      "loss": 2.0312,
      "num_input_tokens_seen": 268192496,
      "step": 263400
    },
    {
      "epoch": 5.187926994940049,
      "grad_norm": 1.8488363027572632,
      "learning_rate": 7.32779623533628e-07,
      "loss": 1.9876,
      "num_input_tokens_seen": 268294600,
      "step": 263500
    },
    {
      "epoch": 5.189895847689551,
      "grad_norm": 2.171096086502075,
      "learning_rate": 7.325971305728071e-07,
      "loss": 2.0159,
      "num_input_tokens_seen": 268395568,
      "step": 263600
    },
    {
      "epoch": 5.191864700439054,
      "grad_norm": 1.946254014968872,
      "learning_rate": 7.324145980608061e-07,
      "loss": 1.9888,
      "num_input_tokens_seen": 268497280,
      "step": 263700
    },
    {
      "epoch": 5.193833553188557,
      "grad_norm": 1.659362554550171,
      "learning_rate": 7.322320260286628e-07,
      "loss": 1.9865,
      "num_input_tokens_seen": 268598872,
      "step": 263800
    },
    {
      "epoch": 5.19580240593806,
      "grad_norm": 2.198559045791626,
      "learning_rate": 7.320494145074225e-07,
      "loss": 2.0033,
      "num_input_tokens_seen": 268701272,
      "step": 263900
    },
    {
      "epoch": 5.197771258687562,
      "grad_norm": 1.7130135297775269,
      "learning_rate": 7.318667635281362e-07,
      "loss": 2.0082,
      "num_input_tokens_seen": 268802096,
      "step": 264000
    },
    {
      "epoch": 5.199740111437066,
      "grad_norm": 2.05407977104187,
      "learning_rate": 7.316840731218626e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 268904496,
      "step": 264100
    },
    {
      "epoch": 5.201708964186569,
      "grad_norm": 1.9083105325698853,
      "learning_rate": 7.315013433196664e-07,
      "loss": 1.9745,
      "num_input_tokens_seen": 269005560,
      "step": 264200
    },
    {
      "epoch": 5.203677816936072,
      "grad_norm": 1.7571542263031006,
      "learning_rate": 7.313185741526194e-07,
      "loss": 1.9806,
      "num_input_tokens_seen": 269107960,
      "step": 264300
    },
    {
      "epoch": 5.205646669685574,
      "grad_norm": 2.0068085193634033,
      "learning_rate": 7.311357656517998e-07,
      "loss": 1.9971,
      "num_input_tokens_seen": 269210360,
      "step": 264400
    },
    {
      "epoch": 5.207615522435077,
      "grad_norm": 2.325728178024292,
      "learning_rate": 7.309529178482929e-07,
      "loss": 2.0123,
      "num_input_tokens_seen": 269312760,
      "step": 264500
    },
    {
      "epoch": 5.20958437518458,
      "grad_norm": 1.9679704904556274,
      "learning_rate": 7.3077003077319e-07,
      "loss": 1.975,
      "num_input_tokens_seen": 269413704,
      "step": 264600
    },
    {
      "epoch": 5.211553227934083,
      "grad_norm": 1.719275951385498,
      "learning_rate": 7.3058710445759e-07,
      "loss": 1.9636,
      "num_input_tokens_seen": 269514696,
      "step": 264700
    },
    {
      "epoch": 5.213522080683585,
      "grad_norm": 1.858886957168579,
      "learning_rate": 7.304041389325977e-07,
      "loss": 1.9479,
      "num_input_tokens_seen": 269616544,
      "step": 264800
    },
    {
      "epoch": 5.215490933433088,
      "grad_norm": 1.9864132404327393,
      "learning_rate": 7.302211342293248e-07,
      "loss": 1.9733,
      "num_input_tokens_seen": 269718400,
      "step": 264900
    },
    {
      "epoch": 5.217459786182592,
      "grad_norm": 1.702999234199524,
      "learning_rate": 7.300380903788899e-07,
      "loss": 1.9468,
      "num_input_tokens_seen": 269820800,
      "step": 265000
    },
    {
      "epoch": 5.2194286389320945,
      "grad_norm": 2.4030091762542725,
      "learning_rate": 7.298550074124177e-07,
      "loss": 1.9844,
      "num_input_tokens_seen": 269920656,
      "step": 265100
    },
    {
      "epoch": 5.221397491681597,
      "grad_norm": 2.1367692947387695,
      "learning_rate": 7.296718853610407e-07,
      "loss": 1.9744,
      "num_input_tokens_seen": 270022280,
      "step": 265200
    },
    {
      "epoch": 5.2233663444311,
      "grad_norm": 1.9486796855926514,
      "learning_rate": 7.294887242558962e-07,
      "loss": 2.0031,
      "num_input_tokens_seen": 270123712,
      "step": 265300
    },
    {
      "epoch": 5.225335197180603,
      "grad_norm": 2.01839280128479,
      "learning_rate": 7.293055241281301e-07,
      "loss": 2.0585,
      "num_input_tokens_seen": 270226112,
      "step": 265400
    },
    {
      "epoch": 5.227304049930106,
      "grad_norm": 1.9012913703918457,
      "learning_rate": 7.291222850088935e-07,
      "loss": 1.9708,
      "num_input_tokens_seen": 270328512,
      "step": 265500
    },
    {
      "epoch": 5.229272902679608,
      "grad_norm": 1.9223334789276123,
      "learning_rate": 7.289390069293448e-07,
      "loss": 2.0117,
      "num_input_tokens_seen": 270430912,
      "step": 265600
    },
    {
      "epoch": 5.231241755429111,
      "grad_norm": 2.0429482460021973,
      "learning_rate": 7.287556899206491e-07,
      "loss": 2.0293,
      "num_input_tokens_seen": 270533312,
      "step": 265700
    },
    {
      "epoch": 5.233210608178615,
      "grad_norm": 1.7632092237472534,
      "learning_rate": 7.285723340139776e-07,
      "loss": 1.991,
      "num_input_tokens_seen": 270635536,
      "step": 265800
    },
    {
      "epoch": 5.2351794609281175,
      "grad_norm": 1.9305258989334106,
      "learning_rate": 7.283889392405086e-07,
      "loss": 1.9637,
      "num_input_tokens_seen": 270737936,
      "step": 265900
    },
    {
      "epoch": 5.23714831367762,
      "grad_norm": 1.7483024597167969,
      "learning_rate": 7.282055056314269e-07,
      "loss": 2.0053,
      "num_input_tokens_seen": 270839576,
      "step": 266000
    },
    {
      "epoch": 5.239117166427123,
      "grad_norm": 1.9088778495788574,
      "learning_rate": 7.280220332179236e-07,
      "loss": 1.9983,
      "num_input_tokens_seen": 270941144,
      "step": 266100
    },
    {
      "epoch": 5.241086019176626,
      "grad_norm": 1.9415194988250732,
      "learning_rate": 7.278385220311969e-07,
      "loss": 2.0006,
      "num_input_tokens_seen": 271041760,
      "step": 266200
    },
    {
      "epoch": 5.2430548719261285,
      "grad_norm": 1.802951693534851,
      "learning_rate": 7.276549721024511e-07,
      "loss": 1.9903,
      "num_input_tokens_seen": 271143440,
      "step": 266300
    },
    {
      "epoch": 5.245023724675631,
      "grad_norm": 1.9440362453460693,
      "learning_rate": 7.274713834628976e-07,
      "loss": 1.9921,
      "num_input_tokens_seen": 271245072,
      "step": 266400
    },
    {
      "epoch": 5.246992577425134,
      "grad_norm": 1.9700713157653809,
      "learning_rate": 7.272877561437538e-07,
      "loss": 1.9831,
      "num_input_tokens_seen": 271346584,
      "step": 266500
    },
    {
      "epoch": 5.248961430174637,
      "grad_norm": 3.628180742263794,
      "learning_rate": 7.271040901762442e-07,
      "loss": 1.9835,
      "num_input_tokens_seen": 271447520,
      "step": 266600
    },
    {
      "epoch": 5.25093028292414,
      "grad_norm": 1.8094944953918457,
      "learning_rate": 7.269203855915994e-07,
      "loss": 1.9918,
      "num_input_tokens_seen": 271549344,
      "step": 266700
    },
    {
      "epoch": 5.252899135673643,
      "grad_norm": 2.1155598163604736,
      "learning_rate": 7.267366424210569e-07,
      "loss": 1.9705,
      "num_input_tokens_seen": 271651744,
      "step": 266800
    },
    {
      "epoch": 5.254867988423146,
      "grad_norm": 2.0874040126800537,
      "learning_rate": 7.265528606958608e-07,
      "loss": 1.9565,
      "num_input_tokens_seen": 271753728,
      "step": 266900
    },
    {
      "epoch": 5.256836841172649,
      "grad_norm": 2.0286948680877686,
      "learning_rate": 7.263690404472617e-07,
      "loss": 1.9715,
      "num_input_tokens_seen": 271856128,
      "step": 267000
    },
    {
      "epoch": 5.2588056939221515,
      "grad_norm": 2.1304943561553955,
      "learning_rate": 7.261851817065166e-07,
      "loss": 1.9564,
      "num_input_tokens_seen": 271957712,
      "step": 267100
    },
    {
      "epoch": 5.260774546671654,
      "grad_norm": 2.0912668704986572,
      "learning_rate": 7.260012845048889e-07,
      "loss": 1.9743,
      "num_input_tokens_seen": 272060112,
      "step": 267200
    },
    {
      "epoch": 5.262743399421157,
      "grad_norm": 1.8552043437957764,
      "learning_rate": 7.258173488736491e-07,
      "loss": 1.9784,
      "num_input_tokens_seen": 272161680,
      "step": 267300
    },
    {
      "epoch": 5.26471225217066,
      "grad_norm": 1.8491747379302979,
      "learning_rate": 7.256333748440738e-07,
      "loss": 1.989,
      "num_input_tokens_seen": 272261816,
      "step": 267400
    },
    {
      "epoch": 5.266681104920163,
      "grad_norm": 2.0953633785247803,
      "learning_rate": 7.254493624474459e-07,
      "loss": 1.9825,
      "num_input_tokens_seen": 272363528,
      "step": 267500
    },
    {
      "epoch": 5.268649957669666,
      "grad_norm": 1.7955294847488403,
      "learning_rate": 7.252653117150558e-07,
      "loss": 1.9691,
      "num_input_tokens_seen": 272465928,
      "step": 267600
    },
    {
      "epoch": 5.270618810419169,
      "grad_norm": 1.9059914350509644,
      "learning_rate": 7.250812226781992e-07,
      "loss": 1.9479,
      "num_input_tokens_seen": 272568328,
      "step": 267700
    },
    {
      "epoch": 5.272587663168672,
      "grad_norm": 1.9488171339035034,
      "learning_rate": 7.248970953681791e-07,
      "loss": 1.9611,
      "num_input_tokens_seen": 272669912,
      "step": 267800
    },
    {
      "epoch": 5.274556515918174,
      "grad_norm": 1.8518226146697998,
      "learning_rate": 7.247129298163047e-07,
      "loss": 1.9795,
      "num_input_tokens_seen": 272772312,
      "step": 267900
    },
    {
      "epoch": 5.276525368667677,
      "grad_norm": 2.598212957382202,
      "learning_rate": 7.245287260538922e-07,
      "loss": 1.9582,
      "num_input_tokens_seen": 272874712,
      "step": 268000
    },
    {
      "epoch": 5.27849422141718,
      "grad_norm": 2.0378260612487793,
      "learning_rate": 7.243444841122633e-07,
      "loss": 2.0094,
      "num_input_tokens_seen": 272976384,
      "step": 268100
    },
    {
      "epoch": 5.280463074166683,
      "grad_norm": 2.052414894104004,
      "learning_rate": 7.241602040227473e-07,
      "loss": 1.96,
      "num_input_tokens_seen": 273078784,
      "step": 268200
    },
    {
      "epoch": 5.282431926916186,
      "grad_norm": 2.1668992042541504,
      "learning_rate": 7.239758858166792e-07,
      "loss": 1.9581,
      "num_input_tokens_seen": 273181184,
      "step": 268300
    },
    {
      "epoch": 5.284400779665689,
      "grad_norm": 1.810660719871521,
      "learning_rate": 7.237915295254009e-07,
      "loss": 2.0088,
      "num_input_tokens_seen": 273283008,
      "step": 268400
    },
    {
      "epoch": 5.286369632415192,
      "grad_norm": 1.9220634698867798,
      "learning_rate": 7.236071351802604e-07,
      "loss": 1.9636,
      "num_input_tokens_seen": 273384560,
      "step": 268500
    },
    {
      "epoch": 5.288338485164695,
      "grad_norm": 1.7055270671844482,
      "learning_rate": 7.234227028126126e-07,
      "loss": 1.9836,
      "num_input_tokens_seen": 273486440,
      "step": 268600
    },
    {
      "epoch": 5.290307337914197,
      "grad_norm": 2.014674663543701,
      "learning_rate": 7.232382324538187e-07,
      "loss": 1.9769,
      "num_input_tokens_seen": 273588264,
      "step": 268700
    },
    {
      "epoch": 5.2922761906637,
      "grad_norm": 1.929937720298767,
      "learning_rate": 7.230537241352461e-07,
      "loss": 1.9866,
      "num_input_tokens_seen": 273690064,
      "step": 268800
    },
    {
      "epoch": 5.294245043413203,
      "grad_norm": 1.5885132551193237,
      "learning_rate": 7.228691778882692e-07,
      "loss": 1.9612,
      "num_input_tokens_seen": 273792464,
      "step": 268900
    },
    {
      "epoch": 5.296213896162706,
      "grad_norm": 1.7675259113311768,
      "learning_rate": 7.226845937442682e-07,
      "loss": 1.9569,
      "num_input_tokens_seen": 273894000,
      "step": 269000
    },
    {
      "epoch": 5.298182748912209,
      "grad_norm": 2.164607286453247,
      "learning_rate": 7.224999717346303e-07,
      "loss": 2.022,
      "num_input_tokens_seen": 273994816,
      "step": 269100
    },
    {
      "epoch": 5.300151601661712,
      "grad_norm": 2.145895004272461,
      "learning_rate": 7.223153118907489e-07,
      "loss": 2.0185,
      "num_input_tokens_seen": 274097216,
      "step": 269200
    },
    {
      "epoch": 5.302120454411215,
      "grad_norm": 1.8865959644317627,
      "learning_rate": 7.221306142440238e-07,
      "loss": 1.9796,
      "num_input_tokens_seen": 274199616,
      "step": 269300
    },
    {
      "epoch": 5.304089307160718,
      "grad_norm": 2.5805323123931885,
      "learning_rate": 7.219458788258613e-07,
      "loss": 2.0091,
      "num_input_tokens_seen": 274301400,
      "step": 269400
    },
    {
      "epoch": 5.30605815991022,
      "grad_norm": 1.778024673461914,
      "learning_rate": 7.21761105667674e-07,
      "loss": 1.988,
      "num_input_tokens_seen": 274403176,
      "step": 269500
    },
    {
      "epoch": 5.308027012659723,
      "grad_norm": 2.467095375061035,
      "learning_rate": 7.215762948008812e-07,
      "loss": 1.9659,
      "num_input_tokens_seen": 274504024,
      "step": 269600
    },
    {
      "epoch": 5.309995865409226,
      "grad_norm": 1.9283493757247925,
      "learning_rate": 7.213914462569081e-07,
      "loss": 1.9774,
      "num_input_tokens_seen": 274606136,
      "step": 269700
    },
    {
      "epoch": 5.311964718158729,
      "grad_norm": 1.9295318126678467,
      "learning_rate": 7.212065600671872e-07,
      "loss": 2.0061,
      "num_input_tokens_seen": 274706624,
      "step": 269800
    },
    {
      "epoch": 5.313933570908231,
      "grad_norm": 0.0,
      "learning_rate": 7.210216362631564e-07,
      "loss": 1.9677,
      "num_input_tokens_seen": 274808480,
      "step": 269900
    },
    {
      "epoch": 5.315902423657735,
      "grad_norm": 1.8951165676116943,
      "learning_rate": 7.208366748762604e-07,
      "loss": 1.9482,
      "num_input_tokens_seen": 274910880,
      "step": 270000
    },
    {
      "epoch": 5.317871276407238,
      "grad_norm": 1.9429322481155396,
      "learning_rate": 7.206516759379506e-07,
      "loss": 2.0014,
      "num_input_tokens_seen": 275012448,
      "step": 270100
    },
    {
      "epoch": 5.3198401291567405,
      "grad_norm": 1.852536916732788,
      "learning_rate": 7.204666394796843e-07,
      "loss": 1.96,
      "num_input_tokens_seen": 275113992,
      "step": 270200
    },
    {
      "epoch": 5.321808981906243,
      "grad_norm": 1.945887804031372,
      "learning_rate": 7.202815655329255e-07,
      "loss": 2.0011,
      "num_input_tokens_seen": 275216392,
      "step": 270300
    },
    {
      "epoch": 5.323777834655746,
      "grad_norm": 2.177349805831909,
      "learning_rate": 7.200964541291443e-07,
      "loss": 1.9628,
      "num_input_tokens_seen": 275318792,
      "step": 270400
    },
    {
      "epoch": 5.325746687405249,
      "grad_norm": 1.929465889930725,
      "learning_rate": 7.199113052998177e-07,
      "loss": 1.9621,
      "num_input_tokens_seen": 275420112,
      "step": 270500
    },
    {
      "epoch": 5.327715540154752,
      "grad_norm": 1.8120075464248657,
      "learning_rate": 7.197261190764283e-07,
      "loss": 2.0163,
      "num_input_tokens_seen": 275522512,
      "step": 270600
    },
    {
      "epoch": 5.329684392904254,
      "grad_norm": 1.9308222532272339,
      "learning_rate": 7.195408954904656e-07,
      "loss": 2.009,
      "num_input_tokens_seen": 275624912,
      "step": 270700
    },
    {
      "epoch": 5.331653245653758,
      "grad_norm": 2.230597496032715,
      "learning_rate": 7.193556345734253e-07,
      "loss": 1.9733,
      "num_input_tokens_seen": 275727312,
      "step": 270800
    },
    {
      "epoch": 5.333622098403261,
      "grad_norm": 1.8992881774902344,
      "learning_rate": 7.191703363568095e-07,
      "loss": 1.99,
      "num_input_tokens_seen": 275828920,
      "step": 270900
    },
    {
      "epoch": 5.3355909511527635,
      "grad_norm": 1.7125335931777954,
      "learning_rate": 7.189850008721264e-07,
      "loss": 1.9107,
      "num_input_tokens_seen": 275931320,
      "step": 271000
    },
    {
      "epoch": 5.337559803902266,
      "grad_norm": 1.9915541410446167,
      "learning_rate": 7.187996281508912e-07,
      "loss": 1.9711,
      "num_input_tokens_seen": 276033720,
      "step": 271100
    },
    {
      "epoch": 5.339528656651769,
      "grad_norm": 1.9223518371582031,
      "learning_rate": 7.186142182246245e-07,
      "loss": 2.0245,
      "num_input_tokens_seen": 276136120,
      "step": 271200
    },
    {
      "epoch": 5.341497509401272,
      "grad_norm": 2.25526762008667,
      "learning_rate": 7.18428771124854e-07,
      "loss": 1.9876,
      "num_input_tokens_seen": 276237960,
      "step": 271300
    },
    {
      "epoch": 5.3434663621507745,
      "grad_norm": 2.335881471633911,
      "learning_rate": 7.182432868831131e-07,
      "loss": 1.9671,
      "num_input_tokens_seen": 276340360,
      "step": 271400
    },
    {
      "epoch": 5.345435214900277,
      "grad_norm": 1.7091542482376099,
      "learning_rate": 7.180577655309422e-07,
      "loss": 1.9718,
      "num_input_tokens_seen": 276442760,
      "step": 271500
    },
    {
      "epoch": 5.34740406764978,
      "grad_norm": 2.0681424140930176,
      "learning_rate": 7.178722070998872e-07,
      "loss": 2.0358,
      "num_input_tokens_seen": 276545112,
      "step": 271600
    },
    {
      "epoch": 5.349372920399284,
      "grad_norm": 1.8593565225601196,
      "learning_rate": 7.176866116215014e-07,
      "loss": 1.9698,
      "num_input_tokens_seen": 276647512,
      "step": 271700
    },
    {
      "epoch": 5.3513417731487865,
      "grad_norm": 2.043499708175659,
      "learning_rate": 7.17500979127343e-07,
      "loss": 2.0253,
      "num_input_tokens_seen": 276749544,
      "step": 271800
    },
    {
      "epoch": 5.353310625898289,
      "grad_norm": 2.8592991828918457,
      "learning_rate": 7.173153096489776e-07,
      "loss": 1.9873,
      "num_input_tokens_seen": 276851944,
      "step": 271900
    },
    {
      "epoch": 5.355279478647792,
      "grad_norm": 2.0717766284942627,
      "learning_rate": 7.171296032179768e-07,
      "loss": 1.9718,
      "num_input_tokens_seen": 276954344,
      "step": 272000
    },
    {
      "epoch": 5.357248331397295,
      "grad_norm": 1.941322922706604,
      "learning_rate": 7.169438598659185e-07,
      "loss": 2.0223,
      "num_input_tokens_seen": 277056200,
      "step": 272100
    },
    {
      "epoch": 5.3592171841467975,
      "grad_norm": 2.0059738159179688,
      "learning_rate": 7.167580796243864e-07,
      "loss": 1.9871,
      "num_input_tokens_seen": 277158600,
      "step": 272200
    },
    {
      "epoch": 5.3611860368963,
      "grad_norm": 3.2781922817230225,
      "learning_rate": 7.165722625249712e-07,
      "loss": 2.0257,
      "num_input_tokens_seen": 277260912,
      "step": 272300
    },
    {
      "epoch": 5.363154889645803,
      "grad_norm": 1.8294517993927002,
      "learning_rate": 7.163864085992693e-07,
      "loss": 2.0039,
      "num_input_tokens_seen": 277361752,
      "step": 272400
    },
    {
      "epoch": 5.365123742395307,
      "grad_norm": 1.742654800415039,
      "learning_rate": 7.162005178788839e-07,
      "loss": 1.9659,
      "num_input_tokens_seen": 277464152,
      "step": 272500
    },
    {
      "epoch": 5.367092595144809,
      "grad_norm": 1.9192051887512207,
      "learning_rate": 7.160145903954238e-07,
      "loss": 1.9592,
      "num_input_tokens_seen": 277566552,
      "step": 272600
    },
    {
      "epoch": 5.369061447894312,
      "grad_norm": 2.0247249603271484,
      "learning_rate": 7.158286261805046e-07,
      "loss": 1.9748,
      "num_input_tokens_seen": 277667784,
      "step": 272700
    },
    {
      "epoch": 5.371030300643815,
      "grad_norm": 2.092761278152466,
      "learning_rate": 7.156426252657478e-07,
      "loss": 1.9688,
      "num_input_tokens_seen": 277770184,
      "step": 272800
    },
    {
      "epoch": 5.372999153393318,
      "grad_norm": 1.821317434310913,
      "learning_rate": 7.154565876827814e-07,
      "loss": 1.9271,
      "num_input_tokens_seen": 277872584,
      "step": 272900
    },
    {
      "epoch": 5.37496800614282,
      "grad_norm": 1.9703651666641235,
      "learning_rate": 7.152705134632396e-07,
      "loss": 1.9833,
      "num_input_tokens_seen": 277974440,
      "step": 273000
    },
    {
      "epoch": 5.376936858892323,
      "grad_norm": 2.015634059906006,
      "learning_rate": 7.150844026387627e-07,
      "loss": 2.0069,
      "num_input_tokens_seen": 278075432,
      "step": 273100
    },
    {
      "epoch": 5.378905711641826,
      "grad_norm": 1.8306469917297363,
      "learning_rate": 7.148982552409972e-07,
      "loss": 1.9866,
      "num_input_tokens_seen": 278177832,
      "step": 273200
    },
    {
      "epoch": 5.380874564391329,
      "grad_norm": 5.945248126983643,
      "learning_rate": 7.147120713015958e-07,
      "loss": 1.9769,
      "num_input_tokens_seen": 278280232,
      "step": 273300
    },
    {
      "epoch": 5.382843417140832,
      "grad_norm": 1.8421549797058105,
      "learning_rate": 7.145258508522178e-07,
      "loss": 1.9737,
      "num_input_tokens_seen": 278382632,
      "step": 273400
    },
    {
      "epoch": 5.384812269890335,
      "grad_norm": 2.0037624835968018,
      "learning_rate": 7.143395939245282e-07,
      "loss": 1.9702,
      "num_input_tokens_seen": 278485032,
      "step": 273500
    },
    {
      "epoch": 5.386781122639838,
      "grad_norm": 1.753377079963684,
      "learning_rate": 7.141533005501985e-07,
      "loss": 2.0035,
      "num_input_tokens_seen": 278587432,
      "step": 273600
    },
    {
      "epoch": 5.388749975389341,
      "grad_norm": 1.9795148372650146,
      "learning_rate": 7.139669707609063e-07,
      "loss": 1.9657,
      "num_input_tokens_seen": 278688496,
      "step": 273700
    },
    {
      "epoch": 5.390718828138843,
      "grad_norm": 1.8461229801177979,
      "learning_rate": 7.137806045883353e-07,
      "loss": 2.0223,
      "num_input_tokens_seen": 278790512,
      "step": 273800
    },
    {
      "epoch": 5.392687680888346,
      "grad_norm": 1.8531826734542847,
      "learning_rate": 7.135942020641757e-07,
      "loss": 1.9799,
      "num_input_tokens_seen": 278892256,
      "step": 273900
    },
    {
      "epoch": 5.394656533637849,
      "grad_norm": 2.0050060749053955,
      "learning_rate": 7.134077632201236e-07,
      "loss": 2.0154,
      "num_input_tokens_seen": 278992488,
      "step": 274000
    },
    {
      "epoch": 5.396625386387352,
      "grad_norm": 1.7922987937927246,
      "learning_rate": 7.132212880878812e-07,
      "loss": 2.0385,
      "num_input_tokens_seen": 279094224,
      "step": 274100
    },
    {
      "epoch": 5.398594239136855,
      "grad_norm": 2.1020994186401367,
      "learning_rate": 7.130347766991572e-07,
      "loss": 1.9438,
      "num_input_tokens_seen": 279195992,
      "step": 274200
    },
    {
      "epoch": 5.400563091886358,
      "grad_norm": 1.7647255659103394,
      "learning_rate": 7.128482290856663e-07,
      "loss": 1.9707,
      "num_input_tokens_seen": 279297160,
      "step": 274300
    },
    {
      "epoch": 5.402531944635861,
      "grad_norm": 2.0838844776153564,
      "learning_rate": 7.126616452791293e-07,
      "loss": 1.9841,
      "num_input_tokens_seen": 279399072,
      "step": 274400
    },
    {
      "epoch": 5.404500797385364,
      "grad_norm": 1.7502622604370117,
      "learning_rate": 7.124750253112732e-07,
      "loss": 1.9721,
      "num_input_tokens_seen": 279501472,
      "step": 274500
    },
    {
      "epoch": 5.406469650134866,
      "grad_norm": 2.0231852531433105,
      "learning_rate": 7.12288369213831e-07,
      "loss": 1.99,
      "num_input_tokens_seen": 279603168,
      "step": 274600
    },
    {
      "epoch": 5.408438502884369,
      "grad_norm": 1.9301915168762207,
      "learning_rate": 7.121016770185423e-07,
      "loss": 1.9967,
      "num_input_tokens_seen": 279705568,
      "step": 274700
    },
    {
      "epoch": 5.410407355633872,
      "grad_norm": 1.983007788658142,
      "learning_rate": 7.119149487571523e-07,
      "loss": 1.9451,
      "num_input_tokens_seen": 279807000,
      "step": 274800
    },
    {
      "epoch": 5.412376208383375,
      "grad_norm": 1.8708869218826294,
      "learning_rate": 7.117281844614126e-07,
      "loss": 1.9594,
      "num_input_tokens_seen": 279909400,
      "step": 274900
    },
    {
      "epoch": 5.414345061132878,
      "grad_norm": 1.9879542589187622,
      "learning_rate": 7.115413841630809e-07,
      "loss": 2.0033,
      "num_input_tokens_seen": 280011800,
      "step": 275000
    },
    {
      "epoch": 5.416313913882381,
      "grad_norm": 1.6958532333374023,
      "learning_rate": 7.113545478939211e-07,
      "loss": 2.0198,
      "num_input_tokens_seen": 280113360,
      "step": 275100
    },
    {
      "epoch": 5.418282766631884,
      "grad_norm": 1.8821911811828613,
      "learning_rate": 7.111676756857033e-07,
      "loss": 1.9338,
      "num_input_tokens_seen": 280215760,
      "step": 275200
    },
    {
      "epoch": 5.4202516193813866,
      "grad_norm": 1.8032538890838623,
      "learning_rate": 7.109807675702029e-07,
      "loss": 2.0059,
      "num_input_tokens_seen": 280317640,
      "step": 275300
    },
    {
      "epoch": 5.422220472130889,
      "grad_norm": 1.7668246030807495,
      "learning_rate": 7.107938235792028e-07,
      "loss": 1.9434,
      "num_input_tokens_seen": 280419664,
      "step": 275400
    },
    {
      "epoch": 5.424189324880392,
      "grad_norm": 1.779638409614563,
      "learning_rate": 7.106068437444908e-07,
      "loss": 1.9485,
      "num_input_tokens_seen": 280521488,
      "step": 275500
    },
    {
      "epoch": 5.426158177629895,
      "grad_norm": 1.7614749670028687,
      "learning_rate": 7.104198280978614e-07,
      "loss": 2.0248,
      "num_input_tokens_seen": 280623888,
      "step": 275600
    },
    {
      "epoch": 5.428127030379398,
      "grad_norm": 2.207742929458618,
      "learning_rate": 7.102327766711149e-07,
      "loss": 1.9829,
      "num_input_tokens_seen": 280725144,
      "step": 275700
    },
    {
      "epoch": 5.430095883128901,
      "grad_norm": 1.989269495010376,
      "learning_rate": 7.100456894960581e-07,
      "loss": 1.9831,
      "num_input_tokens_seen": 280826584,
      "step": 275800
    },
    {
      "epoch": 5.432064735878404,
      "grad_norm": 2.120246171951294,
      "learning_rate": 7.098585666045032e-07,
      "loss": 1.972,
      "num_input_tokens_seen": 280928320,
      "step": 275900
    },
    {
      "epoch": 5.434033588627907,
      "grad_norm": 1.8878870010375977,
      "learning_rate": 7.096714080282691e-07,
      "loss": 1.9603,
      "num_input_tokens_seen": 281030720,
      "step": 276000
    },
    {
      "epoch": 5.4360024413774095,
      "grad_norm": 2.058067560195923,
      "learning_rate": 7.094842137991804e-07,
      "loss": 1.9911,
      "num_input_tokens_seen": 281133120,
      "step": 276100
    },
    {
      "epoch": 5.437971294126912,
      "grad_norm": 1.9082746505737305,
      "learning_rate": 7.092969839490683e-07,
      "loss": 2.0134,
      "num_input_tokens_seen": 281234952,
      "step": 276200
    },
    {
      "epoch": 5.439940146876415,
      "grad_norm": 2.6323392391204834,
      "learning_rate": 7.091097185097692e-07,
      "loss": 1.9603,
      "num_input_tokens_seen": 281336728,
      "step": 276300
    },
    {
      "epoch": 5.441908999625918,
      "grad_norm": 2.902038812637329,
      "learning_rate": 7.089224175131263e-07,
      "loss": 2.0159,
      "num_input_tokens_seen": 281438664,
      "step": 276400
    },
    {
      "epoch": 5.4438778523754205,
      "grad_norm": 2.4794204235076904,
      "learning_rate": 7.087350809909883e-07,
      "loss": 1.9942,
      "num_input_tokens_seen": 281539616,
      "step": 276500
    },
    {
      "epoch": 5.445846705124923,
      "grad_norm": 1.9060282707214355,
      "learning_rate": 7.085477089752106e-07,
      "loss": 1.9664,
      "num_input_tokens_seen": 281641496,
      "step": 276600
    },
    {
      "epoch": 5.447815557874427,
      "grad_norm": 1.8969968557357788,
      "learning_rate": 7.083603014976536e-07,
      "loss": 2.0323,
      "num_input_tokens_seen": 281742592,
      "step": 276700
    },
    {
      "epoch": 5.44978441062393,
      "grad_norm": 1.7251744270324707,
      "learning_rate": 7.08172858590185e-07,
      "loss": 2.0215,
      "num_input_tokens_seen": 281844384,
      "step": 276800
    },
    {
      "epoch": 5.4517532633734325,
      "grad_norm": 1.742017149925232,
      "learning_rate": 7.079853802846773e-07,
      "loss": 1.9482,
      "num_input_tokens_seen": 281946136,
      "step": 276900
    },
    {
      "epoch": 5.453722116122935,
      "grad_norm": 2.905759811401367,
      "learning_rate": 7.077978666130103e-07,
      "loss": 1.9417,
      "num_input_tokens_seen": 282048536,
      "step": 277000
    },
    {
      "epoch": 5.455690968872438,
      "grad_norm": 1.920806884765625,
      "learning_rate": 7.076103176070685e-07,
      "loss": 1.9708,
      "num_input_tokens_seen": 282150936,
      "step": 277100
    },
    {
      "epoch": 5.457659821621941,
      "grad_norm": 1.9064234495162964,
      "learning_rate": 7.074227332987432e-07,
      "loss": 1.9656,
      "num_input_tokens_seen": 282252712,
      "step": 277200
    },
    {
      "epoch": 5.4596286743714435,
      "grad_norm": 1.929622769355774,
      "learning_rate": 7.072351137199317e-07,
      "loss": 2.004,
      "num_input_tokens_seen": 282355112,
      "step": 277300
    },
    {
      "epoch": 5.461597527120946,
      "grad_norm": 1.8693277835845947,
      "learning_rate": 7.070474589025371e-07,
      "loss": 1.9707,
      "num_input_tokens_seen": 282456920,
      "step": 277400
    },
    {
      "epoch": 5.46356637987045,
      "grad_norm": 1.8607865571975708,
      "learning_rate": 7.068597688784683e-07,
      "loss": 1.995,
      "num_input_tokens_seen": 282557656,
      "step": 277500
    },
    {
      "epoch": 5.465535232619953,
      "grad_norm": 2.2434775829315186,
      "learning_rate": 7.066720436796405e-07,
      "loss": 1.9817,
      "num_input_tokens_seen": 282660056,
      "step": 277600
    },
    {
      "epoch": 5.467504085369455,
      "grad_norm": 1.734452486038208,
      "learning_rate": 7.064842833379749e-07,
      "loss": 1.9861,
      "num_input_tokens_seen": 282760936,
      "step": 277700
    },
    {
      "epoch": 5.469472938118958,
      "grad_norm": 1.921830177307129,
      "learning_rate": 7.062964878853985e-07,
      "loss": 1.994,
      "num_input_tokens_seen": 282863336,
      "step": 277800
    },
    {
      "epoch": 5.471441790868461,
      "grad_norm": 1.9665969610214233,
      "learning_rate": 7.061086573538444e-07,
      "loss": 1.9454,
      "num_input_tokens_seen": 282964928,
      "step": 277900
    },
    {
      "epoch": 5.473410643617964,
      "grad_norm": 1.9603980779647827,
      "learning_rate": 7.059207917752512e-07,
      "loss": 1.9857,
      "num_input_tokens_seen": 283066736,
      "step": 278000
    },
    {
      "epoch": 5.4753794963674665,
      "grad_norm": 1.923351764678955,
      "learning_rate": 7.057328911815644e-07,
      "loss": 1.9627,
      "num_input_tokens_seen": 283167920,
      "step": 278100
    },
    {
      "epoch": 5.477348349116969,
      "grad_norm": 1.6758246421813965,
      "learning_rate": 7.055449556047347e-07,
      "loss": 2.0062,
      "num_input_tokens_seen": 283269752,
      "step": 278200
    },
    {
      "epoch": 5.479317201866472,
      "grad_norm": 2.052189826965332,
      "learning_rate": 7.053569850767189e-07,
      "loss": 1.9848,
      "num_input_tokens_seen": 283370360,
      "step": 278300
    },
    {
      "epoch": 5.481286054615976,
      "grad_norm": 2.209536552429199,
      "learning_rate": 7.051689796294799e-07,
      "loss": 2.0248,
      "num_input_tokens_seen": 283471992,
      "step": 278400
    },
    {
      "epoch": 5.483254907365478,
      "grad_norm": 1.984963297843933,
      "learning_rate": 7.049809392949863e-07,
      "loss": 1.9933,
      "num_input_tokens_seen": 283574392,
      "step": 278500
    },
    {
      "epoch": 5.485223760114981,
      "grad_norm": 1.7700587511062622,
      "learning_rate": 7.04792864105213e-07,
      "loss": 2.0123,
      "num_input_tokens_seen": 283676200,
      "step": 278600
    },
    {
      "epoch": 5.487192612864484,
      "grad_norm": 4.987680435180664,
      "learning_rate": 7.046047540921403e-07,
      "loss": 1.9983,
      "num_input_tokens_seen": 283777928,
      "step": 278700
    },
    {
      "epoch": 5.489161465613987,
      "grad_norm": 2.0759313106536865,
      "learning_rate": 7.044166092877551e-07,
      "loss": 1.9652,
      "num_input_tokens_seen": 283880328,
      "step": 278800
    },
    {
      "epoch": 5.491130318363489,
      "grad_norm": 1.8324307203292847,
      "learning_rate": 7.042284297240496e-07,
      "loss": 2.0007,
      "num_input_tokens_seen": 283980984,
      "step": 278900
    },
    {
      "epoch": 5.493099171112992,
      "grad_norm": 1.8795981407165527,
      "learning_rate": 7.040402154330218e-07,
      "loss": 2.0202,
      "num_input_tokens_seen": 284083384,
      "step": 279000
    },
    {
      "epoch": 5.495068023862495,
      "grad_norm": 1.7920551300048828,
      "learning_rate": 7.038519664466767e-07,
      "loss": 2.0162,
      "num_input_tokens_seen": 284185312,
      "step": 279100
    },
    {
      "epoch": 5.497036876611999,
      "grad_norm": 1.9927595853805542,
      "learning_rate": 7.036636827970238e-07,
      "loss": 1.9652,
      "num_input_tokens_seen": 284287072,
      "step": 279200
    },
    {
      "epoch": 5.499005729361501,
      "grad_norm": 1.9401782751083374,
      "learning_rate": 7.034753645160796e-07,
      "loss": 2.0085,
      "num_input_tokens_seen": 284389472,
      "step": 279300
    },
    {
      "epoch": 5.500974582111004,
      "grad_norm": 1.854786992073059,
      "learning_rate": 7.032870116358658e-07,
      "loss": 2.0226,
      "num_input_tokens_seen": 284491160,
      "step": 279400
    },
    {
      "epoch": 5.502943434860507,
      "grad_norm": 1.9346381425857544,
      "learning_rate": 7.030986241884104e-07,
      "loss": 1.996,
      "num_input_tokens_seen": 284593560,
      "step": 279500
    },
    {
      "epoch": 5.50491228761001,
      "grad_norm": 2.018704891204834,
      "learning_rate": 7.029102022057467e-07,
      "loss": 1.9927,
      "num_input_tokens_seen": 284695960,
      "step": 279600
    },
    {
      "epoch": 5.506881140359512,
      "grad_norm": 1.7590324878692627,
      "learning_rate": 7.027217457199149e-07,
      "loss": 2.0303,
      "num_input_tokens_seen": 284797496,
      "step": 279700
    },
    {
      "epoch": 5.508849993109015,
      "grad_norm": 1.8581695556640625,
      "learning_rate": 7.025332547629598e-07,
      "loss": 2.0225,
      "num_input_tokens_seen": 284898824,
      "step": 279800
    },
    {
      "epoch": 5.510818845858518,
      "grad_norm": 1.8246877193450928,
      "learning_rate": 7.023447293669332e-07,
      "loss": 1.9667,
      "num_input_tokens_seen": 285000416,
      "step": 279900
    },
    {
      "epoch": 5.512787698608021,
      "grad_norm": 1.817454218864441,
      "learning_rate": 7.021561695638917e-07,
      "loss": 1.9313,
      "num_input_tokens_seen": 285102816,
      "step": 280000
    },
    {
      "epoch": 5.514756551357524,
      "grad_norm": 2.2149713039398193,
      "learning_rate": 7.019675753858987e-07,
      "loss": 2.0217,
      "num_input_tokens_seen": 285204384,
      "step": 280100
    },
    {
      "epoch": 5.516725404107027,
      "grad_norm": 2.0778284072875977,
      "learning_rate": 7.01778946865023e-07,
      "loss": 1.9988,
      "num_input_tokens_seen": 285306048,
      "step": 280200
    },
    {
      "epoch": 5.51869425685653,
      "grad_norm": 1.8702492713928223,
      "learning_rate": 7.015902840333394e-07,
      "loss": 2.0129,
      "num_input_tokens_seen": 285408448,
      "step": 280300
    },
    {
      "epoch": 5.520663109606033,
      "grad_norm": 1.7011958360671997,
      "learning_rate": 7.014015869229283e-07,
      "loss": 1.9638,
      "num_input_tokens_seen": 285510352,
      "step": 280400
    },
    {
      "epoch": 5.522631962355535,
      "grad_norm": 2.153801441192627,
      "learning_rate": 7.01212855565876e-07,
      "loss": 2.0367,
      "num_input_tokens_seen": 285611592,
      "step": 280500
    },
    {
      "epoch": 5.524600815105038,
      "grad_norm": 1.9386348724365234,
      "learning_rate": 7.010240899942747e-07,
      "loss": 1.97,
      "num_input_tokens_seen": 285713376,
      "step": 280600
    },
    {
      "epoch": 5.526569667854541,
      "grad_norm": 1.938071846961975,
      "learning_rate": 7.008352902402224e-07,
      "loss": 2.0261,
      "num_input_tokens_seen": 285815128,
      "step": 280700
    },
    {
      "epoch": 5.5285385206040445,
      "grad_norm": 1.8405362367630005,
      "learning_rate": 7.00646456335823e-07,
      "loss": 1.9813,
      "num_input_tokens_seen": 285916920,
      "step": 280800
    },
    {
      "epoch": 5.530507373353547,
      "grad_norm": 1.87881338596344,
      "learning_rate": 7.004575883131859e-07,
      "loss": 1.9776,
      "num_input_tokens_seen": 286019320,
      "step": 280900
    },
    {
      "epoch": 5.53247622610305,
      "grad_norm": 2.1490933895111084,
      "learning_rate": 7.002686862044266e-07,
      "loss": 1.9733,
      "num_input_tokens_seen": 286121720,
      "step": 281000
    },
    {
      "epoch": 5.534445078852553,
      "grad_norm": 2.014657974243164,
      "learning_rate": 7.000797500416665e-07,
      "loss": 1.9733,
      "num_input_tokens_seen": 286224120,
      "step": 281100
    },
    {
      "epoch": 5.5364139316020555,
      "grad_norm": 2.160707712173462,
      "learning_rate": 6.998907798570325e-07,
      "loss": 1.998,
      "num_input_tokens_seen": 286325336,
      "step": 281200
    },
    {
      "epoch": 5.538382784351558,
      "grad_norm": 2.2819178104400635,
      "learning_rate": 6.997017756826572e-07,
      "loss": 1.9691,
      "num_input_tokens_seen": 286427256,
      "step": 281300
    },
    {
      "epoch": 5.540351637101061,
      "grad_norm": 2.0642409324645996,
      "learning_rate": 6.995127375506794e-07,
      "loss": 1.9654,
      "num_input_tokens_seen": 286529656,
      "step": 281400
    },
    {
      "epoch": 5.542320489850564,
      "grad_norm": 2.2802693843841553,
      "learning_rate": 6.993236654932433e-07,
      "loss": 1.9904,
      "num_input_tokens_seen": 286631064,
      "step": 281500
    },
    {
      "epoch": 5.544289342600067,
      "grad_norm": 1.6591362953186035,
      "learning_rate": 6.991345595424991e-07,
      "loss": 1.9814,
      "num_input_tokens_seen": 286733464,
      "step": 281600
    },
    {
      "epoch": 5.546258195349569,
      "grad_norm": 1.9292830228805542,
      "learning_rate": 6.989454197306027e-07,
      "loss": 2.0289,
      "num_input_tokens_seen": 286835120,
      "step": 281700
    },
    {
      "epoch": 5.548227048099073,
      "grad_norm": 2.751769542694092,
      "learning_rate": 6.987562460897155e-07,
      "loss": 1.979,
      "num_input_tokens_seen": 286937520,
      "step": 281800
    },
    {
      "epoch": 5.550195900848576,
      "grad_norm": 2.3896677494049072,
      "learning_rate": 6.98567038652005e-07,
      "loss": 1.9442,
      "num_input_tokens_seen": 287039920,
      "step": 281900
    },
    {
      "epoch": 5.5521647535980785,
      "grad_norm": 1.998288869857788,
      "learning_rate": 6.983777974496446e-07,
      "loss": 1.9655,
      "num_input_tokens_seen": 287142320,
      "step": 282000
    },
    {
      "epoch": 5.554133606347581,
      "grad_norm": 1.8365534543991089,
      "learning_rate": 6.981885225148127e-07,
      "loss": 1.9509,
      "num_input_tokens_seen": 287244720,
      "step": 282100
    },
    {
      "epoch": 5.556102459097084,
      "grad_norm": 1.9905551671981812,
      "learning_rate": 6.979992138796941e-07,
      "loss": 2.0188,
      "num_input_tokens_seen": 287346560,
      "step": 282200
    },
    {
      "epoch": 5.558071311846587,
      "grad_norm": 2.0359601974487305,
      "learning_rate": 6.978098715764794e-07,
      "loss": 2.0305,
      "num_input_tokens_seen": 287447432,
      "step": 282300
    },
    {
      "epoch": 5.5600401645960895,
      "grad_norm": 2.004528522491455,
      "learning_rate": 6.976204956373645e-07,
      "loss": 1.9655,
      "num_input_tokens_seen": 287549832,
      "step": 282400
    },
    {
      "epoch": 5.562009017345593,
      "grad_norm": 1.9784183502197266,
      "learning_rate": 6.974310860945509e-07,
      "loss": 1.9744,
      "num_input_tokens_seen": 287651728,
      "step": 282500
    },
    {
      "epoch": 5.563977870095096,
      "grad_norm": 1.8037654161453247,
      "learning_rate": 6.972416429802465e-07,
      "loss": 1.9985,
      "num_input_tokens_seen": 287754128,
      "step": 282600
    },
    {
      "epoch": 5.565946722844599,
      "grad_norm": 1.8915928602218628,
      "learning_rate": 6.970521663266642e-07,
      "loss": 1.9789,
      "num_input_tokens_seen": 287855848,
      "step": 282700
    },
    {
      "epoch": 5.567915575594101,
      "grad_norm": 1.8359853029251099,
      "learning_rate": 6.968626561660231e-07,
      "loss": 1.9898,
      "num_input_tokens_seen": 287958248,
      "step": 282800
    },
    {
      "epoch": 5.569884428343604,
      "grad_norm": 1.9419841766357422,
      "learning_rate": 6.966731125305476e-07,
      "loss": 1.9427,
      "num_input_tokens_seen": 288060648,
      "step": 282900
    },
    {
      "epoch": 5.571853281093107,
      "grad_norm": 4.6133294105529785,
      "learning_rate": 6.964835354524684e-07,
      "loss": 2.001,
      "num_input_tokens_seen": 288161920,
      "step": 283000
    },
    {
      "epoch": 5.57382213384261,
      "grad_norm": 1.8769828081130981,
      "learning_rate": 6.962939249640209e-07,
      "loss": 1.9891,
      "num_input_tokens_seen": 288264320,
      "step": 283100
    },
    {
      "epoch": 5.5757909865921125,
      "grad_norm": 2.1598141193389893,
      "learning_rate": 6.961042810974474e-07,
      "loss": 1.9779,
      "num_input_tokens_seen": 288365376,
      "step": 283200
    },
    {
      "epoch": 5.577759839341615,
      "grad_norm": 1.9685794115066528,
      "learning_rate": 6.959146038849945e-07,
      "loss": 1.9889,
      "num_input_tokens_seen": 288467776,
      "step": 283300
    },
    {
      "epoch": 5.579728692091119,
      "grad_norm": 5.988781929016113,
      "learning_rate": 6.957248933589161e-07,
      "loss": 2.011,
      "num_input_tokens_seen": 288569528,
      "step": 283400
    },
    {
      "epoch": 5.581697544840622,
      "grad_norm": 1.9712460041046143,
      "learning_rate": 6.955351495514701e-07,
      "loss": 1.9834,
      "num_input_tokens_seen": 288671928,
      "step": 283500
    },
    {
      "epoch": 5.583666397590124,
      "grad_norm": 1.7132648229599,
      "learning_rate": 6.953453724949212e-07,
      "loss": 2.0101,
      "num_input_tokens_seen": 288774328,
      "step": 283600
    },
    {
      "epoch": 5.585635250339627,
      "grad_norm": 2.162921905517578,
      "learning_rate": 6.951555622215393e-07,
      "loss": 2.0175,
      "num_input_tokens_seen": 288875096,
      "step": 283700
    },
    {
      "epoch": 5.58760410308913,
      "grad_norm": 2.0548007488250732,
      "learning_rate": 6.949657187636003e-07,
      "loss": 2.0156,
      "num_input_tokens_seen": 288977016,
      "step": 283800
    },
    {
      "epoch": 5.589572955838633,
      "grad_norm": 2.019831418991089,
      "learning_rate": 6.947758421533849e-07,
      "loss": 1.9681,
      "num_input_tokens_seen": 289078216,
      "step": 283900
    },
    {
      "epoch": 5.591541808588135,
      "grad_norm": 2.0104689598083496,
      "learning_rate": 6.945859324231806e-07,
      "loss": 1.9827,
      "num_input_tokens_seen": 289180616,
      "step": 284000
    },
    {
      "epoch": 5.593510661337638,
      "grad_norm": 2.0804810523986816,
      "learning_rate": 6.943959896052796e-07,
      "loss": 2.0707,
      "num_input_tokens_seen": 289280832,
      "step": 284100
    },
    {
      "epoch": 5.595479514087142,
      "grad_norm": 1.8354921340942383,
      "learning_rate": 6.942060137319802e-07,
      "loss": 2.032,
      "num_input_tokens_seen": 289383232,
      "step": 284200
    },
    {
      "epoch": 5.597448366836645,
      "grad_norm": 2.3384897708892822,
      "learning_rate": 6.940160048355861e-07,
      "loss": 1.9833,
      "num_input_tokens_seen": 289485632,
      "step": 284300
    },
    {
      "epoch": 5.599417219586147,
      "grad_norm": 1.866071343421936,
      "learning_rate": 6.938259629484069e-07,
      "loss": 2.0212,
      "num_input_tokens_seen": 289587584,
      "step": 284400
    },
    {
      "epoch": 5.60138607233565,
      "grad_norm": 1.974873661994934,
      "learning_rate": 6.936358881027575e-07,
      "loss": 1.9549,
      "num_input_tokens_seen": 289688536,
      "step": 284500
    },
    {
      "epoch": 5.603354925085153,
      "grad_norm": 2.2453863620758057,
      "learning_rate": 6.934457803309585e-07,
      "loss": 1.9611,
      "num_input_tokens_seen": 289790936,
      "step": 284600
    },
    {
      "epoch": 5.605323777834656,
      "grad_norm": 1.9215408563613892,
      "learning_rate": 6.93255639665336e-07,
      "loss": 1.9793,
      "num_input_tokens_seen": 289893336,
      "step": 284700
    },
    {
      "epoch": 5.607292630584158,
      "grad_norm": 2.227081537246704,
      "learning_rate": 6.930654661382223e-07,
      "loss": 2.0022,
      "num_input_tokens_seen": 289994992,
      "step": 284800
    },
    {
      "epoch": 5.609261483333661,
      "grad_norm": 2.0311901569366455,
      "learning_rate": 6.928752597819541e-07,
      "loss": 2.0112,
      "num_input_tokens_seen": 290096792,
      "step": 284900
    },
    {
      "epoch": 5.611230336083164,
      "grad_norm": 1.6489031314849854,
      "learning_rate": 6.92685020628875e-07,
      "loss": 1.9136,
      "num_input_tokens_seen": 290199192,
      "step": 285000
    },
    {
      "epoch": 5.6131991888326676,
      "grad_norm": 2.2252256870269775,
      "learning_rate": 6.924947487113333e-07,
      "loss": 1.9666,
      "num_input_tokens_seen": 290300824,
      "step": 285100
    },
    {
      "epoch": 5.61516804158217,
      "grad_norm": 1.861838459968567,
      "learning_rate": 6.923044440616831e-07,
      "loss": 1.974,
      "num_input_tokens_seen": 290402608,
      "step": 285200
    },
    {
      "epoch": 5.617136894331673,
      "grad_norm": 2.2326159477233887,
      "learning_rate": 6.921141067122842e-07,
      "loss": 2.0539,
      "num_input_tokens_seen": 290504248,
      "step": 285300
    },
    {
      "epoch": 5.619105747081176,
      "grad_norm": 1.9460946321487427,
      "learning_rate": 6.919237366955017e-07,
      "loss": 2.0213,
      "num_input_tokens_seen": 290606024,
      "step": 285400
    },
    {
      "epoch": 5.621074599830679,
      "grad_norm": 1.9875304698944092,
      "learning_rate": 6.917333340437067e-07,
      "loss": 1.9549,
      "num_input_tokens_seen": 290707120,
      "step": 285500
    },
    {
      "epoch": 5.623043452580181,
      "grad_norm": 2.336749315261841,
      "learning_rate": 6.915428987892753e-07,
      "loss": 2.0025,
      "num_input_tokens_seen": 290809520,
      "step": 285600
    },
    {
      "epoch": 5.625012305329684,
      "grad_norm": 2.5193111896514893,
      "learning_rate": 6.913524309645897e-07,
      "loss": 1.996,
      "num_input_tokens_seen": 290910528,
      "step": 285700
    },
    {
      "epoch": 5.626981158079188,
      "grad_norm": 1.9238159656524658,
      "learning_rate": 6.911619306020368e-07,
      "loss": 1.9955,
      "num_input_tokens_seen": 291012456,
      "step": 285800
    },
    {
      "epoch": 5.6289500108286905,
      "grad_norm": 1.997316837310791,
      "learning_rate": 6.909713977340102e-07,
      "loss": 1.9883,
      "num_input_tokens_seen": 291113584,
      "step": 285900
    },
    {
      "epoch": 5.630918863578193,
      "grad_norm": 2.033946990966797,
      "learning_rate": 6.90780832392908e-07,
      "loss": 2.0086,
      "num_input_tokens_seen": 291215224,
      "step": 286000
    },
    {
      "epoch": 5.632887716327696,
      "grad_norm": 1.861080288887024,
      "learning_rate": 6.905902346111346e-07,
      "loss": 2.0076,
      "num_input_tokens_seen": 291315200,
      "step": 286100
    },
    {
      "epoch": 5.634856569077199,
      "grad_norm": 2.0090553760528564,
      "learning_rate": 6.90399604421099e-07,
      "loss": 1.9033,
      "num_input_tokens_seen": 291417600,
      "step": 286200
    },
    {
      "epoch": 5.6368254218267015,
      "grad_norm": 1.9354575872421265,
      "learning_rate": 6.902089418552169e-07,
      "loss": 1.9776,
      "num_input_tokens_seen": 291519184,
      "step": 286300
    },
    {
      "epoch": 5.638794274576204,
      "grad_norm": 1.853554368019104,
      "learning_rate": 6.900182469459082e-07,
      "loss": 1.9789,
      "num_input_tokens_seen": 291621584,
      "step": 286400
    },
    {
      "epoch": 5.640763127325707,
      "grad_norm": 1.9268100261688232,
      "learning_rate": 6.898275197255996e-07,
      "loss": 1.9895,
      "num_input_tokens_seen": 291723360,
      "step": 286500
    },
    {
      "epoch": 5.64273198007521,
      "grad_norm": 8.665900230407715,
      "learning_rate": 6.896367602267221e-07,
      "loss": 2.0115,
      "num_input_tokens_seen": 291825400,
      "step": 286600
    },
    {
      "epoch": 5.644700832824713,
      "grad_norm": 2.1298089027404785,
      "learning_rate": 6.894459684817133e-07,
      "loss": 1.9229,
      "num_input_tokens_seen": 291927800,
      "step": 286700
    },
    {
      "epoch": 5.646669685574216,
      "grad_norm": 1.7074607610702515,
      "learning_rate": 6.89255144523015e-07,
      "loss": 1.9744,
      "num_input_tokens_seen": 292030200,
      "step": 286800
    },
    {
      "epoch": 5.648638538323719,
      "grad_norm": 1.8821911811828613,
      "learning_rate": 6.89064288383076e-07,
      "loss": 1.9755,
      "num_input_tokens_seen": 292131136,
      "step": 286900
    },
    {
      "epoch": 5.650607391073222,
      "grad_norm": 1.807858943939209,
      "learning_rate": 6.888734000943493e-07,
      "loss": 2.0074,
      "num_input_tokens_seen": 292232536,
      "step": 287000
    },
    {
      "epoch": 5.6525762438227245,
      "grad_norm": 2.4728779792785645,
      "learning_rate": 6.886824796892939e-07,
      "loss": 1.9892,
      "num_input_tokens_seen": 292334296,
      "step": 287100
    },
    {
      "epoch": 5.654545096572227,
      "grad_norm": 1.7749592065811157,
      "learning_rate": 6.88491527200374e-07,
      "loss": 1.967,
      "num_input_tokens_seen": 292436696,
      "step": 287200
    },
    {
      "epoch": 5.65651394932173,
      "grad_norm": 2.0146358013153076,
      "learning_rate": 6.883005426600601e-07,
      "loss": 1.9608,
      "num_input_tokens_seen": 292538504,
      "step": 287300
    },
    {
      "epoch": 5.658482802071233,
      "grad_norm": 2.08988356590271,
      "learning_rate": 6.88109526100827e-07,
      "loss": 1.9634,
      "num_input_tokens_seen": 292640904,
      "step": 287400
    },
    {
      "epoch": 5.660451654820736,
      "grad_norm": 1.7986937761306763,
      "learning_rate": 6.879184775551556e-07,
      "loss": 1.9478,
      "num_input_tokens_seen": 292743304,
      "step": 287500
    },
    {
      "epoch": 5.662420507570239,
      "grad_norm": 1.8585352897644043,
      "learning_rate": 6.87727397055532e-07,
      "loss": 1.988,
      "num_input_tokens_seen": 292844304,
      "step": 287600
    },
    {
      "epoch": 5.664389360319742,
      "grad_norm": 1.8912304639816284,
      "learning_rate": 6.875362846344479e-07,
      "loss": 1.986,
      "num_input_tokens_seen": 292945192,
      "step": 287700
    },
    {
      "epoch": 5.666358213069245,
      "grad_norm": 1.752672791481018,
      "learning_rate": 6.873451403244004e-07,
      "loss": 1.9173,
      "num_input_tokens_seen": 293047176,
      "step": 287800
    },
    {
      "epoch": 5.6683270658187475,
      "grad_norm": 2.038039207458496,
      "learning_rate": 6.871539641578919e-07,
      "loss": 2.0048,
      "num_input_tokens_seen": 293148896,
      "step": 287900
    },
    {
      "epoch": 5.67029591856825,
      "grad_norm": 1.8853589296340942,
      "learning_rate": 6.869627561674303e-07,
      "loss": 1.9975,
      "num_input_tokens_seen": 293251296,
      "step": 288000
    },
    {
      "epoch": 5.672264771317753,
      "grad_norm": 2.0846149921417236,
      "learning_rate": 6.86771516385529e-07,
      "loss": 1.9658,
      "num_input_tokens_seen": 293353696,
      "step": 288100
    },
    {
      "epoch": 5.674233624067256,
      "grad_norm": 1.9113783836364746,
      "learning_rate": 6.865802448447064e-07,
      "loss": 1.9763,
      "num_input_tokens_seen": 293454416,
      "step": 288200
    },
    {
      "epoch": 5.6762024768167585,
      "grad_norm": 1.9036356210708618,
      "learning_rate": 6.863889415774872e-07,
      "loss": 2.0191,
      "num_input_tokens_seen": 293555320,
      "step": 288300
    },
    {
      "epoch": 5.678171329566261,
      "grad_norm": 2.454892635345459,
      "learning_rate": 6.861976066164004e-07,
      "loss": 1.9844,
      "num_input_tokens_seen": 293657720,
      "step": 288400
    },
    {
      "epoch": 5.680140182315765,
      "grad_norm": 1.8267836570739746,
      "learning_rate": 6.86006239993981e-07,
      "loss": 2.0506,
      "num_input_tokens_seen": 293760120,
      "step": 288500
    },
    {
      "epoch": 5.682109035065268,
      "grad_norm": 2.0792393684387207,
      "learning_rate": 6.858148417427693e-07,
      "loss": 1.961,
      "num_input_tokens_seen": 293862520,
      "step": 288600
    },
    {
      "epoch": 5.68407788781477,
      "grad_norm": 1.9794563055038452,
      "learning_rate": 6.856234118953112e-07,
      "loss": 1.9825,
      "num_input_tokens_seen": 293964288,
      "step": 288700
    },
    {
      "epoch": 5.686046740564273,
      "grad_norm": 2.4365408420562744,
      "learning_rate": 6.854319504841575e-07,
      "loss": 1.9683,
      "num_input_tokens_seen": 294066688,
      "step": 288800
    },
    {
      "epoch": 5.688015593313776,
      "grad_norm": 1.8940140008926392,
      "learning_rate": 6.852404575418647e-07,
      "loss": 2.0323,
      "num_input_tokens_seen": 294168744,
      "step": 288900
    },
    {
      "epoch": 5.689984446063279,
      "grad_norm": 1.882495403289795,
      "learning_rate": 6.850489331009942e-07,
      "loss": 2.0024,
      "num_input_tokens_seen": 294269424,
      "step": 289000
    },
    {
      "epoch": 5.6919532988127814,
      "grad_norm": 2.120837688446045,
      "learning_rate": 6.848573771941137e-07,
      "loss": 1.9956,
      "num_input_tokens_seen": 294371160,
      "step": 289100
    },
    {
      "epoch": 5.693922151562285,
      "grad_norm": 2.1796810626983643,
      "learning_rate": 6.846657898537951e-07,
      "loss": 2.0071,
      "num_input_tokens_seen": 294473560,
      "step": 289200
    },
    {
      "epoch": 5.695891004311788,
      "grad_norm": 1.8778839111328125,
      "learning_rate": 6.844741711126165e-07,
      "loss": 1.9922,
      "num_input_tokens_seen": 294575816,
      "step": 289300
    },
    {
      "epoch": 5.697859857061291,
      "grad_norm": 2.173163414001465,
      "learning_rate": 6.842825210031612e-07,
      "loss": 1.9967,
      "num_input_tokens_seen": 294678216,
      "step": 289400
    },
    {
      "epoch": 5.699828709810793,
      "grad_norm": 2.1021974086761475,
      "learning_rate": 6.840908395580174e-07,
      "loss": 1.9757,
      "num_input_tokens_seen": 294780616,
      "step": 289500
    },
    {
      "epoch": 5.701797562560296,
      "grad_norm": 1.8285257816314697,
      "learning_rate": 6.838991268097791e-07,
      "loss": 1.9713,
      "num_input_tokens_seen": 294883016,
      "step": 289600
    },
    {
      "epoch": 5.703766415309799,
      "grad_norm": 1.738957166671753,
      "learning_rate": 6.837073827910453e-07,
      "loss": 1.9793,
      "num_input_tokens_seen": 294984552,
      "step": 289700
    },
    {
      "epoch": 5.705735268059302,
      "grad_norm": 1.7961680889129639,
      "learning_rate": 6.835156075344206e-07,
      "loss": 1.9934,
      "num_input_tokens_seen": 295085672,
      "step": 289800
    },
    {
      "epoch": 5.707704120808804,
      "grad_norm": 1.8706741333007812,
      "learning_rate": 6.833238010725146e-07,
      "loss": 1.9639,
      "num_input_tokens_seen": 295188072,
      "step": 289900
    },
    {
      "epoch": 5.709672973558307,
      "grad_norm": 1.7334696054458618,
      "learning_rate": 6.831319634379426e-07,
      "loss": 1.9379,
      "num_input_tokens_seen": 295290472,
      "step": 290000
    },
    {
      "epoch": 5.711641826307811,
      "grad_norm": 1.8002521991729736,
      "learning_rate": 6.829400946633245e-07,
      "loss": 1.9655,
      "num_input_tokens_seen": 295392408,
      "step": 290100
    },
    {
      "epoch": 5.713610679057314,
      "grad_norm": 1.7615959644317627,
      "learning_rate": 6.827481947812865e-07,
      "loss": 1.9791,
      "num_input_tokens_seen": 295493256,
      "step": 290200
    },
    {
      "epoch": 5.715579531806816,
      "grad_norm": 2.7020058631896973,
      "learning_rate": 6.825562638244593e-07,
      "loss": 1.9498,
      "num_input_tokens_seen": 295594432,
      "step": 290300
    },
    {
      "epoch": 5.717548384556319,
      "grad_norm": 1.7554099559783936,
      "learning_rate": 6.823643018254794e-07,
      "loss": 1.9931,
      "num_input_tokens_seen": 295696832,
      "step": 290400
    },
    {
      "epoch": 5.719517237305822,
      "grad_norm": 1.9533312320709229,
      "learning_rate": 6.821723088169879e-07,
      "loss": 1.9763,
      "num_input_tokens_seen": 295798408,
      "step": 290500
    },
    {
      "epoch": 5.721486090055325,
      "grad_norm": 2.0266520977020264,
      "learning_rate": 6.819802848316318e-07,
      "loss": 1.9509,
      "num_input_tokens_seen": 295900808,
      "step": 290600
    },
    {
      "epoch": 5.723454942804827,
      "grad_norm": 2.1518337726593018,
      "learning_rate": 6.817882299020631e-07,
      "loss": 1.9939,
      "num_input_tokens_seen": 296002648,
      "step": 290700
    },
    {
      "epoch": 5.72542379555433,
      "grad_norm": 1.8498213291168213,
      "learning_rate": 6.815961440609393e-07,
      "loss": 2.0672,
      "num_input_tokens_seen": 296103424,
      "step": 290800
    },
    {
      "epoch": 5.727392648303834,
      "grad_norm": 2.0848705768585205,
      "learning_rate": 6.814040273409229e-07,
      "loss": 1.978,
      "num_input_tokens_seen": 296204248,
      "step": 290900
    },
    {
      "epoch": 5.7293615010533365,
      "grad_norm": 1.7975051403045654,
      "learning_rate": 6.812118797746817e-07,
      "loss": 2.104,
      "num_input_tokens_seen": 296305504,
      "step": 291000
    },
    {
      "epoch": 5.731330353802839,
      "grad_norm": 1.978773593902588,
      "learning_rate": 6.810197013948885e-07,
      "loss": 1.9739,
      "num_input_tokens_seen": 296407904,
      "step": 291100
    },
    {
      "epoch": 5.733299206552342,
      "grad_norm": 1.8710181713104248,
      "learning_rate": 6.808274922342221e-07,
      "loss": 1.9535,
      "num_input_tokens_seen": 296507648,
      "step": 291200
    },
    {
      "epoch": 5.735268059301845,
      "grad_norm": 1.9824811220169067,
      "learning_rate": 6.806352523253655e-07,
      "loss": 1.9836,
      "num_input_tokens_seen": 296609736,
      "step": 291300
    },
    {
      "epoch": 5.737236912051348,
      "grad_norm": 1.8356417417526245,
      "learning_rate": 6.804429817010082e-07,
      "loss": 2.0004,
      "num_input_tokens_seen": 296710632,
      "step": 291400
    },
    {
      "epoch": 5.73920576480085,
      "grad_norm": 2.2513487339019775,
      "learning_rate": 6.802506803938436e-07,
      "loss": 2.0035,
      "num_input_tokens_seen": 296811680,
      "step": 291500
    },
    {
      "epoch": 5.741174617550353,
      "grad_norm": 2.213818073272705,
      "learning_rate": 6.800583484365712e-07,
      "loss": 1.9697,
      "num_input_tokens_seen": 296912344,
      "step": 291600
    },
    {
      "epoch": 5.743143470299856,
      "grad_norm": 1.8014044761657715,
      "learning_rate": 6.798659858618951e-07,
      "loss": 1.9509,
      "num_input_tokens_seen": 297013824,
      "step": 291700
    },
    {
      "epoch": 5.7451123230493595,
      "grad_norm": 1.8372653722763062,
      "learning_rate": 6.796735927025255e-07,
      "loss": 2.0054,
      "num_input_tokens_seen": 297116224,
      "step": 291800
    },
    {
      "epoch": 5.747081175798862,
      "grad_norm": 1.8215463161468506,
      "learning_rate": 6.794811689911766e-07,
      "loss": 2.0063,
      "num_input_tokens_seen": 297217696,
      "step": 291900
    },
    {
      "epoch": 5.749050028548365,
      "grad_norm": 1.8149486780166626,
      "learning_rate": 6.792887147605689e-07,
      "loss": 1.9432,
      "num_input_tokens_seen": 297319328,
      "step": 292000
    },
    {
      "epoch": 5.751018881297868,
      "grad_norm": 1.895106315612793,
      "learning_rate": 6.790962300434273e-07,
      "loss": 1.983,
      "num_input_tokens_seen": 297421288,
      "step": 292100
    },
    {
      "epoch": 5.7529877340473705,
      "grad_norm": 1.8608224391937256,
      "learning_rate": 6.789037148724826e-07,
      "loss": 1.9708,
      "num_input_tokens_seen": 297523688,
      "step": 292200
    },
    {
      "epoch": 5.754956586796873,
      "grad_norm": 2.3314852714538574,
      "learning_rate": 6.787111692804698e-07,
      "loss": 1.9471,
      "num_input_tokens_seen": 297624984,
      "step": 292300
    },
    {
      "epoch": 5.756925439546376,
      "grad_norm": 2.004523277282715,
      "learning_rate": 6.785185933001302e-07,
      "loss": 2.043,
      "num_input_tokens_seen": 297727384,
      "step": 292400
    },
    {
      "epoch": 5.75889429229588,
      "grad_norm": 1.8335539102554321,
      "learning_rate": 6.783259869642094e-07,
      "loss": 1.924,
      "num_input_tokens_seen": 297829080,
      "step": 292500
    },
    {
      "epoch": 5.760863145045382,
      "grad_norm": 1.7369937896728516,
      "learning_rate": 6.781333503054587e-07,
      "loss": 1.9722,
      "num_input_tokens_seen": 297931480,
      "step": 292600
    },
    {
      "epoch": 5.762831997794885,
      "grad_norm": 1.826288104057312,
      "learning_rate": 6.779406833566341e-07,
      "loss": 2.0215,
      "num_input_tokens_seen": 298033880,
      "step": 292700
    },
    {
      "epoch": 5.764800850544388,
      "grad_norm": 1.8336760997772217,
      "learning_rate": 6.777479861504973e-07,
      "loss": 1.9846,
      "num_input_tokens_seen": 298135456,
      "step": 292800
    },
    {
      "epoch": 5.766769703293891,
      "grad_norm": 2.0768070220947266,
      "learning_rate": 6.775552587198144e-07,
      "loss": 1.981,
      "num_input_tokens_seen": 298237856,
      "step": 292900
    },
    {
      "epoch": 5.7687385560433935,
      "grad_norm": 2.022739887237549,
      "learning_rate": 6.773625010973576e-07,
      "loss": 2.046,
      "num_input_tokens_seen": 298338064,
      "step": 293000
    },
    {
      "epoch": 5.770707408792896,
      "grad_norm": 1.801545262336731,
      "learning_rate": 6.771697133159033e-07,
      "loss": 2.0324,
      "num_input_tokens_seen": 298438440,
      "step": 293100
    },
    {
      "epoch": 5.772676261542399,
      "grad_norm": 2.091960906982422,
      "learning_rate": 6.769768954082335e-07,
      "loss": 1.9919,
      "num_input_tokens_seen": 298540840,
      "step": 293200
    },
    {
      "epoch": 5.774645114291902,
      "grad_norm": 2.9989755153656006,
      "learning_rate": 6.767840474071354e-07,
      "loss": 2.0144,
      "num_input_tokens_seen": 298643168,
      "step": 293300
    },
    {
      "epoch": 5.7766139670414045,
      "grad_norm": 1.9307326078414917,
      "learning_rate": 6.76591169345401e-07,
      "loss": 1.9693,
      "num_input_tokens_seen": 298745568,
      "step": 293400
    },
    {
      "epoch": 5.778582819790908,
      "grad_norm": 1.7557299137115479,
      "learning_rate": 6.763982612558279e-07,
      "loss": 1.9826,
      "num_input_tokens_seen": 298847968,
      "step": 293500
    },
    {
      "epoch": 5.780551672540411,
      "grad_norm": 2.4569568634033203,
      "learning_rate": 6.762053231712181e-07,
      "loss": 2.0128,
      "num_input_tokens_seen": 298948504,
      "step": 293600
    },
    {
      "epoch": 5.782520525289914,
      "grad_norm": 2.366541624069214,
      "learning_rate": 6.760123551243795e-07,
      "loss": 2.0273,
      "num_input_tokens_seen": 299049216,
      "step": 293700
    },
    {
      "epoch": 5.784489378039416,
      "grad_norm": 2.058173179626465,
      "learning_rate": 6.758193571481242e-07,
      "loss": 1.9591,
      "num_input_tokens_seen": 299151016,
      "step": 293800
    },
    {
      "epoch": 5.786458230788919,
      "grad_norm": 1.9351673126220703,
      "learning_rate": 6.756263292752703e-07,
      "loss": 2.005,
      "num_input_tokens_seen": 299252600,
      "step": 293900
    },
    {
      "epoch": 5.788427083538422,
      "grad_norm": 2.0175981521606445,
      "learning_rate": 6.754332715386403e-07,
      "loss": 1.9876,
      "num_input_tokens_seen": 299354504,
      "step": 294000
    },
    {
      "epoch": 5.790395936287925,
      "grad_norm": 1.9232161045074463,
      "learning_rate": 6.752401839710623e-07,
      "loss": 1.9929,
      "num_input_tokens_seen": 299456624,
      "step": 294100
    },
    {
      "epoch": 5.792364789037428,
      "grad_norm": 2.2448158264160156,
      "learning_rate": 6.750470666053689e-07,
      "loss": 1.9901,
      "num_input_tokens_seen": 299559024,
      "step": 294200
    },
    {
      "epoch": 5.794333641786931,
      "grad_norm": 1.9506829977035522,
      "learning_rate": 6.748539194743983e-07,
      "loss": 1.9729,
      "num_input_tokens_seen": 299660120,
      "step": 294300
    },
    {
      "epoch": 5.796302494536434,
      "grad_norm": 4.236745357513428,
      "learning_rate": 6.746607426109935e-07,
      "loss": 2.0085,
      "num_input_tokens_seen": 299761664,
      "step": 294400
    },
    {
      "epoch": 5.798271347285937,
      "grad_norm": 2.086859941482544,
      "learning_rate": 6.744675360480026e-07,
      "loss": 1.9446,
      "num_input_tokens_seen": 299864064,
      "step": 294500
    },
    {
      "epoch": 5.800240200035439,
      "grad_norm": 3.206986427307129,
      "learning_rate": 6.742742998182786e-07,
      "loss": 1.9954,
      "num_input_tokens_seen": 299966464,
      "step": 294600
    },
    {
      "epoch": 5.802209052784942,
      "grad_norm": 2.541658401489258,
      "learning_rate": 6.740810339546801e-07,
      "loss": 1.9703,
      "num_input_tokens_seen": 300068864,
      "step": 294700
    },
    {
      "epoch": 5.804177905534445,
      "grad_norm": 1.910529375076294,
      "learning_rate": 6.738877384900698e-07,
      "loss": 1.9419,
      "num_input_tokens_seen": 300171264,
      "step": 294800
    },
    {
      "epoch": 5.806146758283948,
      "grad_norm": 2.054974317550659,
      "learning_rate": 6.736944134573162e-07,
      "loss": 1.9759,
      "num_input_tokens_seen": 300273664,
      "step": 294900
    },
    {
      "epoch": 5.80811561103345,
      "grad_norm": 1.8165878057479858,
      "learning_rate": 6.735010588892926e-07,
      "loss": 2.006,
      "num_input_tokens_seen": 300375192,
      "step": 295000
    },
    {
      "epoch": 5.810084463782953,
      "grad_norm": 1.8727779388427734,
      "learning_rate": 6.733076748188775e-07,
      "loss": 1.9794,
      "num_input_tokens_seen": 300475400,
      "step": 295100
    },
    {
      "epoch": 5.812053316532457,
      "grad_norm": 1.6799296140670776,
      "learning_rate": 6.731142612789539e-07,
      "loss": 1.9873,
      "num_input_tokens_seen": 300577360,
      "step": 295200
    },
    {
      "epoch": 5.81402216928196,
      "grad_norm": 2.0564098358154297,
      "learning_rate": 6.729208183024102e-07,
      "loss": 2.0005,
      "num_input_tokens_seen": 300679760,
      "step": 295300
    },
    {
      "epoch": 5.815991022031462,
      "grad_norm": 1.930026650428772,
      "learning_rate": 6.727273459221399e-07,
      "loss": 1.9458,
      "num_input_tokens_seen": 300782160,
      "step": 295400
    },
    {
      "epoch": 5.817959874780965,
      "grad_norm": 2.0773508548736572,
      "learning_rate": 6.725338441710413e-07,
      "loss": 1.9488,
      "num_input_tokens_seen": 300884560,
      "step": 295500
    },
    {
      "epoch": 5.819928727530468,
      "grad_norm": 2.026898145675659,
      "learning_rate": 6.723403130820177e-07,
      "loss": 1.9835,
      "num_input_tokens_seen": 300986424,
      "step": 295600
    },
    {
      "epoch": 5.821897580279971,
      "grad_norm": 1.7822896242141724,
      "learning_rate": 6.721467526879775e-07,
      "loss": 1.9954,
      "num_input_tokens_seen": 301088824,
      "step": 295700
    },
    {
      "epoch": 5.823866433029473,
      "grad_norm": 2.170503616333008,
      "learning_rate": 6.719531630218339e-07,
      "loss": 1.9929,
      "num_input_tokens_seen": 301191224,
      "step": 295800
    },
    {
      "epoch": 5.825835285778977,
      "grad_norm": 1.8892829418182373,
      "learning_rate": 6.717595441165053e-07,
      "loss": 1.9756,
      "num_input_tokens_seen": 301293624,
      "step": 295900
    },
    {
      "epoch": 5.82780413852848,
      "grad_norm": 1.8213739395141602,
      "learning_rate": 6.715658960049148e-07,
      "loss": 1.9704,
      "num_input_tokens_seen": 301395480,
      "step": 296000
    },
    {
      "epoch": 5.8297729912779825,
      "grad_norm": 3.8382740020751953,
      "learning_rate": 6.71372218719991e-07,
      "loss": 1.9791,
      "num_input_tokens_seen": 301497128,
      "step": 296100
    },
    {
      "epoch": 5.831741844027485,
      "grad_norm": 1.7518540620803833,
      "learning_rate": 6.711785122946665e-07,
      "loss": 2.0157,
      "num_input_tokens_seen": 301598712,
      "step": 296200
    },
    {
      "epoch": 5.833710696776988,
      "grad_norm": 3.1765055656433105,
      "learning_rate": 6.709847767618799e-07,
      "loss": 2.0557,
      "num_input_tokens_seen": 301701112,
      "step": 296300
    },
    {
      "epoch": 5.835679549526491,
      "grad_norm": 1.9587994813919067,
      "learning_rate": 6.707910121545741e-07,
      "loss": 1.9887,
      "num_input_tokens_seen": 301801120,
      "step": 296400
    },
    {
      "epoch": 5.837648402275994,
      "grad_norm": 2.064427137374878,
      "learning_rate": 6.705972185056974e-07,
      "loss": 1.973,
      "num_input_tokens_seen": 301902472,
      "step": 296500
    },
    {
      "epoch": 5.839617255025496,
      "grad_norm": 2.107823371887207,
      "learning_rate": 6.704033958482023e-07,
      "loss": 2.0214,
      "num_input_tokens_seen": 302002760,
      "step": 296600
    },
    {
      "epoch": 5.841586107774999,
      "grad_norm": 1.9233169555664062,
      "learning_rate": 6.702095442150473e-07,
      "loss": 2.0106,
      "num_input_tokens_seen": 302105160,
      "step": 296700
    },
    {
      "epoch": 5.843554960524503,
      "grad_norm": 1.7996037006378174,
      "learning_rate": 6.700156636391946e-07,
      "loss": 2.0012,
      "num_input_tokens_seen": 302206168,
      "step": 296800
    },
    {
      "epoch": 5.8455238132740055,
      "grad_norm": 1.813899278640747,
      "learning_rate": 6.698217541536126e-07,
      "loss": 1.9686,
      "num_input_tokens_seen": 302307472,
      "step": 296900
    },
    {
      "epoch": 5.847492666023508,
      "grad_norm": 1.7590276002883911,
      "learning_rate": 6.696278157912734e-07,
      "loss": 1.9898,
      "num_input_tokens_seen": 302409872,
      "step": 297000
    },
    {
      "epoch": 5.849461518773011,
      "grad_norm": 1.7217687368392944,
      "learning_rate": 6.69433848585155e-07,
      "loss": 1.9624,
      "num_input_tokens_seen": 302512272,
      "step": 297100
    },
    {
      "epoch": 5.851430371522514,
      "grad_norm": 1.771230697631836,
      "learning_rate": 6.692398525682396e-07,
      "loss": 1.9856,
      "num_input_tokens_seen": 302613304,
      "step": 297200
    },
    {
      "epoch": 5.8533992242720165,
      "grad_norm": 2.3142309188842773,
      "learning_rate": 6.690458277735147e-07,
      "loss": 1.9771,
      "num_input_tokens_seen": 302715704,
      "step": 297300
    },
    {
      "epoch": 5.855368077021519,
      "grad_norm": 2.4750351905822754,
      "learning_rate": 6.688517742339726e-07,
      "loss": 1.9781,
      "num_input_tokens_seen": 302817512,
      "step": 297400
    },
    {
      "epoch": 5.857336929771022,
      "grad_norm": 2.156783103942871,
      "learning_rate": 6.686576919826104e-07,
      "loss": 1.9988,
      "num_input_tokens_seen": 302918136,
      "step": 297500
    },
    {
      "epoch": 5.859305782520526,
      "grad_norm": 1.8108576536178589,
      "learning_rate": 6.684635810524303e-07,
      "loss": 1.9895,
      "num_input_tokens_seen": 303019304,
      "step": 297600
    },
    {
      "epoch": 5.8612746352700285,
      "grad_norm": 1.674936056137085,
      "learning_rate": 6.682694414764391e-07,
      "loss": 1.986,
      "num_input_tokens_seen": 303121704,
      "step": 297700
    },
    {
      "epoch": 5.863243488019531,
      "grad_norm": 1.9008970260620117,
      "learning_rate": 6.680752732876487e-07,
      "loss": 1.9639,
      "num_input_tokens_seen": 303221752,
      "step": 297800
    },
    {
      "epoch": 5.865212340769034,
      "grad_norm": 1.8959661722183228,
      "learning_rate": 6.678810765190756e-07,
      "loss": 2.0142,
      "num_input_tokens_seen": 303323392,
      "step": 297900
    },
    {
      "epoch": 5.867181193518537,
      "grad_norm": 1.9603074789047241,
      "learning_rate": 6.676868512037414e-07,
      "loss": 1.9939,
      "num_input_tokens_seen": 303425792,
      "step": 298000
    },
    {
      "epoch": 5.8691500462680395,
      "grad_norm": 1.7947169542312622,
      "learning_rate": 6.674925973746725e-07,
      "loss": 1.966,
      "num_input_tokens_seen": 303527608,
      "step": 298100
    },
    {
      "epoch": 5.871118899017542,
      "grad_norm": 2.029937744140625,
      "learning_rate": 6.672983150649003e-07,
      "loss": 2.004,
      "num_input_tokens_seen": 303629160,
      "step": 298200
    },
    {
      "epoch": 5.873087751767045,
      "grad_norm": 2.1655921936035156,
      "learning_rate": 6.671040043074605e-07,
      "loss": 1.9543,
      "num_input_tokens_seen": 303731464,
      "step": 298300
    },
    {
      "epoch": 5.875056604516548,
      "grad_norm": 1.9153037071228027,
      "learning_rate": 6.669096651353941e-07,
      "loss": 1.9276,
      "num_input_tokens_seen": 303833104,
      "step": 298400
    },
    {
      "epoch": 5.877025457266051,
      "grad_norm": 1.830942153930664,
      "learning_rate": 6.66715297581747e-07,
      "loss": 1.9594,
      "num_input_tokens_seen": 303935504,
      "step": 298500
    },
    {
      "epoch": 5.878994310015554,
      "grad_norm": 1.5987343788146973,
      "learning_rate": 6.665209016795698e-07,
      "loss": 1.9259,
      "num_input_tokens_seen": 304037304,
      "step": 298600
    },
    {
      "epoch": 5.880963162765057,
      "grad_norm": 2.0710251331329346,
      "learning_rate": 6.663264774619177e-07,
      "loss": 2.0013,
      "num_input_tokens_seen": 304138928,
      "step": 298700
    },
    {
      "epoch": 5.88293201551456,
      "grad_norm": 1.8655959367752075,
      "learning_rate": 6.66132024961851e-07,
      "loss": 1.9836,
      "num_input_tokens_seen": 304241328,
      "step": 298800
    },
    {
      "epoch": 5.8849008682640624,
      "grad_norm": 2.1698358058929443,
      "learning_rate": 6.659375442124346e-07,
      "loss": 1.9933,
      "num_input_tokens_seen": 304343728,
      "step": 298900
    },
    {
      "epoch": 5.886869721013565,
      "grad_norm": 1.8794077634811401,
      "learning_rate": 6.657430352467384e-07,
      "loss": 2.0005,
      "num_input_tokens_seen": 304446128,
      "step": 299000
    },
    {
      "epoch": 5.888838573763068,
      "grad_norm": 1.8287192583084106,
      "learning_rate": 6.655484980978369e-07,
      "loss": 2.0001,
      "num_input_tokens_seen": 304547992,
      "step": 299100
    },
    {
      "epoch": 5.890807426512572,
      "grad_norm": 2.011369228363037,
      "learning_rate": 6.653539327988099e-07,
      "loss": 2.0099,
      "num_input_tokens_seen": 304649976,
      "step": 299200
    },
    {
      "epoch": 5.892776279262074,
      "grad_norm": 2.1407384872436523,
      "learning_rate": 6.65159339382741e-07,
      "loss": 1.9703,
      "num_input_tokens_seen": 304751824,
      "step": 299300
    },
    {
      "epoch": 5.894745132011577,
      "grad_norm": 2.0718042850494385,
      "learning_rate": 6.649647178827196e-07,
      "loss": 2.0043,
      "num_input_tokens_seen": 304854224,
      "step": 299400
    },
    {
      "epoch": 5.89671398476108,
      "grad_norm": 2.4641711711883545,
      "learning_rate": 6.647700683318393e-07,
      "loss": 1.9998,
      "num_input_tokens_seen": 304956624,
      "step": 299500
    },
    {
      "epoch": 5.898682837510583,
      "grad_norm": 1.9418220520019531,
      "learning_rate": 6.645753907631985e-07,
      "loss": 1.9958,
      "num_input_tokens_seen": 305059024,
      "step": 299600
    },
    {
      "epoch": 5.900651690260085,
      "grad_norm": 2.018656015396118,
      "learning_rate": 6.643806852099008e-07,
      "loss": 1.9783,
      "num_input_tokens_seen": 305159496,
      "step": 299700
    },
    {
      "epoch": 5.902620543009588,
      "grad_norm": 1.9650826454162598,
      "learning_rate": 6.641859517050539e-07,
      "loss": 1.9963,
      "num_input_tokens_seen": 305261896,
      "step": 299800
    },
    {
      "epoch": 5.904589395759091,
      "grad_norm": 1.747464895248413,
      "learning_rate": 6.639911902817706e-07,
      "loss": 2.0045,
      "num_input_tokens_seen": 305364296,
      "step": 299900
    },
    {
      "epoch": 5.906558248508594,
      "grad_norm": 2.020000696182251,
      "learning_rate": 6.637964009731688e-07,
      "loss": 1.9278,
      "num_input_tokens_seen": 305466144,
      "step": 300000
    },
    {
      "epoch": 5.908527101258096,
      "grad_norm": 1.8583580255508423,
      "learning_rate": 6.636015838123705e-07,
      "loss": 1.9974,
      "num_input_tokens_seen": 305568544,
      "step": 300100
    },
    {
      "epoch": 5.9104959540076,
      "grad_norm": 1.9425022602081299,
      "learning_rate": 6.634067388325027e-07,
      "loss": 1.9641,
      "num_input_tokens_seen": 305670424,
      "step": 300200
    },
    {
      "epoch": 5.912464806757103,
      "grad_norm": 1.978251338005066,
      "learning_rate": 6.632118660666971e-07,
      "loss": 1.9954,
      "num_input_tokens_seen": 305772208,
      "step": 300300
    },
    {
      "epoch": 5.914433659506606,
      "grad_norm": 2.0952775478363037,
      "learning_rate": 6.630169655480903e-07,
      "loss": 2.0143,
      "num_input_tokens_seen": 305874256,
      "step": 300400
    },
    {
      "epoch": 5.916402512256108,
      "grad_norm": 2.00018572807312,
      "learning_rate": 6.628220373098234e-07,
      "loss": 1.9657,
      "num_input_tokens_seen": 305975320,
      "step": 300500
    },
    {
      "epoch": 5.918371365005611,
      "grad_norm": 2.504668951034546,
      "learning_rate": 6.626270813850425e-07,
      "loss": 1.9915,
      "num_input_tokens_seen": 306077192,
      "step": 300600
    },
    {
      "epoch": 5.920340217755114,
      "grad_norm": 1.8748530149459839,
      "learning_rate": 6.624320978068981e-07,
      "loss": 1.9738,
      "num_input_tokens_seen": 306178816,
      "step": 300700
    },
    {
      "epoch": 5.922309070504617,
      "grad_norm": 1.890470266342163,
      "learning_rate": 6.622370866085454e-07,
      "loss": 1.9791,
      "num_input_tokens_seen": 306281216,
      "step": 300800
    },
    {
      "epoch": 5.92427792325412,
      "grad_norm": 2.4180619716644287,
      "learning_rate": 6.620420478231445e-07,
      "loss": 1.9553,
      "num_input_tokens_seen": 306382984,
      "step": 300900
    },
    {
      "epoch": 5.926246776003623,
      "grad_norm": 2.6123709678649902,
      "learning_rate": 6.618469814838601e-07,
      "loss": 2.0037,
      "num_input_tokens_seen": 306483880,
      "step": 301000
    },
    {
      "epoch": 5.928215628753126,
      "grad_norm": 2.3167307376861572,
      "learning_rate": 6.616518876238616e-07,
      "loss": 1.9711,
      "num_input_tokens_seen": 306585752,
      "step": 301100
    },
    {
      "epoch": 5.930184481502629,
      "grad_norm": 1.7533416748046875,
      "learning_rate": 6.614567662763231e-07,
      "loss": 1.9962,
      "num_input_tokens_seen": 306688152,
      "step": 301200
    },
    {
      "epoch": 5.932153334252131,
      "grad_norm": 1.8543781042099,
      "learning_rate": 6.612616174744231e-07,
      "loss": 1.9595,
      "num_input_tokens_seen": 306790552,
      "step": 301300
    },
    {
      "epoch": 5.934122187001634,
      "grad_norm": 2.121992826461792,
      "learning_rate": 6.610664412513452e-07,
      "loss": 1.9946,
      "num_input_tokens_seen": 306891424,
      "step": 301400
    },
    {
      "epoch": 5.936091039751137,
      "grad_norm": 1.914199709892273,
      "learning_rate": 6.608712376402777e-07,
      "loss": 1.9761,
      "num_input_tokens_seen": 306993384,
      "step": 301500
    },
    {
      "epoch": 5.93805989250064,
      "grad_norm": 1.9082857370376587,
      "learning_rate": 6.606760066744129e-07,
      "loss": 2.0071,
      "num_input_tokens_seen": 307094568,
      "step": 301600
    },
    {
      "epoch": 5.940028745250142,
      "grad_norm": 2.240058660507202,
      "learning_rate": 6.604807483869485e-07,
      "loss": 2.0056,
      "num_input_tokens_seen": 307196744,
      "step": 301700
    },
    {
      "epoch": 5.941997597999645,
      "grad_norm": 1.9677985906600952,
      "learning_rate": 6.602854628110863e-07,
      "loss": 2.002,
      "num_input_tokens_seen": 307298552,
      "step": 301800
    },
    {
      "epoch": 5.943966450749149,
      "grad_norm": 2.0020642280578613,
      "learning_rate": 6.600901499800331e-07,
      "loss": 2.0117,
      "num_input_tokens_seen": 307400952,
      "step": 301900
    },
    {
      "epoch": 5.9459353034986515,
      "grad_norm": 1.9213149547576904,
      "learning_rate": 6.598948099270001e-07,
      "loss": 1.9945,
      "num_input_tokens_seen": 307501816,
      "step": 302000
    },
    {
      "epoch": 5.947904156248154,
      "grad_norm": 1.8820880651474,
      "learning_rate": 6.596994426852034e-07,
      "loss": 2.0174,
      "num_input_tokens_seen": 307604216,
      "step": 302100
    },
    {
      "epoch": 5.949873008997657,
      "grad_norm": 1.952135682106018,
      "learning_rate": 6.595040482878634e-07,
      "loss": 1.9804,
      "num_input_tokens_seen": 307706408,
      "step": 302200
    },
    {
      "epoch": 5.95184186174716,
      "grad_norm": 1.8562573194503784,
      "learning_rate": 6.593086267682052e-07,
      "loss": 1.9871,
      "num_input_tokens_seen": 307808168,
      "step": 302300
    },
    {
      "epoch": 5.9538107144966625,
      "grad_norm": 2.650224208831787,
      "learning_rate": 6.591131781594587e-07,
      "loss": 2.0571,
      "num_input_tokens_seen": 307907840,
      "step": 302400
    },
    {
      "epoch": 5.955779567246165,
      "grad_norm": 1.6823968887329102,
      "learning_rate": 6.589177024948584e-07,
      "loss": 1.982,
      "num_input_tokens_seen": 308010240,
      "step": 302500
    },
    {
      "epoch": 5.957748419995669,
      "grad_norm": 2.0597550868988037,
      "learning_rate": 6.587221998076429e-07,
      "loss": 1.97,
      "num_input_tokens_seen": 308112640,
      "step": 302600
    },
    {
      "epoch": 5.959717272745172,
      "grad_norm": 7.394330978393555,
      "learning_rate": 6.585266701310563e-07,
      "loss": 1.9622,
      "num_input_tokens_seen": 308213864,
      "step": 302700
    },
    {
      "epoch": 5.9616861254946745,
      "grad_norm": 2.0766587257385254,
      "learning_rate": 6.583311134983464e-07,
      "loss": 1.9822,
      "num_input_tokens_seen": 308315672,
      "step": 302800
    },
    {
      "epoch": 5.963654978244177,
      "grad_norm": 2.0015435218811035,
      "learning_rate": 6.581355299427662e-07,
      "loss": 1.9894,
      "num_input_tokens_seen": 308418072,
      "step": 302900
    },
    {
      "epoch": 5.96562383099368,
      "grad_norm": 1.8748551607131958,
      "learning_rate": 6.579399194975729e-07,
      "loss": 1.9726,
      "num_input_tokens_seen": 308520472,
      "step": 303000
    },
    {
      "epoch": 5.967592683743183,
      "grad_norm": 1.9935256242752075,
      "learning_rate": 6.577442821960284e-07,
      "loss": 2.0136,
      "num_input_tokens_seen": 308622208,
      "step": 303100
    },
    {
      "epoch": 5.9695615364926855,
      "grad_norm": 1.722451090812683,
      "learning_rate": 6.575486180713992e-07,
      "loss": 2.0166,
      "num_input_tokens_seen": 308723224,
      "step": 303200
    },
    {
      "epoch": 5.971530389242188,
      "grad_norm": 1.9466898441314697,
      "learning_rate": 6.573529271569562e-07,
      "loss": 2.0004,
      "num_input_tokens_seen": 308824872,
      "step": 303300
    },
    {
      "epoch": 5.973499241991691,
      "grad_norm": 1.8579930067062378,
      "learning_rate": 6.571572094859752e-07,
      "loss": 1.9879,
      "num_input_tokens_seen": 308926504,
      "step": 303400
    },
    {
      "epoch": 5.975468094741195,
      "grad_norm": 1.9859167337417603,
      "learning_rate": 6.569614650917362e-07,
      "loss": 1.9813,
      "num_input_tokens_seen": 309027640,
      "step": 303500
    },
    {
      "epoch": 5.977436947490697,
      "grad_norm": 1.9646979570388794,
      "learning_rate": 6.56765694007524e-07,
      "loss": 2.0483,
      "num_input_tokens_seen": 309129488,
      "step": 303600
    },
    {
      "epoch": 5.9794058002402,
      "grad_norm": 2.0104897022247314,
      "learning_rate": 6.565698962666277e-07,
      "loss": 1.9652,
      "num_input_tokens_seen": 309231336,
      "step": 303700
    },
    {
      "epoch": 5.981374652989703,
      "grad_norm": 1.8535076379776,
      "learning_rate": 6.563740719023412e-07,
      "loss": 2.0091,
      "num_input_tokens_seen": 309333736,
      "step": 303800
    },
    {
      "epoch": 5.983343505739206,
      "grad_norm": 2.1762800216674805,
      "learning_rate": 6.561782209479626e-07,
      "loss": 1.9728,
      "num_input_tokens_seen": 309436136,
      "step": 303900
    },
    {
      "epoch": 5.9853123584887085,
      "grad_norm": 1.9892023801803589,
      "learning_rate": 6.559823434367948e-07,
      "loss": 2.0493,
      "num_input_tokens_seen": 309538312,
      "step": 304000
    },
    {
      "epoch": 5.987281211238211,
      "grad_norm": 2.135478973388672,
      "learning_rate": 6.557864394021453e-07,
      "loss": 1.9785,
      "num_input_tokens_seen": 309640232,
      "step": 304100
    },
    {
      "epoch": 5.989250063987714,
      "grad_norm": 1.9356056451797485,
      "learning_rate": 6.555905088773255e-07,
      "loss": 1.9847,
      "num_input_tokens_seen": 309741984,
      "step": 304200
    },
    {
      "epoch": 5.991218916737218,
      "grad_norm": 1.9058266878128052,
      "learning_rate": 6.55394551895652e-07,
      "loss": 1.9464,
      "num_input_tokens_seen": 309843072,
      "step": 304300
    },
    {
      "epoch": 5.99318776948672,
      "grad_norm": 1.9438482522964478,
      "learning_rate": 6.551985684904457e-07,
      "loss": 1.9916,
      "num_input_tokens_seen": 309943736,
      "step": 304400
    },
    {
      "epoch": 5.995156622236223,
      "grad_norm": 2.3376619815826416,
      "learning_rate": 6.550025586950319e-07,
      "loss": 2.0215,
      "num_input_tokens_seen": 310046136,
      "step": 304500
    },
    {
      "epoch": 5.997125474985726,
      "grad_norm": 1.8232134580612183,
      "learning_rate": 6.548065225427402e-07,
      "loss": 1.9955,
      "num_input_tokens_seen": 310148536,
      "step": 304600
    },
    {
      "epoch": 5.999094327735229,
      "grad_norm": 2.128138303756714,
      "learning_rate": 6.546104600669051e-07,
      "loss": 2.0323,
      "num_input_tokens_seen": 310249568,
      "step": 304700
    },
    {
      "epoch": 6.001063180484731,
      "grad_norm": 1.7412737607955933,
      "learning_rate": 6.544143713008652e-07,
      "loss": 1.9671,
      "num_input_tokens_seen": 310351192,
      "step": 304800
    },
    {
      "epoch": 6.003032033234234,
      "grad_norm": 1.8386776447296143,
      "learning_rate": 6.542182562779641e-07,
      "loss": 1.9604,
      "num_input_tokens_seen": 310452128,
      "step": 304900
    },
    {
      "epoch": 6.005000885983737,
      "grad_norm": 2.1488161087036133,
      "learning_rate": 6.54022115031549e-07,
      "loss": 2.0128,
      "num_input_tokens_seen": 310552416,
      "step": 305000
    },
    {
      "epoch": 6.006969738733241,
      "grad_norm": 1.9873698949813843,
      "learning_rate": 6.538259475949726e-07,
      "loss": 1.9888,
      "num_input_tokens_seen": 310654816,
      "step": 305100
    },
    {
      "epoch": 6.008938591482743,
      "grad_norm": 2.0063347816467285,
      "learning_rate": 6.53629754001591e-07,
      "loss": 1.9771,
      "num_input_tokens_seen": 310757216,
      "step": 305200
    },
    {
      "epoch": 6.010907444232246,
      "grad_norm": 1.8294849395751953,
      "learning_rate": 6.534335342847657e-07,
      "loss": 2.0169,
      "num_input_tokens_seen": 310856768,
      "step": 305300
    },
    {
      "epoch": 6.012876296981749,
      "grad_norm": 1.804033875465393,
      "learning_rate": 6.532372884778618e-07,
      "loss": 1.9687,
      "num_input_tokens_seen": 310959168,
      "step": 305400
    },
    {
      "epoch": 6.014845149731252,
      "grad_norm": 1.8964186906814575,
      "learning_rate": 6.530410166142495e-07,
      "loss": 1.9883,
      "num_input_tokens_seen": 311060792,
      "step": 305500
    },
    {
      "epoch": 6.016814002480754,
      "grad_norm": 1.8022997379302979,
      "learning_rate": 6.528447187273032e-07,
      "loss": 1.9433,
      "num_input_tokens_seen": 311163192,
      "step": 305600
    },
    {
      "epoch": 6.018782855230257,
      "grad_norm": 2.142498254776001,
      "learning_rate": 6.526483948504016e-07,
      "loss": 1.964,
      "num_input_tokens_seen": 311263936,
      "step": 305700
    },
    {
      "epoch": 6.02075170797976,
      "grad_norm": 1.9424124956130981,
      "learning_rate": 6.52452045016928e-07,
      "loss": 2.0609,
      "num_input_tokens_seen": 311364776,
      "step": 305800
    },
    {
      "epoch": 6.022720560729263,
      "grad_norm": 1.9188220500946045,
      "learning_rate": 6.522556692602698e-07,
      "loss": 2.0215,
      "num_input_tokens_seen": 311465584,
      "step": 305900
    },
    {
      "epoch": 6.024689413478766,
      "grad_norm": 1.7720577716827393,
      "learning_rate": 6.520592676138196e-07,
      "loss": 1.9621,
      "num_input_tokens_seen": 311567984,
      "step": 306000
    },
    {
      "epoch": 6.026658266228269,
      "grad_norm": 1.8154102563858032,
      "learning_rate": 6.518628401109731e-07,
      "loss": 1.998,
      "num_input_tokens_seen": 311669728,
      "step": 306100
    },
    {
      "epoch": 6.028627118977772,
      "grad_norm": 2.2377007007598877,
      "learning_rate": 6.516663867851318e-07,
      "loss": 1.9819,
      "num_input_tokens_seen": 311770080,
      "step": 306200
    },
    {
      "epoch": 6.030595971727275,
      "grad_norm": 1.8667947053909302,
      "learning_rate": 6.514699076697005e-07,
      "loss": 1.9614,
      "num_input_tokens_seen": 311872480,
      "step": 306300
    },
    {
      "epoch": 6.032564824476777,
      "grad_norm": 1.9634737968444824,
      "learning_rate": 6.512734027980889e-07,
      "loss": 2.0079,
      "num_input_tokens_seen": 311974880,
      "step": 306400
    },
    {
      "epoch": 6.03453367722628,
      "grad_norm": 2.067988157272339,
      "learning_rate": 6.51076872203711e-07,
      "loss": 2.0224,
      "num_input_tokens_seen": 312076840,
      "step": 306500
    },
    {
      "epoch": 6.036502529975783,
      "grad_norm": 1.9597209692001343,
      "learning_rate": 6.508803159199854e-07,
      "loss": 1.962,
      "num_input_tokens_seen": 312179240,
      "step": 306600
    },
    {
      "epoch": 6.038471382725286,
      "grad_norm": 1.8099509477615356,
      "learning_rate": 6.506837339803346e-07,
      "loss": 1.9511,
      "num_input_tokens_seen": 312281640,
      "step": 306700
    },
    {
      "epoch": 6.040440235474789,
      "grad_norm": 2.1230101585388184,
      "learning_rate": 6.504871264181857e-07,
      "loss": 2.016,
      "num_input_tokens_seen": 312384040,
      "step": 306800
    },
    {
      "epoch": 6.042409088224292,
      "grad_norm": 1.8817248344421387,
      "learning_rate": 6.502904932669702e-07,
      "loss": 2.0034,
      "num_input_tokens_seen": 312484688,
      "step": 306900
    },
    {
      "epoch": 6.044377940973795,
      "grad_norm": 1.9372673034667969,
      "learning_rate": 6.500938345601242e-07,
      "loss": 2.0577,
      "num_input_tokens_seen": 312586792,
      "step": 307000
    },
    {
      "epoch": 6.0463467937232975,
      "grad_norm": 1.7838410139083862,
      "learning_rate": 6.498971503310872e-07,
      "loss": 1.9655,
      "num_input_tokens_seen": 312688880,
      "step": 307100
    },
    {
      "epoch": 6.0483156464728,
      "grad_norm": 2.113298177719116,
      "learning_rate": 6.497004406133044e-07,
      "loss": 2.0385,
      "num_input_tokens_seen": 312790280,
      "step": 307200
    },
    {
      "epoch": 6.050284499222303,
      "grad_norm": 1.816819190979004,
      "learning_rate": 6.495037054402242e-07,
      "loss": 1.9808,
      "num_input_tokens_seen": 312892680,
      "step": 307300
    },
    {
      "epoch": 6.052253351971806,
      "grad_norm": 1.963083028793335,
      "learning_rate": 6.493069448452998e-07,
      "loss": 1.9673,
      "num_input_tokens_seen": 312994440,
      "step": 307400
    },
    {
      "epoch": 6.054222204721309,
      "grad_norm": 1.9467344284057617,
      "learning_rate": 6.491101588619888e-07,
      "loss": 1.9874,
      "num_input_tokens_seen": 313096840,
      "step": 307500
    },
    {
      "epoch": 6.056191057470811,
      "grad_norm": 1.747800588607788,
      "learning_rate": 6.489133475237528e-07,
      "loss": 1.9606,
      "num_input_tokens_seen": 313198624,
      "step": 307600
    },
    {
      "epoch": 6.058159910220315,
      "grad_norm": 2.129979133605957,
      "learning_rate": 6.487165108640581e-07,
      "loss": 2.0203,
      "num_input_tokens_seen": 313300320,
      "step": 307700
    },
    {
      "epoch": 6.060128762969818,
      "grad_norm": 1.9538705348968506,
      "learning_rate": 6.485196489163752e-07,
      "loss": 2.0202,
      "num_input_tokens_seen": 313401664,
      "step": 307800
    },
    {
      "epoch": 6.0620976157193205,
      "grad_norm": 2.107147216796875,
      "learning_rate": 6.483227617141786e-07,
      "loss": 1.9897,
      "num_input_tokens_seen": 313504064,
      "step": 307900
    },
    {
      "epoch": 6.064066468468823,
      "grad_norm": 2.035672903060913,
      "learning_rate": 6.481258492909474e-07,
      "loss": 2.024,
      "num_input_tokens_seen": 313604280,
      "step": 308000
    },
    {
      "epoch": 6.066035321218326,
      "grad_norm": 1.7072129249572754,
      "learning_rate": 6.479289116801648e-07,
      "loss": 2.0033,
      "num_input_tokens_seen": 313705928,
      "step": 308100
    },
    {
      "epoch": 6.068004173967829,
      "grad_norm": 1.9450758695602417,
      "learning_rate": 6.477319489153185e-07,
      "loss": 1.9879,
      "num_input_tokens_seen": 313808328,
      "step": 308200
    },
    {
      "epoch": 6.0699730267173315,
      "grad_norm": 2.076735019683838,
      "learning_rate": 6.475349610299004e-07,
      "loss": 1.9798,
      "num_input_tokens_seen": 313909912,
      "step": 308300
    },
    {
      "epoch": 6.071941879466834,
      "grad_norm": 1.7819191217422485,
      "learning_rate": 6.473379480574065e-07,
      "loss": 2.0272,
      "num_input_tokens_seen": 314011856,
      "step": 308400
    },
    {
      "epoch": 6.073910732216338,
      "grad_norm": 2.2660250663757324,
      "learning_rate": 6.471409100313372e-07,
      "loss": 2.0186,
      "num_input_tokens_seen": 314114256,
      "step": 308500
    },
    {
      "epoch": 6.075879584965841,
      "grad_norm": 1.8634023666381836,
      "learning_rate": 6.469438469851971e-07,
      "loss": 2.01,
      "num_input_tokens_seen": 314215208,
      "step": 308600
    },
    {
      "epoch": 6.077848437715343,
      "grad_norm": 1.7328294515609741,
      "learning_rate": 6.467467589524954e-07,
      "loss": 1.9789,
      "num_input_tokens_seen": 314317608,
      "step": 308700
    },
    {
      "epoch": 6.079817290464846,
      "grad_norm": 2.0241761207580566,
      "learning_rate": 6.465496459667449e-07,
      "loss": 2.0178,
      "num_input_tokens_seen": 314418184,
      "step": 308800
    },
    {
      "epoch": 6.081786143214349,
      "grad_norm": 1.7773127555847168,
      "learning_rate": 6.463525080614631e-07,
      "loss": 2.0023,
      "num_input_tokens_seen": 314520584,
      "step": 308900
    },
    {
      "epoch": 6.083754995963852,
      "grad_norm": 2.274423122406006,
      "learning_rate": 6.46155345270172e-07,
      "loss": 1.9614,
      "num_input_tokens_seen": 314622984,
      "step": 309000
    },
    {
      "epoch": 6.0857238487133545,
      "grad_norm": 2.167804479598999,
      "learning_rate": 6.459581576263969e-07,
      "loss": 2.0152,
      "num_input_tokens_seen": 314724968,
      "step": 309100
    },
    {
      "epoch": 6.087692701462857,
      "grad_norm": 1.6423709392547607,
      "learning_rate": 6.457609451636683e-07,
      "loss": 2.0155,
      "num_input_tokens_seen": 314827368,
      "step": 309200
    },
    {
      "epoch": 6.089661554212361,
      "grad_norm": 1.932984709739685,
      "learning_rate": 6.455637079155203e-07,
      "loss": 1.9914,
      "num_input_tokens_seen": 314929080,
      "step": 309300
    },
    {
      "epoch": 6.091630406961864,
      "grad_norm": 2.155951976776123,
      "learning_rate": 6.453664459154916e-07,
      "loss": 1.9567,
      "num_input_tokens_seen": 315031480,
      "step": 309400
    },
    {
      "epoch": 6.093599259711366,
      "grad_norm": 1.9752286672592163,
      "learning_rate": 6.451691591971248e-07,
      "loss": 1.9582,
      "num_input_tokens_seen": 315133880,
      "step": 309500
    },
    {
      "epoch": 6.095568112460869,
      "grad_norm": 1.9401935338974,
      "learning_rate": 6.449718477939669e-07,
      "loss": 1.9556,
      "num_input_tokens_seen": 315236280,
      "step": 309600
    },
    {
      "epoch": 6.097536965210372,
      "grad_norm": 1.7289832830429077,
      "learning_rate": 6.44774511739569e-07,
      "loss": 1.9876,
      "num_input_tokens_seen": 315336760,
      "step": 309700
    },
    {
      "epoch": 6.099505817959875,
      "grad_norm": 1.9301186800003052,
      "learning_rate": 6.445771510674865e-07,
      "loss": 1.988,
      "num_input_tokens_seen": 315439160,
      "step": 309800
    },
    {
      "epoch": 6.101474670709377,
      "grad_norm": 1.912955641746521,
      "learning_rate": 6.443797658112788e-07,
      "loss": 1.9711,
      "num_input_tokens_seen": 315541112,
      "step": 309900
    },
    {
      "epoch": 6.10344352345888,
      "grad_norm": 1.7604267597198486,
      "learning_rate": 6.441823560045098e-07,
      "loss": 1.9465,
      "num_input_tokens_seen": 315642496,
      "step": 310000
    },
    {
      "epoch": 6.105412376208383,
      "grad_norm": 1.8021234273910522,
      "learning_rate": 6.439849216807472e-07,
      "loss": 1.9585,
      "num_input_tokens_seen": 315744896,
      "step": 310100
    },
    {
      "epoch": 6.107381228957887,
      "grad_norm": 2.0396640300750732,
      "learning_rate": 6.437874628735631e-07,
      "loss": 2.0367,
      "num_input_tokens_seen": 315846560,
      "step": 310200
    },
    {
      "epoch": 6.109350081707389,
      "grad_norm": 1.698716163635254,
      "learning_rate": 6.435899796165337e-07,
      "loss": 1.9712,
      "num_input_tokens_seen": 315948960,
      "step": 310300
    },
    {
      "epoch": 6.111318934456892,
      "grad_norm": 1.9908227920532227,
      "learning_rate": 6.433924719432393e-07,
      "loss": 1.9535,
      "num_input_tokens_seen": 316050872,
      "step": 310400
    },
    {
      "epoch": 6.113287787206395,
      "grad_norm": 1.7665883302688599,
      "learning_rate": 6.431949398872646e-07,
      "loss": 1.9809,
      "num_input_tokens_seen": 316151576,
      "step": 310500
    },
    {
      "epoch": 6.115256639955898,
      "grad_norm": 1.7452491521835327,
      "learning_rate": 6.429973834821979e-07,
      "loss": 2.0128,
      "num_input_tokens_seen": 316253384,
      "step": 310600
    },
    {
      "epoch": 6.1172254927054,
      "grad_norm": 2.101358652114868,
      "learning_rate": 6.427998027616324e-07,
      "loss": 1.9806,
      "num_input_tokens_seen": 316355784,
      "step": 310700
    },
    {
      "epoch": 6.119194345454903,
      "grad_norm": 1.9029123783111572,
      "learning_rate": 6.426021977591648e-07,
      "loss": 1.9777,
      "num_input_tokens_seen": 316456696,
      "step": 310800
    },
    {
      "epoch": 6.121163198204406,
      "grad_norm": 1.7739269733428955,
      "learning_rate": 6.424045685083963e-07,
      "loss": 1.9684,
      "num_input_tokens_seen": 316558008,
      "step": 310900
    },
    {
      "epoch": 6.1231320509539096,
      "grad_norm": 2.0812501907348633,
      "learning_rate": 6.42206915042932e-07,
      "loss": 2.0026,
      "num_input_tokens_seen": 316659848,
      "step": 311000
    },
    {
      "epoch": 6.125100903703412,
      "grad_norm": 1.91690194606781,
      "learning_rate": 6.420092373963814e-07,
      "loss": 1.9969,
      "num_input_tokens_seen": 316761712,
      "step": 311100
    },
    {
      "epoch": 6.127069756452915,
      "grad_norm": 1.6721659898757935,
      "learning_rate": 6.418115356023575e-07,
      "loss": 1.9864,
      "num_input_tokens_seen": 316864112,
      "step": 311200
    },
    {
      "epoch": 6.129038609202418,
      "grad_norm": 4.432291507720947,
      "learning_rate": 6.416138096944782e-07,
      "loss": 2.0291,
      "num_input_tokens_seen": 316965184,
      "step": 311300
    },
    {
      "epoch": 6.131007461951921,
      "grad_norm": 1.8571569919586182,
      "learning_rate": 6.414160597063648e-07,
      "loss": 2.0362,
      "num_input_tokens_seen": 317067584,
      "step": 311400
    },
    {
      "epoch": 6.132976314701423,
      "grad_norm": 2.183241128921509,
      "learning_rate": 6.412182856716435e-07,
      "loss": 2.0357,
      "num_input_tokens_seen": 317169984,
      "step": 311500
    },
    {
      "epoch": 6.134945167450926,
      "grad_norm": 1.953399658203125,
      "learning_rate": 6.410204876239435e-07,
      "loss": 1.9955,
      "num_input_tokens_seen": 317272384,
      "step": 311600
    },
    {
      "epoch": 6.136914020200429,
      "grad_norm": 1.638275384902954,
      "learning_rate": 6.408226655968992e-07,
      "loss": 1.9822,
      "num_input_tokens_seen": 317373952,
      "step": 311700
    },
    {
      "epoch": 6.138882872949932,
      "grad_norm": 1.8970870971679688,
      "learning_rate": 6.406248196241482e-07,
      "loss": 1.9544,
      "num_input_tokens_seen": 317476352,
      "step": 311800
    },
    {
      "epoch": 6.140851725699435,
      "grad_norm": 2.006563186645508,
      "learning_rate": 6.404269497393329e-07,
      "loss": 1.9694,
      "num_input_tokens_seen": 317578752,
      "step": 311900
    },
    {
      "epoch": 6.142820578448938,
      "grad_norm": 2.508770704269409,
      "learning_rate": 6.402290559760991e-07,
      "loss": 1.9487,
      "num_input_tokens_seen": 317681152,
      "step": 312000
    },
    {
      "epoch": 6.144789431198441,
      "grad_norm": 1.7859289646148682,
      "learning_rate": 6.400311383680972e-07,
      "loss": 2.0625,
      "num_input_tokens_seen": 317782736,
      "step": 312100
    },
    {
      "epoch": 6.1467582839479435,
      "grad_norm": 1.9423662424087524,
      "learning_rate": 6.398331969489812e-07,
      "loss": 2.0034,
      "num_input_tokens_seen": 317882104,
      "step": 312200
    },
    {
      "epoch": 6.148727136697446,
      "grad_norm": 1.9616044759750366,
      "learning_rate": 6.396352317524094e-07,
      "loss": 2.0182,
      "num_input_tokens_seen": 317983328,
      "step": 312300
    },
    {
      "epoch": 6.150695989446949,
      "grad_norm": 1.9331483840942383,
      "learning_rate": 6.394372428120442e-07,
      "loss": 2.0063,
      "num_input_tokens_seen": 318085728,
      "step": 312400
    },
    {
      "epoch": 6.152664842196452,
      "grad_norm": 2.319206953048706,
      "learning_rate": 6.392392301615521e-07,
      "loss": 1.9624,
      "num_input_tokens_seen": 318187600,
      "step": 312500
    },
    {
      "epoch": 6.154633694945955,
      "grad_norm": 2.0272469520568848,
      "learning_rate": 6.390411938346032e-07,
      "loss": 1.9519,
      "num_input_tokens_seen": 318289240,
      "step": 312600
    },
    {
      "epoch": 6.156602547695458,
      "grad_norm": 5.838209629058838,
      "learning_rate": 6.388431338648719e-07,
      "loss": 1.9764,
      "num_input_tokens_seen": 318390368,
      "step": 312700
    },
    {
      "epoch": 6.158571400444961,
      "grad_norm": 1.9024221897125244,
      "learning_rate": 6.386450502860369e-07,
      "loss": 2.0118,
      "num_input_tokens_seen": 318491344,
      "step": 312800
    },
    {
      "epoch": 6.160540253194464,
      "grad_norm": 2.111111879348755,
      "learning_rate": 6.384469431317805e-07,
      "loss": 2.008,
      "num_input_tokens_seen": 318593744,
      "step": 312900
    },
    {
      "epoch": 6.1625091059439665,
      "grad_norm": 1.8804786205291748,
      "learning_rate": 6.382488124357889e-07,
      "loss": 1.9992,
      "num_input_tokens_seen": 318696144,
      "step": 313000
    },
    {
      "epoch": 6.164477958693469,
      "grad_norm": 1.8802192211151123,
      "learning_rate": 6.380506582317531e-07,
      "loss": 1.9387,
      "num_input_tokens_seen": 318798544,
      "step": 313100
    },
    {
      "epoch": 6.166446811442972,
      "grad_norm": 2.6956980228424072,
      "learning_rate": 6.37852480553367e-07,
      "loss": 1.9605,
      "num_input_tokens_seen": 318900944,
      "step": 313200
    },
    {
      "epoch": 6.168415664192475,
      "grad_norm": 2.159541606903076,
      "learning_rate": 6.376542794343295e-07,
      "loss": 1.9499,
      "num_input_tokens_seen": 319001936,
      "step": 313300
    },
    {
      "epoch": 6.1703845169419775,
      "grad_norm": 1.9137659072875977,
      "learning_rate": 6.374560549083426e-07,
      "loss": 1.9951,
      "num_input_tokens_seen": 319103568,
      "step": 313400
    },
    {
      "epoch": 6.172353369691481,
      "grad_norm": 1.8772330284118652,
      "learning_rate": 6.372578070091131e-07,
      "loss": 1.9354,
      "num_input_tokens_seen": 319205968,
      "step": 313500
    },
    {
      "epoch": 6.174322222440984,
      "grad_norm": 2.034302234649658,
      "learning_rate": 6.370595357703511e-07,
      "loss": 2.0053,
      "num_input_tokens_seen": 319308368,
      "step": 313600
    },
    {
      "epoch": 6.176291075190487,
      "grad_norm": 2.2465462684631348,
      "learning_rate": 6.36861241225771e-07,
      "loss": 1.9863,
      "num_input_tokens_seen": 319410128,
      "step": 313700
    },
    {
      "epoch": 6.1782599279399895,
      "grad_norm": 1.771889090538025,
      "learning_rate": 6.366629234090913e-07,
      "loss": 1.9867,
      "num_input_tokens_seen": 319512528,
      "step": 313800
    },
    {
      "epoch": 6.180228780689492,
      "grad_norm": 1.7891899347305298,
      "learning_rate": 6.364645823540341e-07,
      "loss": 1.9278,
      "num_input_tokens_seen": 319614928,
      "step": 313900
    },
    {
      "epoch": 6.182197633438995,
      "grad_norm": 1.8706449270248413,
      "learning_rate": 6.362662180943257e-07,
      "loss": 2.0377,
      "num_input_tokens_seen": 319716376,
      "step": 314000
    },
    {
      "epoch": 6.184166486188498,
      "grad_norm": 1.7015857696533203,
      "learning_rate": 6.360678306636962e-07,
      "loss": 1.9887,
      "num_input_tokens_seen": 319817640,
      "step": 314100
    },
    {
      "epoch": 6.1861353389380005,
      "grad_norm": 1.8962303400039673,
      "learning_rate": 6.358694200958797e-07,
      "loss": 2.0208,
      "num_input_tokens_seen": 319919264,
      "step": 314200
    },
    {
      "epoch": 6.188104191687504,
      "grad_norm": 2.0260705947875977,
      "learning_rate": 6.356709864246143e-07,
      "loss": 1.9916,
      "num_input_tokens_seen": 320021568,
      "step": 314300
    },
    {
      "epoch": 6.190073044437007,
      "grad_norm": 2.3741517066955566,
      "learning_rate": 6.354725296836421e-07,
      "loss": 2.0141,
      "num_input_tokens_seen": 320123920,
      "step": 314400
    },
    {
      "epoch": 6.19204189718651,
      "grad_norm": 2.0920419692993164,
      "learning_rate": 6.352740499067086e-07,
      "loss": 2.0139,
      "num_input_tokens_seen": 320226024,
      "step": 314500
    },
    {
      "epoch": 6.194010749936012,
      "grad_norm": 1.7310185432434082,
      "learning_rate": 6.350755471275642e-07,
      "loss": 1.9685,
      "num_input_tokens_seen": 320328424,
      "step": 314600
    },
    {
      "epoch": 6.195979602685515,
      "grad_norm": 1.8541840314865112,
      "learning_rate": 6.34877021379962e-07,
      "loss": 1.966,
      "num_input_tokens_seen": 320430824,
      "step": 314700
    },
    {
      "epoch": 6.197948455435018,
      "grad_norm": 1.7966785430908203,
      "learning_rate": 6.346784726976599e-07,
      "loss": 1.9815,
      "num_input_tokens_seen": 320533224,
      "step": 314800
    },
    {
      "epoch": 6.199917308184521,
      "grad_norm": 1.8406530618667603,
      "learning_rate": 6.344799011144196e-07,
      "loss": 2.0185,
      "num_input_tokens_seen": 320634864,
      "step": 314900
    },
    {
      "epoch": 6.2018861609340235,
      "grad_norm": 1.9197044372558594,
      "learning_rate": 6.342813066640063e-07,
      "loss": 2.0046,
      "num_input_tokens_seen": 320736832,
      "step": 315000
    },
    {
      "epoch": 6.203855013683526,
      "grad_norm": 1.9414684772491455,
      "learning_rate": 6.340826893801895e-07,
      "loss": 1.9641,
      "num_input_tokens_seen": 320839232,
      "step": 315100
    },
    {
      "epoch": 6.20582386643303,
      "grad_norm": 1.6983762979507446,
      "learning_rate": 6.338840492967422e-07,
      "loss": 1.9799,
      "num_input_tokens_seen": 320940808,
      "step": 315200
    },
    {
      "epoch": 6.207792719182533,
      "grad_norm": 2.312974214553833,
      "learning_rate": 6.336853864474416e-07,
      "loss": 1.9989,
      "num_input_tokens_seen": 321043208,
      "step": 315300
    },
    {
      "epoch": 6.209761571932035,
      "grad_norm": 1.7660846710205078,
      "learning_rate": 6.334867008660685e-07,
      "loss": 1.9933,
      "num_input_tokens_seen": 321145608,
      "step": 315400
    },
    {
      "epoch": 6.211730424681538,
      "grad_norm": 2.285482406616211,
      "learning_rate": 6.332879925864079e-07,
      "loss": 1.9407,
      "num_input_tokens_seen": 321248008,
      "step": 315500
    },
    {
      "epoch": 6.213699277431041,
      "grad_norm": 2.0247650146484375,
      "learning_rate": 6.330892616422483e-07,
      "loss": 2.0361,
      "num_input_tokens_seen": 321350408,
      "step": 315600
    },
    {
      "epoch": 6.215668130180544,
      "grad_norm": 1.7272166013717651,
      "learning_rate": 6.328905080673821e-07,
      "loss": 2.0177,
      "num_input_tokens_seen": 321451936,
      "step": 315700
    },
    {
      "epoch": 6.217636982930046,
      "grad_norm": 1.9985069036483765,
      "learning_rate": 6.326917318956061e-07,
      "loss": 2.0181,
      "num_input_tokens_seen": 321554336,
      "step": 315800
    },
    {
      "epoch": 6.219605835679549,
      "grad_norm": 1.9732002019882202,
      "learning_rate": 6.324929331607201e-07,
      "loss": 1.9909,
      "num_input_tokens_seen": 321655032,
      "step": 315900
    },
    {
      "epoch": 6.221574688429053,
      "grad_norm": 2.13789439201355,
      "learning_rate": 6.322941118965285e-07,
      "loss": 1.9734,
      "num_input_tokens_seen": 321757432,
      "step": 316000
    },
    {
      "epoch": 6.223543541178556,
      "grad_norm": 1.8527015447616577,
      "learning_rate": 6.320952681368389e-07,
      "loss": 1.9818,
      "num_input_tokens_seen": 321859072,
      "step": 316100
    },
    {
      "epoch": 6.225512393928058,
      "grad_norm": 1.8170995712280273,
      "learning_rate": 6.318964019154632e-07,
      "loss": 1.9601,
      "num_input_tokens_seen": 321960168,
      "step": 316200
    },
    {
      "epoch": 6.227481246677561,
      "grad_norm": 2.16269850730896,
      "learning_rate": 6.316975132662165e-07,
      "loss": 2.0435,
      "num_input_tokens_seen": 322060616,
      "step": 316300
    },
    {
      "epoch": 6.229450099427064,
      "grad_norm": 1.6585129499435425,
      "learning_rate": 6.314986022229189e-07,
      "loss": 1.9044,
      "num_input_tokens_seen": 322162184,
      "step": 316400
    },
    {
      "epoch": 6.231418952176567,
      "grad_norm": 1.984542965888977,
      "learning_rate": 6.312996688193927e-07,
      "loss": 1.993,
      "num_input_tokens_seen": 322263808,
      "step": 316500
    },
    {
      "epoch": 6.233387804926069,
      "grad_norm": 1.7414846420288086,
      "learning_rate": 6.311007130894656e-07,
      "loss": 1.9723,
      "num_input_tokens_seen": 322366208,
      "step": 316600
    },
    {
      "epoch": 6.235356657675572,
      "grad_norm": 1.8732280731201172,
      "learning_rate": 6.309017350669677e-07,
      "loss": 1.9883,
      "num_input_tokens_seen": 322467584,
      "step": 316700
    },
    {
      "epoch": 6.237325510425075,
      "grad_norm": 1.8613649606704712,
      "learning_rate": 6.307027347857338e-07,
      "loss": 1.9783,
      "num_input_tokens_seen": 322569432,
      "step": 316800
    },
    {
      "epoch": 6.2392943631745785,
      "grad_norm": 1.6686112880706787,
      "learning_rate": 6.305037122796027e-07,
      "loss": 1.9398,
      "num_input_tokens_seen": 322671832,
      "step": 316900
    },
    {
      "epoch": 6.241263215924081,
      "grad_norm": 2.0981709957122803,
      "learning_rate": 6.303046675824156e-07,
      "loss": 2.0154,
      "num_input_tokens_seen": 322774232,
      "step": 317000
    },
    {
      "epoch": 6.243232068673584,
      "grad_norm": 1.8230525255203247,
      "learning_rate": 6.301056007280189e-07,
      "loss": 1.956,
      "num_input_tokens_seen": 322876632,
      "step": 317100
    },
    {
      "epoch": 6.245200921423087,
      "grad_norm": 2.0213985443115234,
      "learning_rate": 6.299065117502623e-07,
      "loss": 1.9618,
      "num_input_tokens_seen": 322979032,
      "step": 317200
    },
    {
      "epoch": 6.24716977417259,
      "grad_norm": 1.8146755695343018,
      "learning_rate": 6.29707400682999e-07,
      "loss": 2.0166,
      "num_input_tokens_seen": 323079712,
      "step": 317300
    },
    {
      "epoch": 6.249138626922092,
      "grad_norm": 1.784530520439148,
      "learning_rate": 6.295082675600862e-07,
      "loss": 1.9586,
      "num_input_tokens_seen": 323182112,
      "step": 317400
    },
    {
      "epoch": 6.251107479671595,
      "grad_norm": 1.963919997215271,
      "learning_rate": 6.293091124153847e-07,
      "loss": 1.9939,
      "num_input_tokens_seen": 323283144,
      "step": 317500
    },
    {
      "epoch": 6.253076332421098,
      "grad_norm": 1.6982088088989258,
      "learning_rate": 6.291099352827596e-07,
      "loss": 2.0027,
      "num_input_tokens_seen": 323385544,
      "step": 317600
    },
    {
      "epoch": 6.2550451851706015,
      "grad_norm": 1.648306965827942,
      "learning_rate": 6.289107361960786e-07,
      "loss": 1.9662,
      "num_input_tokens_seen": 323487352,
      "step": 317700
    },
    {
      "epoch": 6.257014037920104,
      "grad_norm": 1.9238020181655884,
      "learning_rate": 6.287115151892144e-07,
      "loss": 2.022,
      "num_input_tokens_seen": 323588072,
      "step": 317800
    },
    {
      "epoch": 6.258982890669607,
      "grad_norm": 2.021799325942993,
      "learning_rate": 6.285122722960426e-07,
      "loss": 1.9903,
      "num_input_tokens_seen": 323689392,
      "step": 317900
    },
    {
      "epoch": 6.26095174341911,
      "grad_norm": 1.8029745817184448,
      "learning_rate": 6.283130075504428e-07,
      "loss": 2.026,
      "num_input_tokens_seen": 323791560,
      "step": 318000
    },
    {
      "epoch": 6.2629205961686125,
      "grad_norm": 1.8561002016067505,
      "learning_rate": 6.281137209862984e-07,
      "loss": 1.984,
      "num_input_tokens_seen": 323893424,
      "step": 318100
    },
    {
      "epoch": 6.264889448918115,
      "grad_norm": 1.7570619583129883,
      "learning_rate": 6.279144126374962e-07,
      "loss": 1.9701,
      "num_input_tokens_seen": 323995824,
      "step": 318200
    },
    {
      "epoch": 6.266858301667618,
      "grad_norm": 2.216931104660034,
      "learning_rate": 6.27715082537927e-07,
      "loss": 1.9989,
      "num_input_tokens_seen": 324096224,
      "step": 318300
    },
    {
      "epoch": 6.268827154417121,
      "grad_norm": 1.9495714902877808,
      "learning_rate": 6.275157307214853e-07,
      "loss": 1.9546,
      "num_input_tokens_seen": 324196616,
      "step": 318400
    },
    {
      "epoch": 6.2707960071666236,
      "grad_norm": 1.943924069404602,
      "learning_rate": 6.273163572220691e-07,
      "loss": 1.9438,
      "num_input_tokens_seen": 324299016,
      "step": 318500
    },
    {
      "epoch": 6.272764859916127,
      "grad_norm": 2.0724143981933594,
      "learning_rate": 6.271169620735801e-07,
      "loss": 1.9761,
      "num_input_tokens_seen": 324401416,
      "step": 318600
    },
    {
      "epoch": 6.27473371266563,
      "grad_norm": 1.738439917564392,
      "learning_rate": 6.269175453099241e-07,
      "loss": 2.0024,
      "num_input_tokens_seen": 324503136,
      "step": 318700
    },
    {
      "epoch": 6.276702565415133,
      "grad_norm": 1.8139582872390747,
      "learning_rate": 6.267181069650096e-07,
      "loss": 1.9696,
      "num_input_tokens_seen": 324605536,
      "step": 318800
    },
    {
      "epoch": 6.2786714181646355,
      "grad_norm": 2.2022266387939453,
      "learning_rate": 6.2651864707275e-07,
      "loss": 1.9885,
      "num_input_tokens_seen": 324707936,
      "step": 318900
    },
    {
      "epoch": 6.280640270914138,
      "grad_norm": 1.9122960567474365,
      "learning_rate": 6.263191656670615e-07,
      "loss": 1.9958,
      "num_input_tokens_seen": 324810336,
      "step": 319000
    },
    {
      "epoch": 6.282609123663641,
      "grad_norm": 1.9304590225219727,
      "learning_rate": 6.261196627818646e-07,
      "loss": 1.9889,
      "num_input_tokens_seen": 324912024,
      "step": 319100
    },
    {
      "epoch": 6.284577976413144,
      "grad_norm": 1.790189266204834,
      "learning_rate": 6.259201384510824e-07,
      "loss": 1.9664,
      "num_input_tokens_seen": 325013592,
      "step": 319200
    },
    {
      "epoch": 6.2865468291626465,
      "grad_norm": 2.089468002319336,
      "learning_rate": 6.25720592708643e-07,
      "loss": 1.9934,
      "num_input_tokens_seen": 325115456,
      "step": 319300
    },
    {
      "epoch": 6.28851568191215,
      "grad_norm": 2.296259880065918,
      "learning_rate": 6.25521025588477e-07,
      "loss": 2.1073,
      "num_input_tokens_seen": 325216088,
      "step": 319400
    },
    {
      "epoch": 6.290484534661653,
      "grad_norm": 2.262141466140747,
      "learning_rate": 6.253214371245195e-07,
      "loss": 2.0009,
      "num_input_tokens_seen": 325318488,
      "step": 319500
    },
    {
      "epoch": 6.292453387411156,
      "grad_norm": 2.2707834243774414,
      "learning_rate": 6.251218273507084e-07,
      "loss": 2.0127,
      "num_input_tokens_seen": 325420888,
      "step": 319600
    },
    {
      "epoch": 6.294422240160658,
      "grad_norm": 1.8361297845840454,
      "learning_rate": 6.249221963009859e-07,
      "loss": 2.0115,
      "num_input_tokens_seen": 325522520,
      "step": 319700
    },
    {
      "epoch": 6.296391092910161,
      "grad_norm": 1.8942452669143677,
      "learning_rate": 6.247225440092976e-07,
      "loss": 1.9458,
      "num_input_tokens_seen": 325624920,
      "step": 319800
    },
    {
      "epoch": 6.298359945659664,
      "grad_norm": 1.986633539199829,
      "learning_rate": 6.245228705095928e-07,
      "loss": 1.9764,
      "num_input_tokens_seen": 325725856,
      "step": 319900
    },
    {
      "epoch": 6.300328798409167,
      "grad_norm": 2.036482810974121,
      "learning_rate": 6.243231758358239e-07,
      "loss": 1.9396,
      "num_input_tokens_seen": 325828256,
      "step": 320000
    },
    {
      "epoch": 6.3022976511586695,
      "grad_norm": 1.6657260656356812,
      "learning_rate": 6.241234600219477e-07,
      "loss": 1.983,
      "num_input_tokens_seen": 325930656,
      "step": 320100
    },
    {
      "epoch": 6.304266503908173,
      "grad_norm": 1.9341163635253906,
      "learning_rate": 6.23923723101924e-07,
      "loss": 1.9626,
      "num_input_tokens_seen": 326033056,
      "step": 320200
    },
    {
      "epoch": 6.306235356657676,
      "grad_norm": 1.9460424184799194,
      "learning_rate": 6.237239651097165e-07,
      "loss": 1.9853,
      "num_input_tokens_seen": 326135456,
      "step": 320300
    },
    {
      "epoch": 6.308204209407179,
      "grad_norm": 1.927511215209961,
      "learning_rate": 6.235241860792921e-07,
      "loss": 1.9929,
      "num_input_tokens_seen": 326237272,
      "step": 320400
    },
    {
      "epoch": 6.310173062156681,
      "grad_norm": 2.020237922668457,
      "learning_rate": 6.233243860446218e-07,
      "loss": 1.951,
      "num_input_tokens_seen": 326339672,
      "step": 320500
    },
    {
      "epoch": 6.312141914906184,
      "grad_norm": 1.974599838256836,
      "learning_rate": 6.231245650396798e-07,
      "loss": 2.0165,
      "num_input_tokens_seen": 326441152,
      "step": 320600
    },
    {
      "epoch": 6.314110767655687,
      "grad_norm": 2.6221978664398193,
      "learning_rate": 6.22924723098444e-07,
      "loss": 1.9924,
      "num_input_tokens_seen": 326543376,
      "step": 320700
    },
    {
      "epoch": 6.31607962040519,
      "grad_norm": 1.7145941257476807,
      "learning_rate": 6.227248602548958e-07,
      "loss": 2.0027,
      "num_input_tokens_seen": 326645776,
      "step": 320800
    },
    {
      "epoch": 6.318048473154692,
      "grad_norm": 1.776474118232727,
      "learning_rate": 6.225249765430203e-07,
      "loss": 1.981,
      "num_input_tokens_seen": 326748176,
      "step": 320900
    },
    {
      "epoch": 6.320017325904196,
      "grad_norm": 1.6626194715499878,
      "learning_rate": 6.22325071996806e-07,
      "loss": 1.9753,
      "num_input_tokens_seen": 326850576,
      "step": 321000
    },
    {
      "epoch": 6.321986178653699,
      "grad_norm": 2.00114107131958,
      "learning_rate": 6.221251466502447e-07,
      "loss": 1.9666,
      "num_input_tokens_seen": 326952464,
      "step": 321100
    },
    {
      "epoch": 6.323955031403202,
      "grad_norm": 1.800858974456787,
      "learning_rate": 6.219252005373326e-07,
      "loss": 1.9883,
      "num_input_tokens_seen": 327054088,
      "step": 321200
    },
    {
      "epoch": 6.325923884152704,
      "grad_norm": 1.948839783668518,
      "learning_rate": 6.217252336920682e-07,
      "loss": 1.9541,
      "num_input_tokens_seen": 327155672,
      "step": 321300
    },
    {
      "epoch": 6.327892736902207,
      "grad_norm": 1.7164121866226196,
      "learning_rate": 6.215252461484546e-07,
      "loss": 2.0029,
      "num_input_tokens_seen": 327255728,
      "step": 321400
    },
    {
      "epoch": 6.32986158965171,
      "grad_norm": 2.1410481929779053,
      "learning_rate": 6.21325237940498e-07,
      "loss": 2.0022,
      "num_input_tokens_seen": 327357848,
      "step": 321500
    },
    {
      "epoch": 6.331830442401213,
      "grad_norm": 2.3601491451263428,
      "learning_rate": 6.211252091022078e-07,
      "loss": 1.9813,
      "num_input_tokens_seen": 327460040,
      "step": 321600
    },
    {
      "epoch": 6.333799295150715,
      "grad_norm": 2.4539542198181152,
      "learning_rate": 6.209251596675975e-07,
      "loss": 1.9681,
      "num_input_tokens_seen": 327561032,
      "step": 321700
    },
    {
      "epoch": 6.335768147900218,
      "grad_norm": 2.0476949214935303,
      "learning_rate": 6.207250896706838e-07,
      "loss": 1.9965,
      "num_input_tokens_seen": 327662536,
      "step": 321800
    },
    {
      "epoch": 6.337737000649722,
      "grad_norm": 3.0949981212615967,
      "learning_rate": 6.205249991454868e-07,
      "loss": 2.0221,
      "num_input_tokens_seen": 327764936,
      "step": 321900
    },
    {
      "epoch": 6.3397058533992245,
      "grad_norm": 1.8338130712509155,
      "learning_rate": 6.203248881260301e-07,
      "loss": 1.9776,
      "num_input_tokens_seen": 327865688,
      "step": 322000
    },
    {
      "epoch": 6.341674706148727,
      "grad_norm": 1.751347303390503,
      "learning_rate": 6.201247566463412e-07,
      "loss": 1.9853,
      "num_input_tokens_seen": 327966528,
      "step": 322100
    },
    {
      "epoch": 6.34364355889823,
      "grad_norm": 2.0570356845855713,
      "learning_rate": 6.199246047404506e-07,
      "loss": 1.9795,
      "num_input_tokens_seen": 328068064,
      "step": 322200
    },
    {
      "epoch": 6.345612411647733,
      "grad_norm": 1.7843410968780518,
      "learning_rate": 6.197244324423925e-07,
      "loss": 1.9746,
      "num_input_tokens_seen": 328170464,
      "step": 322300
    },
    {
      "epoch": 6.347581264397236,
      "grad_norm": 1.8098764419555664,
      "learning_rate": 6.195242397862043e-07,
      "loss": 1.9834,
      "num_input_tokens_seen": 328272120,
      "step": 322400
    },
    {
      "epoch": 6.349550117146738,
      "grad_norm": 2.0858922004699707,
      "learning_rate": 6.193240268059275e-07,
      "loss": 1.9511,
      "num_input_tokens_seen": 328373880,
      "step": 322500
    },
    {
      "epoch": 6.351518969896241,
      "grad_norm": 1.7641922235488892,
      "learning_rate": 6.191237935356062e-07,
      "loss": 1.9747,
      "num_input_tokens_seen": 328475704,
      "step": 322600
    },
    {
      "epoch": 6.353487822645745,
      "grad_norm": 1.9336637258529663,
      "learning_rate": 6.189235400092885e-07,
      "loss": 1.9615,
      "num_input_tokens_seen": 328578104,
      "step": 322700
    },
    {
      "epoch": 6.3554566753952475,
      "grad_norm": 1.7192039489746094,
      "learning_rate": 6.187232662610259e-07,
      "loss": 1.9806,
      "num_input_tokens_seen": 328679928,
      "step": 322800
    },
    {
      "epoch": 6.35742552814475,
      "grad_norm": 1.8622206449508667,
      "learning_rate": 6.185229723248731e-07,
      "loss": 1.9712,
      "num_input_tokens_seen": 328781864,
      "step": 322900
    },
    {
      "epoch": 6.359394380894253,
      "grad_norm": 6.362214088439941,
      "learning_rate": 6.183226582348888e-07,
      "loss": 1.9722,
      "num_input_tokens_seen": 328883384,
      "step": 323000
    },
    {
      "epoch": 6.361363233643756,
      "grad_norm": 2.1649420261383057,
      "learning_rate": 6.181223240251343e-07,
      "loss": 1.9865,
      "num_input_tokens_seen": 328985784,
      "step": 323100
    },
    {
      "epoch": 6.3633320863932585,
      "grad_norm": 3.7690231800079346,
      "learning_rate": 6.17921969729675e-07,
      "loss": 1.9655,
      "num_input_tokens_seen": 329087640,
      "step": 323200
    },
    {
      "epoch": 6.365300939142761,
      "grad_norm": 1.8056912422180176,
      "learning_rate": 6.177215953825792e-07,
      "loss": 1.9763,
      "num_input_tokens_seen": 329190040,
      "step": 323300
    },
    {
      "epoch": 6.367269791892264,
      "grad_norm": 1.8254534006118774,
      "learning_rate": 6.175212010179191e-07,
      "loss": 1.9809,
      "num_input_tokens_seen": 329291960,
      "step": 323400
    },
    {
      "epoch": 6.369238644641767,
      "grad_norm": 1.7076184749603271,
      "learning_rate": 6.173207866697699e-07,
      "loss": 2.0051,
      "num_input_tokens_seen": 329392688,
      "step": 323500
    },
    {
      "epoch": 6.3712074973912705,
      "grad_norm": 1.8180934190750122,
      "learning_rate": 6.171203523722106e-07,
      "loss": 2.0257,
      "num_input_tokens_seen": 329494464,
      "step": 323600
    },
    {
      "epoch": 6.373176350140773,
      "grad_norm": 1.7735439538955688,
      "learning_rate": 6.169198981593231e-07,
      "loss": 1.958,
      "num_input_tokens_seen": 329596000,
      "step": 323700
    },
    {
      "epoch": 6.375145202890276,
      "grad_norm": 1.832324743270874,
      "learning_rate": 6.167194240651931e-07,
      "loss": 2.0022,
      "num_input_tokens_seen": 329698400,
      "step": 323800
    },
    {
      "epoch": 6.377114055639779,
      "grad_norm": 1.8855369091033936,
      "learning_rate": 6.165189301239094e-07,
      "loss": 1.9617,
      "num_input_tokens_seen": 329800088,
      "step": 323900
    },
    {
      "epoch": 6.3790829083892815,
      "grad_norm": 2.000969409942627,
      "learning_rate": 6.163184163695646e-07,
      "loss": 1.9507,
      "num_input_tokens_seen": 329901672,
      "step": 324000
    },
    {
      "epoch": 6.381051761138784,
      "grad_norm": 1.8009623289108276,
      "learning_rate": 6.16117882836254e-07,
      "loss": 2.0282,
      "num_input_tokens_seen": 330003256,
      "step": 324100
    },
    {
      "epoch": 6.383020613888287,
      "grad_norm": 2.3676700592041016,
      "learning_rate": 6.159173295580769e-07,
      "loss": 1.9634,
      "num_input_tokens_seen": 330105136,
      "step": 324200
    },
    {
      "epoch": 6.38498946663779,
      "grad_norm": 2.029550552368164,
      "learning_rate": 6.157167565691356e-07,
      "loss": 1.9662,
      "num_input_tokens_seen": 330207536,
      "step": 324300
    },
    {
      "epoch": 6.386958319387293,
      "grad_norm": 2.4419150352478027,
      "learning_rate": 6.155161639035359e-07,
      "loss": 2.033,
      "num_input_tokens_seen": 330308104,
      "step": 324400
    },
    {
      "epoch": 6.388927172136796,
      "grad_norm": 1.8690277338027954,
      "learning_rate": 6.153155515953867e-07,
      "loss": 1.9661,
      "num_input_tokens_seen": 330410504,
      "step": 324500
    },
    {
      "epoch": 6.390896024886299,
      "grad_norm": 1.9267312288284302,
      "learning_rate": 6.151149196788007e-07,
      "loss": 1.9667,
      "num_input_tokens_seen": 330511512,
      "step": 324600
    },
    {
      "epoch": 6.392864877635802,
      "grad_norm": 2.4113211631774902,
      "learning_rate": 6.149142681878934e-07,
      "loss": 2.0224,
      "num_input_tokens_seen": 330613056,
      "step": 324700
    },
    {
      "epoch": 6.3948337303853044,
      "grad_norm": 1.7615582942962646,
      "learning_rate": 6.147135971567841e-07,
      "loss": 1.9618,
      "num_input_tokens_seen": 330715456,
      "step": 324800
    },
    {
      "epoch": 6.396802583134807,
      "grad_norm": 1.7147644758224487,
      "learning_rate": 6.145129066195951e-07,
      "loss": 1.9512,
      "num_input_tokens_seen": 330817856,
      "step": 324900
    },
    {
      "epoch": 6.39877143588431,
      "grad_norm": 2.0383174419403076,
      "learning_rate": 6.143121966104522e-07,
      "loss": 1.9715,
      "num_input_tokens_seen": 330919608,
      "step": 325000
    },
    {
      "epoch": 6.400740288633813,
      "grad_norm": 2.049348831176758,
      "learning_rate": 6.141114671634844e-07,
      "loss": 1.9778,
      "num_input_tokens_seen": 331022008,
      "step": 325100
    },
    {
      "epoch": 6.4027091413833155,
      "grad_norm": 1.7986857891082764,
      "learning_rate": 6.139107183128241e-07,
      "loss": 1.9972,
      "num_input_tokens_seen": 331122512,
      "step": 325200
    },
    {
      "epoch": 6.404677994132819,
      "grad_norm": 2.0284037590026855,
      "learning_rate": 6.13709950092607e-07,
      "loss": 1.9836,
      "num_input_tokens_seen": 331224008,
      "step": 325300
    },
    {
      "epoch": 6.406646846882322,
      "grad_norm": 2.012815475463867,
      "learning_rate": 6.135091625369718e-07,
      "loss": 1.989,
      "num_input_tokens_seen": 331325904,
      "step": 325400
    },
    {
      "epoch": 6.408615699631825,
      "grad_norm": 1.916353464126587,
      "learning_rate": 6.133083556800609e-07,
      "loss": 1.9713,
      "num_input_tokens_seen": 331427776,
      "step": 325500
    },
    {
      "epoch": 6.410584552381327,
      "grad_norm": 2.093794345855713,
      "learning_rate": 6.1310752955602e-07,
      "loss": 2.0217,
      "num_input_tokens_seen": 331530176,
      "step": 325600
    },
    {
      "epoch": 6.41255340513083,
      "grad_norm": 1.9516007900238037,
      "learning_rate": 6.129066841989974e-07,
      "loss": 1.9553,
      "num_input_tokens_seen": 331631640,
      "step": 325700
    },
    {
      "epoch": 6.414522257880333,
      "grad_norm": 1.891526222229004,
      "learning_rate": 6.127058196431457e-07,
      "loss": 1.9457,
      "num_input_tokens_seen": 331734040,
      "step": 325800
    },
    {
      "epoch": 6.416491110629836,
      "grad_norm": 2.3124842643737793,
      "learning_rate": 6.125049359226199e-07,
      "loss": 1.9858,
      "num_input_tokens_seen": 331835920,
      "step": 325900
    },
    {
      "epoch": 6.418459963379339,
      "grad_norm": 1.8841344118118286,
      "learning_rate": 6.123040330715785e-07,
      "loss": 1.9908,
      "num_input_tokens_seen": 331938320,
      "step": 326000
    },
    {
      "epoch": 6.420428816128842,
      "grad_norm": 1.7686660289764404,
      "learning_rate": 6.121031111241837e-07,
      "loss": 1.9863,
      "num_input_tokens_seen": 332040720,
      "step": 326100
    },
    {
      "epoch": 6.422397668878345,
      "grad_norm": 1.8571674823760986,
      "learning_rate": 6.119021701146003e-07,
      "loss": 1.9864,
      "num_input_tokens_seen": 332142592,
      "step": 326200
    },
    {
      "epoch": 6.424366521627848,
      "grad_norm": 1.7800183296203613,
      "learning_rate": 6.117012100769967e-07,
      "loss": 1.9638,
      "num_input_tokens_seen": 332244992,
      "step": 326300
    },
    {
      "epoch": 6.42633537437735,
      "grad_norm": 2.127934455871582,
      "learning_rate": 6.115002310455445e-07,
      "loss": 1.9432,
      "num_input_tokens_seen": 332346752,
      "step": 326400
    },
    {
      "epoch": 6.428304227126853,
      "grad_norm": 1.9178853034973145,
      "learning_rate": 6.112992330544184e-07,
      "loss": 1.9676,
      "num_input_tokens_seen": 332448616,
      "step": 326500
    },
    {
      "epoch": 6.430273079876356,
      "grad_norm": 2.1226372718811035,
      "learning_rate": 6.110982161377964e-07,
      "loss": 2.0212,
      "num_input_tokens_seen": 332551016,
      "step": 326600
    },
    {
      "epoch": 6.432241932625859,
      "grad_norm": 1.699806571006775,
      "learning_rate": 6.1089718032986e-07,
      "loss": 1.996,
      "num_input_tokens_seen": 332653416,
      "step": 326700
    },
    {
      "epoch": 6.434210785375361,
      "grad_norm": 1.8614561557769775,
      "learning_rate": 6.106961256647932e-07,
      "loss": 1.9219,
      "num_input_tokens_seen": 332755064,
      "step": 326800
    },
    {
      "epoch": 6.436179638124865,
      "grad_norm": 1.6365993022918701,
      "learning_rate": 6.10495052176784e-07,
      "loss": 2.0384,
      "num_input_tokens_seen": 332855528,
      "step": 326900
    },
    {
      "epoch": 6.438148490874368,
      "grad_norm": 1.9996424913406372,
      "learning_rate": 6.102939599000231e-07,
      "loss": 1.9601,
      "num_input_tokens_seen": 332957248,
      "step": 327000
    },
    {
      "epoch": 6.440117343623871,
      "grad_norm": 1.8351149559020996,
      "learning_rate": 6.100928488687046e-07,
      "loss": 1.9764,
      "num_input_tokens_seen": 333059648,
      "step": 327100
    },
    {
      "epoch": 6.442086196373373,
      "grad_norm": 2.0287539958953857,
      "learning_rate": 6.098917191170257e-07,
      "loss": 2.0064,
      "num_input_tokens_seen": 333161352,
      "step": 327200
    },
    {
      "epoch": 6.444055049122876,
      "grad_norm": 2.030189037322998,
      "learning_rate": 6.096905706791869e-07,
      "loss": 1.9555,
      "num_input_tokens_seen": 333263152,
      "step": 327300
    },
    {
      "epoch": 6.446023901872379,
      "grad_norm": 2.1728546619415283,
      "learning_rate": 6.094894035893916e-07,
      "loss": 1.946,
      "num_input_tokens_seen": 333365552,
      "step": 327400
    },
    {
      "epoch": 6.447992754621882,
      "grad_norm": 1.9530166387557983,
      "learning_rate": 6.092882178818468e-07,
      "loss": 1.9646,
      "num_input_tokens_seen": 333467288,
      "step": 327500
    },
    {
      "epoch": 6.449961607371384,
      "grad_norm": 1.7757498025894165,
      "learning_rate": 6.090870135907623e-07,
      "loss": 2.0014,
      "num_input_tokens_seen": 333569056,
      "step": 327600
    },
    {
      "epoch": 6.451930460120888,
      "grad_norm": 1.9455411434173584,
      "learning_rate": 6.088857907503513e-07,
      "loss": 1.9777,
      "num_input_tokens_seen": 333670616,
      "step": 327700
    },
    {
      "epoch": 6.453899312870391,
      "grad_norm": 1.9355013370513916,
      "learning_rate": 6.086845493948299e-07,
      "loss": 1.9434,
      "num_input_tokens_seen": 333773016,
      "step": 327800
    },
    {
      "epoch": 6.4558681656198935,
      "grad_norm": 2.1994781494140625,
      "learning_rate": 6.084832895584174e-07,
      "loss": 1.9661,
      "num_input_tokens_seen": 333875416,
      "step": 327900
    },
    {
      "epoch": 6.457837018369396,
      "grad_norm": 2.1956112384796143,
      "learning_rate": 6.082820112753367e-07,
      "loss": 2.0031,
      "num_input_tokens_seen": 333976984,
      "step": 328000
    },
    {
      "epoch": 6.459805871118899,
      "grad_norm": 1.8465567827224731,
      "learning_rate": 6.080807145798133e-07,
      "loss": 2.0056,
      "num_input_tokens_seen": 334078616,
      "step": 328100
    },
    {
      "epoch": 6.461774723868402,
      "grad_norm": 1.9540822505950928,
      "learning_rate": 6.078793995060758e-07,
      "loss": 2.0113,
      "num_input_tokens_seen": 334178672,
      "step": 328200
    },
    {
      "epoch": 6.4637435766179046,
      "grad_norm": 1.91953706741333,
      "learning_rate": 6.076780660883565e-07,
      "loss": 1.9309,
      "num_input_tokens_seen": 334280160,
      "step": 328300
    },
    {
      "epoch": 6.465712429367407,
      "grad_norm": 2.0835020542144775,
      "learning_rate": 6.074767143608901e-07,
      "loss": 1.9724,
      "num_input_tokens_seen": 334382128,
      "step": 328400
    },
    {
      "epoch": 6.46768128211691,
      "grad_norm": 1.8535174131393433,
      "learning_rate": 6.072753443579151e-07,
      "loss": 2.0113,
      "num_input_tokens_seen": 334483960,
      "step": 328500
    },
    {
      "epoch": 6.469650134866414,
      "grad_norm": 1.879845380783081,
      "learning_rate": 6.070739561136724e-07,
      "loss": 2.0152,
      "num_input_tokens_seen": 334585072,
      "step": 328600
    },
    {
      "epoch": 6.4716189876159165,
      "grad_norm": 2.101710081100464,
      "learning_rate": 6.068725496624068e-07,
      "loss": 1.9484,
      "num_input_tokens_seen": 334687472,
      "step": 328700
    },
    {
      "epoch": 6.473587840365419,
      "grad_norm": 2.185837507247925,
      "learning_rate": 6.066711250383653e-07,
      "loss": 1.9737,
      "num_input_tokens_seen": 334789160,
      "step": 328800
    },
    {
      "epoch": 6.475556693114922,
      "grad_norm": 1.8057156801223755,
      "learning_rate": 6.064696822757988e-07,
      "loss": 1.9503,
      "num_input_tokens_seen": 334891560,
      "step": 328900
    },
    {
      "epoch": 6.477525545864425,
      "grad_norm": 1.943167805671692,
      "learning_rate": 6.062682214089608e-07,
      "loss": 1.97,
      "num_input_tokens_seen": 334993960,
      "step": 329000
    },
    {
      "epoch": 6.4794943986139275,
      "grad_norm": 1.9640806913375854,
      "learning_rate": 6.060667424721081e-07,
      "loss": 2.0122,
      "num_input_tokens_seen": 335096360,
      "step": 329100
    },
    {
      "epoch": 6.48146325136343,
      "grad_norm": 2.066638708114624,
      "learning_rate": 6.058652454995004e-07,
      "loss": 1.9992,
      "num_input_tokens_seen": 335198072,
      "step": 329200
    },
    {
      "epoch": 6.483432104112933,
      "grad_norm": 2.1574697494506836,
      "learning_rate": 6.056637305254007e-07,
      "loss": 1.9994,
      "num_input_tokens_seen": 335299576,
      "step": 329300
    },
    {
      "epoch": 6.485400956862437,
      "grad_norm": 2.0422778129577637,
      "learning_rate": 6.054621975840748e-07,
      "loss": 1.9752,
      "num_input_tokens_seen": 335401360,
      "step": 329400
    },
    {
      "epoch": 6.487369809611939,
      "grad_norm": 2.1192007064819336,
      "learning_rate": 6.052606467097917e-07,
      "loss": 2.0178,
      "num_input_tokens_seen": 335503760,
      "step": 329500
    },
    {
      "epoch": 6.489338662361442,
      "grad_norm": 1.8772376775741577,
      "learning_rate": 6.050590779368234e-07,
      "loss": 1.967,
      "num_input_tokens_seen": 335605520,
      "step": 329600
    },
    {
      "epoch": 6.491307515110945,
      "grad_norm": 1.9499151706695557,
      "learning_rate": 6.048574912994451e-07,
      "loss": 1.9705,
      "num_input_tokens_seen": 335707920,
      "step": 329700
    },
    {
      "epoch": 6.493276367860448,
      "grad_norm": 2.202011823654175,
      "learning_rate": 6.046558868319347e-07,
      "loss": 1.9384,
      "num_input_tokens_seen": 335810320,
      "step": 329800
    },
    {
      "epoch": 6.4952452206099505,
      "grad_norm": 1.7260181903839111,
      "learning_rate": 6.044542645685735e-07,
      "loss": 1.9509,
      "num_input_tokens_seen": 335912720,
      "step": 329900
    },
    {
      "epoch": 6.497214073359453,
      "grad_norm": 1.5298411846160889,
      "learning_rate": 6.042526245436455e-07,
      "loss": 1.9802,
      "num_input_tokens_seen": 336014264,
      "step": 330000
    },
    {
      "epoch": 6.499182926108956,
      "grad_norm": 2.025395631790161,
      "learning_rate": 6.040509667914382e-07,
      "loss": 2.009,
      "num_input_tokens_seen": 336116576,
      "step": 330100
    },
    {
      "epoch": 6.501151778858459,
      "grad_norm": 2.1979007720947266,
      "learning_rate": 6.038492913462417e-07,
      "loss": 1.9367,
      "num_input_tokens_seen": 336218976,
      "step": 330200
    },
    {
      "epoch": 6.503120631607962,
      "grad_norm": 2.0642919540405273,
      "learning_rate": 6.036475982423488e-07,
      "loss": 1.9565,
      "num_input_tokens_seen": 336319424,
      "step": 330300
    },
    {
      "epoch": 6.505089484357465,
      "grad_norm": 1.9608566761016846,
      "learning_rate": 6.034458875140563e-07,
      "loss": 1.9525,
      "num_input_tokens_seen": 336420968,
      "step": 330400
    },
    {
      "epoch": 6.507058337106968,
      "grad_norm": 1.8452330827713013,
      "learning_rate": 6.032441591956628e-07,
      "loss": 1.9453,
      "num_input_tokens_seen": 336523368,
      "step": 330500
    },
    {
      "epoch": 6.509027189856471,
      "grad_norm": 2.0573019981384277,
      "learning_rate": 6.030424133214711e-07,
      "loss": 1.9638,
      "num_input_tokens_seen": 336624960,
      "step": 330600
    },
    {
      "epoch": 6.510996042605973,
      "grad_norm": 1.7164239883422852,
      "learning_rate": 6.02840649925786e-07,
      "loss": 1.9829,
      "num_input_tokens_seen": 336727360,
      "step": 330700
    },
    {
      "epoch": 6.512964895355476,
      "grad_norm": 1.9546500444412231,
      "learning_rate": 6.026388690429158e-07,
      "loss": 2.0137,
      "num_input_tokens_seen": 336828952,
      "step": 330800
    },
    {
      "epoch": 6.514933748104979,
      "grad_norm": 1.9857194423675537,
      "learning_rate": 6.024370707071717e-07,
      "loss": 1.948,
      "num_input_tokens_seen": 336930544,
      "step": 330900
    },
    {
      "epoch": 6.516902600854483,
      "grad_norm": 2.175293207168579,
      "learning_rate": 6.022352549528675e-07,
      "loss": 1.9875,
      "num_input_tokens_seen": 337032312,
      "step": 331000
    },
    {
      "epoch": 6.518871453603985,
      "grad_norm": 2.0272483825683594,
      "learning_rate": 6.020334218143206e-07,
      "loss": 2.0037,
      "num_input_tokens_seen": 337133152,
      "step": 331100
    },
    {
      "epoch": 6.520840306353488,
      "grad_norm": 1.8052414655685425,
      "learning_rate": 6.01831571325851e-07,
      "loss": 1.9783,
      "num_input_tokens_seen": 337234616,
      "step": 331200
    },
    {
      "epoch": 6.522809159102991,
      "grad_norm": 2.2724173069000244,
      "learning_rate": 6.016297035217816e-07,
      "loss": 1.9855,
      "num_input_tokens_seen": 337337016,
      "step": 331300
    },
    {
      "epoch": 6.524778011852494,
      "grad_norm": 1.8018869161605835,
      "learning_rate": 6.014278184364382e-07,
      "loss": 2.0007,
      "num_input_tokens_seen": 337439128,
      "step": 331400
    },
    {
      "epoch": 6.526746864601996,
      "grad_norm": 2.0168981552124023,
      "learning_rate": 6.012259161041497e-07,
      "loss": 1.9533,
      "num_input_tokens_seen": 337540664,
      "step": 331500
    },
    {
      "epoch": 6.528715717351499,
      "grad_norm": 1.9451590776443481,
      "learning_rate": 6.010239965592482e-07,
      "loss": 2.0135,
      "num_input_tokens_seen": 337641112,
      "step": 331600
    },
    {
      "epoch": 6.530684570101002,
      "grad_norm": 2.2448904514312744,
      "learning_rate": 6.008220598360679e-07,
      "loss": 1.9419,
      "num_input_tokens_seen": 337742960,
      "step": 331700
    },
    {
      "epoch": 6.532653422850505,
      "grad_norm": 1.8448795080184937,
      "learning_rate": 6.006201059689468e-07,
      "loss": 1.982,
      "num_input_tokens_seen": 337845360,
      "step": 331800
    },
    {
      "epoch": 6.534622275600007,
      "grad_norm": 1.9624658823013306,
      "learning_rate": 6.004181349922253e-07,
      "loss": 2.0188,
      "num_input_tokens_seen": 337947008,
      "step": 331900
    },
    {
      "epoch": 6.536591128349511,
      "grad_norm": 1.9757168292999268,
      "learning_rate": 6.002161469402469e-07,
      "loss": 1.9757,
      "num_input_tokens_seen": 338049408,
      "step": 332000
    },
    {
      "epoch": 6.538559981099014,
      "grad_norm": 1.7830297946929932,
      "learning_rate": 6.000141418473581e-07,
      "loss": 1.9633,
      "num_input_tokens_seen": 338150376,
      "step": 332100
    },
    {
      "epoch": 6.540528833848517,
      "grad_norm": 5.492066860198975,
      "learning_rate": 5.99812119747908e-07,
      "loss": 2.0123,
      "num_input_tokens_seen": 338251064,
      "step": 332200
    },
    {
      "epoch": 6.542497686598019,
      "grad_norm": 2.2998900413513184,
      "learning_rate": 5.996100806762487e-07,
      "loss": 2.0188,
      "num_input_tokens_seen": 338353016,
      "step": 332300
    },
    {
      "epoch": 6.544466539347522,
      "grad_norm": 1.9289929866790771,
      "learning_rate": 5.994080246667354e-07,
      "loss": 1.9853,
      "num_input_tokens_seen": 338455416,
      "step": 332400
    },
    {
      "epoch": 6.546435392097025,
      "grad_norm": 1.9939790964126587,
      "learning_rate": 5.992059517537259e-07,
      "loss": 1.969,
      "num_input_tokens_seen": 338557264,
      "step": 332500
    },
    {
      "epoch": 6.548404244846528,
      "grad_norm": 3.035196542739868,
      "learning_rate": 5.990038619715811e-07,
      "loss": 2.0201,
      "num_input_tokens_seen": 338658312,
      "step": 332600
    },
    {
      "epoch": 6.550373097596031,
      "grad_norm": 2.042783498764038,
      "learning_rate": 5.988017553546646e-07,
      "loss": 2.0123,
      "num_input_tokens_seen": 338759352,
      "step": 332700
    },
    {
      "epoch": 6.552341950345534,
      "grad_norm": 2.005141258239746,
      "learning_rate": 5.98599631937343e-07,
      "loss": 2.0006,
      "num_input_tokens_seen": 338861536,
      "step": 332800
    },
    {
      "epoch": 6.554310803095037,
      "grad_norm": 1.9736496210098267,
      "learning_rate": 5.983974917539855e-07,
      "loss": 1.9879,
      "num_input_tokens_seen": 338963936,
      "step": 332900
    },
    {
      "epoch": 6.5562796558445395,
      "grad_norm": 1.995004415512085,
      "learning_rate": 5.981953348389645e-07,
      "loss": 2.0082,
      "num_input_tokens_seen": 339066336,
      "step": 333000
    },
    {
      "epoch": 6.558248508594042,
      "grad_norm": 2.825413227081299,
      "learning_rate": 5.97993161226655e-07,
      "loss": 1.9956,
      "num_input_tokens_seen": 339168080,
      "step": 333100
    },
    {
      "epoch": 6.560217361343545,
      "grad_norm": 10.762933731079102,
      "learning_rate": 5.977909709514349e-07,
      "loss": 1.9451,
      "num_input_tokens_seen": 339270480,
      "step": 333200
    },
    {
      "epoch": 6.562186214093048,
      "grad_norm": 2.0744829177856445,
      "learning_rate": 5.975887640476852e-07,
      "loss": 2.0153,
      "num_input_tokens_seen": 339370784,
      "step": 333300
    },
    {
      "epoch": 6.564155066842551,
      "grad_norm": 1.7472552061080933,
      "learning_rate": 5.97386540549789e-07,
      "loss": 1.9787,
      "num_input_tokens_seen": 339472264,
      "step": 333400
    },
    {
      "epoch": 6.566123919592053,
      "grad_norm": 1.8035584688186646,
      "learning_rate": 5.971843004921332e-07,
      "loss": 1.9873,
      "num_input_tokens_seen": 339573752,
      "step": 333500
    },
    {
      "epoch": 6.568092772341556,
      "grad_norm": 1.9936245679855347,
      "learning_rate": 5.969820439091066e-07,
      "loss": 2.0252,
      "num_input_tokens_seen": 339676152,
      "step": 333600
    },
    {
      "epoch": 6.57006162509106,
      "grad_norm": 2.026432991027832,
      "learning_rate": 5.967797708351016e-07,
      "loss": 1.9662,
      "num_input_tokens_seen": 339778552,
      "step": 333700
    },
    {
      "epoch": 6.5720304778405625,
      "grad_norm": 2.013417959213257,
      "learning_rate": 5.965774813045128e-07,
      "loss": 1.9947,
      "num_input_tokens_seen": 339880824,
      "step": 333800
    },
    {
      "epoch": 6.573999330590065,
      "grad_norm": 1.8835017681121826,
      "learning_rate": 5.963751753517378e-07,
      "loss": 1.9986,
      "num_input_tokens_seen": 339983224,
      "step": 333900
    },
    {
      "epoch": 6.575968183339568,
      "grad_norm": 2.0019731521606445,
      "learning_rate": 5.961728530111771e-07,
      "loss": 1.9697,
      "num_input_tokens_seen": 340084880,
      "step": 334000
    },
    {
      "epoch": 6.577937036089071,
      "grad_norm": 1.8406524658203125,
      "learning_rate": 5.95970514317234e-07,
      "loss": 2.0124,
      "num_input_tokens_seen": 340187280,
      "step": 334100
    },
    {
      "epoch": 6.5799058888385735,
      "grad_norm": 2.345460891723633,
      "learning_rate": 5.957681593043144e-07,
      "loss": 1.9799,
      "num_input_tokens_seen": 340288896,
      "step": 334200
    },
    {
      "epoch": 6.581874741588076,
      "grad_norm": 1.9273667335510254,
      "learning_rate": 5.95565788006827e-07,
      "loss": 1.9683,
      "num_input_tokens_seen": 340391296,
      "step": 334300
    },
    {
      "epoch": 6.58384359433758,
      "grad_norm": 1.9310986995697021,
      "learning_rate": 5.953634004591833e-07,
      "loss": 1.9453,
      "num_input_tokens_seen": 340491568,
      "step": 334400
    },
    {
      "epoch": 6.585812447087083,
      "grad_norm": 1.782335638999939,
      "learning_rate": 5.951609966957978e-07,
      "loss": 1.9641,
      "num_input_tokens_seen": 340592192,
      "step": 334500
    },
    {
      "epoch": 6.5877812998365854,
      "grad_norm": 1.9793280363082886,
      "learning_rate": 5.949585767510874e-07,
      "loss": 1.9991,
      "num_input_tokens_seen": 340694592,
      "step": 334600
    },
    {
      "epoch": 6.589750152586088,
      "grad_norm": 1.798757791519165,
      "learning_rate": 5.94756140659472e-07,
      "loss": 1.9617,
      "num_input_tokens_seen": 340796992,
      "step": 334700
    },
    {
      "epoch": 6.591719005335591,
      "grad_norm": 1.9851182699203491,
      "learning_rate": 5.94553688455374e-07,
      "loss": 2.0403,
      "num_input_tokens_seen": 340899008,
      "step": 334800
    },
    {
      "epoch": 6.593687858085094,
      "grad_norm": 1.9218271970748901,
      "learning_rate": 5.943512201732189e-07,
      "loss": 2.0017,
      "num_input_tokens_seen": 341001408,
      "step": 334900
    },
    {
      "epoch": 6.5956567108345965,
      "grad_norm": 1.775185465812683,
      "learning_rate": 5.941487358474344e-07,
      "loss": 1.9468,
      "num_input_tokens_seen": 341102960,
      "step": 335000
    },
    {
      "epoch": 6.597625563584099,
      "grad_norm": 2.445472002029419,
      "learning_rate": 5.939462355124516e-07,
      "loss": 2.0083,
      "num_input_tokens_seen": 341204744,
      "step": 335100
    },
    {
      "epoch": 6.599594416333602,
      "grad_norm": 4.558581829071045,
      "learning_rate": 5.937437192027037e-07,
      "loss": 2.0086,
      "num_input_tokens_seen": 341305304,
      "step": 335200
    },
    {
      "epoch": 6.601563269083106,
      "grad_norm": 2.0440187454223633,
      "learning_rate": 5.935411869526273e-07,
      "loss": 1.9781,
      "num_input_tokens_seen": 341407704,
      "step": 335300
    },
    {
      "epoch": 6.603532121832608,
      "grad_norm": 1.8654133081436157,
      "learning_rate": 5.933386387966608e-07,
      "loss": 1.9822,
      "num_input_tokens_seen": 341510104,
      "step": 335400
    },
    {
      "epoch": 6.605500974582111,
      "grad_norm": 1.9177144765853882,
      "learning_rate": 5.931360747692461e-07,
      "loss": 2.0323,
      "num_input_tokens_seen": 341612504,
      "step": 335500
    },
    {
      "epoch": 6.607469827331614,
      "grad_norm": 1.9475739002227783,
      "learning_rate": 5.929334949048274e-07,
      "loss": 1.9899,
      "num_input_tokens_seen": 341714080,
      "step": 335600
    },
    {
      "epoch": 6.609438680081117,
      "grad_norm": 1.7756541967391968,
      "learning_rate": 5.927308992378518e-07,
      "loss": 2.0027,
      "num_input_tokens_seen": 341816480,
      "step": 335700
    },
    {
      "epoch": 6.611407532830619,
      "grad_norm": 2.0020506381988525,
      "learning_rate": 5.925282878027688e-07,
      "loss": 2.004,
      "num_input_tokens_seen": 341918184,
      "step": 335800
    },
    {
      "epoch": 6.613376385580122,
      "grad_norm": 1.7956039905548096,
      "learning_rate": 5.92325660634031e-07,
      "loss": 2.0263,
      "num_input_tokens_seen": 342019904,
      "step": 335900
    },
    {
      "epoch": 6.615345238329625,
      "grad_norm": 2.057025194168091,
      "learning_rate": 5.921230177660932e-07,
      "loss": 2.0009,
      "num_input_tokens_seen": 342122304,
      "step": 336000
    },
    {
      "epoch": 6.617314091079129,
      "grad_norm": 1.800400733947754,
      "learning_rate": 5.919203592334132e-07,
      "loss": 2.002,
      "num_input_tokens_seen": 342224272,
      "step": 336100
    },
    {
      "epoch": 6.619282943828631,
      "grad_norm": 2.387866973876953,
      "learning_rate": 5.917176850704515e-07,
      "loss": 1.9729,
      "num_input_tokens_seen": 342326672,
      "step": 336200
    },
    {
      "epoch": 6.621251796578134,
      "grad_norm": 1.7869130373001099,
      "learning_rate": 5.91514995311671e-07,
      "loss": 2.0465,
      "num_input_tokens_seen": 342426264,
      "step": 336300
    },
    {
      "epoch": 6.623220649327637,
      "grad_norm": 2.065978527069092,
      "learning_rate": 5.913122899915375e-07,
      "loss": 1.9889,
      "num_input_tokens_seen": 342528032,
      "step": 336400
    },
    {
      "epoch": 6.62518950207714,
      "grad_norm": 1.9413907527923584,
      "learning_rate": 5.911095691445192e-07,
      "loss": 1.9909,
      "num_input_tokens_seen": 342629840,
      "step": 336500
    },
    {
      "epoch": 6.627158354826642,
      "grad_norm": 1.8402358293533325,
      "learning_rate": 5.909068328050871e-07,
      "loss": 1.9538,
      "num_input_tokens_seen": 342732240,
      "step": 336600
    },
    {
      "epoch": 6.629127207576145,
      "grad_norm": 1.9476884603500366,
      "learning_rate": 5.907040810077149e-07,
      "loss": 1.9459,
      "num_input_tokens_seen": 342834640,
      "step": 336700
    },
    {
      "epoch": 6.631096060325648,
      "grad_norm": 1.9057573080062866,
      "learning_rate": 5.905013137868786e-07,
      "loss": 1.948,
      "num_input_tokens_seen": 342937040,
      "step": 336800
    },
    {
      "epoch": 6.633064913075151,
      "grad_norm": 1.87544584274292,
      "learning_rate": 5.902985311770574e-07,
      "loss": 1.994,
      "num_input_tokens_seen": 343038568,
      "step": 336900
    },
    {
      "epoch": 6.635033765824654,
      "grad_norm": 1.833757758140564,
      "learning_rate": 5.900957332127324e-07,
      "loss": 1.9739,
      "num_input_tokens_seen": 343140968,
      "step": 337000
    },
    {
      "epoch": 6.637002618574157,
      "grad_norm": 1.8252379894256592,
      "learning_rate": 5.898929199283879e-07,
      "loss": 2.0175,
      "num_input_tokens_seen": 343241880,
      "step": 337100
    },
    {
      "epoch": 6.63897147132366,
      "grad_norm": 1.8570109605789185,
      "learning_rate": 5.896900913585108e-07,
      "loss": 1.9806,
      "num_input_tokens_seen": 343344232,
      "step": 337200
    },
    {
      "epoch": 6.640940324073163,
      "grad_norm": 1.637611985206604,
      "learning_rate": 5.894872475375898e-07,
      "loss": 1.9897,
      "num_input_tokens_seen": 343445784,
      "step": 337300
    },
    {
      "epoch": 6.642909176822665,
      "grad_norm": 1.9490511417388916,
      "learning_rate": 5.892843885001174e-07,
      "loss": 1.9749,
      "num_input_tokens_seen": 343548184,
      "step": 337400
    },
    {
      "epoch": 6.644878029572168,
      "grad_norm": 2.083259344100952,
      "learning_rate": 5.890815142805877e-07,
      "loss": 2.027,
      "num_input_tokens_seen": 343650584,
      "step": 337500
    },
    {
      "epoch": 6.646846882321671,
      "grad_norm": 1.9399176836013794,
      "learning_rate": 5.88878624913498e-07,
      "loss": 1.9785,
      "num_input_tokens_seen": 343752984,
      "step": 337600
    },
    {
      "epoch": 6.6488157350711745,
      "grad_norm": 2.3614978790283203,
      "learning_rate": 5.886757204333474e-07,
      "loss": 1.9714,
      "num_input_tokens_seen": 343853312,
      "step": 337700
    },
    {
      "epoch": 6.650784587820677,
      "grad_norm": 2.3670032024383545,
      "learning_rate": 5.884728008746388e-07,
      "loss": 1.9606,
      "num_input_tokens_seen": 343955088,
      "step": 337800
    },
    {
      "epoch": 6.65275344057018,
      "grad_norm": 1.8101786375045776,
      "learning_rate": 5.882698662718765e-07,
      "loss": 1.9502,
      "num_input_tokens_seen": 344057488,
      "step": 337900
    },
    {
      "epoch": 6.654722293319683,
      "grad_norm": 1.8339052200317383,
      "learning_rate": 5.880669166595679e-07,
      "loss": 1.9567,
      "num_input_tokens_seen": 344159272,
      "step": 338000
    },
    {
      "epoch": 6.6566911460691855,
      "grad_norm": 1.7340331077575684,
      "learning_rate": 5.878639520722228e-07,
      "loss": 1.9406,
      "num_input_tokens_seen": 344261216,
      "step": 338100
    },
    {
      "epoch": 6.658659998818688,
      "grad_norm": 1.6825907230377197,
      "learning_rate": 5.87660972544354e-07,
      "loss": 1.9812,
      "num_input_tokens_seen": 344363616,
      "step": 338200
    },
    {
      "epoch": 6.660628851568191,
      "grad_norm": 1.6769567728042603,
      "learning_rate": 5.874579781104761e-07,
      "loss": 1.9939,
      "num_input_tokens_seen": 344466016,
      "step": 338300
    },
    {
      "epoch": 6.662597704317694,
      "grad_norm": 1.7573374509811401,
      "learning_rate": 5.872549688051066e-07,
      "loss": 1.9588,
      "num_input_tokens_seen": 344567736,
      "step": 338400
    },
    {
      "epoch": 6.664566557067197,
      "grad_norm": 1.8433133363723755,
      "learning_rate": 5.870519446627656e-07,
      "loss": 2.0107,
      "num_input_tokens_seen": 344666912,
      "step": 338500
    },
    {
      "epoch": 6.666535409816699,
      "grad_norm": 1.5975202322006226,
      "learning_rate": 5.868489057179757e-07,
      "loss": 1.9665,
      "num_input_tokens_seen": 344768880,
      "step": 338600
    },
    {
      "epoch": 6.668504262566203,
      "grad_norm": 1.6781902313232422,
      "learning_rate": 5.86645852005262e-07,
      "loss": 1.9924,
      "num_input_tokens_seen": 344870384,
      "step": 338700
    },
    {
      "epoch": 6.670473115315706,
      "grad_norm": 1.949173092842102,
      "learning_rate": 5.864427835591518e-07,
      "loss": 1.9494,
      "num_input_tokens_seen": 344971976,
      "step": 338800
    },
    {
      "epoch": 6.6724419680652085,
      "grad_norm": 1.9719985723495483,
      "learning_rate": 5.862397004141754e-07,
      "loss": 2.0234,
      "num_input_tokens_seen": 345072800,
      "step": 338900
    },
    {
      "epoch": 6.674410820814711,
      "grad_norm": 1.8209240436553955,
      "learning_rate": 5.860366026048654e-07,
      "loss": 2.0197,
      "num_input_tokens_seen": 345173656,
      "step": 339000
    },
    {
      "epoch": 6.676379673564214,
      "grad_norm": 2.11574649810791,
      "learning_rate": 5.858334901657565e-07,
      "loss": 1.9947,
      "num_input_tokens_seen": 345275320,
      "step": 339100
    },
    {
      "epoch": 6.678348526313717,
      "grad_norm": 1.870963215827942,
      "learning_rate": 5.85630363131387e-07,
      "loss": 2.0067,
      "num_input_tokens_seen": 345376992,
      "step": 339200
    },
    {
      "epoch": 6.6803173790632195,
      "grad_norm": 1.9651352167129517,
      "learning_rate": 5.85427221536296e-07,
      "loss": 2.0146,
      "num_input_tokens_seen": 345479392,
      "step": 339300
    },
    {
      "epoch": 6.682286231812723,
      "grad_norm": 2.019988536834717,
      "learning_rate": 5.852240654150268e-07,
      "loss": 2.0072,
      "num_input_tokens_seen": 345581216,
      "step": 339400
    },
    {
      "epoch": 6.684255084562226,
      "grad_norm": 1.876074194908142,
      "learning_rate": 5.850208948021241e-07,
      "loss": 2.0149,
      "num_input_tokens_seen": 345683072,
      "step": 339500
    },
    {
      "epoch": 6.686223937311729,
      "grad_norm": 1.852657437324524,
      "learning_rate": 5.848177097321354e-07,
      "loss": 1.9422,
      "num_input_tokens_seen": 345785472,
      "step": 339600
    },
    {
      "epoch": 6.6881927900612315,
      "grad_norm": 1.9414300918579102,
      "learning_rate": 5.846145102396104e-07,
      "loss": 1.916,
      "num_input_tokens_seen": 345887872,
      "step": 339700
    },
    {
      "epoch": 6.690161642810734,
      "grad_norm": 1.7268099784851074,
      "learning_rate": 5.844112963591017e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 345990272,
      "step": 339800
    },
    {
      "epoch": 6.692130495560237,
      "grad_norm": 1.8921703100204468,
      "learning_rate": 5.842080681251638e-07,
      "loss": 2.0061,
      "num_input_tokens_seen": 346091328,
      "step": 339900
    },
    {
      "epoch": 6.69409934830974,
      "grad_norm": 2.14228892326355,
      "learning_rate": 5.840048255723544e-07,
      "loss": 2.013,
      "num_input_tokens_seen": 346192800,
      "step": 340000
    },
    {
      "epoch": 6.6960682010592425,
      "grad_norm": 1.9993531703948975,
      "learning_rate": 5.838015687352327e-07,
      "loss": 1.9896,
      "num_input_tokens_seen": 346294408,
      "step": 340100
    },
    {
      "epoch": 6.698037053808745,
      "grad_norm": 1.7952618598937988,
      "learning_rate": 5.835982976483613e-07,
      "loss": 1.9382,
      "num_input_tokens_seen": 346396808,
      "step": 340200
    },
    {
      "epoch": 6.700005906558248,
      "grad_norm": 1.634080410003662,
      "learning_rate": 5.833950123463043e-07,
      "loss": 1.9993,
      "num_input_tokens_seen": 346497864,
      "step": 340300
    },
    {
      "epoch": 6.701974759307752,
      "grad_norm": 2.08182692527771,
      "learning_rate": 5.83191712863629e-07,
      "loss": 1.961,
      "num_input_tokens_seen": 346600264,
      "step": 340400
    },
    {
      "epoch": 6.703943612057254,
      "grad_norm": 2.3181722164154053,
      "learning_rate": 5.829883992349045e-07,
      "loss": 1.985,
      "num_input_tokens_seen": 346701840,
      "step": 340500
    },
    {
      "epoch": 6.705912464806757,
      "grad_norm": 2.2417986392974854,
      "learning_rate": 5.827850714947028e-07,
      "loss": 1.9871,
      "num_input_tokens_seen": 346803280,
      "step": 340600
    },
    {
      "epoch": 6.70788131755626,
      "grad_norm": 2.027480363845825,
      "learning_rate": 5.825817296775978e-07,
      "loss": 1.9796,
      "num_input_tokens_seen": 346905576,
      "step": 340700
    },
    {
      "epoch": 6.709850170305763,
      "grad_norm": 1.8036468029022217,
      "learning_rate": 5.823783738181662e-07,
      "loss": 1.9752,
      "num_input_tokens_seen": 347007240,
      "step": 340800
    },
    {
      "epoch": 6.7118190230552655,
      "grad_norm": 2.3290023803710938,
      "learning_rate": 5.821750039509869e-07,
      "loss": 1.9913,
      "num_input_tokens_seen": 347109120,
      "step": 340900
    },
    {
      "epoch": 6.713787875804768,
      "grad_norm": 1.868285059928894,
      "learning_rate": 5.819716201106415e-07,
      "loss": 1.9771,
      "num_input_tokens_seen": 347211152,
      "step": 341000
    },
    {
      "epoch": 6.715756728554272,
      "grad_norm": 1.776397705078125,
      "learning_rate": 5.817682223317133e-07,
      "loss": 1.9602,
      "num_input_tokens_seen": 347313448,
      "step": 341100
    },
    {
      "epoch": 6.717725581303775,
      "grad_norm": 1.8410203456878662,
      "learning_rate": 5.815648106487885e-07,
      "loss": 2.0098,
      "num_input_tokens_seen": 347414544,
      "step": 341200
    },
    {
      "epoch": 6.719694434053277,
      "grad_norm": 1.9899818897247314,
      "learning_rate": 5.813613850964557e-07,
      "loss": 2.0203,
      "num_input_tokens_seen": 347515848,
      "step": 341300
    },
    {
      "epoch": 6.72166328680278,
      "grad_norm": 1.7764811515808105,
      "learning_rate": 5.811579457093056e-07,
      "loss": 2.0277,
      "num_input_tokens_seen": 347618248,
      "step": 341400
    },
    {
      "epoch": 6.723632139552283,
      "grad_norm": 1.8589383363723755,
      "learning_rate": 5.809544925219313e-07,
      "loss": 1.9778,
      "num_input_tokens_seen": 347720104,
      "step": 341500
    },
    {
      "epoch": 6.725600992301786,
      "grad_norm": 2.182234287261963,
      "learning_rate": 5.807510255689283e-07,
      "loss": 1.994,
      "num_input_tokens_seen": 347822504,
      "step": 341600
    },
    {
      "epoch": 6.727569845051288,
      "grad_norm": 1.8717601299285889,
      "learning_rate": 5.805475448848947e-07,
      "loss": 2.0046,
      "num_input_tokens_seen": 347924560,
      "step": 341700
    },
    {
      "epoch": 6.729538697800791,
      "grad_norm": 1.7180418968200684,
      "learning_rate": 5.803440505044302e-07,
      "loss": 1.9474,
      "num_input_tokens_seen": 348025520,
      "step": 341800
    },
    {
      "epoch": 6.731507550550294,
      "grad_norm": 1.8298027515411377,
      "learning_rate": 5.801405424621379e-07,
      "loss": 2.0128,
      "num_input_tokens_seen": 348126256,
      "step": 341900
    },
    {
      "epoch": 6.733476403299798,
      "grad_norm": 2.12819504737854,
      "learning_rate": 5.79937020792622e-07,
      "loss": 1.9743,
      "num_input_tokens_seen": 348228104,
      "step": 342000
    },
    {
      "epoch": 6.7354452560493,
      "grad_norm": 2.0564088821411133,
      "learning_rate": 5.797334855304898e-07,
      "loss": 1.9498,
      "num_input_tokens_seen": 348330504,
      "step": 342100
    },
    {
      "epoch": 6.737414108798803,
      "grad_norm": 2.293470859527588,
      "learning_rate": 5.795299367103511e-07,
      "loss": 1.9674,
      "num_input_tokens_seen": 348432904,
      "step": 342200
    },
    {
      "epoch": 6.739382961548306,
      "grad_norm": 1.7751963138580322,
      "learning_rate": 5.793263743668175e-07,
      "loss": 1.9997,
      "num_input_tokens_seen": 348535304,
      "step": 342300
    },
    {
      "epoch": 6.741351814297809,
      "grad_norm": 1.7971038818359375,
      "learning_rate": 5.79122798534503e-07,
      "loss": 1.9741,
      "num_input_tokens_seen": 348636496,
      "step": 342400
    },
    {
      "epoch": 6.743320667047311,
      "grad_norm": 1.8907898664474487,
      "learning_rate": 5.789192092480239e-07,
      "loss": 2.0057,
      "num_input_tokens_seen": 348736832,
      "step": 342500
    },
    {
      "epoch": 6.745289519796814,
      "grad_norm": 3.2937254905700684,
      "learning_rate": 5.787156065419989e-07,
      "loss": 1.9634,
      "num_input_tokens_seen": 348837472,
      "step": 342600
    },
    {
      "epoch": 6.747258372546317,
      "grad_norm": 1.9441038370132446,
      "learning_rate": 5.78511990451049e-07,
      "loss": 1.9541,
      "num_input_tokens_seen": 348939224,
      "step": 342700
    },
    {
      "epoch": 6.7492272252958205,
      "grad_norm": 1.8163578510284424,
      "learning_rate": 5.783083610097973e-07,
      "loss": 2.0377,
      "num_input_tokens_seen": 349040752,
      "step": 342800
    },
    {
      "epoch": 6.751196078045323,
      "grad_norm": 1.8017921447753906,
      "learning_rate": 5.781047182528693e-07,
      "loss": 1.9621,
      "num_input_tokens_seen": 349142576,
      "step": 342900
    },
    {
      "epoch": 6.753164930794826,
      "grad_norm": 2.1576220989227295,
      "learning_rate": 5.779010622148926e-07,
      "loss": 2.0212,
      "num_input_tokens_seen": 349244976,
      "step": 343000
    },
    {
      "epoch": 6.755133783544329,
      "grad_norm": 2.0112767219543457,
      "learning_rate": 5.776973929304976e-07,
      "loss": 1.9123,
      "num_input_tokens_seen": 349346712,
      "step": 343100
    },
    {
      "epoch": 6.757102636293832,
      "grad_norm": 1.8547894954681396,
      "learning_rate": 5.774937104343159e-07,
      "loss": 1.994,
      "num_input_tokens_seen": 349447256,
      "step": 343200
    },
    {
      "epoch": 6.759071489043334,
      "grad_norm": 2.012669324874878,
      "learning_rate": 5.772900147609825e-07,
      "loss": 2.0143,
      "num_input_tokens_seen": 349548592,
      "step": 343300
    },
    {
      "epoch": 6.761040341792837,
      "grad_norm": 1.7234598398208618,
      "learning_rate": 5.77086305945134e-07,
      "loss": 1.9973,
      "num_input_tokens_seen": 349650312,
      "step": 343400
    },
    {
      "epoch": 6.76300919454234,
      "grad_norm": 1.903673768043518,
      "learning_rate": 5.768825840214095e-07,
      "loss": 1.9502,
      "num_input_tokens_seen": 349752712,
      "step": 343500
    },
    {
      "epoch": 6.764978047291843,
      "grad_norm": 1.9555567502975464,
      "learning_rate": 5.766788490244498e-07,
      "loss": 2.0053,
      "num_input_tokens_seen": 349853808,
      "step": 343600
    },
    {
      "epoch": 6.766946900041346,
      "grad_norm": 2.0925943851470947,
      "learning_rate": 5.764751009888987e-07,
      "loss": 1.9851,
      "num_input_tokens_seen": 349955768,
      "step": 343700
    },
    {
      "epoch": 6.768915752790849,
      "grad_norm": 1.8598756790161133,
      "learning_rate": 5.762713399494015e-07,
      "loss": 1.9581,
      "num_input_tokens_seen": 350057696,
      "step": 343800
    },
    {
      "epoch": 6.770884605540352,
      "grad_norm": 2.1003823280334473,
      "learning_rate": 5.760675659406065e-07,
      "loss": 1.9688,
      "num_input_tokens_seen": 350160096,
      "step": 343900
    },
    {
      "epoch": 6.7728534582898545,
      "grad_norm": 1.919238805770874,
      "learning_rate": 5.758637789971635e-07,
      "loss": 2.0069,
      "num_input_tokens_seen": 350261160,
      "step": 344000
    },
    {
      "epoch": 6.774822311039357,
      "grad_norm": 1.9554997682571411,
      "learning_rate": 5.756599791537247e-07,
      "loss": 2.0035,
      "num_input_tokens_seen": 350361944,
      "step": 344100
    },
    {
      "epoch": 6.77679116378886,
      "grad_norm": 1.9567841291427612,
      "learning_rate": 5.754561664449444e-07,
      "loss": 1.9541,
      "num_input_tokens_seen": 350464344,
      "step": 344200
    },
    {
      "epoch": 6.778760016538363,
      "grad_norm": 1.673003077507019,
      "learning_rate": 5.752523409054796e-07,
      "loss": 2.0018,
      "num_input_tokens_seen": 350566144,
      "step": 344300
    },
    {
      "epoch": 6.7807288692878664,
      "grad_norm": 2.2396624088287354,
      "learning_rate": 5.750485025699889e-07,
      "loss": 2.0351,
      "num_input_tokens_seen": 350667904,
      "step": 344400
    },
    {
      "epoch": 6.782697722037369,
      "grad_norm": 2.2762765884399414,
      "learning_rate": 5.748446514731333e-07,
      "loss": 1.9764,
      "num_input_tokens_seen": 350769064,
      "step": 344500
    },
    {
      "epoch": 6.784666574786872,
      "grad_norm": 1.8437024354934692,
      "learning_rate": 5.74640787649576e-07,
      "loss": 1.9674,
      "num_input_tokens_seen": 350871184,
      "step": 344600
    },
    {
      "epoch": 6.786635427536375,
      "grad_norm": 1.8720802068710327,
      "learning_rate": 5.744369111339824e-07,
      "loss": 1.9963,
      "num_input_tokens_seen": 350973584,
      "step": 344700
    },
    {
      "epoch": 6.7886042802858775,
      "grad_norm": 3.721076726913452,
      "learning_rate": 5.742330219610197e-07,
      "loss": 2.0007,
      "num_input_tokens_seen": 351075984,
      "step": 344800
    },
    {
      "epoch": 6.79057313303538,
      "grad_norm": 2.1021668910980225,
      "learning_rate": 5.740291201653579e-07,
      "loss": 1.9819,
      "num_input_tokens_seen": 351178384,
      "step": 344900
    },
    {
      "epoch": 6.792541985784883,
      "grad_norm": 1.7631405591964722,
      "learning_rate": 5.738252057816684e-07,
      "loss": 1.9915,
      "num_input_tokens_seen": 351279120,
      "step": 345000
    },
    {
      "epoch": 6.794510838534386,
      "grad_norm": 1.955552339553833,
      "learning_rate": 5.736212788446257e-07,
      "loss": 1.9725,
      "num_input_tokens_seen": 351380592,
      "step": 345100
    },
    {
      "epoch": 6.7964796912838885,
      "grad_norm": 1.7008463144302368,
      "learning_rate": 5.734173393889051e-07,
      "loss": 2.03,
      "num_input_tokens_seen": 351482440,
      "step": 345200
    },
    {
      "epoch": 6.798448544033391,
      "grad_norm": 3.0899970531463623,
      "learning_rate": 5.732133874491853e-07,
      "loss": 2.0021,
      "num_input_tokens_seen": 351584840,
      "step": 345300
    },
    {
      "epoch": 6.800417396782895,
      "grad_norm": 1.8752034902572632,
      "learning_rate": 5.730094230601466e-07,
      "loss": 1.9792,
      "num_input_tokens_seen": 351687240,
      "step": 345400
    },
    {
      "epoch": 6.802386249532398,
      "grad_norm": 1.8636212348937988,
      "learning_rate": 5.728054462564711e-07,
      "loss": 1.9815,
      "num_input_tokens_seen": 351789640,
      "step": 345500
    },
    {
      "epoch": 6.8043551022819,
      "grad_norm": 1.7927242517471313,
      "learning_rate": 5.726014570728437e-07,
      "loss": 2.0082,
      "num_input_tokens_seen": 351892040,
      "step": 345600
    },
    {
      "epoch": 6.806323955031403,
      "grad_norm": 7.441971302032471,
      "learning_rate": 5.723974555439508e-07,
      "loss": 1.9791,
      "num_input_tokens_seen": 351993888,
      "step": 345700
    },
    {
      "epoch": 6.808292807780906,
      "grad_norm": 1.948264479637146,
      "learning_rate": 5.721934417044813e-07,
      "loss": 1.9926,
      "num_input_tokens_seen": 352096288,
      "step": 345800
    },
    {
      "epoch": 6.810261660530409,
      "grad_norm": 1.810975193977356,
      "learning_rate": 5.719894155891258e-07,
      "loss": 2.0123,
      "num_input_tokens_seen": 352198240,
      "step": 345900
    },
    {
      "epoch": 6.8122305132799115,
      "grad_norm": 1.8330130577087402,
      "learning_rate": 5.717853772325774e-07,
      "loss": 2.0488,
      "num_input_tokens_seen": 352300088,
      "step": 346000
    },
    {
      "epoch": 6.814199366029415,
      "grad_norm": 1.7867487668991089,
      "learning_rate": 5.71581326669531e-07,
      "loss": 2.0212,
      "num_input_tokens_seen": 352402488,
      "step": 346100
    },
    {
      "epoch": 6.816168218778918,
      "grad_norm": 1.878553867340088,
      "learning_rate": 5.713772639346839e-07,
      "loss": 2.006,
      "num_input_tokens_seen": 352503584,
      "step": 346200
    },
    {
      "epoch": 6.818137071528421,
      "grad_norm": 1.945835828781128,
      "learning_rate": 5.71173189062735e-07,
      "loss": 2.0214,
      "num_input_tokens_seen": 352605112,
      "step": 346300
    },
    {
      "epoch": 6.820105924277923,
      "grad_norm": 1.8979483842849731,
      "learning_rate": 5.709691020883856e-07,
      "loss": 1.9857,
      "num_input_tokens_seen": 352707512,
      "step": 346400
    },
    {
      "epoch": 6.822074777027426,
      "grad_norm": 1.9538261890411377,
      "learning_rate": 5.707650030463392e-07,
      "loss": 1.9477,
      "num_input_tokens_seen": 352809688,
      "step": 346500
    },
    {
      "epoch": 6.824043629776929,
      "grad_norm": 1.7969032526016235,
      "learning_rate": 5.705608919713006e-07,
      "loss": 1.9765,
      "num_input_tokens_seen": 352912088,
      "step": 346600
    },
    {
      "epoch": 6.826012482526432,
      "grad_norm": 3.273768186569214,
      "learning_rate": 5.703567688979776e-07,
      "loss": 1.9902,
      "num_input_tokens_seen": 353013616,
      "step": 346700
    },
    {
      "epoch": 6.827981335275934,
      "grad_norm": 2.1750895977020264,
      "learning_rate": 5.701526338610794e-07,
      "loss": 1.9447,
      "num_input_tokens_seen": 353116016,
      "step": 346800
    },
    {
      "epoch": 6.829950188025437,
      "grad_norm": 2.028327465057373,
      "learning_rate": 5.699484868953175e-07,
      "loss": 1.9477,
      "num_input_tokens_seen": 353218416,
      "step": 346900
    },
    {
      "epoch": 6.83191904077494,
      "grad_norm": 1.7221735715866089,
      "learning_rate": 5.697443280354055e-07,
      "loss": 1.9551,
      "num_input_tokens_seen": 353320312,
      "step": 347000
    },
    {
      "epoch": 6.833887893524444,
      "grad_norm": 2.1486196517944336,
      "learning_rate": 5.695401573160586e-07,
      "loss": 1.9697,
      "num_input_tokens_seen": 353422712,
      "step": 347100
    },
    {
      "epoch": 6.835856746273946,
      "grad_norm": 2.0063304901123047,
      "learning_rate": 5.693359747719946e-07,
      "loss": 1.9567,
      "num_input_tokens_seen": 353525112,
      "step": 347200
    },
    {
      "epoch": 6.837825599023449,
      "grad_norm": 2.1901392936706543,
      "learning_rate": 5.691317804379326e-07,
      "loss": 1.9624,
      "num_input_tokens_seen": 353627248,
      "step": 347300
    },
    {
      "epoch": 6.839794451772952,
      "grad_norm": 2.088301420211792,
      "learning_rate": 5.689275743485949e-07,
      "loss": 2.0254,
      "num_input_tokens_seen": 353729648,
      "step": 347400
    },
    {
      "epoch": 6.841763304522455,
      "grad_norm": 1.982779860496521,
      "learning_rate": 5.687233565387041e-07,
      "loss": 1.9797,
      "num_input_tokens_seen": 353831512,
      "step": 347500
    },
    {
      "epoch": 6.843732157271957,
      "grad_norm": 1.9566974639892578,
      "learning_rate": 5.685191270429866e-07,
      "loss": 2.0576,
      "num_input_tokens_seen": 353932304,
      "step": 347600
    },
    {
      "epoch": 6.84570101002146,
      "grad_norm": 2.1104142665863037,
      "learning_rate": 5.68314885896169e-07,
      "loss": 2.0094,
      "num_input_tokens_seen": 354033616,
      "step": 347700
    },
    {
      "epoch": 6.847669862770964,
      "grad_norm": 1.9614967107772827,
      "learning_rate": 5.681106331329817e-07,
      "loss": 2.01,
      "num_input_tokens_seen": 354136008,
      "step": 347800
    },
    {
      "epoch": 6.8496387155204665,
      "grad_norm": 2.188655138015747,
      "learning_rate": 5.679063687881554e-07,
      "loss": 1.9384,
      "num_input_tokens_seen": 354237600,
      "step": 347900
    },
    {
      "epoch": 6.851607568269969,
      "grad_norm": 2.3352487087249756,
      "learning_rate": 5.67702092896424e-07,
      "loss": 1.9875,
      "num_input_tokens_seen": 354338600,
      "step": 348000
    },
    {
      "epoch": 6.853576421019472,
      "grad_norm": 1.7377607822418213,
      "learning_rate": 5.674978054925226e-07,
      "loss": 2.0607,
      "num_input_tokens_seen": 354439248,
      "step": 348100
    },
    {
      "epoch": 6.855545273768975,
      "grad_norm": 1.6657297611236572,
      "learning_rate": 5.672935066111888e-07,
      "loss": 1.9402,
      "num_input_tokens_seen": 354540184,
      "step": 348200
    },
    {
      "epoch": 6.857514126518478,
      "grad_norm": 2.0258216857910156,
      "learning_rate": 5.670891962871616e-07,
      "loss": 1.9366,
      "num_input_tokens_seen": 354642584,
      "step": 348300
    },
    {
      "epoch": 6.85948297926798,
      "grad_norm": 1.7575098276138306,
      "learning_rate": 5.668848745551826e-07,
      "loss": 1.9895,
      "num_input_tokens_seen": 354744464,
      "step": 348400
    },
    {
      "epoch": 6.861451832017483,
      "grad_norm": 1.8955113887786865,
      "learning_rate": 5.666805414499947e-07,
      "loss": 2.0603,
      "num_input_tokens_seen": 354846864,
      "step": 348500
    },
    {
      "epoch": 6.863420684766986,
      "grad_norm": 2.3453078269958496,
      "learning_rate": 5.664761970063432e-07,
      "loss": 1.9852,
      "num_input_tokens_seen": 354948904,
      "step": 348600
    },
    {
      "epoch": 6.8653895375164895,
      "grad_norm": 1.8201344013214111,
      "learning_rate": 5.662718412589749e-07,
      "loss": 1.9727,
      "num_input_tokens_seen": 355051304,
      "step": 348700
    },
    {
      "epoch": 6.867358390265992,
      "grad_norm": 1.9629918336868286,
      "learning_rate": 5.660674742426393e-07,
      "loss": 1.9834,
      "num_input_tokens_seen": 355152888,
      "step": 348800
    },
    {
      "epoch": 6.869327243015495,
      "grad_norm": 1.9738911390304565,
      "learning_rate": 5.658630959920867e-07,
      "loss": 1.9556,
      "num_input_tokens_seen": 355255288,
      "step": 348900
    },
    {
      "epoch": 6.871296095764998,
      "grad_norm": 2.033052921295166,
      "learning_rate": 5.656587065420703e-07,
      "loss": 1.988,
      "num_input_tokens_seen": 355357136,
      "step": 349000
    },
    {
      "epoch": 6.8732649485145005,
      "grad_norm": 2.0622310638427734,
      "learning_rate": 5.654543059273446e-07,
      "loss": 1.9965,
      "num_input_tokens_seen": 355457976,
      "step": 349100
    },
    {
      "epoch": 6.875233801264003,
      "grad_norm": 1.981715202331543,
      "learning_rate": 5.652498941826662e-07,
      "loss": 1.9976,
      "num_input_tokens_seen": 355560376,
      "step": 349200
    },
    {
      "epoch": 6.877202654013506,
      "grad_norm": 1.9295616149902344,
      "learning_rate": 5.650454713427936e-07,
      "loss": 2.0092,
      "num_input_tokens_seen": 355662776,
      "step": 349300
    },
    {
      "epoch": 6.879171506763009,
      "grad_norm": 1.954331636428833,
      "learning_rate": 5.648410374424872e-07,
      "loss": 2.0167,
      "num_input_tokens_seen": 355764680,
      "step": 349400
    },
    {
      "epoch": 6.8811403595125125,
      "grad_norm": 3.1415114402770996,
      "learning_rate": 5.646365925165094e-07,
      "loss": 1.9845,
      "num_input_tokens_seen": 355866312,
      "step": 349500
    },
    {
      "epoch": 6.883109212262015,
      "grad_norm": 1.667959451675415,
      "learning_rate": 5.644321365996241e-07,
      "loss": 1.9968,
      "num_input_tokens_seen": 355966392,
      "step": 349600
    },
    {
      "epoch": 6.885078065011518,
      "grad_norm": 2.055408239364624,
      "learning_rate": 5.642276697265975e-07,
      "loss": 1.9822,
      "num_input_tokens_seen": 356067968,
      "step": 349700
    },
    {
      "epoch": 6.887046917761021,
      "grad_norm": 1.9663994312286377,
      "learning_rate": 5.640231919321974e-07,
      "loss": 1.9615,
      "num_input_tokens_seen": 356170368,
      "step": 349800
    },
    {
      "epoch": 6.8890157705105235,
      "grad_norm": 1.9548956155776978,
      "learning_rate": 5.638187032511935e-07,
      "loss": 1.9239,
      "num_input_tokens_seen": 356272768,
      "step": 349900
    },
    {
      "epoch": 6.890984623260026,
      "grad_norm": 2.039267063140869,
      "learning_rate": 5.636142037183574e-07,
      "loss": 1.9914,
      "num_input_tokens_seen": 356375168,
      "step": 350000
    },
    {
      "epoch": 6.892953476009529,
      "grad_norm": 1.8091191053390503,
      "learning_rate": 5.634096933684625e-07,
      "loss": 1.9485,
      "num_input_tokens_seen": 356477568,
      "step": 350100
    },
    {
      "epoch": 6.894922328759032,
      "grad_norm": 1.9435220956802368,
      "learning_rate": 5.632051722362838e-07,
      "loss": 1.9729,
      "num_input_tokens_seen": 356579120,
      "step": 350200
    },
    {
      "epoch": 6.8968911815085345,
      "grad_norm": 2.0470776557922363,
      "learning_rate": 5.630006403565989e-07,
      "loss": 1.9926,
      "num_input_tokens_seen": 356680896,
      "step": 350300
    },
    {
      "epoch": 6.898860034258038,
      "grad_norm": 1.8169724941253662,
      "learning_rate": 5.627960977641863e-07,
      "loss": 1.9603,
      "num_input_tokens_seen": 356783296,
      "step": 350400
    },
    {
      "epoch": 6.900828887007541,
      "grad_norm": 2.2178609371185303,
      "learning_rate": 5.625915444938271e-07,
      "loss": 2.0383,
      "num_input_tokens_seen": 356885128,
      "step": 350500
    },
    {
      "epoch": 6.902797739757044,
      "grad_norm": 1.917288899421692,
      "learning_rate": 5.623869805803038e-07,
      "loss": 2.0111,
      "num_input_tokens_seen": 356987392,
      "step": 350600
    },
    {
      "epoch": 6.9047665925065465,
      "grad_norm": 1.8889974355697632,
      "learning_rate": 5.621824060584005e-07,
      "loss": 2.0086,
      "num_input_tokens_seen": 357089792,
      "step": 350700
    },
    {
      "epoch": 6.906735445256049,
      "grad_norm": 1.655066728591919,
      "learning_rate": 5.619778209629034e-07,
      "loss": 1.9734,
      "num_input_tokens_seen": 357190952,
      "step": 350800
    },
    {
      "epoch": 6.908704298005552,
      "grad_norm": 1.8650329113006592,
      "learning_rate": 5.617732253286009e-07,
      "loss": 1.9931,
      "num_input_tokens_seen": 357293352,
      "step": 350900
    },
    {
      "epoch": 6.910673150755055,
      "grad_norm": 1.9264100790023804,
      "learning_rate": 5.615686191902822e-07,
      "loss": 1.954,
      "num_input_tokens_seen": 357395216,
      "step": 351000
    },
    {
      "epoch": 6.912642003504558,
      "grad_norm": 2.1717875003814697,
      "learning_rate": 5.613640025827392e-07,
      "loss": 2.0366,
      "num_input_tokens_seen": 357496904,
      "step": 351100
    },
    {
      "epoch": 6.914610856254061,
      "grad_norm": 1.8024847507476807,
      "learning_rate": 5.61159375540765e-07,
      "loss": 1.9804,
      "num_input_tokens_seen": 357598344,
      "step": 351200
    },
    {
      "epoch": 6.916579709003564,
      "grad_norm": 2.213294506072998,
      "learning_rate": 5.60954738099155e-07,
      "loss": 2.0139,
      "num_input_tokens_seen": 357700200,
      "step": 351300
    },
    {
      "epoch": 6.918548561753067,
      "grad_norm": 1.8208956718444824,
      "learning_rate": 5.607500902927057e-07,
      "loss": 2.0063,
      "num_input_tokens_seen": 357802600,
      "step": 351400
    },
    {
      "epoch": 6.920517414502569,
      "grad_norm": 1.8078587055206299,
      "learning_rate": 5.605454321562161e-07,
      "loss": 2.0143,
      "num_input_tokens_seen": 357905000,
      "step": 351500
    },
    {
      "epoch": 6.922486267252072,
      "grad_norm": 1.9979687929153442,
      "learning_rate": 5.603407637244864e-07,
      "loss": 1.9555,
      "num_input_tokens_seen": 358007400,
      "step": 351600
    },
    {
      "epoch": 6.924455120001575,
      "grad_norm": 1.9737186431884766,
      "learning_rate": 5.601360850323188e-07,
      "loss": 1.9822,
      "num_input_tokens_seen": 358109800,
      "step": 351700
    },
    {
      "epoch": 6.926423972751078,
      "grad_norm": 2.0724594593048096,
      "learning_rate": 5.59931396114517e-07,
      "loss": 1.9898,
      "num_input_tokens_seen": 358212200,
      "step": 351800
    },
    {
      "epoch": 6.92839282550058,
      "grad_norm": 2.0351979732513428,
      "learning_rate": 5.59726697005887e-07,
      "loss": 1.983,
      "num_input_tokens_seen": 358314600,
      "step": 351900
    },
    {
      "epoch": 6.930361678250083,
      "grad_norm": 1.9320485591888428,
      "learning_rate": 5.595219877412357e-07,
      "loss": 2.0139,
      "num_input_tokens_seen": 358415216,
      "step": 352000
    },
    {
      "epoch": 6.932330530999587,
      "grad_norm": 2.074334144592285,
      "learning_rate": 5.593172683553726e-07,
      "loss": 2.0202,
      "num_input_tokens_seen": 358517144,
      "step": 352100
    },
    {
      "epoch": 6.93429938374909,
      "grad_norm": 1.8210471868515015,
      "learning_rate": 5.591125388831083e-07,
      "loss": 1.9623,
      "num_input_tokens_seen": 358618752,
      "step": 352200
    },
    {
      "epoch": 6.936268236498592,
      "grad_norm": 1.9182939529418945,
      "learning_rate": 5.589077993592552e-07,
      "loss": 1.9568,
      "num_input_tokens_seen": 358720296,
      "step": 352300
    },
    {
      "epoch": 6.938237089248095,
      "grad_norm": 1.9479706287384033,
      "learning_rate": 5.587030498186277e-07,
      "loss": 2.0339,
      "num_input_tokens_seen": 358822200,
      "step": 352400
    },
    {
      "epoch": 6.940205941997598,
      "grad_norm": 1.7668222188949585,
      "learning_rate": 5.584982902960418e-07,
      "loss": 1.9545,
      "num_input_tokens_seen": 358924104,
      "step": 352500
    },
    {
      "epoch": 6.942174794747101,
      "grad_norm": 1.8232539892196655,
      "learning_rate": 5.58293520826315e-07,
      "loss": 2.0338,
      "num_input_tokens_seen": 359026024,
      "step": 352600
    },
    {
      "epoch": 6.944143647496603,
      "grad_norm": 1.8361282348632812,
      "learning_rate": 5.580887414442667e-07,
      "loss": 1.974,
      "num_input_tokens_seen": 359128112,
      "step": 352700
    },
    {
      "epoch": 6.946112500246107,
      "grad_norm": 1.7626744508743286,
      "learning_rate": 5.578839521847178e-07,
      "loss": 1.9398,
      "num_input_tokens_seen": 359230512,
      "step": 352800
    },
    {
      "epoch": 6.94808135299561,
      "grad_norm": 2.0083327293395996,
      "learning_rate": 5.576791530824911e-07,
      "loss": 1.9806,
      "num_input_tokens_seen": 359332912,
      "step": 352900
    },
    {
      "epoch": 6.950050205745113,
      "grad_norm": 2.031496524810791,
      "learning_rate": 5.574743441724108e-07,
      "loss": 1.9822,
      "num_input_tokens_seen": 359434016,
      "step": 353000
    },
    {
      "epoch": 6.952019058494615,
      "grad_norm": 2.3042380809783936,
      "learning_rate": 5.572695254893031e-07,
      "loss": 2.0024,
      "num_input_tokens_seen": 359536416,
      "step": 353100
    },
    {
      "epoch": 6.953987911244118,
      "grad_norm": 1.96910560131073,
      "learning_rate": 5.570646970679957e-07,
      "loss": 1.9881,
      "num_input_tokens_seen": 359638816,
      "step": 353200
    },
    {
      "epoch": 6.955956763993621,
      "grad_norm": 2.1909523010253906,
      "learning_rate": 5.568598589433178e-07,
      "loss": 1.9824,
      "num_input_tokens_seen": 359740496,
      "step": 353300
    },
    {
      "epoch": 6.957925616743124,
      "grad_norm": 1.9911549091339111,
      "learning_rate": 5.566550111501003e-07,
      "loss": 1.9757,
      "num_input_tokens_seen": 359842272,
      "step": 353400
    },
    {
      "epoch": 6.959894469492626,
      "grad_norm": 2.0698816776275635,
      "learning_rate": 5.564501537231763e-07,
      "loss": 1.9465,
      "num_input_tokens_seen": 359944672,
      "step": 353500
    },
    {
      "epoch": 6.961863322242129,
      "grad_norm": 2.081211566925049,
      "learning_rate": 5.562452866973797e-07,
      "loss": 1.9346,
      "num_input_tokens_seen": 360045192,
      "step": 353600
    },
    {
      "epoch": 6.963832174991633,
      "grad_norm": 1.645334243774414,
      "learning_rate": 5.560404101075463e-07,
      "loss": 2.0468,
      "num_input_tokens_seen": 360147592,
      "step": 353700
    },
    {
      "epoch": 6.9658010277411355,
      "grad_norm": 2.0219058990478516,
      "learning_rate": 5.558355239885141e-07,
      "loss": 1.9634,
      "num_input_tokens_seen": 360249384,
      "step": 353800
    },
    {
      "epoch": 6.967769880490638,
      "grad_norm": 1.7644826173782349,
      "learning_rate": 5.556306283751217e-07,
      "loss": 1.9686,
      "num_input_tokens_seen": 360350376,
      "step": 353900
    },
    {
      "epoch": 6.969738733240141,
      "grad_norm": 2.12752103805542,
      "learning_rate": 5.554257233022104e-07,
      "loss": 1.9833,
      "num_input_tokens_seen": 360452776,
      "step": 354000
    },
    {
      "epoch": 6.971707585989644,
      "grad_norm": 2.0328681468963623,
      "learning_rate": 5.552208088046222e-07,
      "loss": 2.0229,
      "num_input_tokens_seen": 360553616,
      "step": 354100
    },
    {
      "epoch": 6.9736764387391466,
      "grad_norm": 2.161731004714966,
      "learning_rate": 5.550158849172013e-07,
      "loss": 2.0487,
      "num_input_tokens_seen": 360655200,
      "step": 354200
    },
    {
      "epoch": 6.975645291488649,
      "grad_norm": 2.0961341857910156,
      "learning_rate": 5.548109516747932e-07,
      "loss": 1.9601,
      "num_input_tokens_seen": 360756976,
      "step": 354300
    },
    {
      "epoch": 6.977614144238152,
      "grad_norm": 1.967660903930664,
      "learning_rate": 5.546060091122448e-07,
      "loss": 1.9854,
      "num_input_tokens_seen": 360858088,
      "step": 354400
    },
    {
      "epoch": 6.979582996987656,
      "grad_norm": 1.7927228212356567,
      "learning_rate": 5.544010572644053e-07,
      "loss": 1.9871,
      "num_input_tokens_seen": 360959016,
      "step": 354500
    },
    {
      "epoch": 6.9815518497371585,
      "grad_norm": 2.0117151737213135,
      "learning_rate": 5.54196096166125e-07,
      "loss": 2.0087,
      "num_input_tokens_seen": 361061416,
      "step": 354600
    },
    {
      "epoch": 6.983520702486661,
      "grad_norm": 1.8553569316864014,
      "learning_rate": 5.539911258522555e-07,
      "loss": 1.952,
      "num_input_tokens_seen": 361162432,
      "step": 354700
    },
    {
      "epoch": 6.985489555236164,
      "grad_norm": 1.8479331731796265,
      "learning_rate": 5.537861463576505e-07,
      "loss": 1.9504,
      "num_input_tokens_seen": 361264208,
      "step": 354800
    },
    {
      "epoch": 6.987458407985667,
      "grad_norm": 1.9098297357559204,
      "learning_rate": 5.535811577171649e-07,
      "loss": 1.9919,
      "num_input_tokens_seen": 361366416,
      "step": 354900
    },
    {
      "epoch": 6.9894272607351695,
      "grad_norm": 1.964388132095337,
      "learning_rate": 5.533761599656556e-07,
      "loss": 1.9851,
      "num_input_tokens_seen": 361466832,
      "step": 355000
    },
    {
      "epoch": 6.991396113484672,
      "grad_norm": 2.710512399673462,
      "learning_rate": 5.531711531379801e-07,
      "loss": 1.9815,
      "num_input_tokens_seen": 361568632,
      "step": 355100
    },
    {
      "epoch": 6.993364966234175,
      "grad_norm": 1.9005932807922363,
      "learning_rate": 5.529661372689987e-07,
      "loss": 1.9938,
      "num_input_tokens_seen": 361670496,
      "step": 355200
    },
    {
      "epoch": 6.995333818983678,
      "grad_norm": 2.0018246173858643,
      "learning_rate": 5.527611123935724e-07,
      "loss": 2.0307,
      "num_input_tokens_seen": 361771968,
      "step": 355300
    },
    {
      "epoch": 6.997302671733181,
      "grad_norm": 2.3461527824401855,
      "learning_rate": 5.525560785465638e-07,
      "loss": 1.9936,
      "num_input_tokens_seen": 361872496,
      "step": 355400
    },
    {
      "epoch": 6.999271524482684,
      "grad_norm": 11.463159561157227,
      "learning_rate": 5.523510357628375e-07,
      "loss": 1.989,
      "num_input_tokens_seen": 361974896,
      "step": 355500
    },
    {
      "epoch": 7.001240377232187,
      "grad_norm": 2.019601821899414,
      "learning_rate": 5.521459840772591e-07,
      "loss": 2.018,
      "num_input_tokens_seen": 362077296,
      "step": 355600
    },
    {
      "epoch": 7.00320922998169,
      "grad_norm": 1.8182095289230347,
      "learning_rate": 5.519409235246958e-07,
      "loss": 2.0031,
      "num_input_tokens_seen": 362179696,
      "step": 355700
    },
    {
      "epoch": 7.0051780827311925,
      "grad_norm": 1.9885685443878174,
      "learning_rate": 5.517358541400166e-07,
      "loss": 1.9816,
      "num_input_tokens_seen": 362282096,
      "step": 355800
    },
    {
      "epoch": 7.007146935480695,
      "grad_norm": 1.6222323179244995,
      "learning_rate": 5.515307759580917e-07,
      "loss": 1.9919,
      "num_input_tokens_seen": 362383640,
      "step": 355900
    },
    {
      "epoch": 7.009115788230198,
      "grad_norm": 2.084733247756958,
      "learning_rate": 5.513256890137932e-07,
      "loss": 2.0344,
      "num_input_tokens_seen": 362485072,
      "step": 356000
    },
    {
      "epoch": 7.011084640979701,
      "grad_norm": 1.9736841917037964,
      "learning_rate": 5.511205933419941e-07,
      "loss": 1.9684,
      "num_input_tokens_seen": 362587472,
      "step": 356100
    },
    {
      "epoch": 7.013053493729204,
      "grad_norm": 1.8116730451583862,
      "learning_rate": 5.509154889775691e-07,
      "loss": 1.9543,
      "num_input_tokens_seen": 362689872,
      "step": 356200
    },
    {
      "epoch": 7.015022346478707,
      "grad_norm": 1.9676889181137085,
      "learning_rate": 5.507103759553947e-07,
      "loss": 2.0076,
      "num_input_tokens_seen": 362792272,
      "step": 356300
    },
    {
      "epoch": 7.01699119922821,
      "grad_norm": 1.8944008350372314,
      "learning_rate": 5.505052543103487e-07,
      "loss": 2.0034,
      "num_input_tokens_seen": 362893760,
      "step": 356400
    },
    {
      "epoch": 7.018960051977713,
      "grad_norm": 1.8716422319412231,
      "learning_rate": 5.503001240773099e-07,
      "loss": 1.9681,
      "num_input_tokens_seen": 362995512,
      "step": 356500
    },
    {
      "epoch": 7.020928904727215,
      "grad_norm": 2.067317008972168,
      "learning_rate": 5.500949852911596e-07,
      "loss": 2.0361,
      "num_input_tokens_seen": 363097352,
      "step": 356600
    },
    {
      "epoch": 7.022897757476718,
      "grad_norm": 1.9743908643722534,
      "learning_rate": 5.498898379867794e-07,
      "loss": 2.0205,
      "num_input_tokens_seen": 363199752,
      "step": 356700
    },
    {
      "epoch": 7.024866610226221,
      "grad_norm": 1.8057843446731567,
      "learning_rate": 5.496846821990532e-07,
      "loss": 1.98,
      "num_input_tokens_seen": 363302152,
      "step": 356800
    },
    {
      "epoch": 7.026835462975724,
      "grad_norm": 2.0450387001037598,
      "learning_rate": 5.494795179628658e-07,
      "loss": 1.9822,
      "num_input_tokens_seen": 363403064,
      "step": 356900
    },
    {
      "epoch": 7.028804315725227,
      "grad_norm": 1.7592406272888184,
      "learning_rate": 5.492743453131036e-07,
      "loss": 2.0479,
      "num_input_tokens_seen": 363504872,
      "step": 357000
    },
    {
      "epoch": 7.03077316847473,
      "grad_norm": 1.8680310249328613,
      "learning_rate": 5.490691642846545e-07,
      "loss": 1.9885,
      "num_input_tokens_seen": 363606808,
      "step": 357100
    },
    {
      "epoch": 7.032742021224233,
      "grad_norm": 1.6684812307357788,
      "learning_rate": 5.488639749124081e-07,
      "loss": 1.9712,
      "num_input_tokens_seen": 363707656,
      "step": 357200
    },
    {
      "epoch": 7.034710873973736,
      "grad_norm": 1.6753751039505005,
      "learning_rate": 5.486587772312547e-07,
      "loss": 1.9733,
      "num_input_tokens_seen": 363809576,
      "step": 357300
    },
    {
      "epoch": 7.036679726723238,
      "grad_norm": 2.127769708633423,
      "learning_rate": 5.484535712760867e-07,
      "loss": 2.0122,
      "num_input_tokens_seen": 363911056,
      "step": 357400
    },
    {
      "epoch": 7.038648579472741,
      "grad_norm": 1.8474689722061157,
      "learning_rate": 5.482483570817977e-07,
      "loss": 2.0123,
      "num_input_tokens_seen": 364012088,
      "step": 357500
    },
    {
      "epoch": 7.040617432222244,
      "grad_norm": 1.979695439338684,
      "learning_rate": 5.480431346832825e-07,
      "loss": 1.9571,
      "num_input_tokens_seen": 364114008,
      "step": 357600
    },
    {
      "epoch": 7.042586284971747,
      "grad_norm": 1.8507744073867798,
      "learning_rate": 5.478379041154374e-07,
      "loss": 2.0002,
      "num_input_tokens_seen": 364215568,
      "step": 357700
    },
    {
      "epoch": 7.044555137721249,
      "grad_norm": 1.8359742164611816,
      "learning_rate": 5.476326654131601e-07,
      "loss": 1.998,
      "num_input_tokens_seen": 364317968,
      "step": 357800
    },
    {
      "epoch": 7.046523990470753,
      "grad_norm": 2.740100860595703,
      "learning_rate": 5.4742741861135e-07,
      "loss": 1.9667,
      "num_input_tokens_seen": 364419264,
      "step": 357900
    },
    {
      "epoch": 7.048492843220256,
      "grad_norm": 1.944488763809204,
      "learning_rate": 5.472221637449073e-07,
      "loss": 1.9726,
      "num_input_tokens_seen": 364521664,
      "step": 358000
    },
    {
      "epoch": 7.050461695969759,
      "grad_norm": 2.1473124027252197,
      "learning_rate": 5.470169008487339e-07,
      "loss": 2.0125,
      "num_input_tokens_seen": 364622504,
      "step": 358100
    },
    {
      "epoch": 7.052430548719261,
      "grad_norm": 1.8698921203613281,
      "learning_rate": 5.468116299577331e-07,
      "loss": 1.9554,
      "num_input_tokens_seen": 364724224,
      "step": 358200
    },
    {
      "epoch": 7.054399401468764,
      "grad_norm": 1.8280047178268433,
      "learning_rate": 5.466063511068095e-07,
      "loss": 1.9793,
      "num_input_tokens_seen": 364825912,
      "step": 358300
    },
    {
      "epoch": 7.056368254218267,
      "grad_norm": 1.7389600276947021,
      "learning_rate": 5.464010643308689e-07,
      "loss": 1.9385,
      "num_input_tokens_seen": 364927736,
      "step": 358400
    },
    {
      "epoch": 7.05833710696777,
      "grad_norm": 2.045398235321045,
      "learning_rate": 5.461957696648186e-07,
      "loss": 1.9591,
      "num_input_tokens_seen": 365030136,
      "step": 358500
    },
    {
      "epoch": 7.060305959717272,
      "grad_norm": 1.9011130332946777,
      "learning_rate": 5.459904671435673e-07,
      "loss": 1.982,
      "num_input_tokens_seen": 365132088,
      "step": 358600
    },
    {
      "epoch": 7.062274812466776,
      "grad_norm": 1.9591118097305298,
      "learning_rate": 5.457851568020252e-07,
      "loss": 2.0108,
      "num_input_tokens_seen": 365233056,
      "step": 358700
    },
    {
      "epoch": 7.064243665216279,
      "grad_norm": 2.0697875022888184,
      "learning_rate": 5.455798386751033e-07,
      "loss": 1.9933,
      "num_input_tokens_seen": 365334920,
      "step": 358800
    },
    {
      "epoch": 7.0662125179657815,
      "grad_norm": 1.8353288173675537,
      "learning_rate": 5.453745127977142e-07,
      "loss": 1.9743,
      "num_input_tokens_seen": 365437320,
      "step": 358900
    },
    {
      "epoch": 7.068181370715284,
      "grad_norm": 1.771433711051941,
      "learning_rate": 5.451691792047719e-07,
      "loss": 1.9767,
      "num_input_tokens_seen": 365539152,
      "step": 359000
    },
    {
      "epoch": 7.070150223464787,
      "grad_norm": 1.89574134349823,
      "learning_rate": 5.449638379311917e-07,
      "loss": 2.0361,
      "num_input_tokens_seen": 365640624,
      "step": 359100
    },
    {
      "epoch": 7.07211907621429,
      "grad_norm": 1.7786600589752197,
      "learning_rate": 5.447584890118901e-07,
      "loss": 1.9738,
      "num_input_tokens_seen": 365743024,
      "step": 359200
    },
    {
      "epoch": 7.074087928963793,
      "grad_norm": 2.0453200340270996,
      "learning_rate": 5.445531324817849e-07,
      "loss": 1.9797,
      "num_input_tokens_seen": 365845424,
      "step": 359300
    },
    {
      "epoch": 7.076056781713295,
      "grad_norm": 1.8308055400848389,
      "learning_rate": 5.443477683757952e-07,
      "loss": 1.9798,
      "num_input_tokens_seen": 365946416,
      "step": 359400
    },
    {
      "epoch": 7.078025634462799,
      "grad_norm": 2.2811360359191895,
      "learning_rate": 5.441423967288416e-07,
      "loss": 1.9628,
      "num_input_tokens_seen": 366047360,
      "step": 359500
    },
    {
      "epoch": 7.079994487212302,
      "grad_norm": 2.0829670429229736,
      "learning_rate": 5.439370175758457e-07,
      "loss": 1.9381,
      "num_input_tokens_seen": 366149760,
      "step": 359600
    },
    {
      "epoch": 7.0819633399618045,
      "grad_norm": 2.0165088176727295,
      "learning_rate": 5.437316309517308e-07,
      "loss": 1.9481,
      "num_input_tokens_seen": 366252160,
      "step": 359700
    },
    {
      "epoch": 7.083932192711307,
      "grad_norm": 5.970108509063721,
      "learning_rate": 5.435262368914207e-07,
      "loss": 1.9574,
      "num_input_tokens_seen": 366352440,
      "step": 359800
    },
    {
      "epoch": 7.08590104546081,
      "grad_norm": 1.8079566955566406,
      "learning_rate": 5.433208354298413e-07,
      "loss": 1.9789,
      "num_input_tokens_seen": 366454840,
      "step": 359900
    },
    {
      "epoch": 7.087869898210313,
      "grad_norm": 1.9190999269485474,
      "learning_rate": 5.43115426601919e-07,
      "loss": 2.0061,
      "num_input_tokens_seen": 366553904,
      "step": 360000
    },
    {
      "epoch": 7.0898387509598155,
      "grad_norm": 1.6828067302703857,
      "learning_rate": 5.429100104425823e-07,
      "loss": 2.0108,
      "num_input_tokens_seen": 366655624,
      "step": 360100
    },
    {
      "epoch": 7.091807603709318,
      "grad_norm": 1.8801361322402954,
      "learning_rate": 5.427045869867601e-07,
      "loss": 1.971,
      "num_input_tokens_seen": 366757520,
      "step": 360200
    },
    {
      "epoch": 7.093776456458821,
      "grad_norm": 1.888176441192627,
      "learning_rate": 5.424991562693831e-07,
      "loss": 1.9798,
      "num_input_tokens_seen": 366859920,
      "step": 360300
    },
    {
      "epoch": 7.095745309208325,
      "grad_norm": 1.9752569198608398,
      "learning_rate": 5.422937183253828e-07,
      "loss": 2.05,
      "num_input_tokens_seen": 366961816,
      "step": 360400
    },
    {
      "epoch": 7.0977141619578275,
      "grad_norm": 3.1653244495391846,
      "learning_rate": 5.420882731896926e-07,
      "loss": 2.0098,
      "num_input_tokens_seen": 367063400,
      "step": 360500
    },
    {
      "epoch": 7.09968301470733,
      "grad_norm": 1.7968275547027588,
      "learning_rate": 5.418828208972465e-07,
      "loss": 1.9966,
      "num_input_tokens_seen": 367165800,
      "step": 360600
    },
    {
      "epoch": 7.101651867456833,
      "grad_norm": 2.205883264541626,
      "learning_rate": 5.416773614829802e-07,
      "loss": 1.9548,
      "num_input_tokens_seen": 367267584,
      "step": 360700
    },
    {
      "epoch": 7.103620720206336,
      "grad_norm": 2.1103649139404297,
      "learning_rate": 5.414718949818299e-07,
      "loss": 1.9843,
      "num_input_tokens_seen": 367369216,
      "step": 360800
    },
    {
      "epoch": 7.1055895729558385,
      "grad_norm": 1.872778296470642,
      "learning_rate": 5.412664214287337e-07,
      "loss": 1.9398,
      "num_input_tokens_seen": 367471616,
      "step": 360900
    },
    {
      "epoch": 7.107558425705341,
      "grad_norm": 1.6096558570861816,
      "learning_rate": 5.410609408586305e-07,
      "loss": 2.0362,
      "num_input_tokens_seen": 367571432,
      "step": 361000
    },
    {
      "epoch": 7.109527278454844,
      "grad_norm": 2.4558537006378174,
      "learning_rate": 5.408554533064607e-07,
      "loss": 1.948,
      "num_input_tokens_seen": 367673160,
      "step": 361100
    },
    {
      "epoch": 7.111496131204348,
      "grad_norm": 1.9940905570983887,
      "learning_rate": 5.406499588071657e-07,
      "loss": 1.9899,
      "num_input_tokens_seen": 367774728,
      "step": 361200
    },
    {
      "epoch": 7.11346498395385,
      "grad_norm": 2.367253541946411,
      "learning_rate": 5.404444573956881e-07,
      "loss": 1.977,
      "num_input_tokens_seen": 367876416,
      "step": 361300
    },
    {
      "epoch": 7.115433836703353,
      "grad_norm": 1.7973729372024536,
      "learning_rate": 5.402389491069714e-07,
      "loss": 1.9838,
      "num_input_tokens_seen": 367978816,
      "step": 361400
    },
    {
      "epoch": 7.117402689452856,
      "grad_norm": 1.813769817352295,
      "learning_rate": 5.400334339759611e-07,
      "loss": 2.0028,
      "num_input_tokens_seen": 368079560,
      "step": 361500
    },
    {
      "epoch": 7.119371542202359,
      "grad_norm": 1.7249197959899902,
      "learning_rate": 5.398279120376028e-07,
      "loss": 1.9971,
      "num_input_tokens_seen": 368181960,
      "step": 361600
    },
    {
      "epoch": 7.121340394951861,
      "grad_norm": 1.7796777486801147,
      "learning_rate": 5.39622383326844e-07,
      "loss": 1.9993,
      "num_input_tokens_seen": 368284360,
      "step": 361700
    },
    {
      "epoch": 7.123309247701364,
      "grad_norm": 1.9053800106048584,
      "learning_rate": 5.394168478786333e-07,
      "loss": 2.0051,
      "num_input_tokens_seen": 368386240,
      "step": 361800
    },
    {
      "epoch": 7.125278100450867,
      "grad_norm": 2.1554689407348633,
      "learning_rate": 5.392113057279199e-07,
      "loss": 2.0154,
      "num_input_tokens_seen": 368488640,
      "step": 361900
    },
    {
      "epoch": 7.12724695320037,
      "grad_norm": 2.1221604347229004,
      "learning_rate": 5.390057569096547e-07,
      "loss": 2.0187,
      "num_input_tokens_seen": 368590752,
      "step": 362000
    },
    {
      "epoch": 7.129215805949873,
      "grad_norm": 2.1467058658599854,
      "learning_rate": 5.388002014587895e-07,
      "loss": 1.9533,
      "num_input_tokens_seen": 368693152,
      "step": 362100
    },
    {
      "epoch": 7.131184658699376,
      "grad_norm": 1.7196701765060425,
      "learning_rate": 5.385946394102774e-07,
      "loss": 2.0007,
      "num_input_tokens_seen": 368795552,
      "step": 362200
    },
    {
      "epoch": 7.133153511448879,
      "grad_norm": 1.777214527130127,
      "learning_rate": 5.383890707990722e-07,
      "loss": 1.9714,
      "num_input_tokens_seen": 368897056,
      "step": 362300
    },
    {
      "epoch": 7.135122364198382,
      "grad_norm": 2.4955103397369385,
      "learning_rate": 5.381834956601296e-07,
      "loss": 2.0043,
      "num_input_tokens_seen": 368998624,
      "step": 362400
    },
    {
      "epoch": 7.137091216947884,
      "grad_norm": 2.059335708618164,
      "learning_rate": 5.379779140284054e-07,
      "loss": 2.024,
      "num_input_tokens_seen": 369100440,
      "step": 362500
    },
    {
      "epoch": 7.139060069697387,
      "grad_norm": 1.8319780826568604,
      "learning_rate": 5.377723259388572e-07,
      "loss": 1.9471,
      "num_input_tokens_seen": 369202840,
      "step": 362600
    },
    {
      "epoch": 7.14102892244689,
      "grad_norm": 3.3601717948913574,
      "learning_rate": 5.375667314264436e-07,
      "loss": 1.968,
      "num_input_tokens_seen": 369304720,
      "step": 362700
    },
    {
      "epoch": 7.142997775196393,
      "grad_norm": 1.861990213394165,
      "learning_rate": 5.373611305261243e-07,
      "loss": 1.9488,
      "num_input_tokens_seen": 369405672,
      "step": 362800
    },
    {
      "epoch": 7.144966627945896,
      "grad_norm": 1.8643226623535156,
      "learning_rate": 5.371555232728597e-07,
      "loss": 2.0228,
      "num_input_tokens_seen": 369507456,
      "step": 362900
    },
    {
      "epoch": 7.146935480695399,
      "grad_norm": 1.7534048557281494,
      "learning_rate": 5.369499097016119e-07,
      "loss": 1.9745,
      "num_input_tokens_seen": 369609304,
      "step": 363000
    },
    {
      "epoch": 7.148904333444902,
      "grad_norm": 2.1406805515289307,
      "learning_rate": 5.367442898473435e-07,
      "loss": 1.975,
      "num_input_tokens_seen": 369711064,
      "step": 363100
    },
    {
      "epoch": 7.150873186194405,
      "grad_norm": 1.6065541505813599,
      "learning_rate": 5.365386637450187e-07,
      "loss": 2.0399,
      "num_input_tokens_seen": 369812864,
      "step": 363200
    },
    {
      "epoch": 7.152842038943907,
      "grad_norm": 2.1287107467651367,
      "learning_rate": 5.363330314296022e-07,
      "loss": 1.9809,
      "num_input_tokens_seen": 369914504,
      "step": 363300
    },
    {
      "epoch": 7.15481089169341,
      "grad_norm": 2.327932834625244,
      "learning_rate": 5.361273929360601e-07,
      "loss": 2.0227,
      "num_input_tokens_seen": 370016904,
      "step": 363400
    },
    {
      "epoch": 7.156779744442913,
      "grad_norm": 1.8001776933670044,
      "learning_rate": 5.359217482993596e-07,
      "loss": 1.9704,
      "num_input_tokens_seen": 370119304,
      "step": 363500
    },
    {
      "epoch": 7.158748597192416,
      "grad_norm": 1.7063366174697876,
      "learning_rate": 5.357160975544687e-07,
      "loss": 1.9766,
      "num_input_tokens_seen": 370221704,
      "step": 363600
    },
    {
      "epoch": 7.160717449941918,
      "grad_norm": 1.834115982055664,
      "learning_rate": 5.355104407363566e-07,
      "loss": 1.9631,
      "num_input_tokens_seen": 370323464,
      "step": 363700
    },
    {
      "epoch": 7.162686302691422,
      "grad_norm": 2.496907949447632,
      "learning_rate": 5.353047778799938e-07,
      "loss": 1.9185,
      "num_input_tokens_seen": 370425232,
      "step": 363800
    },
    {
      "epoch": 7.164655155440925,
      "grad_norm": 1.6020030975341797,
      "learning_rate": 5.350991090203511e-07,
      "loss": 2.0283,
      "num_input_tokens_seen": 370527632,
      "step": 363900
    },
    {
      "epoch": 7.1666240081904276,
      "grad_norm": 2.2522361278533936,
      "learning_rate": 5.348934341924011e-07,
      "loss": 1.9769,
      "num_input_tokens_seen": 370628736,
      "step": 364000
    },
    {
      "epoch": 7.16859286093993,
      "grad_norm": 1.9184983968734741,
      "learning_rate": 5.346877534311169e-07,
      "loss": 2.0164,
      "num_input_tokens_seen": 370730432,
      "step": 364100
    },
    {
      "epoch": 7.170561713689433,
      "grad_norm": 1.8637442588806152,
      "learning_rate": 5.344820667714727e-07,
      "loss": 1.9886,
      "num_input_tokens_seen": 370832832,
      "step": 364200
    },
    {
      "epoch": 7.172530566438936,
      "grad_norm": 1.794509768486023,
      "learning_rate": 5.342763742484439e-07,
      "loss": 2.0217,
      "num_input_tokens_seen": 370934408,
      "step": 364300
    },
    {
      "epoch": 7.174499419188439,
      "grad_norm": 1.8493796586990356,
      "learning_rate": 5.340706758970069e-07,
      "loss": 1.9404,
      "num_input_tokens_seen": 371036352,
      "step": 364400
    },
    {
      "epoch": 7.176468271937941,
      "grad_norm": 2.63429856300354,
      "learning_rate": 5.338649717521387e-07,
      "loss": 1.9922,
      "num_input_tokens_seen": 371137976,
      "step": 364500
    },
    {
      "epoch": 7.178437124687445,
      "grad_norm": 2.149338722229004,
      "learning_rate": 5.336592618488176e-07,
      "loss": 1.9469,
      "num_input_tokens_seen": 371240376,
      "step": 364600
    },
    {
      "epoch": 7.180405977436948,
      "grad_norm": 1.8350015878677368,
      "learning_rate": 5.33453546222023e-07,
      "loss": 1.9582,
      "num_input_tokens_seen": 371342776,
      "step": 364700
    },
    {
      "epoch": 7.1823748301864505,
      "grad_norm": 2.76448655128479,
      "learning_rate": 5.33247824906735e-07,
      "loss": 2.0503,
      "num_input_tokens_seen": 371445176,
      "step": 364800
    },
    {
      "epoch": 7.184343682935953,
      "grad_norm": 4.350588321685791,
      "learning_rate": 5.330420979379349e-07,
      "loss": 1.9721,
      "num_input_tokens_seen": 371546968,
      "step": 364900
    },
    {
      "epoch": 7.186312535685456,
      "grad_norm": 1.8414084911346436,
      "learning_rate": 5.328363653506047e-07,
      "loss": 1.9958,
      "num_input_tokens_seen": 371649368,
      "step": 365000
    },
    {
      "epoch": 7.188281388434959,
      "grad_norm": 1.9045623540878296,
      "learning_rate": 5.326306271797274e-07,
      "loss": 2.0003,
      "num_input_tokens_seen": 371751400,
      "step": 365100
    },
    {
      "epoch": 7.1902502411844615,
      "grad_norm": 1.8278273344039917,
      "learning_rate": 5.324248834602874e-07,
      "loss": 2.0071,
      "num_input_tokens_seen": 371853800,
      "step": 365200
    },
    {
      "epoch": 7.192219093933964,
      "grad_norm": 1.9607822895050049,
      "learning_rate": 5.322191342272692e-07,
      "loss": 1.9618,
      "num_input_tokens_seen": 371955704,
      "step": 365300
    },
    {
      "epoch": 7.194187946683468,
      "grad_norm": 2.0386648178100586,
      "learning_rate": 5.320133795156591e-07,
      "loss": 1.9965,
      "num_input_tokens_seen": 372058104,
      "step": 365400
    },
    {
      "epoch": 7.196156799432971,
      "grad_norm": 1.765509009361267,
      "learning_rate": 5.318076193604439e-07,
      "loss": 1.9829,
      "num_input_tokens_seen": 372160504,
      "step": 365500
    },
    {
      "epoch": 7.1981256521824735,
      "grad_norm": 2.096997022628784,
      "learning_rate": 5.316018537966112e-07,
      "loss": 2.0115,
      "num_input_tokens_seen": 372262128,
      "step": 365600
    },
    {
      "epoch": 7.200094504931976,
      "grad_norm": 1.8946949243545532,
      "learning_rate": 5.313960828591498e-07,
      "loss": 1.9878,
      "num_input_tokens_seen": 372364528,
      "step": 365700
    },
    {
      "epoch": 7.202063357681479,
      "grad_norm": 1.7764822244644165,
      "learning_rate": 5.311903065830494e-07,
      "loss": 1.9765,
      "num_input_tokens_seen": 372466584,
      "step": 365800
    },
    {
      "epoch": 7.204032210430982,
      "grad_norm": 19.783960342407227,
      "learning_rate": 5.309845250033004e-07,
      "loss": 2.0381,
      "num_input_tokens_seen": 372564808,
      "step": 365900
    },
    {
      "epoch": 7.2060010631804845,
      "grad_norm": 16.93162727355957,
      "learning_rate": 5.307787381548943e-07,
      "loss": 1.9692,
      "num_input_tokens_seen": 372665648,
      "step": 366000
    },
    {
      "epoch": 7.207969915929987,
      "grad_norm": 2.239443778991699,
      "learning_rate": 5.305729460728233e-07,
      "loss": 2.0435,
      "num_input_tokens_seen": 372767024,
      "step": 366100
    },
    {
      "epoch": 7.209938768679491,
      "grad_norm": 1.760794758796692,
      "learning_rate": 5.303671487920807e-07,
      "loss": 1.9828,
      "num_input_tokens_seen": 372868616,
      "step": 366200
    },
    {
      "epoch": 7.211907621428994,
      "grad_norm": 1.9899064302444458,
      "learning_rate": 5.301613463476606e-07,
      "loss": 2.0166,
      "num_input_tokens_seen": 372969136,
      "step": 366300
    },
    {
      "epoch": 7.213876474178496,
      "grad_norm": 1.6896626949310303,
      "learning_rate": 5.299555387745579e-07,
      "loss": 1.9703,
      "num_input_tokens_seen": 373070784,
      "step": 366400
    },
    {
      "epoch": 7.215845326927999,
      "grad_norm": 2.2399237155914307,
      "learning_rate": 5.297497261077686e-07,
      "loss": 1.9703,
      "num_input_tokens_seen": 373173184,
      "step": 366500
    },
    {
      "epoch": 7.217814179677502,
      "grad_norm": 2.2764811515808105,
      "learning_rate": 5.295439083822891e-07,
      "loss": 2.0189,
      "num_input_tokens_seen": 373274960,
      "step": 366600
    },
    {
      "epoch": 7.219783032427005,
      "grad_norm": 3.2627921104431152,
      "learning_rate": 5.293380856331173e-07,
      "loss": 1.9366,
      "num_input_tokens_seen": 373377360,
      "step": 366700
    },
    {
      "epoch": 7.2217518851765075,
      "grad_norm": 2.1788229942321777,
      "learning_rate": 5.291322578952514e-07,
      "loss": 2.0299,
      "num_input_tokens_seen": 373479760,
      "step": 366800
    },
    {
      "epoch": 7.22372073792601,
      "grad_norm": 2.0692787170410156,
      "learning_rate": 5.28926425203691e-07,
      "loss": 2.0199,
      "num_input_tokens_seen": 373582160,
      "step": 366900
    },
    {
      "epoch": 7.225689590675513,
      "grad_norm": 2.1803781986236572,
      "learning_rate": 5.287205875934357e-07,
      "loss": 2.0034,
      "num_input_tokens_seen": 373683944,
      "step": 367000
    },
    {
      "epoch": 7.227658443425017,
      "grad_norm": 1.8231624364852905,
      "learning_rate": 5.285147450994868e-07,
      "loss": 1.9944,
      "num_input_tokens_seen": 373786344,
      "step": 367100
    },
    {
      "epoch": 7.229627296174519,
      "grad_norm": 2.024059534072876,
      "learning_rate": 5.283088977568459e-07,
      "loss": 1.9888,
      "num_input_tokens_seen": 373887952,
      "step": 367200
    },
    {
      "epoch": 7.231596148924022,
      "grad_norm": 1.960951566696167,
      "learning_rate": 5.281030456005159e-07,
      "loss": 1.9572,
      "num_input_tokens_seen": 373990352,
      "step": 367300
    },
    {
      "epoch": 7.233565001673525,
      "grad_norm": 1.7448127269744873,
      "learning_rate": 5.278971886654999e-07,
      "loss": 1.9775,
      "num_input_tokens_seen": 374091760,
      "step": 367400
    },
    {
      "epoch": 7.235533854423028,
      "grad_norm": 1.9026989936828613,
      "learning_rate": 5.276913269868024e-07,
      "loss": 1.982,
      "num_input_tokens_seen": 374193528,
      "step": 367500
    },
    {
      "epoch": 7.23750270717253,
      "grad_norm": 2.157411575317383,
      "learning_rate": 5.274854605994281e-07,
      "loss": 1.9915,
      "num_input_tokens_seen": 374294808,
      "step": 367600
    },
    {
      "epoch": 7.239471559922033,
      "grad_norm": 2.068803310394287,
      "learning_rate": 5.272795895383831e-07,
      "loss": 1.9947,
      "num_input_tokens_seen": 374396448,
      "step": 367700
    },
    {
      "epoch": 7.241440412671536,
      "grad_norm": 2.757269859313965,
      "learning_rate": 5.27073713838674e-07,
      "loss": 1.9894,
      "num_input_tokens_seen": 374498288,
      "step": 367800
    },
    {
      "epoch": 7.24340926542104,
      "grad_norm": 1.7236621379852295,
      "learning_rate": 5.268678335353084e-07,
      "loss": 1.9905,
      "num_input_tokens_seen": 374599984,
      "step": 367900
    },
    {
      "epoch": 7.245378118170542,
      "grad_norm": 2.0685951709747314,
      "learning_rate": 5.266619486632941e-07,
      "loss": 1.9732,
      "num_input_tokens_seen": 374702384,
      "step": 368000
    },
    {
      "epoch": 7.247346970920045,
      "grad_norm": 1.782249093055725,
      "learning_rate": 5.264560592576404e-07,
      "loss": 1.9993,
      "num_input_tokens_seen": 374804784,
      "step": 368100
    },
    {
      "epoch": 7.249315823669548,
      "grad_norm": 2.0508511066436768,
      "learning_rate": 5.26250165353357e-07,
      "loss": 2.0149,
      "num_input_tokens_seen": 374906256,
      "step": 368200
    },
    {
      "epoch": 7.251284676419051,
      "grad_norm": 1.8285428285598755,
      "learning_rate": 5.260442669854544e-07,
      "loss": 1.958,
      "num_input_tokens_seen": 375008656,
      "step": 368300
    },
    {
      "epoch": 7.253253529168553,
      "grad_norm": 1.7665964365005493,
      "learning_rate": 5.258383641889438e-07,
      "loss": 1.9831,
      "num_input_tokens_seen": 375109680,
      "step": 368400
    },
    {
      "epoch": 7.255222381918056,
      "grad_norm": 1.7450065612792969,
      "learning_rate": 5.256324569988373e-07,
      "loss": 1.988,
      "num_input_tokens_seen": 375212080,
      "step": 368500
    },
    {
      "epoch": 7.257191234667559,
      "grad_norm": 1.9241122007369995,
      "learning_rate": 5.254265454501476e-07,
      "loss": 1.9472,
      "num_input_tokens_seen": 375314000,
      "step": 368600
    },
    {
      "epoch": 7.259160087417062,
      "grad_norm": 1.618250846862793,
      "learning_rate": 5.252206295778884e-07,
      "loss": 1.983,
      "num_input_tokens_seen": 375415704,
      "step": 368700
    },
    {
      "epoch": 7.261128940166565,
      "grad_norm": 2.255683422088623,
      "learning_rate": 5.250147094170738e-07,
      "loss": 2.0033,
      "num_input_tokens_seen": 375518104,
      "step": 368800
    },
    {
      "epoch": 7.263097792916068,
      "grad_norm": 1.918139100074768,
      "learning_rate": 5.24808785002719e-07,
      "loss": 1.9646,
      "num_input_tokens_seen": 375620504,
      "step": 368900
    },
    {
      "epoch": 7.265066645665571,
      "grad_norm": 2.014235496520996,
      "learning_rate": 5.246028563698394e-07,
      "loss": 1.9885,
      "num_input_tokens_seen": 375722712,
      "step": 369000
    },
    {
      "epoch": 7.267035498415074,
      "grad_norm": 2.120821714401245,
      "learning_rate": 5.243969235534517e-07,
      "loss": 1.9718,
      "num_input_tokens_seen": 375824536,
      "step": 369100
    },
    {
      "epoch": 7.269004351164576,
      "grad_norm": 1.9699757099151611,
      "learning_rate": 5.241909865885728e-07,
      "loss": 1.9586,
      "num_input_tokens_seen": 375926936,
      "step": 369200
    },
    {
      "epoch": 7.270973203914079,
      "grad_norm": 1.9826579093933105,
      "learning_rate": 5.239850455102208e-07,
      "loss": 1.9874,
      "num_input_tokens_seen": 376028128,
      "step": 369300
    },
    {
      "epoch": 7.272942056663582,
      "grad_norm": 1.8360637426376343,
      "learning_rate": 5.23779100353414e-07,
      "loss": 1.9952,
      "num_input_tokens_seen": 376129192,
      "step": 369400
    },
    {
      "epoch": 7.274910909413085,
      "grad_norm": 1.8102113008499146,
      "learning_rate": 5.235731511531718e-07,
      "loss": 1.9623,
      "num_input_tokens_seen": 376231592,
      "step": 369500
    },
    {
      "epoch": 7.276879762162588,
      "grad_norm": 1.8897552490234375,
      "learning_rate": 5.23367197944514e-07,
      "loss": 2.0113,
      "num_input_tokens_seen": 376333552,
      "step": 369600
    },
    {
      "epoch": 7.278848614912091,
      "grad_norm": 2.0573432445526123,
      "learning_rate": 5.231612407624614e-07,
      "loss": 1.9529,
      "num_input_tokens_seen": 376435672,
      "step": 369700
    },
    {
      "epoch": 7.280817467661594,
      "grad_norm": 1.9561767578125,
      "learning_rate": 5.229552796420351e-07,
      "loss": 2.0072,
      "num_input_tokens_seen": 376536784,
      "step": 369800
    },
    {
      "epoch": 7.2827863204110965,
      "grad_norm": 1.959929347038269,
      "learning_rate": 5.227493146182571e-07,
      "loss": 2.0258,
      "num_input_tokens_seen": 376637344,
      "step": 369900
    },
    {
      "epoch": 7.284755173160599,
      "grad_norm": 2.0031423568725586,
      "learning_rate": 5.225433457261501e-07,
      "loss": 1.9895,
      "num_input_tokens_seen": 376738952,
      "step": 370000
    },
    {
      "epoch": 7.286724025910102,
      "grad_norm": 2.1231939792633057,
      "learning_rate": 5.223373730007371e-07,
      "loss": 1.9619,
      "num_input_tokens_seen": 376840224,
      "step": 370100
    },
    {
      "epoch": 7.288692878659605,
      "grad_norm": 2.225644826889038,
      "learning_rate": 5.221313964770424e-07,
      "loss": 1.9698,
      "num_input_tokens_seen": 376941584,
      "step": 370200
    },
    {
      "epoch": 7.290661731409108,
      "grad_norm": 2.6365232467651367,
      "learning_rate": 5.219254161900903e-07,
      "loss": 1.9748,
      "num_input_tokens_seen": 377043984,
      "step": 370300
    },
    {
      "epoch": 7.29263058415861,
      "grad_norm": 1.9064011573791504,
      "learning_rate": 5.217194321749064e-07,
      "loss": 1.9599,
      "num_input_tokens_seen": 377146384,
      "step": 370400
    },
    {
      "epoch": 7.294599436908114,
      "grad_norm": 2.127610921859741,
      "learning_rate": 5.215134444665157e-07,
      "loss": 2.0045,
      "num_input_tokens_seen": 377248784,
      "step": 370500
    },
    {
      "epoch": 7.296568289657617,
      "grad_norm": 1.8829401731491089,
      "learning_rate": 5.213074530999457e-07,
      "loss": 1.9863,
      "num_input_tokens_seen": 377349416,
      "step": 370600
    },
    {
      "epoch": 7.2985371424071195,
      "grad_norm": 2.1090333461761475,
      "learning_rate": 5.211014581102227e-07,
      "loss": 1.9816,
      "num_input_tokens_seen": 377450296,
      "step": 370700
    },
    {
      "epoch": 7.300505995156622,
      "grad_norm": 1.9458856582641602,
      "learning_rate": 5.208954595323749e-07,
      "loss": 2.0244,
      "num_input_tokens_seen": 377550640,
      "step": 370800
    },
    {
      "epoch": 7.302474847906125,
      "grad_norm": 2.082719326019287,
      "learning_rate": 5.206894574014303e-07,
      "loss": 1.9991,
      "num_input_tokens_seen": 377652896,
      "step": 370900
    },
    {
      "epoch": 7.304443700655628,
      "grad_norm": 16.064754486083984,
      "learning_rate": 5.204834517524182e-07,
      "loss": 1.9665,
      "num_input_tokens_seen": 377753784,
      "step": 371000
    },
    {
      "epoch": 7.3064125534051305,
      "grad_norm": 1.8369344472885132,
      "learning_rate": 5.202774426203678e-07,
      "loss": 1.9223,
      "num_input_tokens_seen": 377856184,
      "step": 371100
    },
    {
      "epoch": 7.308381406154634,
      "grad_norm": 1.8641718626022339,
      "learning_rate": 5.200714300403093e-07,
      "loss": 1.9927,
      "num_input_tokens_seen": 377957056,
      "step": 371200
    },
    {
      "epoch": 7.310350258904137,
      "grad_norm": 1.9015660285949707,
      "learning_rate": 5.198654140472733e-07,
      "loss": 1.9718,
      "num_input_tokens_seen": 378059456,
      "step": 371300
    },
    {
      "epoch": 7.31231911165364,
      "grad_norm": 1.9375883340835571,
      "learning_rate": 5.196593946762914e-07,
      "loss": 1.9936,
      "num_input_tokens_seen": 378161320,
      "step": 371400
    },
    {
      "epoch": 7.314287964403142,
      "grad_norm": 1.999479055404663,
      "learning_rate": 5.194533719623951e-07,
      "loss": 2.0122,
      "num_input_tokens_seen": 378262448,
      "step": 371500
    },
    {
      "epoch": 7.316256817152645,
      "grad_norm": 1.7307672500610352,
      "learning_rate": 5.192473459406171e-07,
      "loss": 1.9676,
      "num_input_tokens_seen": 378364848,
      "step": 371600
    },
    {
      "epoch": 7.318225669902148,
      "grad_norm": 2.3091683387756348,
      "learning_rate": 5.190413166459901e-07,
      "loss": 1.9577,
      "num_input_tokens_seen": 378466480,
      "step": 371700
    },
    {
      "epoch": 7.320194522651651,
      "grad_norm": 2.060296058654785,
      "learning_rate": 5.188352841135478e-07,
      "loss": 1.9613,
      "num_input_tokens_seen": 378568256,
      "step": 371800
    },
    {
      "epoch": 7.3221633754011535,
      "grad_norm": 1.7498630285263062,
      "learning_rate": 5.186292483783244e-07,
      "loss": 1.9721,
      "num_input_tokens_seen": 378669696,
      "step": 371900
    },
    {
      "epoch": 7.324132228150656,
      "grad_norm": 2.043344736099243,
      "learning_rate": 5.184232094753545e-07,
      "loss": 2.0076,
      "num_input_tokens_seen": 378772096,
      "step": 372000
    },
    {
      "epoch": 7.32610108090016,
      "grad_norm": 1.6896330118179321,
      "learning_rate": 5.182171674396732e-07,
      "loss": 1.9804,
      "num_input_tokens_seen": 378873680,
      "step": 372100
    },
    {
      "epoch": 7.328069933649663,
      "grad_norm": 1.9267531633377075,
      "learning_rate": 5.180111223063162e-07,
      "loss": 1.9435,
      "num_input_tokens_seen": 378976080,
      "step": 372200
    },
    {
      "epoch": 7.330038786399165,
      "grad_norm": 1.8462328910827637,
      "learning_rate": 5.178050741103196e-07,
      "loss": 1.9769,
      "num_input_tokens_seen": 379077664,
      "step": 372300
    },
    {
      "epoch": 7.332007639148668,
      "grad_norm": 1.8400561809539795,
      "learning_rate": 5.175990228867205e-07,
      "loss": 1.9737,
      "num_input_tokens_seen": 379179144,
      "step": 372400
    },
    {
      "epoch": 7.333976491898171,
      "grad_norm": 3.141594886779785,
      "learning_rate": 5.173929686705559e-07,
      "loss": 1.9448,
      "num_input_tokens_seen": 379281544,
      "step": 372500
    },
    {
      "epoch": 7.335945344647674,
      "grad_norm": 2.2036776542663574,
      "learning_rate": 5.171869114968639e-07,
      "loss": 2.0001,
      "num_input_tokens_seen": 379382264,
      "step": 372600
    },
    {
      "epoch": 7.337914197397176,
      "grad_norm": 2.153395414352417,
      "learning_rate": 5.169808514006822e-07,
      "loss": 1.983,
      "num_input_tokens_seen": 379484168,
      "step": 372700
    },
    {
      "epoch": 7.339883050146679,
      "grad_norm": 2.3387227058410645,
      "learning_rate": 5.167747884170502e-07,
      "loss": 1.9849,
      "num_input_tokens_seen": 379586568,
      "step": 372800
    },
    {
      "epoch": 7.341851902896183,
      "grad_norm": 2.0397276878356934,
      "learning_rate": 5.165687225810068e-07,
      "loss": 2.0222,
      "num_input_tokens_seen": 379688032,
      "step": 372900
    },
    {
      "epoch": 7.343820755645686,
      "grad_norm": 1.8556203842163086,
      "learning_rate": 5.163626539275919e-07,
      "loss": 2.0037,
      "num_input_tokens_seen": 379789992,
      "step": 373000
    },
    {
      "epoch": 7.345789608395188,
      "grad_norm": 1.7890570163726807,
      "learning_rate": 5.161565824918457e-07,
      "loss": 2.0557,
      "num_input_tokens_seen": 379891664,
      "step": 373100
    },
    {
      "epoch": 7.347758461144691,
      "grad_norm": 1.9766457080841064,
      "learning_rate": 5.15950508308809e-07,
      "loss": 2.013,
      "num_input_tokens_seen": 379992776,
      "step": 373200
    },
    {
      "epoch": 7.349727313894194,
      "grad_norm": 1.9591894149780273,
      "learning_rate": 5.157444314135229e-07,
      "loss": 1.9999,
      "num_input_tokens_seen": 380093768,
      "step": 373300
    },
    {
      "epoch": 7.351696166643697,
      "grad_norm": 2.112759828567505,
      "learning_rate": 5.15538351841029e-07,
      "loss": 1.9896,
      "num_input_tokens_seen": 380195320,
      "step": 373400
    },
    {
      "epoch": 7.353665019393199,
      "grad_norm": 2.1614561080932617,
      "learning_rate": 5.153322696263693e-07,
      "loss": 2.0226,
      "num_input_tokens_seen": 380296872,
      "step": 373500
    },
    {
      "epoch": 7.355633872142702,
      "grad_norm": 1.9520407915115356,
      "learning_rate": 5.151261848045867e-07,
      "loss": 1.9622,
      "num_input_tokens_seen": 380399272,
      "step": 373600
    },
    {
      "epoch": 7.357602724892205,
      "grad_norm": 2.2021682262420654,
      "learning_rate": 5.149200974107237e-07,
      "loss": 1.9745,
      "num_input_tokens_seen": 380499280,
      "step": 373700
    },
    {
      "epoch": 7.3595715776417086,
      "grad_norm": 2.7257330417633057,
      "learning_rate": 5.147140074798242e-07,
      "loss": 1.9719,
      "num_input_tokens_seen": 380601680,
      "step": 373800
    },
    {
      "epoch": 7.361540430391211,
      "grad_norm": 14.789177894592285,
      "learning_rate": 5.145079150469317e-07,
      "loss": 1.9557,
      "num_input_tokens_seen": 380704080,
      "step": 373900
    },
    {
      "epoch": 7.363509283140714,
      "grad_norm": 1.8463948965072632,
      "learning_rate": 5.143018201470907e-07,
      "loss": 2.0062,
      "num_input_tokens_seen": 380806480,
      "step": 374000
    },
    {
      "epoch": 7.365478135890217,
      "grad_norm": 1.9206358194351196,
      "learning_rate": 5.140957228153458e-07,
      "loss": 1.9592,
      "num_input_tokens_seen": 380908880,
      "step": 374100
    },
    {
      "epoch": 7.36744698863972,
      "grad_norm": 1.9654873609542847,
      "learning_rate": 5.13889623086742e-07,
      "loss": 2.0127,
      "num_input_tokens_seen": 381010488,
      "step": 374200
    },
    {
      "epoch": 7.369415841389222,
      "grad_norm": 1.9551275968551636,
      "learning_rate": 5.13683520996325e-07,
      "loss": 1.9735,
      "num_input_tokens_seen": 381112880,
      "step": 374300
    },
    {
      "epoch": 7.371384694138725,
      "grad_norm": 2.2033655643463135,
      "learning_rate": 5.134774165791406e-07,
      "loss": 1.9619,
      "num_input_tokens_seen": 381214616,
      "step": 374400
    },
    {
      "epoch": 7.373353546888228,
      "grad_norm": 6.106286525726318,
      "learning_rate": 5.132713098702354e-07,
      "loss": 1.9567,
      "num_input_tokens_seen": 381316440,
      "step": 374500
    },
    {
      "epoch": 7.3753223996377315,
      "grad_norm": 4.4674072265625,
      "learning_rate": 5.130652009046557e-07,
      "loss": 1.9488,
      "num_input_tokens_seen": 381418840,
      "step": 374600
    },
    {
      "epoch": 7.377291252387234,
      "grad_norm": 1.7868443727493286,
      "learning_rate": 5.12859089717449e-07,
      "loss": 1.9884,
      "num_input_tokens_seen": 381520672,
      "step": 374700
    },
    {
      "epoch": 7.379260105136737,
      "grad_norm": 1.8189754486083984,
      "learning_rate": 5.126529763436622e-07,
      "loss": 1.9947,
      "num_input_tokens_seen": 381623072,
      "step": 374800
    },
    {
      "epoch": 7.38122895788624,
      "grad_norm": 2.129924774169922,
      "learning_rate": 5.124468608183437e-07,
      "loss": 1.9858,
      "num_input_tokens_seen": 381725472,
      "step": 374900
    },
    {
      "epoch": 7.3831978106357425,
      "grad_norm": 2.500575304031372,
      "learning_rate": 5.122407431765415e-07,
      "loss": 1.9967,
      "num_input_tokens_seen": 381826600,
      "step": 375000
    },
    {
      "epoch": 7.385166663385245,
      "grad_norm": 1.8880819082260132,
      "learning_rate": 5.120346234533042e-07,
      "loss": 1.9229,
      "num_input_tokens_seen": 381929000,
      "step": 375100
    },
    {
      "epoch": 7.387135516134748,
      "grad_norm": 1.848900318145752,
      "learning_rate": 5.118285016836806e-07,
      "loss": 1.983,
      "num_input_tokens_seen": 382031400,
      "step": 375200
    },
    {
      "epoch": 7.389104368884251,
      "grad_norm": 1.8943896293640137,
      "learning_rate": 5.116223779027202e-07,
      "loss": 1.9336,
      "num_input_tokens_seen": 382133800,
      "step": 375300
    },
    {
      "epoch": 7.391073221633754,
      "grad_norm": 1.9485511779785156,
      "learning_rate": 5.114162521454724e-07,
      "loss": 1.9777,
      "num_input_tokens_seen": 382236200,
      "step": 375400
    },
    {
      "epoch": 7.393042074383257,
      "grad_norm": 1.9208974838256836,
      "learning_rate": 5.112101244469872e-07,
      "loss": 1.9951,
      "num_input_tokens_seen": 382338600,
      "step": 375500
    },
    {
      "epoch": 7.39501092713276,
      "grad_norm": 1.8350746631622314,
      "learning_rate": 5.110039948423148e-07,
      "loss": 1.9833,
      "num_input_tokens_seen": 382439760,
      "step": 375600
    },
    {
      "epoch": 7.396979779882263,
      "grad_norm": 1.8583351373672485,
      "learning_rate": 5.107978633665061e-07,
      "loss": 2.0236,
      "num_input_tokens_seen": 382541456,
      "step": 375700
    },
    {
      "epoch": 7.3989486326317655,
      "grad_norm": 2.133082389831543,
      "learning_rate": 5.105917300546114e-07,
      "loss": 1.9868,
      "num_input_tokens_seen": 382643720,
      "step": 375800
    },
    {
      "epoch": 7.400917485381268,
      "grad_norm": 1.9567073583602905,
      "learning_rate": 5.103855949416828e-07,
      "loss": 2.014,
      "num_input_tokens_seen": 382745352,
      "step": 375900
    },
    {
      "epoch": 7.402886338130771,
      "grad_norm": 2.0236613750457764,
      "learning_rate": 5.101794580627711e-07,
      "loss": 1.9616,
      "num_input_tokens_seen": 382847752,
      "step": 376000
    },
    {
      "epoch": 7.404855190880274,
      "grad_norm": 1.8273890018463135,
      "learning_rate": 5.099733194529285e-07,
      "loss": 1.988,
      "num_input_tokens_seen": 382949488,
      "step": 376100
    },
    {
      "epoch": 7.4068240436297765,
      "grad_norm": 1.6872155666351318,
      "learning_rate": 5.09767179147207e-07,
      "loss": 1.9586,
      "num_input_tokens_seen": 383051888,
      "step": 376200
    },
    {
      "epoch": 7.40879289637928,
      "grad_norm": 1.778103232383728,
      "learning_rate": 5.095610371806593e-07,
      "loss": 2.0053,
      "num_input_tokens_seen": 383152944,
      "step": 376300
    },
    {
      "epoch": 7.410761749128783,
      "grad_norm": 2.087170124053955,
      "learning_rate": 5.093548935883374e-07,
      "loss": 2.0005,
      "num_input_tokens_seen": 383255344,
      "step": 376400
    },
    {
      "epoch": 7.412730601878286,
      "grad_norm": 1.8912262916564941,
      "learning_rate": 5.091487484052952e-07,
      "loss": 2.0023,
      "num_input_tokens_seen": 383357744,
      "step": 376500
    },
    {
      "epoch": 7.4146994546277885,
      "grad_norm": 1.964666485786438,
      "learning_rate": 5.089426016665854e-07,
      "loss": 1.9713,
      "num_input_tokens_seen": 383458144,
      "step": 376600
    },
    {
      "epoch": 7.416668307377291,
      "grad_norm": 1.9928014278411865,
      "learning_rate": 5.087364534072615e-07,
      "loss": 1.9759,
      "num_input_tokens_seen": 383560544,
      "step": 376700
    },
    {
      "epoch": 7.418637160126794,
      "grad_norm": 1.9334039688110352,
      "learning_rate": 5.085303036623773e-07,
      "loss": 2.0017,
      "num_input_tokens_seen": 383662720,
      "step": 376800
    },
    {
      "epoch": 7.420606012876297,
      "grad_norm": 2.05747389793396,
      "learning_rate": 5.08324152466987e-07,
      "loss": 2.0033,
      "num_input_tokens_seen": 383764568,
      "step": 376900
    },
    {
      "epoch": 7.4225748656257995,
      "grad_norm": 1.861793041229248,
      "learning_rate": 5.081179998561448e-07,
      "loss": 1.9831,
      "num_input_tokens_seen": 383866968,
      "step": 377000
    },
    {
      "epoch": 7.424543718375302,
      "grad_norm": 1.9125750064849854,
      "learning_rate": 5.079118458649053e-07,
      "loss": 2.0068,
      "num_input_tokens_seen": 383968736,
      "step": 377100
    },
    {
      "epoch": 7.426512571124806,
      "grad_norm": 2.0038869380950928,
      "learning_rate": 5.077056905283231e-07,
      "loss": 1.991,
      "num_input_tokens_seen": 384071136,
      "step": 377200
    },
    {
      "epoch": 7.428481423874309,
      "grad_norm": 1.9794186353683472,
      "learning_rate": 5.074995338814531e-07,
      "loss": 1.9528,
      "num_input_tokens_seen": 384173536,
      "step": 377300
    },
    {
      "epoch": 7.430450276623811,
      "grad_norm": 1.9688060283660889,
      "learning_rate": 5.072933759593506e-07,
      "loss": 1.9709,
      "num_input_tokens_seen": 384274624,
      "step": 377400
    },
    {
      "epoch": 7.432419129373314,
      "grad_norm": 2.9141573905944824,
      "learning_rate": 5.070872167970712e-07,
      "loss": 1.9723,
      "num_input_tokens_seen": 384377024,
      "step": 377500
    },
    {
      "epoch": 7.434387982122817,
      "grad_norm": 1.8954172134399414,
      "learning_rate": 5.068810564296702e-07,
      "loss": 1.9585,
      "num_input_tokens_seen": 384479424,
      "step": 377600
    },
    {
      "epoch": 7.43635683487232,
      "grad_norm": 1.6338471174240112,
      "learning_rate": 5.066748948922039e-07,
      "loss": 1.9816,
      "num_input_tokens_seen": 384581824,
      "step": 377700
    },
    {
      "epoch": 7.4383256876218224,
      "grad_norm": 2.2469611167907715,
      "learning_rate": 5.064687322197277e-07,
      "loss": 1.9873,
      "num_input_tokens_seen": 384683360,
      "step": 377800
    },
    {
      "epoch": 7.440294540371326,
      "grad_norm": 1.7577176094055176,
      "learning_rate": 5.062625684472981e-07,
      "loss": 2.0197,
      "num_input_tokens_seen": 384784528,
      "step": 377900
    },
    {
      "epoch": 7.442263393120829,
      "grad_norm": 1.6048203706741333,
      "learning_rate": 5.060564036099718e-07,
      "loss": 1.9432,
      "num_input_tokens_seen": 384886928,
      "step": 378000
    },
    {
      "epoch": 7.444232245870332,
      "grad_norm": 1.8339276313781738,
      "learning_rate": 5.05850237742805e-07,
      "loss": 1.9394,
      "num_input_tokens_seen": 384989328,
      "step": 378100
    },
    {
      "epoch": 7.446201098619834,
      "grad_norm": 1.7405064105987549,
      "learning_rate": 5.056440708808547e-07,
      "loss": 1.9962,
      "num_input_tokens_seen": 385091728,
      "step": 378200
    },
    {
      "epoch": 7.448169951369337,
      "grad_norm": 2.010094165802002,
      "learning_rate": 5.054379030591777e-07,
      "loss": 1.9904,
      "num_input_tokens_seen": 385193256,
      "step": 378300
    },
    {
      "epoch": 7.45013880411884,
      "grad_norm": 1.825519323348999,
      "learning_rate": 5.052317343128312e-07,
      "loss": 2.005,
      "num_input_tokens_seen": 385294232,
      "step": 378400
    },
    {
      "epoch": 7.452107656868343,
      "grad_norm": 1.987107515335083,
      "learning_rate": 5.050255646768723e-07,
      "loss": 1.9861,
      "num_input_tokens_seen": 385396632,
      "step": 378500
    },
    {
      "epoch": 7.454076509617845,
      "grad_norm": 1.8715169429779053,
      "learning_rate": 5.048193941863586e-07,
      "loss": 1.9449,
      "num_input_tokens_seen": 385499032,
      "step": 378600
    },
    {
      "epoch": 7.456045362367348,
      "grad_norm": 1.9312773942947388,
      "learning_rate": 5.046132228763476e-07,
      "loss": 2.0466,
      "num_input_tokens_seen": 385601432,
      "step": 378700
    },
    {
      "epoch": 7.458014215116852,
      "grad_norm": 1.7639271020889282,
      "learning_rate": 5.044070507818968e-07,
      "loss": 1.9658,
      "num_input_tokens_seen": 385702968,
      "step": 378800
    },
    {
      "epoch": 7.459983067866355,
      "grad_norm": 1.7868900299072266,
      "learning_rate": 5.042008779380642e-07,
      "loss": 1.98,
      "num_input_tokens_seen": 385805368,
      "step": 378900
    },
    {
      "epoch": 7.461951920615857,
      "grad_norm": 2.0073390007019043,
      "learning_rate": 5.039947043799078e-07,
      "loss": 1.9442,
      "num_input_tokens_seen": 385906328,
      "step": 379000
    },
    {
      "epoch": 7.46392077336536,
      "grad_norm": 3.7053864002227783,
      "learning_rate": 5.037885301424857e-07,
      "loss": 1.9677,
      "num_input_tokens_seen": 386007952,
      "step": 379100
    },
    {
      "epoch": 7.465889626114863,
      "grad_norm": 2.0953264236450195,
      "learning_rate": 5.035823552608561e-07,
      "loss": 2.0439,
      "num_input_tokens_seen": 386109376,
      "step": 379200
    },
    {
      "epoch": 7.467858478864366,
      "grad_norm": 2.27166748046875,
      "learning_rate": 5.033761797700769e-07,
      "loss": 2.0019,
      "num_input_tokens_seen": 386211112,
      "step": 379300
    },
    {
      "epoch": 7.469827331613868,
      "grad_norm": 1.8703604936599731,
      "learning_rate": 5.031700037052071e-07,
      "loss": 1.9618,
      "num_input_tokens_seen": 386313512,
      "step": 379400
    },
    {
      "epoch": 7.471796184363371,
      "grad_norm": 1.7375736236572266,
      "learning_rate": 5.029638271013047e-07,
      "loss": 1.9814,
      "num_input_tokens_seen": 386415912,
      "step": 379500
    },
    {
      "epoch": 7.473765037112875,
      "grad_norm": 1.7769442796707153,
      "learning_rate": 5.027576499934286e-07,
      "loss": 1.9812,
      "num_input_tokens_seen": 386518312,
      "step": 379600
    },
    {
      "epoch": 7.4757338898623775,
      "grad_norm": 1.9671293497085571,
      "learning_rate": 5.025514724166373e-07,
      "loss": 2.0277,
      "num_input_tokens_seen": 386620112,
      "step": 379700
    },
    {
      "epoch": 7.47770274261188,
      "grad_norm": 1.70367431640625,
      "learning_rate": 5.023452944059896e-07,
      "loss": 1.9941,
      "num_input_tokens_seen": 386721640,
      "step": 379800
    },
    {
      "epoch": 7.479671595361383,
      "grad_norm": 1.8726248741149902,
      "learning_rate": 5.021391159965444e-07,
      "loss": 2.0435,
      "num_input_tokens_seen": 386824040,
      "step": 379900
    },
    {
      "epoch": 7.481640448110886,
      "grad_norm": 1.744035005569458,
      "learning_rate": 5.019329372233607e-07,
      "loss": 1.9917,
      "num_input_tokens_seen": 386925888,
      "step": 380000
    },
    {
      "epoch": 7.483609300860389,
      "grad_norm": 2.1568663120269775,
      "learning_rate": 5.017267581214971e-07,
      "loss": 1.9669,
      "num_input_tokens_seen": 387028288,
      "step": 380100
    },
    {
      "epoch": 7.485578153609891,
      "grad_norm": 1.7321265935897827,
      "learning_rate": 5.015205787260131e-07,
      "loss": 2.0032,
      "num_input_tokens_seen": 387129536,
      "step": 380200
    },
    {
      "epoch": 7.487547006359394,
      "grad_norm": 2.007984161376953,
      "learning_rate": 5.013143990719674e-07,
      "loss": 1.98,
      "num_input_tokens_seen": 387231232,
      "step": 380300
    },
    {
      "epoch": 7.489515859108897,
      "grad_norm": 2.950364589691162,
      "learning_rate": 5.011082191944193e-07,
      "loss": 1.9772,
      "num_input_tokens_seen": 387333184,
      "step": 380400
    },
    {
      "epoch": 7.4914847118584005,
      "grad_norm": 2.139864683151245,
      "learning_rate": 5.009020391284277e-07,
      "loss": 1.9928,
      "num_input_tokens_seen": 387434408,
      "step": 380500
    },
    {
      "epoch": 7.493453564607903,
      "grad_norm": 1.9698947668075562,
      "learning_rate": 5.006958589090521e-07,
      "loss": 1.9968,
      "num_input_tokens_seen": 387536808,
      "step": 380600
    },
    {
      "epoch": 7.495422417357406,
      "grad_norm": 1.985487461090088,
      "learning_rate": 5.004896785713515e-07,
      "loss": 1.9626,
      "num_input_tokens_seen": 387639208,
      "step": 380700
    },
    {
      "epoch": 7.497391270106909,
      "grad_norm": 2.3485677242279053,
      "learning_rate": 5.002834981503852e-07,
      "loss": 2.0093,
      "num_input_tokens_seen": 387741112,
      "step": 380800
    },
    {
      "epoch": 7.4993601228564115,
      "grad_norm": 3.384591817855835,
      "learning_rate": 5.000773176812125e-07,
      "loss": 2.0292,
      "num_input_tokens_seen": 387842600,
      "step": 380900
    },
    {
      "epoch": 7.501328975605914,
      "grad_norm": 1.7954802513122559,
      "learning_rate": 4.998711371988923e-07,
      "loss": 1.9622,
      "num_input_tokens_seen": 387943632,
      "step": 381000
    },
    {
      "epoch": 7.503297828355417,
      "grad_norm": 1.9172871112823486,
      "learning_rate": 4.996649567384845e-07,
      "loss": 2.0529,
      "num_input_tokens_seen": 388044400,
      "step": 381100
    },
    {
      "epoch": 7.50526668110492,
      "grad_norm": 3.3087704181671143,
      "learning_rate": 4.994587763350477e-07,
      "loss": 1.9565,
      "num_input_tokens_seen": 388145936,
      "step": 381200
    },
    {
      "epoch": 7.507235533854423,
      "grad_norm": 1.7594445943832397,
      "learning_rate": 4.992525960236417e-07,
      "loss": 1.9969,
      "num_input_tokens_seen": 388248336,
      "step": 381300
    },
    {
      "epoch": 7.509204386603926,
      "grad_norm": 1.5387449264526367,
      "learning_rate": 4.990464158393252e-07,
      "loss": 1.956,
      "num_input_tokens_seen": 388350736,
      "step": 381400
    },
    {
      "epoch": 7.511173239353429,
      "grad_norm": 2.319288969039917,
      "learning_rate": 4.988402358171577e-07,
      "loss": 1.9539,
      "num_input_tokens_seen": 388452512,
      "step": 381500
    },
    {
      "epoch": 7.513142092102932,
      "grad_norm": 2.0414860248565674,
      "learning_rate": 4.986340559921984e-07,
      "loss": 2.0061,
      "num_input_tokens_seen": 388554248,
      "step": 381600
    },
    {
      "epoch": 7.5151109448524345,
      "grad_norm": 5.795549392700195,
      "learning_rate": 4.984278763995062e-07,
      "loss": 1.9892,
      "num_input_tokens_seen": 388656648,
      "step": 381700
    },
    {
      "epoch": 7.517079797601937,
      "grad_norm": 1.7130744457244873,
      "learning_rate": 4.982216970741406e-07,
      "loss": 1.98,
      "num_input_tokens_seen": 388757640,
      "step": 381800
    },
    {
      "epoch": 7.51904865035144,
      "grad_norm": 2.2430741786956787,
      "learning_rate": 4.980155180511602e-07,
      "loss": 2.0105,
      "num_input_tokens_seen": 388860040,
      "step": 381900
    },
    {
      "epoch": 7.521017503100943,
      "grad_norm": 2.157201051712036,
      "learning_rate": 4.978093393656243e-07,
      "loss": 2.0139,
      "num_input_tokens_seen": 388960760,
      "step": 382000
    },
    {
      "epoch": 7.5229863558504455,
      "grad_norm": 2.068794012069702,
      "learning_rate": 4.976031610525918e-07,
      "loss": 1.9737,
      "num_input_tokens_seen": 389062648,
      "step": 382100
    },
    {
      "epoch": 7.524955208599949,
      "grad_norm": 2.1377761363983154,
      "learning_rate": 4.973969831471215e-07,
      "loss": 2.0052,
      "num_input_tokens_seen": 389164152,
      "step": 382200
    },
    {
      "epoch": 7.526924061349452,
      "grad_norm": 1.9068855047225952,
      "learning_rate": 4.971908056842723e-07,
      "loss": 1.9984,
      "num_input_tokens_seen": 389266552,
      "step": 382300
    },
    {
      "epoch": 7.528892914098955,
      "grad_norm": 1.7767568826675415,
      "learning_rate": 4.969846286991027e-07,
      "loss": 1.9902,
      "num_input_tokens_seen": 389368064,
      "step": 382400
    },
    {
      "epoch": 7.530861766848457,
      "grad_norm": 1.9303306341171265,
      "learning_rate": 4.967784522266718e-07,
      "loss": 1.966,
      "num_input_tokens_seen": 389469832,
      "step": 382500
    },
    {
      "epoch": 7.53283061959796,
      "grad_norm": 2.213029623031616,
      "learning_rate": 4.965722763020377e-07,
      "loss": 2.0059,
      "num_input_tokens_seen": 389570504,
      "step": 382600
    },
    {
      "epoch": 7.534799472347463,
      "grad_norm": 1.8339900970458984,
      "learning_rate": 4.963661009602594e-07,
      "loss": 1.9822,
      "num_input_tokens_seen": 389672904,
      "step": 382700
    },
    {
      "epoch": 7.536768325096966,
      "grad_norm": 1.7456740140914917,
      "learning_rate": 4.961599262363947e-07,
      "loss": 2.0026,
      "num_input_tokens_seen": 389773800,
      "step": 382800
    },
    {
      "epoch": 7.538737177846469,
      "grad_norm": 2.0507090091705322,
      "learning_rate": 4.959537521655025e-07,
      "loss": 2.0021,
      "num_input_tokens_seen": 389875648,
      "step": 382900
    },
    {
      "epoch": 7.540706030595972,
      "grad_norm": 2.145613431930542,
      "learning_rate": 4.957475787826402e-07,
      "loss": 1.9656,
      "num_input_tokens_seen": 389977280,
      "step": 383000
    },
    {
      "epoch": 7.542674883345475,
      "grad_norm": 1.9413028955459595,
      "learning_rate": 4.955414061228667e-07,
      "loss": 1.9959,
      "num_input_tokens_seen": 390078728,
      "step": 383100
    },
    {
      "epoch": 7.544643736094978,
      "grad_norm": 1.96881103515625,
      "learning_rate": 4.95335234221239e-07,
      "loss": 2.0271,
      "num_input_tokens_seen": 390181128,
      "step": 383200
    },
    {
      "epoch": 7.54661258884448,
      "grad_norm": 1.736331582069397,
      "learning_rate": 4.951290631128159e-07,
      "loss": 1.9793,
      "num_input_tokens_seen": 390283296,
      "step": 383300
    },
    {
      "epoch": 7.548581441593983,
      "grad_norm": 1.6167480945587158,
      "learning_rate": 4.949228928326541e-07,
      "loss": 1.9816,
      "num_input_tokens_seen": 390385424,
      "step": 383400
    },
    {
      "epoch": 7.550550294343486,
      "grad_norm": 1.916285514831543,
      "learning_rate": 4.947167234158116e-07,
      "loss": 1.9783,
      "num_input_tokens_seen": 390487824,
      "step": 383500
    },
    {
      "epoch": 7.552519147092989,
      "grad_norm": 2.083465099334717,
      "learning_rate": 4.945105548973459e-07,
      "loss": 1.9687,
      "num_input_tokens_seen": 390589808,
      "step": 383600
    },
    {
      "epoch": 7.554487999842491,
      "grad_norm": 1.9483658075332642,
      "learning_rate": 4.94304387312314e-07,
      "loss": 2.0251,
      "num_input_tokens_seen": 390690984,
      "step": 383700
    },
    {
      "epoch": 7.556456852591994,
      "grad_norm": 1.8473742008209229,
      "learning_rate": 4.94098220695773e-07,
      "loss": 1.9761,
      "num_input_tokens_seen": 390792944,
      "step": 383800
    },
    {
      "epoch": 7.558425705341498,
      "grad_norm": 2.0336999893188477,
      "learning_rate": 4.938920550827797e-07,
      "loss": 2.0295,
      "num_input_tokens_seen": 390892968,
      "step": 383900
    },
    {
      "epoch": 7.560394558091001,
      "grad_norm": 1.696427345275879,
      "learning_rate": 4.936858905083911e-07,
      "loss": 1.9938,
      "num_input_tokens_seen": 390995368,
      "step": 384000
    },
    {
      "epoch": 7.562363410840503,
      "grad_norm": 2.0347323417663574,
      "learning_rate": 4.934797270076633e-07,
      "loss": 1.9942,
      "num_input_tokens_seen": 391097768,
      "step": 384100
    },
    {
      "epoch": 7.564332263590006,
      "grad_norm": 1.7522668838500977,
      "learning_rate": 4.932735646156531e-07,
      "loss": 1.974,
      "num_input_tokens_seen": 391199616,
      "step": 384200
    },
    {
      "epoch": 7.566301116339509,
      "grad_norm": 1.8831778764724731,
      "learning_rate": 4.930674033674164e-07,
      "loss": 1.9561,
      "num_input_tokens_seen": 391301424,
      "step": 384300
    },
    {
      "epoch": 7.568269969089012,
      "grad_norm": 1.9880766868591309,
      "learning_rate": 4.928612432980095e-07,
      "loss": 1.9855,
      "num_input_tokens_seen": 391402448,
      "step": 384400
    },
    {
      "epoch": 7.570238821838514,
      "grad_norm": 1.921152949333191,
      "learning_rate": 4.926550844424879e-07,
      "loss": 1.9937,
      "num_input_tokens_seen": 391504024,
      "step": 384500
    },
    {
      "epoch": 7.572207674588018,
      "grad_norm": 1.8683234453201294,
      "learning_rate": 4.92448926835907e-07,
      "loss": 1.9792,
      "num_input_tokens_seen": 391605728,
      "step": 384600
    },
    {
      "epoch": 7.574176527337521,
      "grad_norm": 1.8247352838516235,
      "learning_rate": 4.922427705133227e-07,
      "loss": 1.9588,
      "num_input_tokens_seen": 391707208,
      "step": 384700
    },
    {
      "epoch": 7.5761453800870235,
      "grad_norm": 1.791880488395691,
      "learning_rate": 4.920366155097896e-07,
      "loss": 1.9357,
      "num_input_tokens_seen": 391809608,
      "step": 384800
    },
    {
      "epoch": 7.578114232836526,
      "grad_norm": 1.5522576570510864,
      "learning_rate": 4.918304618603631e-07,
      "loss": 1.9654,
      "num_input_tokens_seen": 391912008,
      "step": 384900
    },
    {
      "epoch": 7.580083085586029,
      "grad_norm": 1.8918328285217285,
      "learning_rate": 4.916243096000975e-07,
      "loss": 1.9744,
      "num_input_tokens_seen": 392014408,
      "step": 385000
    },
    {
      "epoch": 7.582051938335532,
      "grad_norm": 1.887258529663086,
      "learning_rate": 4.914181587640474e-07,
      "loss": 2.0231,
      "num_input_tokens_seen": 392116808,
      "step": 385100
    },
    {
      "epoch": 7.584020791085035,
      "grad_norm": 1.9370651245117188,
      "learning_rate": 4.912120093872671e-07,
      "loss": 2.009,
      "num_input_tokens_seen": 392219208,
      "step": 385200
    },
    {
      "epoch": 7.585989643834537,
      "grad_norm": 1.9454143047332764,
      "learning_rate": 4.910058615048106e-07,
      "loss": 2.024,
      "num_input_tokens_seen": 392321608,
      "step": 385300
    },
    {
      "epoch": 7.58795849658404,
      "grad_norm": 2.1019110679626465,
      "learning_rate": 4.907997151517312e-07,
      "loss": 1.9752,
      "num_input_tokens_seen": 392424008,
      "step": 385400
    },
    {
      "epoch": 7.589927349333543,
      "grad_norm": 2.675548791885376,
      "learning_rate": 4.905935703630827e-07,
      "loss": 2.0203,
      "num_input_tokens_seen": 392525792,
      "step": 385500
    },
    {
      "epoch": 7.5918962020830465,
      "grad_norm": 1.801592469215393,
      "learning_rate": 4.903874271739183e-07,
      "loss": 2.0493,
      "num_input_tokens_seen": 392627240,
      "step": 385600
    },
    {
      "epoch": 7.593865054832549,
      "grad_norm": 1.9246045351028442,
      "learning_rate": 4.901812856192907e-07,
      "loss": 1.9721,
      "num_input_tokens_seen": 392728072,
      "step": 385700
    },
    {
      "epoch": 7.595833907582052,
      "grad_norm": 1.9143521785736084,
      "learning_rate": 4.89975145734253e-07,
      "loss": 1.958,
      "num_input_tokens_seen": 392829648,
      "step": 385800
    },
    {
      "epoch": 7.597802760331555,
      "grad_norm": 2.024423837661743,
      "learning_rate": 4.897690075538569e-07,
      "loss": 1.9792,
      "num_input_tokens_seen": 392932048,
      "step": 385900
    },
    {
      "epoch": 7.5997716130810575,
      "grad_norm": 1.9278972148895264,
      "learning_rate": 4.89562871113155e-07,
      "loss": 1.9674,
      "num_input_tokens_seen": 393033064,
      "step": 386000
    },
    {
      "epoch": 7.60174046583056,
      "grad_norm": 2.0642013549804688,
      "learning_rate": 4.893567364471987e-07,
      "loss": 1.9564,
      "num_input_tokens_seen": 393134944,
      "step": 386100
    },
    {
      "epoch": 7.603709318580063,
      "grad_norm": 2.1293320655822754,
      "learning_rate": 4.891506035910397e-07,
      "loss": 1.9998,
      "num_input_tokens_seen": 393235272,
      "step": 386200
    },
    {
      "epoch": 7.605678171329567,
      "grad_norm": 1.9950129985809326,
      "learning_rate": 4.889444725797289e-07,
      "loss": 1.9973,
      "num_input_tokens_seen": 393337672,
      "step": 386300
    },
    {
      "epoch": 7.6076470240790695,
      "grad_norm": 4.244419574737549,
      "learning_rate": 4.887383434483174e-07,
      "loss": 1.9796,
      "num_input_tokens_seen": 393438960,
      "step": 386400
    },
    {
      "epoch": 7.609615876828572,
      "grad_norm": 1.8216191530227661,
      "learning_rate": 4.885322162318555e-07,
      "loss": 1.9711,
      "num_input_tokens_seen": 393541000,
      "step": 386500
    },
    {
      "epoch": 7.611584729578075,
      "grad_norm": 1.9653093814849854,
      "learning_rate": 4.883260909653936e-07,
      "loss": 1.9856,
      "num_input_tokens_seen": 393643400,
      "step": 386600
    },
    {
      "epoch": 7.613553582327578,
      "grad_norm": 1.881364345550537,
      "learning_rate": 4.881199676839815e-07,
      "loss": 2.0015,
      "num_input_tokens_seen": 393745800,
      "step": 386700
    },
    {
      "epoch": 7.6155224350770805,
      "grad_norm": 2.109998941421509,
      "learning_rate": 4.879138464226685e-07,
      "loss": 1.9924,
      "num_input_tokens_seen": 393847664,
      "step": 386800
    },
    {
      "epoch": 7.617491287826583,
      "grad_norm": 1.764736533164978,
      "learning_rate": 4.877077272165042e-07,
      "loss": 1.9937,
      "num_input_tokens_seen": 393949472,
      "step": 386900
    },
    {
      "epoch": 7.619460140576086,
      "grad_norm": 1.9379056692123413,
      "learning_rate": 4.875016101005371e-07,
      "loss": 1.9419,
      "num_input_tokens_seen": 394051872,
      "step": 387000
    },
    {
      "epoch": 7.621428993325589,
      "grad_norm": 2.092463254928589,
      "learning_rate": 4.87295495109816e-07,
      "loss": 2.0024,
      "num_input_tokens_seen": 394154272,
      "step": 387100
    },
    {
      "epoch": 7.623397846075092,
      "grad_norm": 3.187513589859009,
      "learning_rate": 4.870893822793885e-07,
      "loss": 1.9831,
      "num_input_tokens_seen": 394256672,
      "step": 387200
    },
    {
      "epoch": 7.625366698824595,
      "grad_norm": 1.91410231590271,
      "learning_rate": 4.868832716443028e-07,
      "loss": 1.953,
      "num_input_tokens_seen": 394359072,
      "step": 387300
    },
    {
      "epoch": 7.627335551574098,
      "grad_norm": 1.9732996225357056,
      "learning_rate": 4.866771632396061e-07,
      "loss": 2.0144,
      "num_input_tokens_seen": 394460712,
      "step": 387400
    },
    {
      "epoch": 7.629304404323601,
      "grad_norm": 2.05133056640625,
      "learning_rate": 4.864710571003452e-07,
      "loss": 1.9703,
      "num_input_tokens_seen": 394562384,
      "step": 387500
    },
    {
      "epoch": 7.6312732570731034,
      "grad_norm": 1.8731197118759155,
      "learning_rate": 4.862649532615674e-07,
      "loss": 1.9779,
      "num_input_tokens_seen": 394664784,
      "step": 387600
    },
    {
      "epoch": 7.633242109822606,
      "grad_norm": 1.89797842502594,
      "learning_rate": 4.86058851758318e-07,
      "loss": 1.9999,
      "num_input_tokens_seen": 394765608,
      "step": 387700
    },
    {
      "epoch": 7.635210962572109,
      "grad_norm": 1.8504283428192139,
      "learning_rate": 4.858527526256436e-07,
      "loss": 1.9336,
      "num_input_tokens_seen": 394868008,
      "step": 387800
    },
    {
      "epoch": 7.637179815321612,
      "grad_norm": 2.03739595413208,
      "learning_rate": 4.856466558985891e-07,
      "loss": 1.979,
      "num_input_tokens_seen": 394969544,
      "step": 387900
    },
    {
      "epoch": 7.639148668071115,
      "grad_norm": 1.8224365711212158,
      "learning_rate": 4.854405616122e-07,
      "loss": 1.9796,
      "num_input_tokens_seen": 395071944,
      "step": 388000
    },
    {
      "epoch": 7.641117520820618,
      "grad_norm": 1.924677848815918,
      "learning_rate": 4.852344698015201e-07,
      "loss": 1.9725,
      "num_input_tokens_seen": 395171824,
      "step": 388100
    },
    {
      "epoch": 7.643086373570121,
      "grad_norm": 1.9571551084518433,
      "learning_rate": 4.850283805015945e-07,
      "loss": 1.9663,
      "num_input_tokens_seen": 395272880,
      "step": 388200
    },
    {
      "epoch": 7.645055226319624,
      "grad_norm": 2.1420321464538574,
      "learning_rate": 4.848222937474662e-07,
      "loss": 1.9934,
      "num_input_tokens_seen": 395375280,
      "step": 388300
    },
    {
      "epoch": 7.647024079069126,
      "grad_norm": 2.246696710586548,
      "learning_rate": 4.846162095741792e-07,
      "loss": 1.9712,
      "num_input_tokens_seen": 395477680,
      "step": 388400
    },
    {
      "epoch": 7.648992931818629,
      "grad_norm": 1.9481399059295654,
      "learning_rate": 4.844101280167757e-07,
      "loss": 1.9636,
      "num_input_tokens_seen": 395579736,
      "step": 388500
    },
    {
      "epoch": 7.650961784568132,
      "grad_norm": 1.8843986988067627,
      "learning_rate": 4.842040491102984e-07,
      "loss": 1.9908,
      "num_input_tokens_seen": 395681648,
      "step": 388600
    },
    {
      "epoch": 7.652930637317635,
      "grad_norm": 2.112291097640991,
      "learning_rate": 4.839979728897896e-07,
      "loss": 2.0163,
      "num_input_tokens_seen": 395782360,
      "step": 388700
    },
    {
      "epoch": 7.654899490067137,
      "grad_norm": 2.6145987510681152,
      "learning_rate": 4.837918993902903e-07,
      "loss": 1.9773,
      "num_input_tokens_seen": 395884760,
      "step": 388800
    },
    {
      "epoch": 7.656868342816641,
      "grad_norm": 1.907364010810852,
      "learning_rate": 4.83585828646842e-07,
      "loss": 2.0128,
      "num_input_tokens_seen": 395986296,
      "step": 388900
    },
    {
      "epoch": 7.658837195566144,
      "grad_norm": 1.9746867418289185,
      "learning_rate": 4.833797606944849e-07,
      "loss": 2.0028,
      "num_input_tokens_seen": 396088696,
      "step": 389000
    },
    {
      "epoch": 7.660806048315647,
      "grad_norm": 7.818428039550781,
      "learning_rate": 4.831736955682594e-07,
      "loss": 2.0206,
      "num_input_tokens_seen": 396186872,
      "step": 389100
    },
    {
      "epoch": 7.662774901065149,
      "grad_norm": 2.0519790649414062,
      "learning_rate": 4.829676333032052e-07,
      "loss": 1.966,
      "num_input_tokens_seen": 396288552,
      "step": 389200
    },
    {
      "epoch": 7.664743753814652,
      "grad_norm": 2.0729544162750244,
      "learning_rate": 4.827615739343611e-07,
      "loss": 1.9273,
      "num_input_tokens_seen": 396390520,
      "step": 389300
    },
    {
      "epoch": 7.666712606564155,
      "grad_norm": 1.708318829536438,
      "learning_rate": 4.82555517496766e-07,
      "loss": 1.9807,
      "num_input_tokens_seen": 396492920,
      "step": 389400
    },
    {
      "epoch": 7.668681459313658,
      "grad_norm": 2.070146083831787,
      "learning_rate": 4.82349464025458e-07,
      "loss": 1.9808,
      "num_input_tokens_seen": 396594784,
      "step": 389500
    },
    {
      "epoch": 7.670650312063161,
      "grad_norm": 1.8097540140151978,
      "learning_rate": 4.821434135554749e-07,
      "loss": 1.9423,
      "num_input_tokens_seen": 396696568,
      "step": 389600
    },
    {
      "epoch": 7.672619164812664,
      "grad_norm": 1.9803106784820557,
      "learning_rate": 4.819373661218535e-07,
      "loss": 1.9883,
      "num_input_tokens_seen": 396797000,
      "step": 389700
    },
    {
      "epoch": 7.674588017562167,
      "grad_norm": 2.093808650970459,
      "learning_rate": 4.817313217596309e-07,
      "loss": 1.9903,
      "num_input_tokens_seen": 396899400,
      "step": 389800
    },
    {
      "epoch": 7.67655687031167,
      "grad_norm": 1.857288122177124,
      "learning_rate": 4.815252805038427e-07,
      "loss": 1.9834,
      "num_input_tokens_seen": 397001352,
      "step": 389900
    },
    {
      "epoch": 7.678525723061172,
      "grad_norm": 1.878383994102478,
      "learning_rate": 4.813192423895249e-07,
      "loss": 1.9938,
      "num_input_tokens_seen": 397103120,
      "step": 390000
    },
    {
      "epoch": 7.680494575810675,
      "grad_norm": 2.043911933898926,
      "learning_rate": 4.811132074517123e-07,
      "loss": 1.9765,
      "num_input_tokens_seen": 397203488,
      "step": 390100
    },
    {
      "epoch": 7.682463428560178,
      "grad_norm": 2.0450711250305176,
      "learning_rate": 4.809071757254396e-07,
      "loss": 1.988,
      "num_input_tokens_seen": 397305888,
      "step": 390200
    },
    {
      "epoch": 7.684432281309681,
      "grad_norm": 1.8975472450256348,
      "learning_rate": 4.807011472457404e-07,
      "loss": 1.9803,
      "num_input_tokens_seen": 397406536,
      "step": 390300
    },
    {
      "epoch": 7.686401134059183,
      "grad_norm": 1.8599958419799805,
      "learning_rate": 4.804951220476485e-07,
      "loss": 1.9725,
      "num_input_tokens_seen": 397508936,
      "step": 390400
    },
    {
      "epoch": 7.688369986808686,
      "grad_norm": 2.0340805053710938,
      "learning_rate": 4.802891001661966e-07,
      "loss": 1.9621,
      "num_input_tokens_seen": 397611336,
      "step": 390500
    },
    {
      "epoch": 7.69033883955819,
      "grad_norm": 1.8969095945358276,
      "learning_rate": 4.800830816364167e-07,
      "loss": 1.9509,
      "num_input_tokens_seen": 397713048,
      "step": 390600
    },
    {
      "epoch": 7.6923076923076925,
      "grad_norm": 3.9885106086730957,
      "learning_rate": 4.798770664933412e-07,
      "loss": 1.9888,
      "num_input_tokens_seen": 397815448,
      "step": 390700
    },
    {
      "epoch": 7.694276545057195,
      "grad_norm": 1.7530561685562134,
      "learning_rate": 4.796710547720004e-07,
      "loss": 1.9934,
      "num_input_tokens_seen": 397917848,
      "step": 390800
    },
    {
      "epoch": 7.696245397806698,
      "grad_norm": 1.9657628536224365,
      "learning_rate": 4.794650465074256e-07,
      "loss": 1.9506,
      "num_input_tokens_seen": 398019352,
      "step": 390900
    },
    {
      "epoch": 7.698214250556201,
      "grad_norm": 1.740928053855896,
      "learning_rate": 4.792590417346462e-07,
      "loss": 2.0059,
      "num_input_tokens_seen": 398121752,
      "step": 391000
    },
    {
      "epoch": 7.7001831033057035,
      "grad_norm": 1.9120116233825684,
      "learning_rate": 4.790530404886921e-07,
      "loss": 1.9725,
      "num_input_tokens_seen": 398224152,
      "step": 391100
    },
    {
      "epoch": 7.702151956055206,
      "grad_norm": 1.888346552848816,
      "learning_rate": 4.788470428045915e-07,
      "loss": 1.9833,
      "num_input_tokens_seen": 398326000,
      "step": 391200
    },
    {
      "epoch": 7.70412080880471,
      "grad_norm": 2.0434091091156006,
      "learning_rate": 4.786410487173732e-07,
      "loss": 2.0046,
      "num_input_tokens_seen": 398427544,
      "step": 391300
    },
    {
      "epoch": 7.706089661554213,
      "grad_norm": 1.8175299167633057,
      "learning_rate": 4.78435058262064e-07,
      "loss": 2.0173,
      "num_input_tokens_seen": 398529944,
      "step": 391400
    },
    {
      "epoch": 7.7080585143037155,
      "grad_norm": 1.7560431957244873,
      "learning_rate": 4.782290714736917e-07,
      "loss": 1.9428,
      "num_input_tokens_seen": 398632344,
      "step": 391500
    },
    {
      "epoch": 7.710027367053218,
      "grad_norm": 2.2701058387756348,
      "learning_rate": 4.780230883872817e-07,
      "loss": 1.9698,
      "num_input_tokens_seen": 398734648,
      "step": 391600
    },
    {
      "epoch": 7.711996219802721,
      "grad_norm": 1.9336497783660889,
      "learning_rate": 4.778171090378603e-07,
      "loss": 2.0544,
      "num_input_tokens_seen": 398836128,
      "step": 391700
    },
    {
      "epoch": 7.713965072552224,
      "grad_norm": 1.927473783493042,
      "learning_rate": 4.776111334604525e-07,
      "loss": 2.0082,
      "num_input_tokens_seen": 398937160,
      "step": 391800
    },
    {
      "epoch": 7.7159339253017265,
      "grad_norm": 1.9075239896774292,
      "learning_rate": 4.774051616900827e-07,
      "loss": 2.0238,
      "num_input_tokens_seen": 399039032,
      "step": 391900
    },
    {
      "epoch": 7.717902778051229,
      "grad_norm": 1.7816892862319946,
      "learning_rate": 4.771991937617745e-07,
      "loss": 2.0645,
      "num_input_tokens_seen": 399141432,
      "step": 392000
    },
    {
      "epoch": 7.719871630800732,
      "grad_norm": 1.7579230070114136,
      "learning_rate": 4.769932297105511e-07,
      "loss": 1.9769,
      "num_input_tokens_seen": 399243832,
      "step": 392100
    },
    {
      "epoch": 7.721840483550235,
      "grad_norm": 1.818023443222046,
      "learning_rate": 4.7678726957143496e-07,
      "loss": 1.9954,
      "num_input_tokens_seen": 399345736,
      "step": 392200
    },
    {
      "epoch": 7.723809336299738,
      "grad_norm": 2.290372610092163,
      "learning_rate": 4.7658131337944773e-07,
      "loss": 1.976,
      "num_input_tokens_seen": 399447264,
      "step": 392300
    },
    {
      "epoch": 7.725778189049241,
      "grad_norm": 1.9023890495300293,
      "learning_rate": 4.763753611696108e-07,
      "loss": 1.9823,
      "num_input_tokens_seen": 399548816,
      "step": 392400
    },
    {
      "epoch": 7.727747041798744,
      "grad_norm": 1.8353853225708008,
      "learning_rate": 4.761694129769443e-07,
      "loss": 2.0116,
      "num_input_tokens_seen": 399648552,
      "step": 392500
    },
    {
      "epoch": 7.729715894548247,
      "grad_norm": 1.8889423608779907,
      "learning_rate": 4.7596346883646813e-07,
      "loss": 1.9611,
      "num_input_tokens_seen": 399748816,
      "step": 392600
    },
    {
      "epoch": 7.7316847472977495,
      "grad_norm": 1.9612642526626587,
      "learning_rate": 4.7575752878320144e-07,
      "loss": 1.9828,
      "num_input_tokens_seen": 399851216,
      "step": 392700
    },
    {
      "epoch": 7.733653600047252,
      "grad_norm": 2.140599250793457,
      "learning_rate": 4.7555159285216247e-07,
      "loss": 2.0269,
      "num_input_tokens_seen": 399950960,
      "step": 392800
    },
    {
      "epoch": 7.735622452796755,
      "grad_norm": 1.8909140825271606,
      "learning_rate": 4.75345661078369e-07,
      "loss": 1.9684,
      "num_input_tokens_seen": 400053360,
      "step": 392900
    },
    {
      "epoch": 7.737591305546259,
      "grad_norm": 1.9300559759140015,
      "learning_rate": 4.751397334968378e-07,
      "loss": 1.9843,
      "num_input_tokens_seen": 400154744,
      "step": 393000
    },
    {
      "epoch": 7.739560158295761,
      "grad_norm": 2.045980215072632,
      "learning_rate": 4.749338101425854e-07,
      "loss": 1.9915,
      "num_input_tokens_seen": 400256528,
      "step": 393100
    },
    {
      "epoch": 7.741529011045264,
      "grad_norm": 1.9019083976745605,
      "learning_rate": 4.74727891050627e-07,
      "loss": 1.9644,
      "num_input_tokens_seen": 400358928,
      "step": 393200
    },
    {
      "epoch": 7.743497863794767,
      "grad_norm": 1.8340952396392822,
      "learning_rate": 4.7452197625597766e-07,
      "loss": 1.9546,
      "num_input_tokens_seen": 400461328,
      "step": 393300
    },
    {
      "epoch": 7.74546671654427,
      "grad_norm": 2.0176641941070557,
      "learning_rate": 4.743160657936513e-07,
      "loss": 1.9918,
      "num_input_tokens_seen": 400563576,
      "step": 393400
    },
    {
      "epoch": 7.747435569293772,
      "grad_norm": 1.8577077388763428,
      "learning_rate": 4.741101596986614e-07,
      "loss": 1.9688,
      "num_input_tokens_seen": 400665368,
      "step": 393500
    },
    {
      "epoch": 7.749404422043275,
      "grad_norm": 2.1358959674835205,
      "learning_rate": 4.739042580060204e-07,
      "loss": 1.9926,
      "num_input_tokens_seen": 400767768,
      "step": 393600
    },
    {
      "epoch": 7.751373274792778,
      "grad_norm": 1.9763044118881226,
      "learning_rate": 4.736983607507401e-07,
      "loss": 1.9844,
      "num_input_tokens_seen": 400867752,
      "step": 393700
    },
    {
      "epoch": 7.753342127542281,
      "grad_norm": 1.8901817798614502,
      "learning_rate": 4.7349246796783196e-07,
      "loss": 2.0072,
      "num_input_tokens_seen": 400970152,
      "step": 393800
    },
    {
      "epoch": 7.755310980291784,
      "grad_norm": 1.8803762197494507,
      "learning_rate": 4.7328657969230597e-07,
      "loss": 1.9447,
      "num_input_tokens_seen": 401072552,
      "step": 393900
    },
    {
      "epoch": 7.757279833041287,
      "grad_norm": 1.8106716871261597,
      "learning_rate": 4.7308069595917183e-07,
      "loss": 1.9471,
      "num_input_tokens_seen": 401174952,
      "step": 394000
    },
    {
      "epoch": 7.75924868579079,
      "grad_norm": 1.8555454015731812,
      "learning_rate": 4.728748168034382e-07,
      "loss": 1.9743,
      "num_input_tokens_seen": 401276432,
      "step": 394100
    },
    {
      "epoch": 7.761217538540293,
      "grad_norm": 1.8103944063186646,
      "learning_rate": 4.726689422601133e-07,
      "loss": 1.977,
      "num_input_tokens_seen": 401377968,
      "step": 394200
    },
    {
      "epoch": 7.763186391289795,
      "grad_norm": 1.7369400262832642,
      "learning_rate": 4.724630723642042e-07,
      "loss": 1.9881,
      "num_input_tokens_seen": 401479664,
      "step": 394300
    },
    {
      "epoch": 7.765155244039298,
      "grad_norm": 1.6876634359359741,
      "learning_rate": 4.722572071507175e-07,
      "loss": 2.0008,
      "num_input_tokens_seen": 401581496,
      "step": 394400
    },
    {
      "epoch": 7.767124096788801,
      "grad_norm": 1.995707392692566,
      "learning_rate": 4.720513466546586e-07,
      "loss": 1.9945,
      "num_input_tokens_seen": 401682520,
      "step": 394500
    },
    {
      "epoch": 7.769092949538304,
      "grad_norm": 1.9053715467453003,
      "learning_rate": 4.718454909110324e-07,
      "loss": 1.9644,
      "num_input_tokens_seen": 401783360,
      "step": 394600
    },
    {
      "epoch": 7.771061802287807,
      "grad_norm": 1.96388840675354,
      "learning_rate": 4.7163963995484316e-07,
      "loss": 1.9831,
      "num_input_tokens_seen": 401884904,
      "step": 394700
    },
    {
      "epoch": 7.77303065503731,
      "grad_norm": 2.2016756534576416,
      "learning_rate": 4.7143379382109386e-07,
      "loss": 2.0089,
      "num_input_tokens_seen": 401987304,
      "step": 394800
    },
    {
      "epoch": 7.774999507786813,
      "grad_norm": 1.8871245384216309,
      "learning_rate": 4.712279525447871e-07,
      "loss": 1.9891,
      "num_input_tokens_seen": 402089704,
      "step": 394900
    },
    {
      "epoch": 7.776968360536316,
      "grad_norm": 1.7828187942504883,
      "learning_rate": 4.710221161609242e-07,
      "loss": 2.0138,
      "num_input_tokens_seen": 402192104,
      "step": 395000
    },
    {
      "epoch": 7.778937213285818,
      "grad_norm": 1.8966572284698486,
      "learning_rate": 4.7081628470450615e-07,
      "loss": 1.9889,
      "num_input_tokens_seen": 402293080,
      "step": 395100
    },
    {
      "epoch": 7.780906066035321,
      "grad_norm": 1.9374722242355347,
      "learning_rate": 4.706104582105326e-07,
      "loss": 1.9743,
      "num_input_tokens_seen": 402394952,
      "step": 395200
    },
    {
      "epoch": 7.782874918784824,
      "grad_norm": 1.7325857877731323,
      "learning_rate": 4.704046367140029e-07,
      "loss": 1.958,
      "num_input_tokens_seen": 402497352,
      "step": 395300
    },
    {
      "epoch": 7.784843771534327,
      "grad_norm": 1.7959673404693604,
      "learning_rate": 4.701988202499149e-07,
      "loss": 1.9683,
      "num_input_tokens_seen": 402599664,
      "step": 395400
    },
    {
      "epoch": 7.786812624283829,
      "grad_norm": 1.869259238243103,
      "learning_rate": 4.6999300885326636e-07,
      "loss": 1.9328,
      "num_input_tokens_seen": 402701568,
      "step": 395500
    },
    {
      "epoch": 7.788781477033333,
      "grad_norm": 1.8868144750595093,
      "learning_rate": 4.697872025590535e-07,
      "loss": 1.9937,
      "num_input_tokens_seen": 402802392,
      "step": 395600
    },
    {
      "epoch": 7.790750329782836,
      "grad_norm": 1.6757222414016724,
      "learning_rate": 4.695814014022719e-07,
      "loss": 2.0285,
      "num_input_tokens_seen": 402904168,
      "step": 395700
    },
    {
      "epoch": 7.7927191825323385,
      "grad_norm": 2.2087771892547607,
      "learning_rate": 4.6937560541791666e-07,
      "loss": 1.9824,
      "num_input_tokens_seen": 403006568,
      "step": 395800
    },
    {
      "epoch": 7.794688035281841,
      "grad_norm": 1.9810230731964111,
      "learning_rate": 4.6916981464098114e-07,
      "loss": 1.9791,
      "num_input_tokens_seen": 403108968,
      "step": 395900
    },
    {
      "epoch": 7.796656888031344,
      "grad_norm": 1.995098352432251,
      "learning_rate": 4.68964029106459e-07,
      "loss": 2.0055,
      "num_input_tokens_seen": 403211368,
      "step": 396000
    },
    {
      "epoch": 7.798625740780847,
      "grad_norm": 2.033402442932129,
      "learning_rate": 4.687582488493415e-07,
      "loss": 1.9772,
      "num_input_tokens_seen": 403312368,
      "step": 396100
    },
    {
      "epoch": 7.80059459353035,
      "grad_norm": 1.8263920545578003,
      "learning_rate": 4.6855247390462073e-07,
      "loss": 1.9769,
      "num_input_tokens_seen": 403414768,
      "step": 396200
    },
    {
      "epoch": 7.802563446279853,
      "grad_norm": 2.0289275646209717,
      "learning_rate": 4.683467043072861e-07,
      "loss": 1.9937,
      "num_input_tokens_seen": 403516520,
      "step": 396300
    },
    {
      "epoch": 7.804532299029356,
      "grad_norm": 2.0917770862579346,
      "learning_rate": 4.681409400923278e-07,
      "loss": 2.0014,
      "num_input_tokens_seen": 403617400,
      "step": 396400
    },
    {
      "epoch": 7.806501151778859,
      "grad_norm": 1.7936569452285767,
      "learning_rate": 4.6793518129473357e-07,
      "loss": 1.9794,
      "num_input_tokens_seen": 403719576,
      "step": 396500
    },
    {
      "epoch": 7.8084700045283615,
      "grad_norm": 1.9994819164276123,
      "learning_rate": 4.6772942794949144e-07,
      "loss": 1.9729,
      "num_input_tokens_seen": 403821376,
      "step": 396600
    },
    {
      "epoch": 7.810438857277864,
      "grad_norm": 2.0388219356536865,
      "learning_rate": 4.6752368009158803e-07,
      "loss": 1.9552,
      "num_input_tokens_seen": 403922984,
      "step": 396700
    },
    {
      "epoch": 7.812407710027367,
      "grad_norm": 1.730810284614563,
      "learning_rate": 4.673179377560088e-07,
      "loss": 1.9948,
      "num_input_tokens_seen": 404024592,
      "step": 396800
    },
    {
      "epoch": 7.81437656277687,
      "grad_norm": 2.40167236328125,
      "learning_rate": 4.671122009777387e-07,
      "loss": 2.0161,
      "num_input_tokens_seen": 404126120,
      "step": 396900
    },
    {
      "epoch": 7.8163454155263725,
      "grad_norm": 1.8635061979293823,
      "learning_rate": 4.669064697917613e-07,
      "loss": 1.9808,
      "num_input_tokens_seen": 404228520,
      "step": 397000
    },
    {
      "epoch": 7.818314268275875,
      "grad_norm": 2.0289618968963623,
      "learning_rate": 4.667007442330597e-07,
      "loss": 1.9991,
      "num_input_tokens_seen": 404330920,
      "step": 397100
    },
    {
      "epoch": 7.820283121025378,
      "grad_norm": 1.9724180698394775,
      "learning_rate": 4.6649502433661566e-07,
      "loss": 2.018,
      "num_input_tokens_seen": 404432512,
      "step": 397200
    },
    {
      "epoch": 7.822251973774882,
      "grad_norm": 1.9349591732025146,
      "learning_rate": 4.6628931013741015e-07,
      "loss": 1.9811,
      "num_input_tokens_seen": 404534912,
      "step": 397300
    },
    {
      "epoch": 7.824220826524384,
      "grad_norm": 1.8570616245269775,
      "learning_rate": 4.66083601670423e-07,
      "loss": 1.9727,
      "num_input_tokens_seen": 404636808,
      "step": 397400
    },
    {
      "epoch": 7.826189679273887,
      "grad_norm": 1.8789377212524414,
      "learning_rate": 4.658778989706335e-07,
      "loss": 1.9598,
      "num_input_tokens_seen": 404739208,
      "step": 397500
    },
    {
      "epoch": 7.82815853202339,
      "grad_norm": 2.2145214080810547,
      "learning_rate": 4.6567220207301933e-07,
      "loss": 1.9881,
      "num_input_tokens_seen": 404841608,
      "step": 397600
    },
    {
      "epoch": 7.830127384772893,
      "grad_norm": 1.8645763397216797,
      "learning_rate": 4.654665110125577e-07,
      "loss": 1.9751,
      "num_input_tokens_seen": 404942592,
      "step": 397700
    },
    {
      "epoch": 7.8320962375223955,
      "grad_norm": 1.8048200607299805,
      "learning_rate": 4.652608258242247e-07,
      "loss": 1.9832,
      "num_input_tokens_seen": 405044224,
      "step": 397800
    },
    {
      "epoch": 7.834065090271898,
      "grad_norm": 2.1996238231658936,
      "learning_rate": 4.6505514654299516e-07,
      "loss": 1.9799,
      "num_input_tokens_seen": 405146624,
      "step": 397900
    },
    {
      "epoch": 7.836033943021402,
      "grad_norm": 2.032247304916382,
      "learning_rate": 4.648494732038433e-07,
      "loss": 2.0608,
      "num_input_tokens_seen": 405248096,
      "step": 398000
    },
    {
      "epoch": 7.838002795770905,
      "grad_norm": 1.9830822944641113,
      "learning_rate": 4.646438058417419e-07,
      "loss": 2.0156,
      "num_input_tokens_seen": 405350312,
      "step": 398100
    },
    {
      "epoch": 7.839971648520407,
      "grad_norm": 6.491024971008301,
      "learning_rate": 4.644381444916633e-07,
      "loss": 1.9995,
      "num_input_tokens_seen": 405451952,
      "step": 398200
    },
    {
      "epoch": 7.84194050126991,
      "grad_norm": 2.343080759048462,
      "learning_rate": 4.642324891885781e-07,
      "loss": 1.9949,
      "num_input_tokens_seen": 405553912,
      "step": 398300
    },
    {
      "epoch": 7.843909354019413,
      "grad_norm": 1.8906807899475098,
      "learning_rate": 4.640268399674566e-07,
      "loss": 1.9698,
      "num_input_tokens_seen": 405656312,
      "step": 398400
    },
    {
      "epoch": 7.845878206768916,
      "grad_norm": 1.6502257585525513,
      "learning_rate": 4.638211968632674e-07,
      "loss": 1.9669,
      "num_input_tokens_seen": 405758712,
      "step": 398500
    },
    {
      "epoch": 7.847847059518418,
      "grad_norm": 2.343722343444824,
      "learning_rate": 4.6361555991097875e-07,
      "loss": 1.961,
      "num_input_tokens_seen": 405861112,
      "step": 398600
    },
    {
      "epoch": 7.849815912267921,
      "grad_norm": 1.8986138105392456,
      "learning_rate": 4.6340992914555705e-07,
      "loss": 1.9415,
      "num_input_tokens_seen": 405963512,
      "step": 398700
    },
    {
      "epoch": 7.851784765017424,
      "grad_norm": 7.905078411102295,
      "learning_rate": 4.632043046019681e-07,
      "loss": 1.9445,
      "num_input_tokens_seen": 406065912,
      "step": 398800
    },
    {
      "epoch": 7.853753617766928,
      "grad_norm": 1.8434330224990845,
      "learning_rate": 4.629986863151773e-07,
      "loss": 1.9951,
      "num_input_tokens_seen": 406167552,
      "step": 398900
    },
    {
      "epoch": 7.85572247051643,
      "grad_norm": 1.826326847076416,
      "learning_rate": 4.6279307432014734e-07,
      "loss": 2.0278,
      "num_input_tokens_seen": 406269176,
      "step": 399000
    },
    {
      "epoch": 7.857691323265933,
      "grad_norm": 1.8038502931594849,
      "learning_rate": 4.625874686518417e-07,
      "loss": 1.9805,
      "num_input_tokens_seen": 406371056,
      "step": 399100
    },
    {
      "epoch": 7.859660176015436,
      "grad_norm": 1.8817472457885742,
      "learning_rate": 4.6238186934522113e-07,
      "loss": 1.9664,
      "num_input_tokens_seen": 406472104,
      "step": 399200
    },
    {
      "epoch": 7.861629028764939,
      "grad_norm": 2.131619691848755,
      "learning_rate": 4.6217627643524663e-07,
      "loss": 1.9796,
      "num_input_tokens_seen": 406573320,
      "step": 399300
    },
    {
      "epoch": 7.863597881514441,
      "grad_norm": 1.8322391510009766,
      "learning_rate": 4.6197068995687704e-07,
      "loss": 1.9649,
      "num_input_tokens_seen": 406674712,
      "step": 399400
    },
    {
      "epoch": 7.865566734263944,
      "grad_norm": 2.3457887172698975,
      "learning_rate": 4.617651099450712e-07,
      "loss": 1.9471,
      "num_input_tokens_seen": 406777112,
      "step": 399500
    },
    {
      "epoch": 7.867535587013447,
      "grad_norm": 2.0252625942230225,
      "learning_rate": 4.6155953643478563e-07,
      "loss": 2.0067,
      "num_input_tokens_seen": 406879512,
      "step": 399600
    },
    {
      "epoch": 7.8695044397629506,
      "grad_norm": 1.6721338033676147,
      "learning_rate": 4.613539694609767e-07,
      "loss": 1.9832,
      "num_input_tokens_seen": 406981272,
      "step": 399700
    },
    {
      "epoch": 7.871473292512453,
      "grad_norm": 2.0794119834899902,
      "learning_rate": 4.611484090585994e-07,
      "loss": 1.9983,
      "num_input_tokens_seen": 407082968,
      "step": 399800
    },
    {
      "epoch": 7.873442145261956,
      "grad_norm": 1.8829747438430786,
      "learning_rate": 4.6094285526260745e-07,
      "loss": 1.9626,
      "num_input_tokens_seen": 407184656,
      "step": 399900
    },
    {
      "epoch": 7.875410998011459,
      "grad_norm": 2.0226032733917236,
      "learning_rate": 4.6073730810795357e-07,
      "loss": 1.9699,
      "num_input_tokens_seen": 407286960,
      "step": 400000
    },
    {
      "epoch": 7.877379850760962,
      "grad_norm": 1.9273746013641357,
      "learning_rate": 4.605317676295892e-07,
      "loss": 1.9637,
      "num_input_tokens_seen": 407389360,
      "step": 400100
    },
    {
      "epoch": 7.879348703510464,
      "grad_norm": 1.977658987045288,
      "learning_rate": 4.60326233862465e-07,
      "loss": 2.0162,
      "num_input_tokens_seen": 407490992,
      "step": 400200
    },
    {
      "epoch": 7.881317556259967,
      "grad_norm": 1.931991696357727,
      "learning_rate": 4.601207068415299e-07,
      "loss": 1.9887,
      "num_input_tokens_seen": 407592552,
      "step": 400300
    },
    {
      "epoch": 7.88328640900947,
      "grad_norm": 1.8200076818466187,
      "learning_rate": 4.5991518660173243e-07,
      "loss": 1.9524,
      "num_input_tokens_seen": 407694568,
      "step": 400400
    },
    {
      "epoch": 7.885255261758973,
      "grad_norm": 2.193967342376709,
      "learning_rate": 4.597096731780193e-07,
      "loss": 1.9761,
      "num_input_tokens_seen": 407796968,
      "step": 400500
    },
    {
      "epoch": 7.887224114508476,
      "grad_norm": 3.000343084335327,
      "learning_rate": 4.5950416660533647e-07,
      "loss": 1.93,
      "num_input_tokens_seen": 407899368,
      "step": 400600
    },
    {
      "epoch": 7.889192967257979,
      "grad_norm": 1.9227417707443237,
      "learning_rate": 4.5929866691862843e-07,
      "loss": 2.0151,
      "num_input_tokens_seen": 408001720,
      "step": 400700
    },
    {
      "epoch": 7.891161820007482,
      "grad_norm": 2.170828104019165,
      "learning_rate": 4.590931741528389e-07,
      "loss": 2.0049,
      "num_input_tokens_seen": 408102688,
      "step": 400800
    },
    {
      "epoch": 7.8931306727569845,
      "grad_norm": 2.095761775970459,
      "learning_rate": 4.588876883429101e-07,
      "loss": 2.0066,
      "num_input_tokens_seen": 408203800,
      "step": 400900
    },
    {
      "epoch": 7.895099525506487,
      "grad_norm": 2.010538339614868,
      "learning_rate": 4.58682209523783e-07,
      "loss": 1.9598,
      "num_input_tokens_seen": 408306200,
      "step": 401000
    },
    {
      "epoch": 7.89706837825599,
      "grad_norm": 2.081390619277954,
      "learning_rate": 4.5847673773039787e-07,
      "loss": 1.976,
      "num_input_tokens_seen": 408407960,
      "step": 401100
    },
    {
      "epoch": 7.899037231005493,
      "grad_norm": 1.9769562482833862,
      "learning_rate": 4.5827127299769316e-07,
      "loss": 2.0141,
      "num_input_tokens_seen": 408510272,
      "step": 401200
    },
    {
      "epoch": 7.901006083754996,
      "grad_norm": 1.8852051496505737,
      "learning_rate": 4.580658153606066e-07,
      "loss": 1.9648,
      "num_input_tokens_seen": 408612672,
      "step": 401300
    },
    {
      "epoch": 7.902974936504499,
      "grad_norm": 1.8560434579849243,
      "learning_rate": 4.5786036485407433e-07,
      "loss": 1.918,
      "num_input_tokens_seen": 408715072,
      "step": 401400
    },
    {
      "epoch": 7.904943789254002,
      "grad_norm": 1.6488037109375,
      "learning_rate": 4.5765492151303164e-07,
      "loss": 1.95,
      "num_input_tokens_seen": 408816776,
      "step": 401500
    },
    {
      "epoch": 7.906912642003505,
      "grad_norm": 1.9223612546920776,
      "learning_rate": 4.574494853724123e-07,
      "loss": 2.0605,
      "num_input_tokens_seen": 408917632,
      "step": 401600
    },
    {
      "epoch": 7.9088814947530075,
      "grad_norm": 1.851230263710022,
      "learning_rate": 4.572440564671491e-07,
      "loss": 1.9802,
      "num_input_tokens_seen": 409019168,
      "step": 401700
    },
    {
      "epoch": 7.91085034750251,
      "grad_norm": 1.7442927360534668,
      "learning_rate": 4.5703863483217376e-07,
      "loss": 1.9523,
      "num_input_tokens_seen": 409121568,
      "step": 401800
    },
    {
      "epoch": 7.912819200252013,
      "grad_norm": 1.6845791339874268,
      "learning_rate": 4.5683322050241577e-07,
      "loss": 1.9634,
      "num_input_tokens_seen": 409223392,
      "step": 401900
    },
    {
      "epoch": 7.914788053001516,
      "grad_norm": 2.3882217407226562,
      "learning_rate": 4.566278135128049e-07,
      "loss": 1.971,
      "num_input_tokens_seen": 409325792,
      "step": 402000
    },
    {
      "epoch": 7.9167569057510185,
      "grad_norm": 1.6533641815185547,
      "learning_rate": 4.5642241389826835e-07,
      "loss": 1.9921,
      "num_input_tokens_seen": 409426512,
      "step": 402100
    },
    {
      "epoch": 7.918725758500521,
      "grad_norm": 2.1795406341552734,
      "learning_rate": 4.562170216937329e-07,
      "loss": 1.9475,
      "num_input_tokens_seen": 409528912,
      "step": 402200
    },
    {
      "epoch": 7.920694611250025,
      "grad_norm": 1.9683079719543457,
      "learning_rate": 4.560116369341235e-07,
      "loss": 1.9598,
      "num_input_tokens_seen": 409630464,
      "step": 402300
    },
    {
      "epoch": 7.922663463999528,
      "grad_norm": 2.0511274337768555,
      "learning_rate": 4.558062596543644e-07,
      "loss": 2.0094,
      "num_input_tokens_seen": 409731408,
      "step": 402400
    },
    {
      "epoch": 7.9246323167490305,
      "grad_norm": 2.065965414047241,
      "learning_rate": 4.556008898893779e-07,
      "loss": 1.9937,
      "num_input_tokens_seen": 409833592,
      "step": 402500
    },
    {
      "epoch": 7.926601169498533,
      "grad_norm": 1.9457801580429077,
      "learning_rate": 4.5539552767408575e-07,
      "loss": 2.0166,
      "num_input_tokens_seen": 409935264,
      "step": 402600
    },
    {
      "epoch": 7.928570022248036,
      "grad_norm": 1.847275733947754,
      "learning_rate": 4.5519017304340777e-07,
      "loss": 2.023,
      "num_input_tokens_seen": 410036296,
      "step": 402700
    },
    {
      "epoch": 7.930538874997539,
      "grad_norm": 1.8351118564605713,
      "learning_rate": 4.5498482603226295e-07,
      "loss": 1.9832,
      "num_input_tokens_seen": 410138696,
      "step": 402800
    },
    {
      "epoch": 7.9325077277470415,
      "grad_norm": 1.958203673362732,
      "learning_rate": 4.5477948667556885e-07,
      "loss": 2.0532,
      "num_input_tokens_seen": 410238312,
      "step": 402900
    },
    {
      "epoch": 7.934476580496545,
      "grad_norm": 1.9280420541763306,
      "learning_rate": 4.545741550082416e-07,
      "loss": 2.0072,
      "num_input_tokens_seen": 410340712,
      "step": 403000
    },
    {
      "epoch": 7.936445433246048,
      "grad_norm": 1.8636482954025269,
      "learning_rate": 4.543688310651962e-07,
      "loss": 1.9968,
      "num_input_tokens_seen": 410443112,
      "step": 403100
    },
    {
      "epoch": 7.938414285995551,
      "grad_norm": 2.1279783248901367,
      "learning_rate": 4.541635148813461e-07,
      "loss": 1.9661,
      "num_input_tokens_seen": 410541816,
      "step": 403200
    },
    {
      "epoch": 7.940383138745053,
      "grad_norm": 1.7693428993225098,
      "learning_rate": 4.539582064916038e-07,
      "loss": 1.9378,
      "num_input_tokens_seen": 410643456,
      "step": 403300
    },
    {
      "epoch": 7.942351991494556,
      "grad_norm": 1.7903947830200195,
      "learning_rate": 4.537529059308801e-07,
      "loss": 2.0237,
      "num_input_tokens_seen": 410744504,
      "step": 403400
    },
    {
      "epoch": 7.944320844244059,
      "grad_norm": 1.720933198928833,
      "learning_rate": 4.5354761323408476e-07,
      "loss": 1.9494,
      "num_input_tokens_seen": 410846904,
      "step": 403500
    },
    {
      "epoch": 7.946289696993562,
      "grad_norm": 1.999996304512024,
      "learning_rate": 4.5334232843612587e-07,
      "loss": 1.9988,
      "num_input_tokens_seen": 410949304,
      "step": 403600
    },
    {
      "epoch": 7.9482585497430644,
      "grad_norm": 2.0352823734283447,
      "learning_rate": 4.531370515719106e-07,
      "loss": 2.0084,
      "num_input_tokens_seen": 411051272,
      "step": 403700
    },
    {
      "epoch": 7.950227402492567,
      "grad_norm": 1.7085314989089966,
      "learning_rate": 4.5293178267634435e-07,
      "loss": 1.9817,
      "num_input_tokens_seen": 411152936,
      "step": 403800
    },
    {
      "epoch": 7.95219625524207,
      "grad_norm": 1.8613543510437012,
      "learning_rate": 4.5272652178433137e-07,
      "loss": 2.0533,
      "num_input_tokens_seen": 411252824,
      "step": 403900
    },
    {
      "epoch": 7.954165107991574,
      "grad_norm": 1.9997822046279907,
      "learning_rate": 4.5252126893077494e-07,
      "loss": 1.9447,
      "num_input_tokens_seen": 411354408,
      "step": 404000
    },
    {
      "epoch": 7.956133960741076,
      "grad_norm": 1.9824864864349365,
      "learning_rate": 4.52316024150576e-07,
      "loss": 1.9597,
      "num_input_tokens_seen": 411456808,
      "step": 404100
    },
    {
      "epoch": 7.958102813490579,
      "grad_norm": 1.8737788200378418,
      "learning_rate": 4.521107874786352e-07,
      "loss": 2.0305,
      "num_input_tokens_seen": 411557928,
      "step": 404200
    },
    {
      "epoch": 7.960071666240082,
      "grad_norm": 1.9790064096450806,
      "learning_rate": 4.5190555894985084e-07,
      "loss": 2.0123,
      "num_input_tokens_seen": 411660328,
      "step": 404300
    },
    {
      "epoch": 7.962040518989585,
      "grad_norm": 2.1659984588623047,
      "learning_rate": 4.517003385991208e-07,
      "loss": 1.9605,
      "num_input_tokens_seen": 411762184,
      "step": 404400
    },
    {
      "epoch": 7.964009371739087,
      "grad_norm": 1.9814157485961914,
      "learning_rate": 4.5149512646134045e-07,
      "loss": 2.0205,
      "num_input_tokens_seen": 411863848,
      "step": 404500
    },
    {
      "epoch": 7.96597822448859,
      "grad_norm": 1.9131046533584595,
      "learning_rate": 4.5128992257140514e-07,
      "loss": 2.0331,
      "num_input_tokens_seen": 411965808,
      "step": 404600
    },
    {
      "epoch": 7.967947077238094,
      "grad_norm": 1.846489667892456,
      "learning_rate": 4.510847269642073e-07,
      "loss": 2.029,
      "num_input_tokens_seen": 412068208,
      "step": 404700
    },
    {
      "epoch": 7.969915929987597,
      "grad_norm": 1.982517123222351,
      "learning_rate": 4.508795396746392e-07,
      "loss": 2.0108,
      "num_input_tokens_seen": 412169984,
      "step": 404800
    },
    {
      "epoch": 7.971884782737099,
      "grad_norm": 1.8219513893127441,
      "learning_rate": 4.5067436073759114e-07,
      "loss": 1.9365,
      "num_input_tokens_seen": 412272384,
      "step": 404900
    },
    {
      "epoch": 7.973853635486602,
      "grad_norm": 1.8535133600234985,
      "learning_rate": 4.504691901879519e-07,
      "loss": 1.999,
      "num_input_tokens_seen": 412374352,
      "step": 405000
    },
    {
      "epoch": 7.975822488236105,
      "grad_norm": 2.036672830581665,
      "learning_rate": 4.5026402806060924e-07,
      "loss": 1.9979,
      "num_input_tokens_seen": 412475384,
      "step": 405100
    },
    {
      "epoch": 7.977791340985608,
      "grad_norm": 2.0089473724365234,
      "learning_rate": 4.5005887439044904e-07,
      "loss": 2.0108,
      "num_input_tokens_seen": 412577784,
      "step": 405200
    },
    {
      "epoch": 7.97976019373511,
      "grad_norm": 3.8883466720581055,
      "learning_rate": 4.498537292123561e-07,
      "loss": 1.9855,
      "num_input_tokens_seen": 412678760,
      "step": 405300
    },
    {
      "epoch": 7.981729046484613,
      "grad_norm": 1.8637185096740723,
      "learning_rate": 4.4964859256121347e-07,
      "loss": 2.001,
      "num_input_tokens_seen": 412781160,
      "step": 405400
    },
    {
      "epoch": 7.983697899234116,
      "grad_norm": 3.9107072353363037,
      "learning_rate": 4.4944346447190307e-07,
      "loss": 1.9783,
      "num_input_tokens_seen": 412883560,
      "step": 405500
    },
    {
      "epoch": 7.9856667519836195,
      "grad_norm": 2.056428909301758,
      "learning_rate": 4.49238344979305e-07,
      "loss": 1.979,
      "num_input_tokens_seen": 412985912,
      "step": 405600
    },
    {
      "epoch": 7.987635604733122,
      "grad_norm": 1.8555965423583984,
      "learning_rate": 4.4903323411829834e-07,
      "loss": 1.9656,
      "num_input_tokens_seen": 413088264,
      "step": 405700
    },
    {
      "epoch": 7.989604457482625,
      "grad_norm": 1.9894907474517822,
      "learning_rate": 4.4882813192376024e-07,
      "loss": 1.9573,
      "num_input_tokens_seen": 413190080,
      "step": 405800
    },
    {
      "epoch": 7.991573310232128,
      "grad_norm": 1.6447981595993042,
      "learning_rate": 4.486230384305667e-07,
      "loss": 1.9908,
      "num_input_tokens_seen": 413291752,
      "step": 405900
    },
    {
      "epoch": 7.993542162981631,
      "grad_norm": 1.8621642589569092,
      "learning_rate": 4.4841795367359223e-07,
      "loss": 1.976,
      "num_input_tokens_seen": 413393568,
      "step": 406000
    },
    {
      "epoch": 7.995511015731133,
      "grad_norm": 1.8075709342956543,
      "learning_rate": 4.4821287768770967e-07,
      "loss": 1.9935,
      "num_input_tokens_seen": 413495424,
      "step": 406100
    },
    {
      "epoch": 7.997479868480636,
      "grad_norm": 1.8301723003387451,
      "learning_rate": 4.480078105077905e-07,
      "loss": 1.9843,
      "num_input_tokens_seen": 413597824,
      "step": 406200
    },
    {
      "epoch": 7.999448721230139,
      "grad_norm": 2.453486680984497,
      "learning_rate": 4.478027521687045e-07,
      "loss": 2.0093,
      "num_input_tokens_seen": 413700224,
      "step": 406300
    },
    {
      "epoch": 8.001417573979642,
      "grad_norm": 1.8768980503082275,
      "learning_rate": 4.475977027053204e-07,
      "loss": 2.0736,
      "num_input_tokens_seen": 413802624,
      "step": 406400
    },
    {
      "epoch": 8.003386426729145,
      "grad_norm": 2.593552350997925,
      "learning_rate": 4.473926621525048e-07,
      "loss": 1.9937,
      "num_input_tokens_seen": 413904352,
      "step": 406500
    },
    {
      "epoch": 8.005355279478648,
      "grad_norm": 1.6389186382293701,
      "learning_rate": 4.471876305451234e-07,
      "loss": 1.9593,
      "num_input_tokens_seen": 414006752,
      "step": 406600
    },
    {
      "epoch": 8.00732413222815,
      "grad_norm": 2.015887498855591,
      "learning_rate": 4.4698260791804e-07,
      "loss": 1.9714,
      "num_input_tokens_seen": 414108384,
      "step": 406700
    },
    {
      "epoch": 8.009292984977654,
      "grad_norm": 1.7495338916778564,
      "learning_rate": 4.4677759430611674e-07,
      "loss": 2.0084,
      "num_input_tokens_seen": 414210496,
      "step": 406800
    },
    {
      "epoch": 8.011261837727156,
      "grad_norm": 1.7535310983657837,
      "learning_rate": 4.46572589744215e-07,
      "loss": 1.9955,
      "num_input_tokens_seen": 414312808,
      "step": 406900
    },
    {
      "epoch": 8.013230690476659,
      "grad_norm": 2.2229084968566895,
      "learning_rate": 4.4636759426719347e-07,
      "loss": 1.954,
      "num_input_tokens_seen": 414414488,
      "step": 407000
    },
    {
      "epoch": 8.015199543226162,
      "grad_norm": 2.1934497356414795,
      "learning_rate": 4.461626079099104e-07,
      "loss": 1.9182,
      "num_input_tokens_seen": 414516888,
      "step": 407100
    },
    {
      "epoch": 8.017168395975665,
      "grad_norm": 1.7163622379302979,
      "learning_rate": 4.459576307072215e-07,
      "loss": 1.9731,
      "num_input_tokens_seen": 414617696,
      "step": 407200
    },
    {
      "epoch": 8.019137248725167,
      "grad_norm": 1.955451250076294,
      "learning_rate": 4.4575266269398203e-07,
      "loss": 1.9797,
      "num_input_tokens_seen": 414719648,
      "step": 407300
    },
    {
      "epoch": 8.02110610147467,
      "grad_norm": 2.0064711570739746,
      "learning_rate": 4.4554770390504443e-07,
      "loss": 1.971,
      "num_input_tokens_seen": 414821432,
      "step": 407400
    },
    {
      "epoch": 8.023074954224173,
      "grad_norm": 1.9356880187988281,
      "learning_rate": 4.453427543752608e-07,
      "loss": 2.0352,
      "num_input_tokens_seen": 414922488,
      "step": 407500
    },
    {
      "epoch": 8.025043806973676,
      "grad_norm": 1.8024967908859253,
      "learning_rate": 4.451378141394805e-07,
      "loss": 1.9791,
      "num_input_tokens_seen": 415024888,
      "step": 407600
    },
    {
      "epoch": 8.02701265972318,
      "grad_norm": 1.9446879625320435,
      "learning_rate": 4.449328832325524e-07,
      "loss": 1.9586,
      "num_input_tokens_seen": 415126456,
      "step": 407700
    },
    {
      "epoch": 8.028981512472683,
      "grad_norm": 1.8077514171600342,
      "learning_rate": 4.447279616893228e-07,
      "loss": 1.9698,
      "num_input_tokens_seen": 415228856,
      "step": 407800
    },
    {
      "epoch": 8.030950365222186,
      "grad_norm": 1.8537914752960205,
      "learning_rate": 4.445230495446372e-07,
      "loss": 2.0053,
      "num_input_tokens_seen": 415331256,
      "step": 407900
    },
    {
      "epoch": 8.032919217971688,
      "grad_norm": 2.114763021469116,
      "learning_rate": 4.443181468333392e-07,
      "loss": 2.045,
      "num_input_tokens_seen": 415431368,
      "step": 408000
    },
    {
      "epoch": 8.034888070721191,
      "grad_norm": 1.8701435327529907,
      "learning_rate": 4.441132535902707e-07,
      "loss": 1.9582,
      "num_input_tokens_seen": 415533312,
      "step": 408100
    },
    {
      "epoch": 8.036856923470694,
      "grad_norm": 2.1332123279571533,
      "learning_rate": 4.439083698502721e-07,
      "loss": 1.9816,
      "num_input_tokens_seen": 415635712,
      "step": 408200
    },
    {
      "epoch": 8.038825776220197,
      "grad_norm": 1.9398605823516846,
      "learning_rate": 4.4370349564818197e-07,
      "loss": 1.9609,
      "num_input_tokens_seen": 415736664,
      "step": 408300
    },
    {
      "epoch": 8.0407946289697,
      "grad_norm": 1.9261415004730225,
      "learning_rate": 4.4349863101883767e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 415839064,
      "step": 408400
    },
    {
      "epoch": 8.042763481719202,
      "grad_norm": 2.070559501647949,
      "learning_rate": 4.432937759970746e-07,
      "loss": 2.0263,
      "num_input_tokens_seen": 415940552,
      "step": 408500
    },
    {
      "epoch": 8.044732334468705,
      "grad_norm": 1.7003812789916992,
      "learning_rate": 4.430889306177267e-07,
      "loss": 1.9846,
      "num_input_tokens_seen": 416042952,
      "step": 408600
    },
    {
      "epoch": 8.046701187218208,
      "grad_norm": 1.6576703786849976,
      "learning_rate": 4.42884094915626e-07,
      "loss": 1.9659,
      "num_input_tokens_seen": 416143776,
      "step": 408700
    },
    {
      "epoch": 8.04867003996771,
      "grad_norm": 1.9099196195602417,
      "learning_rate": 4.4267926892560334e-07,
      "loss": 2.0357,
      "num_input_tokens_seen": 416244952,
      "step": 408800
    },
    {
      "epoch": 8.050638892717213,
      "grad_norm": 1.7863417863845825,
      "learning_rate": 4.424744526824874e-07,
      "loss": 1.9849,
      "num_input_tokens_seen": 416347352,
      "step": 408900
    },
    {
      "epoch": 8.052607745466716,
      "grad_norm": 1.963246464729309,
      "learning_rate": 4.422696462211056e-07,
      "loss": 2.0101,
      "num_input_tokens_seen": 416449752,
      "step": 409000
    },
    {
      "epoch": 8.054576598216219,
      "grad_norm": 1.973564624786377,
      "learning_rate": 4.4206484957628367e-07,
      "loss": 1.9676,
      "num_input_tokens_seen": 416551520,
      "step": 409100
    },
    {
      "epoch": 8.056545450965721,
      "grad_norm": 1.9066550731658936,
      "learning_rate": 4.4186006278284535e-07,
      "loss": 2.0223,
      "num_input_tokens_seen": 416652472,
      "step": 409200
    },
    {
      "epoch": 8.058514303715224,
      "grad_norm": 1.968870997428894,
      "learning_rate": 4.41655285875613e-07,
      "loss": 1.9441,
      "num_input_tokens_seen": 416754208,
      "step": 409300
    },
    {
      "epoch": 8.060483156464729,
      "grad_norm": 1.8548153638839722,
      "learning_rate": 4.4145051888940715e-07,
      "loss": 1.9885,
      "num_input_tokens_seen": 416856608,
      "step": 409400
    },
    {
      "epoch": 8.062452009214232,
      "grad_norm": 1.7558403015136719,
      "learning_rate": 4.412457618590468e-07,
      "loss": 1.9663,
      "num_input_tokens_seen": 416959008,
      "step": 409500
    },
    {
      "epoch": 8.064420861963734,
      "grad_norm": 2.3564465045928955,
      "learning_rate": 4.4104101481934896e-07,
      "loss": 2.0136,
      "num_input_tokens_seen": 417060840,
      "step": 409600
    },
    {
      "epoch": 8.066389714713237,
      "grad_norm": 1.9404773712158203,
      "learning_rate": 4.4083627780512945e-07,
      "loss": 1.9905,
      "num_input_tokens_seen": 417162384,
      "step": 409700
    },
    {
      "epoch": 8.06835856746274,
      "grad_norm": 1.9789031744003296,
      "learning_rate": 4.4063155085120173e-07,
      "loss": 1.9415,
      "num_input_tokens_seen": 417264280,
      "step": 409800
    },
    {
      "epoch": 8.070327420212243,
      "grad_norm": 2.336280345916748,
      "learning_rate": 4.4042683399237796e-07,
      "loss": 2.0524,
      "num_input_tokens_seen": 417363904,
      "step": 409900
    },
    {
      "epoch": 8.072296272961745,
      "grad_norm": 1.9985854625701904,
      "learning_rate": 4.402221272634689e-07,
      "loss": 1.9839,
      "num_input_tokens_seen": 417466304,
      "step": 410000
    },
    {
      "epoch": 8.074265125711248,
      "grad_norm": 1.961057186126709,
      "learning_rate": 4.4001743069928265e-07,
      "loss": 1.9513,
      "num_input_tokens_seen": 417567960,
      "step": 410100
    },
    {
      "epoch": 8.07623397846075,
      "grad_norm": 3.8790998458862305,
      "learning_rate": 4.398127443346267e-07,
      "loss": 2.0032,
      "num_input_tokens_seen": 417670360,
      "step": 410200
    },
    {
      "epoch": 8.078202831210254,
      "grad_norm": 1.961729884147644,
      "learning_rate": 4.3960806820430555e-07,
      "loss": 1.9924,
      "num_input_tokens_seen": 417772224,
      "step": 410300
    },
    {
      "epoch": 8.080171683959756,
      "grad_norm": 1.9560363292694092,
      "learning_rate": 4.394034023431233e-07,
      "loss": 2.0176,
      "num_input_tokens_seen": 417874624,
      "step": 410400
    },
    {
      "epoch": 8.08214053670926,
      "grad_norm": 1.913643479347229,
      "learning_rate": 4.3919874678588126e-07,
      "loss": 1.9908,
      "num_input_tokens_seen": 417976104,
      "step": 410500
    },
    {
      "epoch": 8.084109389458762,
      "grad_norm": 1.811996579170227,
      "learning_rate": 4.389941015673796e-07,
      "loss": 1.9808,
      "num_input_tokens_seen": 418077208,
      "step": 410600
    },
    {
      "epoch": 8.086078242208265,
      "grad_norm": 2.0247504711151123,
      "learning_rate": 4.3878946672241634e-07,
      "loss": 2.0238,
      "num_input_tokens_seen": 418179608,
      "step": 410700
    },
    {
      "epoch": 8.088047094957767,
      "grad_norm": 1.8828580379486084,
      "learning_rate": 4.385848422857881e-07,
      "loss": 1.9857,
      "num_input_tokens_seen": 418281152,
      "step": 410800
    },
    {
      "epoch": 8.09001594770727,
      "grad_norm": 2.2555432319641113,
      "learning_rate": 4.3838022829228933e-07,
      "loss": 2.0139,
      "num_input_tokens_seen": 418383552,
      "step": 410900
    },
    {
      "epoch": 8.091984800456775,
      "grad_norm": 1.7268421649932861,
      "learning_rate": 4.38175624776713e-07,
      "loss": 2.0021,
      "num_input_tokens_seen": 418485144,
      "step": 411000
    },
    {
      "epoch": 8.093953653206277,
      "grad_norm": 1.8772838115692139,
      "learning_rate": 4.379710317738503e-07,
      "loss": 1.9899,
      "num_input_tokens_seen": 418586160,
      "step": 411100
    },
    {
      "epoch": 8.09592250595578,
      "grad_norm": 1.8884308338165283,
      "learning_rate": 4.3776644931849044e-07,
      "loss": 1.9949,
      "num_input_tokens_seen": 418687712,
      "step": 411200
    },
    {
      "epoch": 8.097891358705283,
      "grad_norm": 1.8476943969726562,
      "learning_rate": 4.37561877445421e-07,
      "loss": 1.9742,
      "num_input_tokens_seen": 418790112,
      "step": 411300
    },
    {
      "epoch": 8.099860211454786,
      "grad_norm": 2.120741128921509,
      "learning_rate": 4.3735731618942764e-07,
      "loss": 1.9978,
      "num_input_tokens_seen": 418892064,
      "step": 411400
    },
    {
      "epoch": 8.101829064204288,
      "grad_norm": 1.7202106714248657,
      "learning_rate": 4.3715276558529436e-07,
      "loss": 1.9588,
      "num_input_tokens_seen": 418993728,
      "step": 411500
    },
    {
      "epoch": 8.103797916953791,
      "grad_norm": 1.9934120178222656,
      "learning_rate": 4.369482256678031e-07,
      "loss": 1.9947,
      "num_input_tokens_seen": 419094808,
      "step": 411600
    },
    {
      "epoch": 8.105766769703294,
      "grad_norm": 2.1794073581695557,
      "learning_rate": 4.3674369647173435e-07,
      "loss": 1.9949,
      "num_input_tokens_seen": 419196312,
      "step": 411700
    },
    {
      "epoch": 8.107735622452797,
      "grad_norm": 1.9189203977584839,
      "learning_rate": 4.3653917803186644e-07,
      "loss": 1.9819,
      "num_input_tokens_seen": 419297152,
      "step": 411800
    },
    {
      "epoch": 8.1097044752023,
      "grad_norm": 2.280561685562134,
      "learning_rate": 4.3633467038297604e-07,
      "loss": 1.992,
      "num_input_tokens_seen": 419399040,
      "step": 411900
    },
    {
      "epoch": 8.111673327951802,
      "grad_norm": 1.8303940296173096,
      "learning_rate": 4.3613017355983803e-07,
      "loss": 1.9362,
      "num_input_tokens_seen": 419500592,
      "step": 412000
    },
    {
      "epoch": 8.113642180701305,
      "grad_norm": 1.8243738412857056,
      "learning_rate": 4.3592568759722513e-07,
      "loss": 2.0041,
      "num_input_tokens_seen": 419602416,
      "step": 412100
    },
    {
      "epoch": 8.115611033450808,
      "grad_norm": 2.0194602012634277,
      "learning_rate": 4.35721212529909e-07,
      "loss": 1.9372,
      "num_input_tokens_seen": 419704816,
      "step": 412200
    },
    {
      "epoch": 8.11757988620031,
      "grad_norm": 1.8387428522109985,
      "learning_rate": 4.3551674839265814e-07,
      "loss": 1.9939,
      "num_input_tokens_seen": 419807216,
      "step": 412300
    },
    {
      "epoch": 8.119548738949813,
      "grad_norm": 1.7517679929733276,
      "learning_rate": 4.3531229522024065e-07,
      "loss": 1.9665,
      "num_input_tokens_seen": 419908792,
      "step": 412400
    },
    {
      "epoch": 8.121517591699316,
      "grad_norm": 1.9406459331512451,
      "learning_rate": 4.351078530474215e-07,
      "loss": 1.9767,
      "num_input_tokens_seen": 420010400,
      "step": 412500
    },
    {
      "epoch": 8.123486444448819,
      "grad_norm": 1.822887659072876,
      "learning_rate": 4.3490342190896485e-07,
      "loss": 1.955,
      "num_input_tokens_seen": 420112136,
      "step": 412600
    },
    {
      "epoch": 8.125455297198323,
      "grad_norm": 2.7356631755828857,
      "learning_rate": 4.346990018396319e-07,
      "loss": 1.9891,
      "num_input_tokens_seen": 420213672,
      "step": 412700
    },
    {
      "epoch": 8.127424149947826,
      "grad_norm": 1.9019370079040527,
      "learning_rate": 4.344945928741834e-07,
      "loss": 1.9666,
      "num_input_tokens_seen": 420314440,
      "step": 412800
    },
    {
      "epoch": 8.129393002697329,
      "grad_norm": 1.843346118927002,
      "learning_rate": 4.3429019504737643e-07,
      "loss": 1.9398,
      "num_input_tokens_seen": 420416840,
      "step": 412900
    },
    {
      "epoch": 8.131361855446832,
      "grad_norm": 1.9066869020462036,
      "learning_rate": 4.340858083939677e-07,
      "loss": 1.9923,
      "num_input_tokens_seen": 420518608,
      "step": 413000
    },
    {
      "epoch": 8.133330708196334,
      "grad_norm": 2.1369450092315674,
      "learning_rate": 4.3388143294871144e-07,
      "loss": 1.9991,
      "num_input_tokens_seen": 420619736,
      "step": 413100
    },
    {
      "epoch": 8.135299560945837,
      "grad_norm": 1.8550385236740112,
      "learning_rate": 4.3367706874635975e-07,
      "loss": 1.9679,
      "num_input_tokens_seen": 420722136,
      "step": 413200
    },
    {
      "epoch": 8.13726841369534,
      "grad_norm": 1.9507849216461182,
      "learning_rate": 4.334727158216632e-07,
      "loss": 1.9271,
      "num_input_tokens_seen": 420823080,
      "step": 413300
    },
    {
      "epoch": 8.139237266444843,
      "grad_norm": 2.0656447410583496,
      "learning_rate": 4.3326837420937007e-07,
      "loss": 2.0056,
      "num_input_tokens_seen": 420925480,
      "step": 413400
    },
    {
      "epoch": 8.141206119194345,
      "grad_norm": 2.0361788272857666,
      "learning_rate": 4.3306404394422716e-07,
      "loss": 1.9977,
      "num_input_tokens_seen": 421026864,
      "step": 413500
    },
    {
      "epoch": 8.143174971943848,
      "grad_norm": 1.8786455392837524,
      "learning_rate": 4.32859725060979e-07,
      "loss": 1.9456,
      "num_input_tokens_seen": 421127064,
      "step": 413600
    },
    {
      "epoch": 8.145143824693351,
      "grad_norm": 2.021986961364746,
      "learning_rate": 4.326554175943684e-07,
      "loss": 1.9837,
      "num_input_tokens_seen": 421228704,
      "step": 413700
    },
    {
      "epoch": 8.147112677442854,
      "grad_norm": 1.950090765953064,
      "learning_rate": 4.3245112157913583e-07,
      "loss": 1.9887,
      "num_input_tokens_seen": 421331104,
      "step": 413800
    },
    {
      "epoch": 8.149081530192356,
      "grad_norm": 1.8825485706329346,
      "learning_rate": 4.3224683705002046e-07,
      "loss": 1.9525,
      "num_input_tokens_seen": 421433128,
      "step": 413900
    },
    {
      "epoch": 8.15105038294186,
      "grad_norm": 1.9467074871063232,
      "learning_rate": 4.3204256404175886e-07,
      "loss": 2.0093,
      "num_input_tokens_seen": 421534480,
      "step": 414000
    },
    {
      "epoch": 8.153019235691362,
      "grad_norm": 1.6391477584838867,
      "learning_rate": 4.318383025890862e-07,
      "loss": 1.9754,
      "num_input_tokens_seen": 421636880,
      "step": 414100
    },
    {
      "epoch": 8.154988088440865,
      "grad_norm": 1.941521167755127,
      "learning_rate": 4.316340527267352e-07,
      "loss": 1.9348,
      "num_input_tokens_seen": 421739280,
      "step": 414200
    },
    {
      "epoch": 8.156956941190368,
      "grad_norm": 2.1100356578826904,
      "learning_rate": 4.3142981448943687e-07,
      "loss": 1.9707,
      "num_input_tokens_seen": 421841680,
      "step": 414300
    },
    {
      "epoch": 8.158925793939872,
      "grad_norm": 1.9965530633926392,
      "learning_rate": 4.312255879119204e-07,
      "loss": 1.9807,
      "num_input_tokens_seen": 421943600,
      "step": 414400
    },
    {
      "epoch": 8.160894646689375,
      "grad_norm": 1.5953651666641235,
      "learning_rate": 4.310213730289125e-07,
      "loss": 2.0153,
      "num_input_tokens_seen": 422045392,
      "step": 414500
    },
    {
      "epoch": 8.162863499438878,
      "grad_norm": 1.9872205257415771,
      "learning_rate": 4.308171698751384e-07,
      "loss": 1.9624,
      "num_input_tokens_seen": 422147040,
      "step": 414600
    },
    {
      "epoch": 8.16483235218838,
      "grad_norm": 2.0076491832733154,
      "learning_rate": 4.3061297848532094e-07,
      "loss": 1.9825,
      "num_input_tokens_seen": 422248016,
      "step": 414700
    },
    {
      "epoch": 8.166801204937883,
      "grad_norm": 1.7215479612350464,
      "learning_rate": 4.3040879889418143e-07,
      "loss": 1.9748,
      "num_input_tokens_seen": 422349664,
      "step": 414800
    },
    {
      "epoch": 8.168770057687386,
      "grad_norm": 2.028359889984131,
      "learning_rate": 4.302046311364385e-07,
      "loss": 2.0006,
      "num_input_tokens_seen": 422452064,
      "step": 414900
    },
    {
      "epoch": 8.170738910436889,
      "grad_norm": 2.7671148777008057,
      "learning_rate": 4.3000047524680935e-07,
      "loss": 1.9735,
      "num_input_tokens_seen": 422552784,
      "step": 415000
    },
    {
      "epoch": 8.172707763186391,
      "grad_norm": 1.802409291267395,
      "learning_rate": 4.2979633126000923e-07,
      "loss": 2.003,
      "num_input_tokens_seen": 422655184,
      "step": 415100
    },
    {
      "epoch": 8.174676615935894,
      "grad_norm": 2.103318452835083,
      "learning_rate": 4.295921992107506e-07,
      "loss": 2.0106,
      "num_input_tokens_seen": 422756176,
      "step": 415200
    },
    {
      "epoch": 8.176645468685397,
      "grad_norm": 2.1737442016601562,
      "learning_rate": 4.2938807913374494e-07,
      "loss": 1.9747,
      "num_input_tokens_seen": 422858040,
      "step": 415300
    },
    {
      "epoch": 8.1786143214349,
      "grad_norm": 11.40088176727295,
      "learning_rate": 4.2918397106370047e-07,
      "loss": 1.9635,
      "num_input_tokens_seen": 422959528,
      "step": 415400
    },
    {
      "epoch": 8.180583174184402,
      "grad_norm": 1.9554022550582886,
      "learning_rate": 4.2897987503532486e-07,
      "loss": 2.0448,
      "num_input_tokens_seen": 423061928,
      "step": 415500
    },
    {
      "epoch": 8.182552026933905,
      "grad_norm": 1.760983943939209,
      "learning_rate": 4.287757910833221e-07,
      "loss": 1.977,
      "num_input_tokens_seen": 423164328,
      "step": 415600
    },
    {
      "epoch": 8.184520879683408,
      "grad_norm": 2.022050619125366,
      "learning_rate": 4.2857171924239565e-07,
      "loss": 1.9657,
      "num_input_tokens_seen": 423265456,
      "step": 415700
    },
    {
      "epoch": 8.18648973243291,
      "grad_norm": 2.083146810531616,
      "learning_rate": 4.283676595472456e-07,
      "loss": 1.9946,
      "num_input_tokens_seen": 423367856,
      "step": 415800
    },
    {
      "epoch": 8.188458585182413,
      "grad_norm": 1.9051077365875244,
      "learning_rate": 4.281636120325712e-07,
      "loss": 2.0006,
      "num_input_tokens_seen": 423470256,
      "step": 415900
    },
    {
      "epoch": 8.190427437931916,
      "grad_norm": 2.152902126312256,
      "learning_rate": 4.2795957673306824e-07,
      "loss": 1.9659,
      "num_input_tokens_seen": 423571888,
      "step": 416000
    },
    {
      "epoch": 8.19239629068142,
      "grad_norm": 1.9484357833862305,
      "learning_rate": 4.2775555368343184e-07,
      "loss": 1.9503,
      "num_input_tokens_seen": 423673592,
      "step": 416100
    },
    {
      "epoch": 8.194365143430923,
      "grad_norm": 1.811044454574585,
      "learning_rate": 4.2755154291835436e-07,
      "loss": 1.9951,
      "num_input_tokens_seen": 423775992,
      "step": 416200
    },
    {
      "epoch": 8.196333996180426,
      "grad_norm": 2.6572468280792236,
      "learning_rate": 4.273475444725258e-07,
      "loss": 1.9828,
      "num_input_tokens_seen": 423877600,
      "step": 416300
    },
    {
      "epoch": 8.198302848929929,
      "grad_norm": 1.873489260673523,
      "learning_rate": 4.2714355838063457e-07,
      "loss": 2.0001,
      "num_input_tokens_seen": 423979040,
      "step": 416400
    },
    {
      "epoch": 8.200271701679432,
      "grad_norm": 1.9626555442810059,
      "learning_rate": 4.269395846773667e-07,
      "loss": 1.9072,
      "num_input_tokens_seen": 424081440,
      "step": 416500
    },
    {
      "epoch": 8.202240554428935,
      "grad_norm": 1.839101791381836,
      "learning_rate": 4.2673562339740634e-07,
      "loss": 1.9907,
      "num_input_tokens_seen": 424182984,
      "step": 416600
    },
    {
      "epoch": 8.204209407178437,
      "grad_norm": 21.5980167388916,
      "learning_rate": 4.2653167457543515e-07,
      "loss": 2.0106,
      "num_input_tokens_seen": 424284424,
      "step": 416700
    },
    {
      "epoch": 8.20617825992794,
      "grad_norm": 2.051440715789795,
      "learning_rate": 4.2632773824613314e-07,
      "loss": 2.037,
      "num_input_tokens_seen": 424386312,
      "step": 416800
    },
    {
      "epoch": 8.208147112677443,
      "grad_norm": 2.139739751815796,
      "learning_rate": 4.261238144441778e-07,
      "loss": 1.9819,
      "num_input_tokens_seen": 424487896,
      "step": 416900
    },
    {
      "epoch": 8.210115965426946,
      "grad_norm": 2.032843828201294,
      "learning_rate": 4.2591990320424464e-07,
      "loss": 1.9742,
      "num_input_tokens_seen": 424590296,
      "step": 417000
    },
    {
      "epoch": 8.212084818176448,
      "grad_norm": 2.0535480976104736,
      "learning_rate": 4.2571600456100727e-07,
      "loss": 2.0141,
      "num_input_tokens_seen": 424692696,
      "step": 417100
    },
    {
      "epoch": 8.214053670925951,
      "grad_norm": 1.7785221338272095,
      "learning_rate": 4.2551211854913666e-07,
      "loss": 1.9624,
      "num_input_tokens_seen": 424795096,
      "step": 417200
    },
    {
      "epoch": 8.216022523675454,
      "grad_norm": 1.574901819229126,
      "learning_rate": 4.253082452033022e-07,
      "loss": 1.9824,
      "num_input_tokens_seen": 424897496,
      "step": 417300
    },
    {
      "epoch": 8.217991376424957,
      "grad_norm": 2.0116829872131348,
      "learning_rate": 4.2510438455817045e-07,
      "loss": 2.01,
      "num_input_tokens_seen": 424998560,
      "step": 417400
    },
    {
      "epoch": 8.21996022917446,
      "grad_norm": 1.813637614250183,
      "learning_rate": 4.2490053664840657e-07,
      "loss": 2.0074,
      "num_input_tokens_seen": 425100960,
      "step": 417500
    },
    {
      "epoch": 8.221929081923962,
      "grad_norm": 1.6019140481948853,
      "learning_rate": 4.246967015086729e-07,
      "loss": 1.9644,
      "num_input_tokens_seen": 425202240,
      "step": 417600
    },
    {
      "epoch": 8.223897934673467,
      "grad_norm": 1.9128168821334839,
      "learning_rate": 4.2449287917363013e-07,
      "loss": 2.0033,
      "num_input_tokens_seen": 425303872,
      "step": 417700
    },
    {
      "epoch": 8.22586678742297,
      "grad_norm": 2.0122196674346924,
      "learning_rate": 4.2428906967793627e-07,
      "loss": 1.9778,
      "num_input_tokens_seen": 425406272,
      "step": 417800
    },
    {
      "epoch": 8.227835640172472,
      "grad_norm": 1.946831226348877,
      "learning_rate": 4.2408527305624763e-07,
      "loss": 1.9131,
      "num_input_tokens_seen": 425508672,
      "step": 417900
    },
    {
      "epoch": 8.229804492921975,
      "grad_norm": 2.8963088989257812,
      "learning_rate": 4.2388148934321797e-07,
      "loss": 1.9759,
      "num_input_tokens_seen": 425611072,
      "step": 418000
    },
    {
      "epoch": 8.231773345671478,
      "grad_norm": 2.354146718978882,
      "learning_rate": 4.236777185734989e-07,
      "loss": 1.9959,
      "num_input_tokens_seen": 425711696,
      "step": 418100
    },
    {
      "epoch": 8.23374219842098,
      "grad_norm": 2.2027013301849365,
      "learning_rate": 4.2347396078174035e-07,
      "loss": 1.981,
      "num_input_tokens_seen": 425813368,
      "step": 418200
    },
    {
      "epoch": 8.235711051170483,
      "grad_norm": 1.856339693069458,
      "learning_rate": 4.23270216002589e-07,
      "loss": 1.9627,
      "num_input_tokens_seen": 425915768,
      "step": 418300
    },
    {
      "epoch": 8.237679903919986,
      "grad_norm": 2.4227607250213623,
      "learning_rate": 4.230664842706906e-07,
      "loss": 1.949,
      "num_input_tokens_seen": 426018168,
      "step": 418400
    },
    {
      "epoch": 8.239648756669489,
      "grad_norm": 3.0098276138305664,
      "learning_rate": 4.228627656206873e-07,
      "loss": 1.9839,
      "num_input_tokens_seen": 426120472,
      "step": 418500
    },
    {
      "epoch": 8.241617609418991,
      "grad_norm": 3.137199640274048,
      "learning_rate": 4.226590600872204e-07,
      "loss": 1.9741,
      "num_input_tokens_seen": 426222872,
      "step": 418600
    },
    {
      "epoch": 8.243586462168494,
      "grad_norm": 2.072751045227051,
      "learning_rate": 4.224553677049277e-07,
      "loss": 2.0482,
      "num_input_tokens_seen": 426323976,
      "step": 418700
    },
    {
      "epoch": 8.245555314917997,
      "grad_norm": 2.0092689990997314,
      "learning_rate": 4.222516885084459e-07,
      "loss": 2.0172,
      "num_input_tokens_seen": 426425728,
      "step": 418800
    },
    {
      "epoch": 8.2475241676675,
      "grad_norm": 1.9350968599319458,
      "learning_rate": 4.2204802253240867e-07,
      "loss": 2.0268,
      "num_input_tokens_seen": 426528128,
      "step": 418900
    },
    {
      "epoch": 8.249493020417002,
      "grad_norm": 1.8077954053878784,
      "learning_rate": 4.218443698114477e-07,
      "loss": 1.9851,
      "num_input_tokens_seen": 426629912,
      "step": 419000
    },
    {
      "epoch": 8.251461873166505,
      "grad_norm": 1.7765130996704102,
      "learning_rate": 4.2164073038019255e-07,
      "loss": 1.9909,
      "num_input_tokens_seen": 426730600,
      "step": 419100
    },
    {
      "epoch": 8.253430725916008,
      "grad_norm": 1.8612784147262573,
      "learning_rate": 4.2143710427327016e-07,
      "loss": 1.9292,
      "num_input_tokens_seen": 426832208,
      "step": 419200
    },
    {
      "epoch": 8.25539957866551,
      "grad_norm": 1.9122129678726196,
      "learning_rate": 4.2123349152530567e-07,
      "loss": 2.0009,
      "num_input_tokens_seen": 426934008,
      "step": 419300
    },
    {
      "epoch": 8.257368431415015,
      "grad_norm": 1.730830192565918,
      "learning_rate": 4.2102989217092155e-07,
      "loss": 1.9699,
      "num_input_tokens_seen": 427036408,
      "step": 419400
    },
    {
      "epoch": 8.259337284164518,
      "grad_norm": 2.005462646484375,
      "learning_rate": 4.2082630624473824e-07,
      "loss": 1.9936,
      "num_input_tokens_seen": 427138808,
      "step": 419500
    },
    {
      "epoch": 8.26130613691402,
      "grad_norm": 1.696393370628357,
      "learning_rate": 4.206227337813737e-07,
      "loss": 1.9563,
      "num_input_tokens_seen": 427241208,
      "step": 419600
    },
    {
      "epoch": 8.263274989663524,
      "grad_norm": 1.833376407623291,
      "learning_rate": 4.204191748154439e-07,
      "loss": 1.9736,
      "num_input_tokens_seen": 427342240,
      "step": 419700
    },
    {
      "epoch": 8.265243842413026,
      "grad_norm": 2.1345181465148926,
      "learning_rate": 4.2021562938156204e-07,
      "loss": 1.9145,
      "num_input_tokens_seen": 427444640,
      "step": 419800
    },
    {
      "epoch": 8.267212695162529,
      "grad_norm": 2.0645036697387695,
      "learning_rate": 4.200120975143396e-07,
      "loss": 2.0103,
      "num_input_tokens_seen": 427546264,
      "step": 419900
    },
    {
      "epoch": 8.269181547912032,
      "grad_norm": 1.9077825546264648,
      "learning_rate": 4.198085792483852e-07,
      "loss": 2.0053,
      "num_input_tokens_seen": 427645224,
      "step": 420000
    },
    {
      "epoch": 8.271150400661535,
      "grad_norm": 1.7693476676940918,
      "learning_rate": 4.196050746183056e-07,
      "loss": 1.9874,
      "num_input_tokens_seen": 427747056,
      "step": 420100
    },
    {
      "epoch": 8.273119253411037,
      "grad_norm": 2.262269973754883,
      "learning_rate": 4.1940158365870505e-07,
      "loss": 2.0144,
      "num_input_tokens_seen": 427847912,
      "step": 420200
    },
    {
      "epoch": 8.27508810616054,
      "grad_norm": 2.2255208492279053,
      "learning_rate": 4.191981064041852e-07,
      "loss": 1.9802,
      "num_input_tokens_seen": 427948560,
      "step": 420300
    },
    {
      "epoch": 8.277056958910043,
      "grad_norm": 1.746009349822998,
      "learning_rate": 4.189946428893461e-07,
      "loss": 1.9744,
      "num_input_tokens_seen": 428050336,
      "step": 420400
    },
    {
      "epoch": 8.279025811659546,
      "grad_norm": 2.0518009662628174,
      "learning_rate": 4.1879119314878445e-07,
      "loss": 1.9827,
      "num_input_tokens_seen": 428151904,
      "step": 420500
    },
    {
      "epoch": 8.280994664409048,
      "grad_norm": 2.1446449756622314,
      "learning_rate": 4.185877572170958e-07,
      "loss": 2.0161,
      "num_input_tokens_seen": 428254304,
      "step": 420600
    },
    {
      "epoch": 8.282963517158551,
      "grad_norm": 2.2428677082061768,
      "learning_rate": 4.1838433512887206e-07,
      "loss": 2.0101,
      "num_input_tokens_seen": 428354696,
      "step": 420700
    },
    {
      "epoch": 8.284932369908054,
      "grad_norm": 1.8926867246627808,
      "learning_rate": 4.181809269187041e-07,
      "loss": 1.9907,
      "num_input_tokens_seen": 428457096,
      "step": 420800
    },
    {
      "epoch": 8.286901222657557,
      "grad_norm": 1.7421817779541016,
      "learning_rate": 4.1797753262117906e-07,
      "loss": 2.0325,
      "num_input_tokens_seen": 428559496,
      "step": 420900
    },
    {
      "epoch": 8.28887007540706,
      "grad_norm": 2.029924154281616,
      "learning_rate": 4.177741522708831e-07,
      "loss": 1.9524,
      "num_input_tokens_seen": 428661456,
      "step": 421000
    },
    {
      "epoch": 8.290838928156564,
      "grad_norm": 1.8868544101715088,
      "learning_rate": 4.1757078590239877e-07,
      "loss": 1.9485,
      "num_input_tokens_seen": 428763120,
      "step": 421100
    },
    {
      "epoch": 8.292807780906067,
      "grad_norm": 2.0351948738098145,
      "learning_rate": 4.173674335503071e-07,
      "loss": 1.9618,
      "num_input_tokens_seen": 428865056,
      "step": 421200
    },
    {
      "epoch": 8.29477663365557,
      "grad_norm": 2.06782865524292,
      "learning_rate": 4.171640952491865e-07,
      "loss": 1.959,
      "num_input_tokens_seen": 428967456,
      "step": 421300
    },
    {
      "epoch": 8.296745486405072,
      "grad_norm": 1.8675702810287476,
      "learning_rate": 4.169607710336127e-07,
      "loss": 2.0106,
      "num_input_tokens_seen": 429069320,
      "step": 421400
    },
    {
      "epoch": 8.298714339154575,
      "grad_norm": 2.083636522293091,
      "learning_rate": 4.167574609381596e-07,
      "loss": 2.0069,
      "num_input_tokens_seen": 429171720,
      "step": 421500
    },
    {
      "epoch": 8.300683191904078,
      "grad_norm": 2.0831527709960938,
      "learning_rate": 4.1655416499739805e-07,
      "loss": 1.9547,
      "num_input_tokens_seen": 429273488,
      "step": 421600
    },
    {
      "epoch": 8.30265204465358,
      "grad_norm": 1.8514652252197266,
      "learning_rate": 4.1635088324589697e-07,
      "loss": 1.9648,
      "num_input_tokens_seen": 429375128,
      "step": 421700
    },
    {
      "epoch": 8.304620897403083,
      "grad_norm": 2.076215982437134,
      "learning_rate": 4.161476157182226e-07,
      "loss": 2.0021,
      "num_input_tokens_seen": 429477528,
      "step": 421800
    },
    {
      "epoch": 8.306589750152586,
      "grad_norm": 1.8311114311218262,
      "learning_rate": 4.15944362448939e-07,
      "loss": 1.9753,
      "num_input_tokens_seen": 429579928,
      "step": 421900
    },
    {
      "epoch": 8.308558602902089,
      "grad_norm": 1.9638522863388062,
      "learning_rate": 4.157411234726075e-07,
      "loss": 2.026,
      "num_input_tokens_seen": 429682328,
      "step": 422000
    },
    {
      "epoch": 8.310527455651592,
      "grad_norm": 1.8673714399337769,
      "learning_rate": 4.155378988237873e-07,
      "loss": 1.9869,
      "num_input_tokens_seen": 429783424,
      "step": 422100
    },
    {
      "epoch": 8.312496308401094,
      "grad_norm": 1.8887158632278442,
      "learning_rate": 4.153346885370352e-07,
      "loss": 1.9707,
      "num_input_tokens_seen": 429885824,
      "step": 422200
    },
    {
      "epoch": 8.314465161150597,
      "grad_norm": 2.0292675495147705,
      "learning_rate": 4.15131492646905e-07,
      "loss": 1.9664,
      "num_input_tokens_seen": 429987752,
      "step": 422300
    },
    {
      "epoch": 8.3164340139001,
      "grad_norm": 1.9890892505645752,
      "learning_rate": 4.149283111879488e-07,
      "loss": 1.9944,
      "num_input_tokens_seen": 430089480,
      "step": 422400
    },
    {
      "epoch": 8.318402866649603,
      "grad_norm": 2.198776960372925,
      "learning_rate": 4.147251441947157e-07,
      "loss": 1.963,
      "num_input_tokens_seen": 430191200,
      "step": 422500
    },
    {
      "epoch": 8.320371719399105,
      "grad_norm": 1.9108864068984985,
      "learning_rate": 4.145219917017526e-07,
      "loss": 1.9883,
      "num_input_tokens_seen": 430293600,
      "step": 422600
    },
    {
      "epoch": 8.322340572148608,
      "grad_norm": 1.9309444427490234,
      "learning_rate": 4.1431885374360386e-07,
      "loss": 1.9865,
      "num_input_tokens_seen": 430393568,
      "step": 422700
    },
    {
      "epoch": 8.324309424898113,
      "grad_norm": 1.9458986520767212,
      "learning_rate": 4.141157303548114e-07,
      "loss": 1.9893,
      "num_input_tokens_seen": 430495968,
      "step": 422800
    },
    {
      "epoch": 8.326278277647615,
      "grad_norm": 2.189464807510376,
      "learning_rate": 4.1391262156991454e-07,
      "loss": 1.9625,
      "num_input_tokens_seen": 430598368,
      "step": 422900
    },
    {
      "epoch": 8.328247130397118,
      "grad_norm": 1.830135464668274,
      "learning_rate": 4.137095274234504e-07,
      "loss": 1.9965,
      "num_input_tokens_seen": 430699680,
      "step": 423000
    },
    {
      "epoch": 8.330215983146621,
      "grad_norm": 2.2409379482269287,
      "learning_rate": 4.1350644794995315e-07,
      "loss": 2.0005,
      "num_input_tokens_seen": 430801344,
      "step": 423100
    },
    {
      "epoch": 8.332184835896124,
      "grad_norm": 2.0005760192871094,
      "learning_rate": 4.133033831839549e-07,
      "loss": 2.0115,
      "num_input_tokens_seen": 430902008,
      "step": 423200
    },
    {
      "epoch": 8.334153688645626,
      "grad_norm": 2.054933786392212,
      "learning_rate": 4.131003331599853e-07,
      "loss": 1.9885,
      "num_input_tokens_seen": 431004408,
      "step": 423300
    },
    {
      "epoch": 8.33612254139513,
      "grad_norm": 1.8333909511566162,
      "learning_rate": 4.128972979125709e-07,
      "loss": 1.9816,
      "num_input_tokens_seen": 431105448,
      "step": 423400
    },
    {
      "epoch": 8.338091394144632,
      "grad_norm": 2.044424533843994,
      "learning_rate": 4.126942774762365e-07,
      "loss": 1.992,
      "num_input_tokens_seen": 431207848,
      "step": 423500
    },
    {
      "epoch": 8.340060246894135,
      "grad_norm": 2.124260187149048,
      "learning_rate": 4.124912718855037e-07,
      "loss": 1.9986,
      "num_input_tokens_seen": 431309032,
      "step": 423600
    },
    {
      "epoch": 8.342029099643637,
      "grad_norm": 2.029766082763672,
      "learning_rate": 4.1228828117489225e-07,
      "loss": 1.9755,
      "num_input_tokens_seen": 431410840,
      "step": 423700
    },
    {
      "epoch": 8.34399795239314,
      "grad_norm": 1.7196077108383179,
      "learning_rate": 4.120853053789185e-07,
      "loss": 1.9747,
      "num_input_tokens_seen": 431512200,
      "step": 423800
    },
    {
      "epoch": 8.345966805142643,
      "grad_norm": 1.8024121522903442,
      "learning_rate": 4.1188234453209735e-07,
      "loss": 1.9553,
      "num_input_tokens_seen": 431614600,
      "step": 423900
    },
    {
      "epoch": 8.347935657892146,
      "grad_norm": 1.8844993114471436,
      "learning_rate": 4.1167939866894e-07,
      "loss": 1.9603,
      "num_input_tokens_seen": 431716448,
      "step": 424000
    },
    {
      "epoch": 8.349904510641649,
      "grad_norm": 2.231053590774536,
      "learning_rate": 4.1147646782395614e-07,
      "loss": 1.9783,
      "num_input_tokens_seen": 431817920,
      "step": 424100
    },
    {
      "epoch": 8.351873363391151,
      "grad_norm": 2.028012990951538,
      "learning_rate": 4.112735520316523e-07,
      "loss": 1.9808,
      "num_input_tokens_seen": 431919736,
      "step": 424200
    },
    {
      "epoch": 8.353842216140654,
      "grad_norm": 1.9216039180755615,
      "learning_rate": 4.110706513265325e-07,
      "loss": 1.9357,
      "num_input_tokens_seen": 432020928,
      "step": 424300
    },
    {
      "epoch": 8.355811068890159,
      "grad_norm": 1.8875792026519775,
      "learning_rate": 4.108677657430986e-07,
      "loss": 1.9918,
      "num_input_tokens_seen": 432121480,
      "step": 424400
    },
    {
      "epoch": 8.357779921639661,
      "grad_norm": 2.23197340965271,
      "learning_rate": 4.106648953158491e-07,
      "loss": 1.9791,
      "num_input_tokens_seen": 432223008,
      "step": 424500
    },
    {
      "epoch": 8.359748774389164,
      "grad_norm": 1.9889016151428223,
      "learning_rate": 4.1046204007928083e-07,
      "loss": 1.9955,
      "num_input_tokens_seen": 432324496,
      "step": 424600
    },
    {
      "epoch": 8.361717627138667,
      "grad_norm": 1.8320786952972412,
      "learning_rate": 4.102592000678873e-07,
      "loss": 1.9879,
      "num_input_tokens_seen": 432426896,
      "step": 424700
    },
    {
      "epoch": 8.36368647988817,
      "grad_norm": 1.7026339769363403,
      "learning_rate": 4.1005637531616006e-07,
      "loss": 1.9809,
      "num_input_tokens_seen": 432529208,
      "step": 424800
    },
    {
      "epoch": 8.365655332637672,
      "grad_norm": 1.7036831378936768,
      "learning_rate": 4.098535658585874e-07,
      "loss": 1.9566,
      "num_input_tokens_seen": 432631608,
      "step": 424900
    },
    {
      "epoch": 8.367624185387175,
      "grad_norm": 1.6798295974731445,
      "learning_rate": 4.096507717296557e-07,
      "loss": 2.0003,
      "num_input_tokens_seen": 432733384,
      "step": 425000
    },
    {
      "epoch": 8.369593038136678,
      "grad_norm": 1.8298149108886719,
      "learning_rate": 4.0944799296384813e-07,
      "loss": 1.9813,
      "num_input_tokens_seen": 432835784,
      "step": 425100
    },
    {
      "epoch": 8.37156189088618,
      "grad_norm": 3.7572927474975586,
      "learning_rate": 4.092452295956456e-07,
      "loss": 2.0267,
      "num_input_tokens_seen": 432936896,
      "step": 425200
    },
    {
      "epoch": 8.373530743635683,
      "grad_norm": 1.8785380125045776,
      "learning_rate": 4.090424816595264e-07,
      "loss": 1.9489,
      "num_input_tokens_seen": 433038616,
      "step": 425300
    },
    {
      "epoch": 8.375499596385186,
      "grad_norm": 1.6943247318267822,
      "learning_rate": 4.0883974918996587e-07,
      "loss": 1.9718,
      "num_input_tokens_seen": 433141016,
      "step": 425400
    },
    {
      "epoch": 8.377468449134689,
      "grad_norm": 1.8722991943359375,
      "learning_rate": 4.0863703222143725e-07,
      "loss": 1.9749,
      "num_input_tokens_seen": 433242080,
      "step": 425500
    },
    {
      "epoch": 8.379437301884192,
      "grad_norm": 1.8121262788772583,
      "learning_rate": 4.084343307884106e-07,
      "loss": 1.9709,
      "num_input_tokens_seen": 433343856,
      "step": 425600
    },
    {
      "epoch": 8.381406154633694,
      "grad_norm": 1.9046111106872559,
      "learning_rate": 4.0823164492535383e-07,
      "loss": 2.0071,
      "num_input_tokens_seen": 433446256,
      "step": 425700
    },
    {
      "epoch": 8.383375007383197,
      "grad_norm": 1.6672890186309814,
      "learning_rate": 4.080289746667317e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 433548656,
      "step": 425800
    },
    {
      "epoch": 8.3853438601327,
      "grad_norm": 1.9436790943145752,
      "learning_rate": 4.0782632004700685e-07,
      "loss": 2.0125,
      "num_input_tokens_seen": 433649784,
      "step": 425900
    },
    {
      "epoch": 8.387312712882203,
      "grad_norm": 1.7911956310272217,
      "learning_rate": 4.0762368110063874e-07,
      "loss": 1.9535,
      "num_input_tokens_seen": 433751712,
      "step": 426000
    },
    {
      "epoch": 8.389281565631707,
      "grad_norm": 2.1295416355133057,
      "learning_rate": 4.0742105786208456e-07,
      "loss": 1.9249,
      "num_input_tokens_seen": 433854112,
      "step": 426100
    },
    {
      "epoch": 8.39125041838121,
      "grad_norm": 1.8713308572769165,
      "learning_rate": 4.072184503657986e-07,
      "loss": 1.9851,
      "num_input_tokens_seen": 433955736,
      "step": 426200
    },
    {
      "epoch": 8.393219271130713,
      "grad_norm": 1.8352092504501343,
      "learning_rate": 4.070158586462325e-07,
      "loss": 2.0063,
      "num_input_tokens_seen": 434057696,
      "step": 426300
    },
    {
      "epoch": 8.395188123880216,
      "grad_norm": 1.8988292217254639,
      "learning_rate": 4.0681328273783565e-07,
      "loss": 2.0578,
      "num_input_tokens_seen": 434159568,
      "step": 426400
    },
    {
      "epoch": 8.397156976629718,
      "grad_norm": 1.7122265100479126,
      "learning_rate": 4.066107226750538e-07,
      "loss": 1.9879,
      "num_input_tokens_seen": 434261768,
      "step": 426500
    },
    {
      "epoch": 8.399125829379221,
      "grad_norm": 1.8604533672332764,
      "learning_rate": 4.0640817849233124e-07,
      "loss": 2.0116,
      "num_input_tokens_seen": 434363464,
      "step": 426600
    },
    {
      "epoch": 8.401094682128724,
      "grad_norm": 1.8360460996627808,
      "learning_rate": 4.062056502241081e-07,
      "loss": 2.0201,
      "num_input_tokens_seen": 434465864,
      "step": 426700
    },
    {
      "epoch": 8.403063534878227,
      "grad_norm": 2.057734727859497,
      "learning_rate": 4.0600313790482347e-07,
      "loss": 2.0158,
      "num_input_tokens_seen": 434567000,
      "step": 426800
    },
    {
      "epoch": 8.40503238762773,
      "grad_norm": 3.415806770324707,
      "learning_rate": 4.058006415689121e-07,
      "loss": 2.0208,
      "num_input_tokens_seen": 434668168,
      "step": 426900
    },
    {
      "epoch": 8.407001240377232,
      "grad_norm": 23.310867309570312,
      "learning_rate": 4.0559816125080733e-07,
      "loss": 2.0415,
      "num_input_tokens_seen": 434769040,
      "step": 427000
    },
    {
      "epoch": 8.408970093126735,
      "grad_norm": 2.0601069927215576,
      "learning_rate": 4.053956969849389e-07,
      "loss": 1.9822,
      "num_input_tokens_seen": 434870896,
      "step": 427100
    },
    {
      "epoch": 8.410938945876238,
      "grad_norm": 1.8918333053588867,
      "learning_rate": 4.0519324880573423e-07,
      "loss": 1.9624,
      "num_input_tokens_seen": 434972520,
      "step": 427200
    },
    {
      "epoch": 8.41290779862574,
      "grad_norm": 2.754359722137451,
      "learning_rate": 4.049908167476181e-07,
      "loss": 1.9915,
      "num_input_tokens_seen": 435074920,
      "step": 427300
    },
    {
      "epoch": 8.414876651375243,
      "grad_norm": 2.1084115505218506,
      "learning_rate": 4.0478840084501207e-07,
      "loss": 1.9835,
      "num_input_tokens_seen": 435177320,
      "step": 427400
    },
    {
      "epoch": 8.416845504124746,
      "grad_norm": 1.9508016109466553,
      "learning_rate": 4.0458600113233566e-07,
      "loss": 1.9841,
      "num_input_tokens_seen": 435279088,
      "step": 427500
    },
    {
      "epoch": 8.418814356874249,
      "grad_norm": 1.9915024042129517,
      "learning_rate": 4.043836176440048e-07,
      "loss": 1.9733,
      "num_input_tokens_seen": 435381488,
      "step": 427600
    },
    {
      "epoch": 8.420783209623753,
      "grad_norm": 1.9296541213989258,
      "learning_rate": 4.041812504144334e-07,
      "loss": 2.0083,
      "num_input_tokens_seen": 435483144,
      "step": 427700
    },
    {
      "epoch": 8.422752062373256,
      "grad_norm": 2.1648142337799072,
      "learning_rate": 4.039788994780321e-07,
      "loss": 1.9969,
      "num_input_tokens_seen": 435584784,
      "step": 427800
    },
    {
      "epoch": 8.424720915122759,
      "grad_norm": 2.1455576419830322,
      "learning_rate": 4.037765648692092e-07,
      "loss": 1.9947,
      "num_input_tokens_seen": 435685320,
      "step": 427900
    },
    {
      "epoch": 8.426689767872261,
      "grad_norm": 1.8843876123428345,
      "learning_rate": 4.0357424662236974e-07,
      "loss": 2.0373,
      "num_input_tokens_seen": 435785712,
      "step": 428000
    },
    {
      "epoch": 8.428658620621764,
      "grad_norm": 1.9977744817733765,
      "learning_rate": 4.0337194477191643e-07,
      "loss": 2.0173,
      "num_input_tokens_seen": 435886168,
      "step": 428100
    },
    {
      "epoch": 8.430627473371267,
      "grad_norm": 1.967344045639038,
      "learning_rate": 4.031696593522488e-07,
      "loss": 1.9716,
      "num_input_tokens_seen": 435988568,
      "step": 428200
    },
    {
      "epoch": 8.43259632612077,
      "grad_norm": 1.6700453758239746,
      "learning_rate": 4.0296739039776394e-07,
      "loss": 1.9326,
      "num_input_tokens_seen": 436090968,
      "step": 428300
    },
    {
      "epoch": 8.434565178870272,
      "grad_norm": 1.8812015056610107,
      "learning_rate": 4.02765137942856e-07,
      "loss": 1.9479,
      "num_input_tokens_seen": 436191576,
      "step": 428400
    },
    {
      "epoch": 8.436534031619775,
      "grad_norm": 1.8461774587631226,
      "learning_rate": 4.0256290202191615e-07,
      "loss": 2.0043,
      "num_input_tokens_seen": 436293976,
      "step": 428500
    },
    {
      "epoch": 8.438502884369278,
      "grad_norm": 2.5795412063598633,
      "learning_rate": 4.023606826693331e-07,
      "loss": 1.9447,
      "num_input_tokens_seen": 436396376,
      "step": 428600
    },
    {
      "epoch": 8.44047173711878,
      "grad_norm": 2.1136581897735596,
      "learning_rate": 4.0215847991949226e-07,
      "loss": 1.969,
      "num_input_tokens_seen": 436498776,
      "step": 428700
    },
    {
      "epoch": 8.442440589868283,
      "grad_norm": 2.1589603424072266,
      "learning_rate": 4.0195629380677695e-07,
      "loss": 1.9825,
      "num_input_tokens_seen": 436599592,
      "step": 428800
    },
    {
      "epoch": 8.444409442617786,
      "grad_norm": 1.8121565580368042,
      "learning_rate": 4.017541243655666e-07,
      "loss": 2.0007,
      "num_input_tokens_seen": 436701992,
      "step": 428900
    },
    {
      "epoch": 8.446378295367289,
      "grad_norm": 2.016279935836792,
      "learning_rate": 4.015519716302391e-07,
      "loss": 1.9881,
      "num_input_tokens_seen": 436804392,
      "step": 429000
    },
    {
      "epoch": 8.448347148116792,
      "grad_norm": 2.1842615604400635,
      "learning_rate": 4.013498356351681e-07,
      "loss": 1.9946,
      "num_input_tokens_seen": 436905456,
      "step": 429100
    },
    {
      "epoch": 8.450316000866295,
      "grad_norm": 1.8275010585784912,
      "learning_rate": 4.011477164147257e-07,
      "loss": 1.9784,
      "num_input_tokens_seen": 437006192,
      "step": 429200
    },
    {
      "epoch": 8.452284853615797,
      "grad_norm": 1.9177550077438354,
      "learning_rate": 4.009456140032804e-07,
      "loss": 1.9949,
      "num_input_tokens_seen": 437108592,
      "step": 429300
    },
    {
      "epoch": 8.4542537063653,
      "grad_norm": 2.019090414047241,
      "learning_rate": 4.0074352843519787e-07,
      "loss": 2.0295,
      "num_input_tokens_seen": 437208776,
      "step": 429400
    },
    {
      "epoch": 8.456222559114805,
      "grad_norm": 1.9815258979797363,
      "learning_rate": 4.0054145974484134e-07,
      "loss": 1.9856,
      "num_input_tokens_seen": 437309936,
      "step": 429500
    },
    {
      "epoch": 8.458191411864307,
      "grad_norm": 1.8791011571884155,
      "learning_rate": 4.003394079665705e-07,
      "loss": 1.9485,
      "num_input_tokens_seen": 437412104,
      "step": 429600
    },
    {
      "epoch": 8.46016026461381,
      "grad_norm": 1.783757209777832,
      "learning_rate": 4.0013737313474297e-07,
      "loss": 2.0115,
      "num_input_tokens_seen": 437513832,
      "step": 429700
    },
    {
      "epoch": 8.462129117363313,
      "grad_norm": 1.8619985580444336,
      "learning_rate": 3.999353552837127e-07,
      "loss": 1.9659,
      "num_input_tokens_seen": 437615416,
      "step": 429800
    },
    {
      "epoch": 8.464097970112816,
      "grad_norm": 1.8177839517593384,
      "learning_rate": 3.997333544478314e-07,
      "loss": 1.9783,
      "num_input_tokens_seen": 437717816,
      "step": 429900
    },
    {
      "epoch": 8.466066822862318,
      "grad_norm": 2.1941258907318115,
      "learning_rate": 3.9953137066144736e-07,
      "loss": 1.9966,
      "num_input_tokens_seen": 437820216,
      "step": 430000
    },
    {
      "epoch": 8.468035675611821,
      "grad_norm": 1.86483895778656,
      "learning_rate": 3.993294039589065e-07,
      "loss": 1.9953,
      "num_input_tokens_seen": 437921048,
      "step": 430100
    },
    {
      "epoch": 8.470004528361324,
      "grad_norm": 1.8158870935440063,
      "learning_rate": 3.991274543745512e-07,
      "loss": 1.9526,
      "num_input_tokens_seen": 438023448,
      "step": 430200
    },
    {
      "epoch": 8.471973381110827,
      "grad_norm": 1.7289897203445435,
      "learning_rate": 3.9892552194272156e-07,
      "loss": 1.9768,
      "num_input_tokens_seen": 438125848,
      "step": 430300
    },
    {
      "epoch": 8.47394223386033,
      "grad_norm": 1.8524589538574219,
      "learning_rate": 3.987236066977545e-07,
      "loss": 2.0123,
      "num_input_tokens_seen": 438227656,
      "step": 430400
    },
    {
      "epoch": 8.475911086609832,
      "grad_norm": 1.8160289525985718,
      "learning_rate": 3.9852170867398385e-07,
      "loss": 2.0069,
      "num_input_tokens_seen": 438329464,
      "step": 430500
    },
    {
      "epoch": 8.477879939359335,
      "grad_norm": 1.9283607006072998,
      "learning_rate": 3.9831982790574074e-07,
      "loss": 1.9812,
      "num_input_tokens_seen": 438430968,
      "step": 430600
    },
    {
      "epoch": 8.479848792108838,
      "grad_norm": 6.773560047149658,
      "learning_rate": 3.9811796442735326e-07,
      "loss": 1.9583,
      "num_input_tokens_seen": 438533368,
      "step": 430700
    },
    {
      "epoch": 8.48181764485834,
      "grad_norm": 1.9884876012802124,
      "learning_rate": 3.979161182731467e-07,
      "loss": 2.0006,
      "num_input_tokens_seen": 438635328,
      "step": 430800
    },
    {
      "epoch": 8.483786497607843,
      "grad_norm": 2.183328628540039,
      "learning_rate": 3.977142894774431e-07,
      "loss": 2.0123,
      "num_input_tokens_seen": 438737728,
      "step": 430900
    },
    {
      "epoch": 8.485755350357346,
      "grad_norm": 1.9174261093139648,
      "learning_rate": 3.97512478074562e-07,
      "loss": 1.9989,
      "num_input_tokens_seen": 438840128,
      "step": 431000
    },
    {
      "epoch": 8.48772420310685,
      "grad_norm": 2.13865065574646,
      "learning_rate": 3.973106840988194e-07,
      "loss": 1.9562,
      "num_input_tokens_seen": 438942184,
      "step": 431100
    },
    {
      "epoch": 8.489693055856353,
      "grad_norm": 1.809452772140503,
      "learning_rate": 3.97108907584529e-07,
      "loss": 1.9784,
      "num_input_tokens_seen": 439044584,
      "step": 431200
    },
    {
      "epoch": 8.491661908605856,
      "grad_norm": 2.4269907474517822,
      "learning_rate": 3.969071485660009e-07,
      "loss": 1.9849,
      "num_input_tokens_seen": 439146984,
      "step": 431300
    },
    {
      "epoch": 8.493630761355359,
      "grad_norm": 1.9628105163574219,
      "learning_rate": 3.967054070775426e-07,
      "loss": 2.0059,
      "num_input_tokens_seen": 439249384,
      "step": 431400
    },
    {
      "epoch": 8.495599614104862,
      "grad_norm": 2.2871105670928955,
      "learning_rate": 3.965036831534589e-07,
      "loss": 1.9808,
      "num_input_tokens_seen": 439351784,
      "step": 431500
    },
    {
      "epoch": 8.497568466854364,
      "grad_norm": 1.9083971977233887,
      "learning_rate": 3.963019768280507e-07,
      "loss": 1.9865,
      "num_input_tokens_seen": 439454184,
      "step": 431600
    },
    {
      "epoch": 8.499537319603867,
      "grad_norm": 2.2168989181518555,
      "learning_rate": 3.9610028813561696e-07,
      "loss": 1.9583,
      "num_input_tokens_seen": 439556088,
      "step": 431700
    },
    {
      "epoch": 8.50150617235337,
      "grad_norm": 2.5334222316741943,
      "learning_rate": 3.9589861711045257e-07,
      "loss": 1.9762,
      "num_input_tokens_seen": 439657728,
      "step": 431800
    },
    {
      "epoch": 8.503475025102873,
      "grad_norm": 1.8706711530685425,
      "learning_rate": 3.956969637868507e-07,
      "loss": 1.9874,
      "num_input_tokens_seen": 439760128,
      "step": 431900
    },
    {
      "epoch": 8.505443877852375,
      "grad_norm": 2.2788848876953125,
      "learning_rate": 3.954953281991e-07,
      "loss": 1.9648,
      "num_input_tokens_seen": 439862528,
      "step": 432000
    },
    {
      "epoch": 8.507412730601878,
      "grad_norm": 2.0718917846679688,
      "learning_rate": 3.9529371038148765e-07,
      "loss": 1.9935,
      "num_input_tokens_seen": 439964928,
      "step": 432100
    },
    {
      "epoch": 8.50938158335138,
      "grad_norm": 2.064072847366333,
      "learning_rate": 3.9509211036829644e-07,
      "loss": 1.9885,
      "num_input_tokens_seen": 440067328,
      "step": 432200
    },
    {
      "epoch": 8.511350436100884,
      "grad_norm": 1.813452124595642,
      "learning_rate": 3.948905281938071e-07,
      "loss": 1.9709,
      "num_input_tokens_seen": 440168432,
      "step": 432300
    },
    {
      "epoch": 8.513319288850386,
      "grad_norm": 1.9435439109802246,
      "learning_rate": 3.94688963892297e-07,
      "loss": 2.0214,
      "num_input_tokens_seen": 440270032,
      "step": 432400
    },
    {
      "epoch": 8.51528814159989,
      "grad_norm": 1.7269937992095947,
      "learning_rate": 3.9448741749804027e-07,
      "loss": 1.9927,
      "num_input_tokens_seen": 440371184,
      "step": 432500
    },
    {
      "epoch": 8.517256994349392,
      "grad_norm": 1.9344353675842285,
      "learning_rate": 3.9428588904530836e-07,
      "loss": 2.0034,
      "num_input_tokens_seen": 440473584,
      "step": 432600
    },
    {
      "epoch": 8.519225847098895,
      "grad_norm": 2.3893494606018066,
      "learning_rate": 3.940843785683692e-07,
      "loss": 1.9497,
      "num_input_tokens_seen": 440575384,
      "step": 432700
    },
    {
      "epoch": 8.5211946998484,
      "grad_norm": 2.681697130203247,
      "learning_rate": 3.9388288610148833e-07,
      "loss": 1.9514,
      "num_input_tokens_seen": 440677712,
      "step": 432800
    },
    {
      "epoch": 8.523163552597902,
      "grad_norm": 1.8612326383590698,
      "learning_rate": 3.9368141167892756e-07,
      "loss": 1.9723,
      "num_input_tokens_seen": 440779680,
      "step": 432900
    },
    {
      "epoch": 8.525132405347405,
      "grad_norm": 1.8434288501739502,
      "learning_rate": 3.9347995533494605e-07,
      "loss": 1.9724,
      "num_input_tokens_seen": 440880920,
      "step": 433000
    },
    {
      "epoch": 8.527101258096907,
      "grad_norm": 2.232220411300659,
      "learning_rate": 3.932785171037996e-07,
      "loss": 1.9915,
      "num_input_tokens_seen": 440983320,
      "step": 433100
    },
    {
      "epoch": 8.52907011084641,
      "grad_norm": 2.323387384414673,
      "learning_rate": 3.9307709701974135e-07,
      "loss": 1.9952,
      "num_input_tokens_seen": 441082520,
      "step": 433200
    },
    {
      "epoch": 8.531038963595913,
      "grad_norm": 1.8952317237854004,
      "learning_rate": 3.928756951170208e-07,
      "loss": 1.9617,
      "num_input_tokens_seen": 441184920,
      "step": 433300
    },
    {
      "epoch": 8.533007816345416,
      "grad_norm": 1.9474390745162964,
      "learning_rate": 3.9267431142988473e-07,
      "loss": 1.9576,
      "num_input_tokens_seen": 441286760,
      "step": 433400
    },
    {
      "epoch": 8.534976669094918,
      "grad_norm": 2.042001485824585,
      "learning_rate": 3.9247294599257683e-07,
      "loss": 1.996,
      "num_input_tokens_seen": 441389160,
      "step": 433500
    },
    {
      "epoch": 8.536945521844421,
      "grad_norm": 1.6450037956237793,
      "learning_rate": 3.922715988393375e-07,
      "loss": 1.9503,
      "num_input_tokens_seen": 441491560,
      "step": 433600
    },
    {
      "epoch": 8.538914374593924,
      "grad_norm": 1.9197132587432861,
      "learning_rate": 3.9207027000440427e-07,
      "loss": 1.9922,
      "num_input_tokens_seen": 441593128,
      "step": 433700
    },
    {
      "epoch": 8.540883227343427,
      "grad_norm": 2.1575756072998047,
      "learning_rate": 3.918689595220112e-07,
      "loss": 1.9514,
      "num_input_tokens_seen": 441694936,
      "step": 433800
    },
    {
      "epoch": 8.54285208009293,
      "grad_norm": 2.118551015853882,
      "learning_rate": 3.916676674263897e-07,
      "loss": 1.9915,
      "num_input_tokens_seen": 441797336,
      "step": 433900
    },
    {
      "epoch": 8.544820932842432,
      "grad_norm": 2.099083185195923,
      "learning_rate": 3.914663937517674e-07,
      "loss": 2.0036,
      "num_input_tokens_seen": 441899256,
      "step": 434000
    },
    {
      "epoch": 8.546789785591935,
      "grad_norm": 1.927480697631836,
      "learning_rate": 3.9126513853236966e-07,
      "loss": 1.9632,
      "num_input_tokens_seen": 442001088,
      "step": 434100
    },
    {
      "epoch": 8.548758638341438,
      "grad_norm": 2.0223116874694824,
      "learning_rate": 3.9106390180241784e-07,
      "loss": 1.9731,
      "num_input_tokens_seen": 442102624,
      "step": 434200
    },
    {
      "epoch": 8.55072749109094,
      "grad_norm": 2.0387868881225586,
      "learning_rate": 3.908626835961306e-07,
      "loss": 2.0017,
      "num_input_tokens_seen": 442204096,
      "step": 434300
    },
    {
      "epoch": 8.552696343840445,
      "grad_norm": 1.9302129745483398,
      "learning_rate": 3.906614839477238e-07,
      "loss": 1.9831,
      "num_input_tokens_seen": 442304768,
      "step": 434400
    },
    {
      "epoch": 8.554665196589948,
      "grad_norm": 2.205199718475342,
      "learning_rate": 3.9046030289140916e-07,
      "loss": 2.0015,
      "num_input_tokens_seen": 442407168,
      "step": 434500
    },
    {
      "epoch": 8.55663404933945,
      "grad_norm": 1.7624105215072632,
      "learning_rate": 3.902591404613965e-07,
      "loss": 2.0227,
      "num_input_tokens_seen": 442507768,
      "step": 434600
    },
    {
      "epoch": 8.558602902088953,
      "grad_norm": 6.508541584014893,
      "learning_rate": 3.90057996691891e-07,
      "loss": 2.0247,
      "num_input_tokens_seen": 442608272,
      "step": 434700
    },
    {
      "epoch": 8.560571754838456,
      "grad_norm": 1.8274939060211182,
      "learning_rate": 3.898568716170962e-07,
      "loss": 1.963,
      "num_input_tokens_seen": 442710216,
      "step": 434800
    },
    {
      "epoch": 8.562540607587959,
      "grad_norm": 2.1489410400390625,
      "learning_rate": 3.8965576527121106e-07,
      "loss": 1.9748,
      "num_input_tokens_seen": 442811976,
      "step": 434900
    },
    {
      "epoch": 8.564509460337462,
      "grad_norm": 2.253645896911621,
      "learning_rate": 3.894546776884327e-07,
      "loss": 2.0141,
      "num_input_tokens_seen": 442913288,
      "step": 435000
    },
    {
      "epoch": 8.566478313086964,
      "grad_norm": 1.892430067062378,
      "learning_rate": 3.892536089029537e-07,
      "loss": 1.9889,
      "num_input_tokens_seen": 443015688,
      "step": 435100
    },
    {
      "epoch": 8.568447165836467,
      "grad_norm": 1.7716001272201538,
      "learning_rate": 3.8905255894896475e-07,
      "loss": 1.9583,
      "num_input_tokens_seen": 443118088,
      "step": 435200
    },
    {
      "epoch": 8.57041601858597,
      "grad_norm": 2.080090045928955,
      "learning_rate": 3.8885152786065204e-07,
      "loss": 2.046,
      "num_input_tokens_seen": 443219720,
      "step": 435300
    },
    {
      "epoch": 8.572384871335473,
      "grad_norm": 2.0503625869750977,
      "learning_rate": 3.886505156721996e-07,
      "loss": 1.9642,
      "num_input_tokens_seen": 443322120,
      "step": 435400
    },
    {
      "epoch": 8.574353724084975,
      "grad_norm": 1.9515348672866821,
      "learning_rate": 3.8844952241778807e-07,
      "loss": 1.9432,
      "num_input_tokens_seen": 443424432,
      "step": 435500
    },
    {
      "epoch": 8.576322576834478,
      "grad_norm": 1.940108060836792,
      "learning_rate": 3.882485481315941e-07,
      "loss": 1.9411,
      "num_input_tokens_seen": 443526096,
      "step": 435600
    },
    {
      "epoch": 8.578291429583981,
      "grad_norm": 1.890031337738037,
      "learning_rate": 3.880475928477922e-07,
      "loss": 1.964,
      "num_input_tokens_seen": 443628200,
      "step": 435700
    },
    {
      "epoch": 8.580260282333484,
      "grad_norm": 1.9414799213409424,
      "learning_rate": 3.878466566005528e-07,
      "loss": 1.9801,
      "num_input_tokens_seen": 443730600,
      "step": 435800
    },
    {
      "epoch": 8.582229135082986,
      "grad_norm": 1.7501819133758545,
      "learning_rate": 3.876457394240435e-07,
      "loss": 1.9669,
      "num_input_tokens_seen": 443832128,
      "step": 435900
    },
    {
      "epoch": 8.58419798783249,
      "grad_norm": 1.8722598552703857,
      "learning_rate": 3.874448413524285e-07,
      "loss": 1.978,
      "num_input_tokens_seen": 443932944,
      "step": 436000
    },
    {
      "epoch": 8.586166840581992,
      "grad_norm": 1.9408739805221558,
      "learning_rate": 3.87243962419869e-07,
      "loss": 2.0172,
      "num_input_tokens_seen": 444035344,
      "step": 436100
    },
    {
      "epoch": 8.588135693331497,
      "grad_norm": 1.9807482957839966,
      "learning_rate": 3.8704310266052257e-07,
      "loss": 1.9877,
      "num_input_tokens_seen": 444137744,
      "step": 436200
    },
    {
      "epoch": 8.590104546081,
      "grad_norm": 1.986048698425293,
      "learning_rate": 3.868422621085439e-07,
      "loss": 1.9757,
      "num_input_tokens_seen": 444240144,
      "step": 436300
    },
    {
      "epoch": 8.592073398830502,
      "grad_norm": 3.6090807914733887,
      "learning_rate": 3.866414407980841e-07,
      "loss": 1.9651,
      "num_input_tokens_seen": 444342008,
      "step": 436400
    },
    {
      "epoch": 8.594042251580005,
      "grad_norm": 1.8517863750457764,
      "learning_rate": 3.8644063876329114e-07,
      "loss": 1.9837,
      "num_input_tokens_seen": 444442616,
      "step": 436500
    },
    {
      "epoch": 8.596011104329508,
      "grad_norm": 1.7607098817825317,
      "learning_rate": 3.862398560383098e-07,
      "loss": 1.9728,
      "num_input_tokens_seen": 444543928,
      "step": 436600
    },
    {
      "epoch": 8.59797995707901,
      "grad_norm": 2.03020977973938,
      "learning_rate": 3.860390926572814e-07,
      "loss": 1.9836,
      "num_input_tokens_seen": 444646328,
      "step": 436700
    },
    {
      "epoch": 8.599948809828513,
      "grad_norm": 1.9390146732330322,
      "learning_rate": 3.858383486543442e-07,
      "loss": 1.9898,
      "num_input_tokens_seen": 444748176,
      "step": 436800
    },
    {
      "epoch": 8.601917662578016,
      "grad_norm": 1.862971544265747,
      "learning_rate": 3.8563762406363284e-07,
      "loss": 1.9241,
      "num_input_tokens_seen": 444849872,
      "step": 436900
    },
    {
      "epoch": 8.603886515327519,
      "grad_norm": 2.3055028915405273,
      "learning_rate": 3.8543691891927896e-07,
      "loss": 2.0357,
      "num_input_tokens_seen": 444950808,
      "step": 437000
    },
    {
      "epoch": 8.605855368077021,
      "grad_norm": 2.1133251190185547,
      "learning_rate": 3.852362332554107e-07,
      "loss": 2.007,
      "num_input_tokens_seen": 445053208,
      "step": 437100
    },
    {
      "epoch": 8.607824220826524,
      "grad_norm": 2.1245574951171875,
      "learning_rate": 3.8503556710615316e-07,
      "loss": 1.9548,
      "num_input_tokens_seen": 445155040,
      "step": 437200
    },
    {
      "epoch": 8.609793073576027,
      "grad_norm": 1.8872050046920776,
      "learning_rate": 3.8483492050562747e-07,
      "loss": 1.9922,
      "num_input_tokens_seen": 445256480,
      "step": 437300
    },
    {
      "epoch": 8.61176192632553,
      "grad_norm": 1.849798321723938,
      "learning_rate": 3.846342934879524e-07,
      "loss": 1.9615,
      "num_input_tokens_seen": 445358880,
      "step": 437400
    },
    {
      "epoch": 8.613730779075032,
      "grad_norm": 2.1888458728790283,
      "learning_rate": 3.8443368608724274e-07,
      "loss": 2.022,
      "num_input_tokens_seen": 445460488,
      "step": 437500
    },
    {
      "epoch": 8.615699631824535,
      "grad_norm": 1.822311282157898,
      "learning_rate": 3.8423309833761e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 445562888,
      "step": 437600
    },
    {
      "epoch": 8.617668484574038,
      "grad_norm": 1.8156375885009766,
      "learning_rate": 3.8403253027316246e-07,
      "loss": 1.9618,
      "num_input_tokens_seen": 445664008,
      "step": 437700
    },
    {
      "epoch": 8.619637337323542,
      "grad_norm": 1.8686310052871704,
      "learning_rate": 3.8383198192800496e-07,
      "loss": 2.0324,
      "num_input_tokens_seen": 445766408,
      "step": 437800
    },
    {
      "epoch": 8.621606190073045,
      "grad_norm": 2.314169406890869,
      "learning_rate": 3.8363145333623924e-07,
      "loss": 1.9561,
      "num_input_tokens_seen": 445868808,
      "step": 437900
    },
    {
      "epoch": 8.623575042822548,
      "grad_norm": 2.126976490020752,
      "learning_rate": 3.834309445319633e-07,
      "loss": 1.9656,
      "num_input_tokens_seen": 445971208,
      "step": 438000
    },
    {
      "epoch": 8.62554389557205,
      "grad_norm": 4.256861209869385,
      "learning_rate": 3.8323045554927215e-07,
      "loss": 1.9902,
      "num_input_tokens_seen": 446072184,
      "step": 438100
    },
    {
      "epoch": 8.627512748321553,
      "grad_norm": 2.1243627071380615,
      "learning_rate": 3.8302998642225705e-07,
      "loss": 1.982,
      "num_input_tokens_seen": 446174584,
      "step": 438200
    },
    {
      "epoch": 8.629481601071056,
      "grad_norm": 4.225976467132568,
      "learning_rate": 3.828295371850062e-07,
      "loss": 1.9624,
      "num_input_tokens_seen": 446276984,
      "step": 438300
    },
    {
      "epoch": 8.631450453820559,
      "grad_norm": 2.3296940326690674,
      "learning_rate": 3.826291078716043e-07,
      "loss": 1.9954,
      "num_input_tokens_seen": 446378328,
      "step": 438400
    },
    {
      "epoch": 8.633419306570062,
      "grad_norm": 3.5202696323394775,
      "learning_rate": 3.8242869851613265e-07,
      "loss": 1.9833,
      "num_input_tokens_seen": 446479984,
      "step": 438500
    },
    {
      "epoch": 8.635388159319564,
      "grad_norm": 1.9449868202209473,
      "learning_rate": 3.8222830915266923e-07,
      "loss": 1.9779,
      "num_input_tokens_seen": 446582384,
      "step": 438600
    },
    {
      "epoch": 8.637357012069067,
      "grad_norm": 1.9325412511825562,
      "learning_rate": 3.820279398152884e-07,
      "loss": 2.0006,
      "num_input_tokens_seen": 446684784,
      "step": 438700
    },
    {
      "epoch": 8.63932586481857,
      "grad_norm": 2.032569169998169,
      "learning_rate": 3.818275905380615e-07,
      "loss": 2.0278,
      "num_input_tokens_seen": 446787184,
      "step": 438800
    },
    {
      "epoch": 8.641294717568073,
      "grad_norm": 2.101090908050537,
      "learning_rate": 3.8162726135505607e-07,
      "loss": 1.9513,
      "num_input_tokens_seen": 446889584,
      "step": 438900
    },
    {
      "epoch": 8.643263570317576,
      "grad_norm": 2.1767234802246094,
      "learning_rate": 3.814269523003365e-07,
      "loss": 1.938,
      "num_input_tokens_seen": 446991984,
      "step": 439000
    },
    {
      "epoch": 8.645232423067078,
      "grad_norm": 1.8542765378952026,
      "learning_rate": 3.812266634079635e-07,
      "loss": 1.9841,
      "num_input_tokens_seen": 447094384,
      "step": 439100
    },
    {
      "epoch": 8.647201275816581,
      "grad_norm": 1.9887007474899292,
      "learning_rate": 3.8102639471199473e-07,
      "loss": 1.9702,
      "num_input_tokens_seen": 447196160,
      "step": 439200
    },
    {
      "epoch": 8.649170128566084,
      "grad_norm": 1.7489919662475586,
      "learning_rate": 3.808261462464839e-07,
      "loss": 2.1175,
      "num_input_tokens_seen": 447297608,
      "step": 439300
    },
    {
      "epoch": 8.651138981315587,
      "grad_norm": 1.914811372756958,
      "learning_rate": 3.8062591804548165e-07,
      "loss": 1.9412,
      "num_input_tokens_seen": 447400008,
      "step": 439400
    },
    {
      "epoch": 8.653107834065091,
      "grad_norm": 1.98484206199646,
      "learning_rate": 3.804257101430355e-07,
      "loss": 2.0423,
      "num_input_tokens_seen": 447502024,
      "step": 439500
    },
    {
      "epoch": 8.655076686814594,
      "grad_norm": 6.037850379943848,
      "learning_rate": 3.802255225731886e-07,
      "loss": 2.0018,
      "num_input_tokens_seen": 447604424,
      "step": 439600
    },
    {
      "epoch": 8.657045539564097,
      "grad_norm": 2.195216178894043,
      "learning_rate": 3.800253553699816e-07,
      "loss": 1.998,
      "num_input_tokens_seen": 447706824,
      "step": 439700
    },
    {
      "epoch": 8.6590143923136,
      "grad_norm": 2.026824951171875,
      "learning_rate": 3.7982520856745063e-07,
      "loss": 2.011,
      "num_input_tokens_seen": 447808600,
      "step": 439800
    },
    {
      "epoch": 8.660983245063102,
      "grad_norm": 1.842523217201233,
      "learning_rate": 3.7962508219962975e-07,
      "loss": 1.9868,
      "num_input_tokens_seen": 447910184,
      "step": 439900
    },
    {
      "epoch": 8.662952097812605,
      "grad_norm": 2.0842556953430176,
      "learning_rate": 3.7942497630054803e-07,
      "loss": 2.0129,
      "num_input_tokens_seen": 448012584,
      "step": 440000
    },
    {
      "epoch": 8.664920950562108,
      "grad_norm": 1.891471028327942,
      "learning_rate": 3.792248909042324e-07,
      "loss": 2.0276,
      "num_input_tokens_seen": 448114984,
      "step": 440100
    },
    {
      "epoch": 8.66688980331161,
      "grad_norm": 1.9226154088974,
      "learning_rate": 3.790248260447051e-07,
      "loss": 2.033,
      "num_input_tokens_seen": 448216312,
      "step": 440200
    },
    {
      "epoch": 8.668858656061113,
      "grad_norm": 1.829014778137207,
      "learning_rate": 3.788247817559861e-07,
      "loss": 1.9778,
      "num_input_tokens_seen": 448317328,
      "step": 440300
    },
    {
      "epoch": 8.670827508810616,
      "grad_norm": 1.9798763990402222,
      "learning_rate": 3.786247580720906e-07,
      "loss": 1.9561,
      "num_input_tokens_seen": 448418960,
      "step": 440400
    },
    {
      "epoch": 8.672796361560119,
      "grad_norm": 2.055565118789673,
      "learning_rate": 3.784247550270313e-07,
      "loss": 1.9836,
      "num_input_tokens_seen": 448521360,
      "step": 440500
    },
    {
      "epoch": 8.674765214309621,
      "grad_norm": 1.8370145559310913,
      "learning_rate": 3.7822477265481725e-07,
      "loss": 2.0119,
      "num_input_tokens_seen": 448623760,
      "step": 440600
    },
    {
      "epoch": 8.676734067059124,
      "grad_norm": 2.077500343322754,
      "learning_rate": 3.7802481098945325e-07,
      "loss": 1.9884,
      "num_input_tokens_seen": 448724848,
      "step": 440700
    },
    {
      "epoch": 8.678702919808627,
      "grad_norm": 4.799295425415039,
      "learning_rate": 3.7782487006494155e-07,
      "loss": 1.943,
      "num_input_tokens_seen": 448827248,
      "step": 440800
    },
    {
      "epoch": 8.68067177255813,
      "grad_norm": 2.258174180984497,
      "learning_rate": 3.776249499152801e-07,
      "loss": 2.0372,
      "num_input_tokens_seen": 448928944,
      "step": 440900
    },
    {
      "epoch": 8.682640625307632,
      "grad_norm": 2.02398419380188,
      "learning_rate": 3.774250505744639e-07,
      "loss": 1.9563,
      "num_input_tokens_seen": 449030248,
      "step": 441000
    },
    {
      "epoch": 8.684609478057137,
      "grad_norm": 2.144974946975708,
      "learning_rate": 3.7722517207648385e-07,
      "loss": 2.0016,
      "num_input_tokens_seen": 449131240,
      "step": 441100
    },
    {
      "epoch": 8.68657833080664,
      "grad_norm": 1.9988431930541992,
      "learning_rate": 3.770253144553279e-07,
      "loss": 2.0404,
      "num_input_tokens_seen": 449233192,
      "step": 441200
    },
    {
      "epoch": 8.688547183556143,
      "grad_norm": 1.8716778755187988,
      "learning_rate": 3.7682547774497985e-07,
      "loss": 2.0045,
      "num_input_tokens_seen": 449335592,
      "step": 441300
    },
    {
      "epoch": 8.690516036305645,
      "grad_norm": 1.728929877281189,
      "learning_rate": 3.766256619794205e-07,
      "loss": 1.9987,
      "num_input_tokens_seen": 449437992,
      "step": 441400
    },
    {
      "epoch": 8.692484889055148,
      "grad_norm": 1.7776018381118774,
      "learning_rate": 3.764258671926267e-07,
      "loss": 2.012,
      "num_input_tokens_seen": 449538968,
      "step": 441500
    },
    {
      "epoch": 8.69445374180465,
      "grad_norm": 2.1455764770507812,
      "learning_rate": 3.762260934185719e-07,
      "loss": 1.9857,
      "num_input_tokens_seen": 449640816,
      "step": 441600
    },
    {
      "epoch": 8.696422594554154,
      "grad_norm": 1.8637111186981201,
      "learning_rate": 3.760263406912261e-07,
      "loss": 1.9827,
      "num_input_tokens_seen": 449742352,
      "step": 441700
    },
    {
      "epoch": 8.698391447303656,
      "grad_norm": 1.8431859016418457,
      "learning_rate": 3.758266090445553e-07,
      "loss": 1.9565,
      "num_input_tokens_seen": 449844104,
      "step": 441800
    },
    {
      "epoch": 8.700360300053159,
      "grad_norm": 2.1522130966186523,
      "learning_rate": 3.7562689851252233e-07,
      "loss": 1.9257,
      "num_input_tokens_seen": 449946504,
      "step": 441900
    },
    {
      "epoch": 8.702329152802662,
      "grad_norm": 1.9919891357421875,
      "learning_rate": 3.754272091290862e-07,
      "loss": 1.9943,
      "num_input_tokens_seen": 450047648,
      "step": 442000
    },
    {
      "epoch": 8.704298005552165,
      "grad_norm": 1.9764221906661987,
      "learning_rate": 3.7522754092820256e-07,
      "loss": 1.9611,
      "num_input_tokens_seen": 450149208,
      "step": 442100
    },
    {
      "epoch": 8.706266858301667,
      "grad_norm": 2.01125431060791,
      "learning_rate": 3.7502789394382316e-07,
      "loss": 1.9854,
      "num_input_tokens_seen": 450251608,
      "step": 442200
    },
    {
      "epoch": 8.70823571105117,
      "grad_norm": 1.8752394914627075,
      "learning_rate": 3.748282682098964e-07,
      "loss": 1.9714,
      "num_input_tokens_seen": 450354008,
      "step": 442300
    },
    {
      "epoch": 8.710204563800673,
      "grad_norm": 2.136979341506958,
      "learning_rate": 3.7462866376036674e-07,
      "loss": 1.9278,
      "num_input_tokens_seen": 450456408,
      "step": 442400
    },
    {
      "epoch": 8.712173416550176,
      "grad_norm": 1.881626009941101,
      "learning_rate": 3.7442908062917533e-07,
      "loss": 1.9564,
      "num_input_tokens_seen": 450558808,
      "step": 442500
    },
    {
      "epoch": 8.714142269299678,
      "grad_norm": 1.8991940021514893,
      "learning_rate": 3.7422951885025986e-07,
      "loss": 1.9451,
      "num_input_tokens_seen": 450661208,
      "step": 442600
    },
    {
      "epoch": 8.716111122049181,
      "grad_norm": 1.857869267463684,
      "learning_rate": 3.740299784575537e-07,
      "loss": 2.01,
      "num_input_tokens_seen": 450763064,
      "step": 442700
    },
    {
      "epoch": 8.718079974798684,
      "grad_norm": 1.9196866750717163,
      "learning_rate": 3.7383045948498754e-07,
      "loss": 1.9864,
      "num_input_tokens_seen": 450865464,
      "step": 442800
    },
    {
      "epoch": 8.720048827548188,
      "grad_norm": 1.865019679069519,
      "learning_rate": 3.7363096196648723e-07,
      "loss": 1.9767,
      "num_input_tokens_seen": 450967864,
      "step": 442900
    },
    {
      "epoch": 8.722017680297691,
      "grad_norm": 2.0263545513153076,
      "learning_rate": 3.734314859359763e-07,
      "loss": 1.9688,
      "num_input_tokens_seen": 451069624,
      "step": 443000
    },
    {
      "epoch": 8.723986533047194,
      "grad_norm": 1.958918571472168,
      "learning_rate": 3.7323203142737336e-07,
      "loss": 1.9957,
      "num_input_tokens_seen": 451170392,
      "step": 443100
    },
    {
      "epoch": 8.725955385796697,
      "grad_norm": 1.7758839130401611,
      "learning_rate": 3.7303259847459457e-07,
      "loss": 1.9405,
      "num_input_tokens_seen": 451272792,
      "step": 443200
    },
    {
      "epoch": 8.7279242385462,
      "grad_norm": 1.9893596172332764,
      "learning_rate": 3.728331871115512e-07,
      "loss": 2.0114,
      "num_input_tokens_seen": 451375192,
      "step": 443300
    },
    {
      "epoch": 8.729893091295702,
      "grad_norm": 1.9820750951766968,
      "learning_rate": 3.726337973721521e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 451477040,
      "step": 443400
    },
    {
      "epoch": 8.731861944045205,
      "grad_norm": 1.8078739643096924,
      "learning_rate": 3.724344292903012e-07,
      "loss": 1.957,
      "num_input_tokens_seen": 451579440,
      "step": 443500
    },
    {
      "epoch": 8.733830796794708,
      "grad_norm": 1.7033534049987793,
      "learning_rate": 3.722350828998998e-07,
      "loss": 1.9845,
      "num_input_tokens_seen": 451681840,
      "step": 443600
    },
    {
      "epoch": 8.73579964954421,
      "grad_norm": 2.0560247898101807,
      "learning_rate": 3.7203575823484497e-07,
      "loss": 2.0014,
      "num_input_tokens_seen": 451782536,
      "step": 443700
    },
    {
      "epoch": 8.737768502293713,
      "grad_norm": 3.4981229305267334,
      "learning_rate": 3.718364553290301e-07,
      "loss": 2.0083,
      "num_input_tokens_seen": 451884344,
      "step": 443800
    },
    {
      "epoch": 8.739737355043216,
      "grad_norm": 2.064596652984619,
      "learning_rate": 3.716371742163451e-07,
      "loss": 1.9539,
      "num_input_tokens_seen": 451986744,
      "step": 443900
    },
    {
      "epoch": 8.741706207792719,
      "grad_norm": 1.8347772359848022,
      "learning_rate": 3.714379149306758e-07,
      "loss": 2.0239,
      "num_input_tokens_seen": 452088856,
      "step": 444000
    },
    {
      "epoch": 8.743675060542222,
      "grad_norm": 2.1077797412872314,
      "learning_rate": 3.7123867750590493e-07,
      "loss": 1.9747,
      "num_input_tokens_seen": 452189624,
      "step": 444100
    },
    {
      "epoch": 8.745643913291724,
      "grad_norm": 1.8972969055175781,
      "learning_rate": 3.7103946197591073e-07,
      "loss": 1.9705,
      "num_input_tokens_seen": 452291248,
      "step": 444200
    },
    {
      "epoch": 8.747612766041227,
      "grad_norm": 1.5250269174575806,
      "learning_rate": 3.708402683745685e-07,
      "loss": 2.0073,
      "num_input_tokens_seen": 452393648,
      "step": 444300
    },
    {
      "epoch": 8.74958161879073,
      "grad_norm": 1.7753756046295166,
      "learning_rate": 3.7064109673574917e-07,
      "loss": 2.003,
      "num_input_tokens_seen": 452494688,
      "step": 444400
    },
    {
      "epoch": 8.751550471540234,
      "grad_norm": 1.8046605587005615,
      "learning_rate": 3.704419470933202e-07,
      "loss": 1.9798,
      "num_input_tokens_seen": 452596408,
      "step": 444500
    },
    {
      "epoch": 8.753519324289737,
      "grad_norm": 1.854323387145996,
      "learning_rate": 3.702428194811455e-07,
      "loss": 1.9931,
      "num_input_tokens_seen": 452698808,
      "step": 444600
    },
    {
      "epoch": 8.75548817703924,
      "grad_norm": 2.2141740322113037,
      "learning_rate": 3.7004371393308484e-07,
      "loss": 2.0221,
      "num_input_tokens_seen": 452799992,
      "step": 444700
    },
    {
      "epoch": 8.757457029788743,
      "grad_norm": 1.8548610210418701,
      "learning_rate": 3.6984463048299465e-07,
      "loss": 2.0111,
      "num_input_tokens_seen": 452900960,
      "step": 444800
    },
    {
      "epoch": 8.759425882538245,
      "grad_norm": 2.267636299133301,
      "learning_rate": 3.696455691647271e-07,
      "loss": 1.9805,
      "num_input_tokens_seen": 453002720,
      "step": 444900
    },
    {
      "epoch": 8.761394735287748,
      "grad_norm": 2.0605995655059814,
      "learning_rate": 3.694465300121312e-07,
      "loss": 2.0197,
      "num_input_tokens_seen": 453104312,
      "step": 445000
    },
    {
      "epoch": 8.763363588037251,
      "grad_norm": 2.4092280864715576,
      "learning_rate": 3.6924751305905157e-07,
      "loss": 1.9399,
      "num_input_tokens_seen": 453205288,
      "step": 445100
    },
    {
      "epoch": 8.765332440786754,
      "grad_norm": 1.8316638469696045,
      "learning_rate": 3.690485183393297e-07,
      "loss": 1.9581,
      "num_input_tokens_seen": 453307688,
      "step": 445200
    },
    {
      "epoch": 8.767301293536256,
      "grad_norm": 2.0612685680389404,
      "learning_rate": 3.6884954588680254e-07,
      "loss": 2.0297,
      "num_input_tokens_seen": 453409512,
      "step": 445300
    },
    {
      "epoch": 8.76927014628576,
      "grad_norm": 2.094058036804199,
      "learning_rate": 3.6865059573530427e-07,
      "loss": 2.013,
      "num_input_tokens_seen": 453511152,
      "step": 445400
    },
    {
      "epoch": 8.771238999035262,
      "grad_norm": 1.5972319841384888,
      "learning_rate": 3.68451667918664e-07,
      "loss": 1.993,
      "num_input_tokens_seen": 453613552,
      "step": 445500
    },
    {
      "epoch": 8.773207851784765,
      "grad_norm": 2.213054895401001,
      "learning_rate": 3.682527624707082e-07,
      "loss": 1.9934,
      "num_input_tokens_seen": 453715472,
      "step": 445600
    },
    {
      "epoch": 8.775176704534267,
      "grad_norm": 2.0300986766815186,
      "learning_rate": 3.6805387942525904e-07,
      "loss": 1.9601,
      "num_input_tokens_seen": 453817872,
      "step": 445700
    },
    {
      "epoch": 8.77714555728377,
      "grad_norm": 5.396900177001953,
      "learning_rate": 3.6785501881613466e-07,
      "loss": 1.971,
      "num_input_tokens_seen": 453919544,
      "step": 445800
    },
    {
      "epoch": 8.779114410033273,
      "grad_norm": 1.8516314029693604,
      "learning_rate": 3.6765618067714997e-07,
      "loss": 2.0277,
      "num_input_tokens_seen": 454021944,
      "step": 445900
    },
    {
      "epoch": 8.781083262782776,
      "grad_norm": 1.9930393695831299,
      "learning_rate": 3.6745736504211533e-07,
      "loss": 1.959,
      "num_input_tokens_seen": 454123072,
      "step": 446000
    },
    {
      "epoch": 8.783052115532278,
      "grad_norm": 1.9945868253707886,
      "learning_rate": 3.67258571944838e-07,
      "loss": 2.0495,
      "num_input_tokens_seen": 454225472,
      "step": 446100
    },
    {
      "epoch": 8.785020968281783,
      "grad_norm": 1.9024966955184937,
      "learning_rate": 3.670598014191207e-07,
      "loss": 2.0057,
      "num_input_tokens_seen": 454327024,
      "step": 446200
    },
    {
      "epoch": 8.786989821031286,
      "grad_norm": 1.992486596107483,
      "learning_rate": 3.668610534987631e-07,
      "loss": 2.0012,
      "num_input_tokens_seen": 454429424,
      "step": 446300
    },
    {
      "epoch": 8.788958673780789,
      "grad_norm": 2.2042038440704346,
      "learning_rate": 3.666623282175603e-07,
      "loss": 1.9778,
      "num_input_tokens_seen": 454531272,
      "step": 446400
    },
    {
      "epoch": 8.790927526530291,
      "grad_norm": 3.83732533454895,
      "learning_rate": 3.66463625609304e-07,
      "loss": 2.0007,
      "num_input_tokens_seen": 454632824,
      "step": 446500
    },
    {
      "epoch": 8.792896379279794,
      "grad_norm": 1.7048379182815552,
      "learning_rate": 3.662649457077819e-07,
      "loss": 1.9609,
      "num_input_tokens_seen": 454735224,
      "step": 446600
    },
    {
      "epoch": 8.794865232029297,
      "grad_norm": 1.6984621286392212,
      "learning_rate": 3.660662885467777e-07,
      "loss": 1.9816,
      "num_input_tokens_seen": 454836544,
      "step": 446700
    },
    {
      "epoch": 8.7968340847788,
      "grad_norm": 2.0316247940063477,
      "learning_rate": 3.658676541600716e-07,
      "loss": 1.9409,
      "num_input_tokens_seen": 454938408,
      "step": 446800
    },
    {
      "epoch": 8.798802937528302,
      "grad_norm": 2.1081910133361816,
      "learning_rate": 3.656690425814395e-07,
      "loss": 1.9658,
      "num_input_tokens_seen": 455038888,
      "step": 446900
    },
    {
      "epoch": 8.800771790277805,
      "grad_norm": 1.9483733177185059,
      "learning_rate": 3.654704538446538e-07,
      "loss": 2.0191,
      "num_input_tokens_seen": 455141288,
      "step": 447000
    },
    {
      "epoch": 8.802740643027308,
      "grad_norm": 3.234713315963745,
      "learning_rate": 3.652718879834827e-07,
      "loss": 1.9853,
      "num_input_tokens_seen": 455243688,
      "step": 447100
    },
    {
      "epoch": 8.80470949577681,
      "grad_norm": 1.7976369857788086,
      "learning_rate": 3.650733450316907e-07,
      "loss": 1.9376,
      "num_input_tokens_seen": 455345344,
      "step": 447200
    },
    {
      "epoch": 8.806678348526313,
      "grad_norm": 1.6544246673583984,
      "learning_rate": 3.6487482502303833e-07,
      "loss": 1.9435,
      "num_input_tokens_seen": 455447744,
      "step": 447300
    },
    {
      "epoch": 8.808647201275816,
      "grad_norm": 1.76790452003479,
      "learning_rate": 3.6467632799128236e-07,
      "loss": 1.981,
      "num_input_tokens_seen": 455550144,
      "step": 447400
    },
    {
      "epoch": 8.810616054025319,
      "grad_norm": 1.8400684595108032,
      "learning_rate": 3.644778539701753e-07,
      "loss": 1.9773,
      "num_input_tokens_seen": 455651888,
      "step": 447500
    },
    {
      "epoch": 8.812584906774822,
      "grad_norm": 1.904646873474121,
      "learning_rate": 3.64279402993466e-07,
      "loss": 1.9986,
      "num_input_tokens_seen": 455753608,
      "step": 447600
    },
    {
      "epoch": 8.814553759524324,
      "grad_norm": 2.317580461502075,
      "learning_rate": 3.640809750949e-07,
      "loss": 1.9769,
      "num_input_tokens_seen": 455856008,
      "step": 447700
    },
    {
      "epoch": 8.816522612273829,
      "grad_norm": 1.7815885543823242,
      "learning_rate": 3.638825703082173e-07,
      "loss": 2.0695,
      "num_input_tokens_seen": 455958408,
      "step": 447800
    },
    {
      "epoch": 8.818491465023332,
      "grad_norm": 1.9623051881790161,
      "learning_rate": 3.636841886671558e-07,
      "loss": 1.9935,
      "num_input_tokens_seen": 456060496,
      "step": 447900
    },
    {
      "epoch": 8.820460317772834,
      "grad_norm": 1.8556016683578491,
      "learning_rate": 3.63485830205448e-07,
      "loss": 1.9769,
      "num_input_tokens_seen": 456161528,
      "step": 448000
    },
    {
      "epoch": 8.822429170522337,
      "grad_norm": 1.9822044372558594,
      "learning_rate": 3.632874949568236e-07,
      "loss": 2.0191,
      "num_input_tokens_seen": 456263512,
      "step": 448100
    },
    {
      "epoch": 8.82439802327184,
      "grad_norm": 1.9502639770507812,
      "learning_rate": 3.6308918295500736e-07,
      "loss": 1.9856,
      "num_input_tokens_seen": 456365032,
      "step": 448200
    },
    {
      "epoch": 8.826366876021343,
      "grad_norm": 3.6109819412231445,
      "learning_rate": 3.6289089423372106e-07,
      "loss": 2.014,
      "num_input_tokens_seen": 456463664,
      "step": 448300
    },
    {
      "epoch": 8.828335728770845,
      "grad_norm": 2.757236957550049,
      "learning_rate": 3.626926288266814e-07,
      "loss": 2.0203,
      "num_input_tokens_seen": 456565664,
      "step": 448400
    },
    {
      "epoch": 8.830304581520348,
      "grad_norm": 1.8749210834503174,
      "learning_rate": 3.624943867676026e-07,
      "loss": 1.9957,
      "num_input_tokens_seen": 456667120,
      "step": 448500
    },
    {
      "epoch": 8.832273434269851,
      "grad_norm": 1.7945029735565186,
      "learning_rate": 3.6229616809019305e-07,
      "loss": 1.9833,
      "num_input_tokens_seen": 456769520,
      "step": 448600
    },
    {
      "epoch": 8.834242287019354,
      "grad_norm": 2.066924810409546,
      "learning_rate": 3.620979728281588e-07,
      "loss": 2.0258,
      "num_input_tokens_seen": 456869608,
      "step": 448700
    },
    {
      "epoch": 8.836211139768857,
      "grad_norm": 1.9594933986663818,
      "learning_rate": 3.6189980101520133e-07,
      "loss": 2.0104,
      "num_input_tokens_seen": 456970768,
      "step": 448800
    },
    {
      "epoch": 8.83817999251836,
      "grad_norm": 1.7667531967163086,
      "learning_rate": 3.6170165268501786e-07,
      "loss": 1.9563,
      "num_input_tokens_seen": 457073168,
      "step": 448900
    },
    {
      "epoch": 8.840148845267862,
      "grad_norm": 1.9170072078704834,
      "learning_rate": 3.615035278713019e-07,
      "loss": 2.0247,
      "num_input_tokens_seen": 457174920,
      "step": 449000
    },
    {
      "epoch": 8.842117698017365,
      "grad_norm": 2.301262617111206,
      "learning_rate": 3.613054266077428e-07,
      "loss": 1.9894,
      "num_input_tokens_seen": 457276104,
      "step": 449100
    },
    {
      "epoch": 8.844086550766868,
      "grad_norm": 2.0854878425598145,
      "learning_rate": 3.611073489280262e-07,
      "loss": 2.0195,
      "num_input_tokens_seen": 457378504,
      "step": 449200
    },
    {
      "epoch": 8.84605540351637,
      "grad_norm": 2.1191959381103516,
      "learning_rate": 3.609092948658333e-07,
      "loss": 2.0333,
      "num_input_tokens_seen": 457479344,
      "step": 449300
    },
    {
      "epoch": 8.848024256265873,
      "grad_norm": 2.0031979084014893,
      "learning_rate": 3.607112644548418e-07,
      "loss": 1.9964,
      "num_input_tokens_seen": 457580864,
      "step": 449400
    },
    {
      "epoch": 8.849993109015376,
      "grad_norm": 2.0754759311676025,
      "learning_rate": 3.605132577287249e-07,
      "loss": 2.0488,
      "num_input_tokens_seen": 457682344,
      "step": 449500
    },
    {
      "epoch": 8.85196196176488,
      "grad_norm": 1.924009919166565,
      "learning_rate": 3.60315274721152e-07,
      "loss": 1.9409,
      "num_input_tokens_seen": 457784744,
      "step": 449600
    },
    {
      "epoch": 8.853930814514383,
      "grad_norm": 2.0600244998931885,
      "learning_rate": 3.601173154657885e-07,
      "loss": 2.0058,
      "num_input_tokens_seen": 457887040,
      "step": 449700
    },
    {
      "epoch": 8.855899667263886,
      "grad_norm": 1.9509413242340088,
      "learning_rate": 3.5991937999629575e-07,
      "loss": 1.976,
      "num_input_tokens_seen": 457988544,
      "step": 449800
    },
    {
      "epoch": 8.857868520013389,
      "grad_norm": 2.0268354415893555,
      "learning_rate": 3.5972146834633094e-07,
      "loss": 1.9861,
      "num_input_tokens_seen": 458090944,
      "step": 449900
    },
    {
      "epoch": 8.859837372762891,
      "grad_norm": 1.9515221118927002,
      "learning_rate": 3.595235805495472e-07,
      "loss": 1.9914,
      "num_input_tokens_seen": 458193344,
      "step": 450000
    },
    {
      "epoch": 8.861806225512394,
      "grad_norm": 1.9911466836929321,
      "learning_rate": 3.593257166395938e-07,
      "loss": 2.0072,
      "num_input_tokens_seen": 458294048,
      "step": 450100
    },
    {
      "epoch": 8.863775078261897,
      "grad_norm": 1.843186616897583,
      "learning_rate": 3.591278766501157e-07,
      "loss": 1.9979,
      "num_input_tokens_seen": 458396272,
      "step": 450200
    },
    {
      "epoch": 8.8657439310114,
      "grad_norm": 1.930181622505188,
      "learning_rate": 3.589300606147542e-07,
      "loss": 2.0004,
      "num_input_tokens_seen": 458497200,
      "step": 450300
    },
    {
      "epoch": 8.867712783760902,
      "grad_norm": 1.896479606628418,
      "learning_rate": 3.5873226856714584e-07,
      "loss": 1.9616,
      "num_input_tokens_seen": 458598408,
      "step": 450400
    },
    {
      "epoch": 8.869681636510405,
      "grad_norm": 2.050766706466675,
      "learning_rate": 3.585345005409238e-07,
      "loss": 1.9723,
      "num_input_tokens_seen": 458700808,
      "step": 450500
    },
    {
      "epoch": 8.871650489259908,
      "grad_norm": 1.9739941358566284,
      "learning_rate": 3.5833675656971675e-07,
      "loss": 1.9593,
      "num_input_tokens_seen": 458803208,
      "step": 450600
    },
    {
      "epoch": 8.87361934200941,
      "grad_norm": 1.6464823484420776,
      "learning_rate": 3.581390366871493e-07,
      "loss": 2.0061,
      "num_input_tokens_seen": 458905608,
      "step": 450700
    },
    {
      "epoch": 8.875588194758913,
      "grad_norm": 1.916754126548767,
      "learning_rate": 3.579413409268424e-07,
      "loss": 1.945,
      "num_input_tokens_seen": 459007408,
      "step": 450800
    },
    {
      "epoch": 8.877557047508416,
      "grad_norm": 1.9122352600097656,
      "learning_rate": 3.5774366932241197e-07,
      "loss": 2.0026,
      "num_input_tokens_seen": 459109368,
      "step": 450900
    },
    {
      "epoch": 8.879525900257919,
      "grad_norm": 1.735640525817871,
      "learning_rate": 3.5754602190747107e-07,
      "loss": 1.956,
      "num_input_tokens_seen": 459211768,
      "step": 451000
    },
    {
      "epoch": 8.881494753007422,
      "grad_norm": 1.6875808238983154,
      "learning_rate": 3.573483987156274e-07,
      "loss": 1.9439,
      "num_input_tokens_seen": 459312560,
      "step": 451100
    },
    {
      "epoch": 8.883463605756926,
      "grad_norm": 1.9245717525482178,
      "learning_rate": 3.5715079978048557e-07,
      "loss": 1.9483,
      "num_input_tokens_seen": 459414960,
      "step": 451200
    },
    {
      "epoch": 8.885432458506429,
      "grad_norm": 1.9336355924606323,
      "learning_rate": 3.5695322513564505e-07,
      "loss": 1.9446,
      "num_input_tokens_seen": 459517360,
      "step": 451300
    },
    {
      "epoch": 8.887401311255932,
      "grad_norm": 2.0105578899383545,
      "learning_rate": 3.567556748147025e-07,
      "loss": 2.0121,
      "num_input_tokens_seen": 459618216,
      "step": 451400
    },
    {
      "epoch": 8.889370164005435,
      "grad_norm": 2.152290105819702,
      "learning_rate": 3.565581488512489e-07,
      "loss": 1.9641,
      "num_input_tokens_seen": 459719928,
      "step": 451500
    },
    {
      "epoch": 8.891339016754937,
      "grad_norm": 1.7344796657562256,
      "learning_rate": 3.5636064727887227e-07,
      "loss": 2.0073,
      "num_input_tokens_seen": 459821088,
      "step": 451600
    },
    {
      "epoch": 8.89330786950444,
      "grad_norm": 2.3190789222717285,
      "learning_rate": 3.5616317013115627e-07,
      "loss": 2.0011,
      "num_input_tokens_seen": 459923008,
      "step": 451700
    },
    {
      "epoch": 8.895276722253943,
      "grad_norm": 2.0082907676696777,
      "learning_rate": 3.5596571744167994e-07,
      "loss": 1.9735,
      "num_input_tokens_seen": 460024416,
      "step": 451800
    },
    {
      "epoch": 8.897245575003446,
      "grad_norm": 1.642967939376831,
      "learning_rate": 3.5576828924401855e-07,
      "loss": 1.9947,
      "num_input_tokens_seen": 460126816,
      "step": 451900
    },
    {
      "epoch": 8.899214427752948,
      "grad_norm": 1.9961860179901123,
      "learning_rate": 3.55570885571743e-07,
      "loss": 1.9902,
      "num_input_tokens_seen": 460228528,
      "step": 452000
    },
    {
      "epoch": 8.901183280502451,
      "grad_norm": 1.9979910850524902,
      "learning_rate": 3.553735064584202e-07,
      "loss": 1.978,
      "num_input_tokens_seen": 460330512,
      "step": 452100
    },
    {
      "epoch": 8.903152133251954,
      "grad_norm": 1.8616020679473877,
      "learning_rate": 3.5517615193761274e-07,
      "loss": 1.9982,
      "num_input_tokens_seen": 460432912,
      "step": 452200
    },
    {
      "epoch": 8.905120986001457,
      "grad_norm": 2.3574740886688232,
      "learning_rate": 3.5497882204287933e-07,
      "loss": 1.9914,
      "num_input_tokens_seen": 460534880,
      "step": 452300
    },
    {
      "epoch": 8.90708983875096,
      "grad_norm": 1.9747241735458374,
      "learning_rate": 3.5478151680777387e-07,
      "loss": 1.9967,
      "num_input_tokens_seen": 460637280,
      "step": 452400
    },
    {
      "epoch": 8.909058691500462,
      "grad_norm": 1.8633328676223755,
      "learning_rate": 3.5458423626584675e-07,
      "loss": 2.0346,
      "num_input_tokens_seen": 460739680,
      "step": 452500
    },
    {
      "epoch": 8.911027544249965,
      "grad_norm": 1.8827803134918213,
      "learning_rate": 3.543869804506437e-07,
      "loss": 2.0016,
      "num_input_tokens_seen": 460840632,
      "step": 452600
    },
    {
      "epoch": 8.912996396999468,
      "grad_norm": 1.8843450546264648,
      "learning_rate": 3.5418974939570644e-07,
      "loss": 1.9751,
      "num_input_tokens_seen": 460943032,
      "step": 452700
    },
    {
      "epoch": 8.91496524974897,
      "grad_norm": 2.1894655227661133,
      "learning_rate": 3.539925431345726e-07,
      "loss": 2.0151,
      "num_input_tokens_seen": 461044488,
      "step": 452800
    },
    {
      "epoch": 8.916934102498475,
      "grad_norm": 2.1176750659942627,
      "learning_rate": 3.537953617007752e-07,
      "loss": 1.9507,
      "num_input_tokens_seen": 461146888,
      "step": 452900
    },
    {
      "epoch": 8.918902955247978,
      "grad_norm": 1.8750076293945312,
      "learning_rate": 3.5359820512784347e-07,
      "loss": 2.0159,
      "num_input_tokens_seen": 461249288,
      "step": 453000
    },
    {
      "epoch": 8.92087180799748,
      "grad_norm": 1.8658981323242188,
      "learning_rate": 3.5340107344930206e-07,
      "loss": 1.9918,
      "num_input_tokens_seen": 461351688,
      "step": 453100
    },
    {
      "epoch": 8.922840660746983,
      "grad_norm": 1.9649168252944946,
      "learning_rate": 3.532039666986718e-07,
      "loss": 1.9929,
      "num_input_tokens_seen": 461454088,
      "step": 453200
    },
    {
      "epoch": 8.924809513496486,
      "grad_norm": 2.112973213195801,
      "learning_rate": 3.5300688490946873e-07,
      "loss": 1.9731,
      "num_input_tokens_seen": 461556488,
      "step": 453300
    },
    {
      "epoch": 8.926778366245989,
      "grad_norm": 1.8056161403656006,
      "learning_rate": 3.5280982811520517e-07,
      "loss": 1.9509,
      "num_input_tokens_seen": 461658888,
      "step": 453400
    },
    {
      "epoch": 8.928747218995492,
      "grad_norm": 2.0748400688171387,
      "learning_rate": 3.5261279634938877e-07,
      "loss": 1.9776,
      "num_input_tokens_seen": 461761288,
      "step": 453500
    },
    {
      "epoch": 8.930716071744994,
      "grad_norm": 1.9434674978256226,
      "learning_rate": 3.5241578964552333e-07,
      "loss": 1.9595,
      "num_input_tokens_seen": 461863096,
      "step": 453600
    },
    {
      "epoch": 8.932684924494497,
      "grad_norm": 1.6872519254684448,
      "learning_rate": 3.522188080371079e-07,
      "loss": 1.9883,
      "num_input_tokens_seen": 461965496,
      "step": 453700
    },
    {
      "epoch": 8.934653777244,
      "grad_norm": 1.85960853099823,
      "learning_rate": 3.5202185155763783e-07,
      "loss": 1.9884,
      "num_input_tokens_seen": 462067896,
      "step": 453800
    },
    {
      "epoch": 8.936622629993503,
      "grad_norm": 2.009847640991211,
      "learning_rate": 3.5182492024060383e-07,
      "loss": 2.0315,
      "num_input_tokens_seen": 462170296,
      "step": 453900
    },
    {
      "epoch": 8.938591482743005,
      "grad_norm": 2.876405715942383,
      "learning_rate": 3.5162801411949234e-07,
      "loss": 1.9948,
      "num_input_tokens_seen": 462272696,
      "step": 454000
    },
    {
      "epoch": 8.940560335492508,
      "grad_norm": 1.9534741640090942,
      "learning_rate": 3.5143113322778574e-07,
      "loss": 2.0359,
      "num_input_tokens_seen": 462374432,
      "step": 454100
    },
    {
      "epoch": 8.94252918824201,
      "grad_norm": 1.8362634181976318,
      "learning_rate": 3.5123427759896167e-07,
      "loss": 1.9737,
      "num_input_tokens_seen": 462476064,
      "step": 454200
    },
    {
      "epoch": 8.944498040991514,
      "grad_norm": 2.1819591522216797,
      "learning_rate": 3.510374472664941e-07,
      "loss": 1.9971,
      "num_input_tokens_seen": 462578464,
      "step": 454300
    },
    {
      "epoch": 8.946466893741016,
      "grad_norm": 1.9293622970581055,
      "learning_rate": 3.508406422638521e-07,
      "loss": 2.0277,
      "num_input_tokens_seen": 462679216,
      "step": 454400
    },
    {
      "epoch": 8.94843574649052,
      "grad_norm": 2.0786991119384766,
      "learning_rate": 3.50643862624501e-07,
      "loss": 2.0097,
      "num_input_tokens_seen": 462781616,
      "step": 454500
    },
    {
      "epoch": 8.950404599240024,
      "grad_norm": 1.8562805652618408,
      "learning_rate": 3.5044710838190107e-07,
      "loss": 1.9723,
      "num_input_tokens_seen": 462883488,
      "step": 454600
    },
    {
      "epoch": 8.952373451989526,
      "grad_norm": 2.069399356842041,
      "learning_rate": 3.502503795695091e-07,
      "loss": 1.9541,
      "num_input_tokens_seen": 462985888,
      "step": 454700
    },
    {
      "epoch": 8.95434230473903,
      "grad_norm": 1.9615408182144165,
      "learning_rate": 3.500536762207771e-07,
      "loss": 2.0205,
      "num_input_tokens_seen": 463088240,
      "step": 454800
    },
    {
      "epoch": 8.956311157488532,
      "grad_norm": 2.12089467048645,
      "learning_rate": 3.4985699836915263e-07,
      "loss": 2.0027,
      "num_input_tokens_seen": 463188960,
      "step": 454900
    },
    {
      "epoch": 8.958280010238035,
      "grad_norm": 1.8620356321334839,
      "learning_rate": 3.496603460480793e-07,
      "loss": 1.9786,
      "num_input_tokens_seen": 463289064,
      "step": 455000
    },
    {
      "epoch": 8.960248862987537,
      "grad_norm": 1.6644980907440186,
      "learning_rate": 3.4946371929099605e-07,
      "loss": 1.9584,
      "num_input_tokens_seen": 463391464,
      "step": 455100
    },
    {
      "epoch": 8.96221771573704,
      "grad_norm": 1.7266654968261719,
      "learning_rate": 3.4926711813133773e-07,
      "loss": 1.9637,
      "num_input_tokens_seen": 463491528,
      "step": 455200
    },
    {
      "epoch": 8.964186568486543,
      "grad_norm": 2.0514442920684814,
      "learning_rate": 3.4907054260253455e-07,
      "loss": 2.0351,
      "num_input_tokens_seen": 463593008,
      "step": 455300
    },
    {
      "epoch": 8.966155421236046,
      "grad_norm": 1.8872276544570923,
      "learning_rate": 3.488739927380127e-07,
      "loss": 1.9595,
      "num_input_tokens_seen": 463694592,
      "step": 455400
    },
    {
      "epoch": 8.968124273985548,
      "grad_norm": 2.1632421016693115,
      "learning_rate": 3.4867746857119365e-07,
      "loss": 1.9792,
      "num_input_tokens_seen": 463796992,
      "step": 455500
    },
    {
      "epoch": 8.970093126735051,
      "grad_norm": 1.9777168035507202,
      "learning_rate": 3.484809701354948e-07,
      "loss": 2.0071,
      "num_input_tokens_seen": 463898024,
      "step": 455600
    },
    {
      "epoch": 8.972061979484554,
      "grad_norm": 2.0591378211975098,
      "learning_rate": 3.4828449746432896e-07,
      "loss": 1.9898,
      "num_input_tokens_seen": 463999600,
      "step": 455700
    },
    {
      "epoch": 8.974030832234057,
      "grad_norm": 1.902843952178955,
      "learning_rate": 3.480880505911046e-07,
      "loss": 1.998,
      "num_input_tokens_seen": 464101824,
      "step": 455800
    },
    {
      "epoch": 8.97599968498356,
      "grad_norm": 1.7229907512664795,
      "learning_rate": 3.478916295492262e-07,
      "loss": 2.0109,
      "num_input_tokens_seen": 464203464,
      "step": 455900
    },
    {
      "epoch": 8.977968537733062,
      "grad_norm": 1.8738839626312256,
      "learning_rate": 3.47695234372093e-07,
      "loss": 2.0182,
      "num_input_tokens_seen": 464305472,
      "step": 456000
    },
    {
      "epoch": 8.979937390482565,
      "grad_norm": 2.030946969985962,
      "learning_rate": 3.474988650931009e-07,
      "loss": 1.9661,
      "num_input_tokens_seen": 464407872,
      "step": 456100
    },
    {
      "epoch": 8.981906243232068,
      "grad_norm": 1.7908202409744263,
      "learning_rate": 3.4730252174564017e-07,
      "loss": 1.9571,
      "num_input_tokens_seen": 464510272,
      "step": 456200
    },
    {
      "epoch": 8.983875095981572,
      "grad_norm": 2.0152580738067627,
      "learning_rate": 3.471062043630981e-07,
      "loss": 2.0023,
      "num_input_tokens_seen": 464612672,
      "step": 456300
    },
    {
      "epoch": 8.985843948731075,
      "grad_norm": 1.8081436157226562,
      "learning_rate": 3.46909912978856e-07,
      "loss": 1.9853,
      "num_input_tokens_seen": 464715064,
      "step": 456400
    },
    {
      "epoch": 8.987812801480578,
      "grad_norm": 1.9940552711486816,
      "learning_rate": 3.467136476262924e-07,
      "loss": 1.9396,
      "num_input_tokens_seen": 464816272,
      "step": 456500
    },
    {
      "epoch": 8.98978165423008,
      "grad_norm": 1.7960771322250366,
      "learning_rate": 3.4651740833877975e-07,
      "loss": 2.019,
      "num_input_tokens_seen": 464917816,
      "step": 456600
    },
    {
      "epoch": 8.991750506979583,
      "grad_norm": 1.9710174798965454,
      "learning_rate": 3.4632119514968744e-07,
      "loss": 2.0066,
      "num_input_tokens_seen": 465018928,
      "step": 456700
    },
    {
      "epoch": 8.993719359729086,
      "grad_norm": 1.9642566442489624,
      "learning_rate": 3.461250080923798e-07,
      "loss": 1.9996,
      "num_input_tokens_seen": 465120168,
      "step": 456800
    },
    {
      "epoch": 8.995688212478589,
      "grad_norm": 1.924302339553833,
      "learning_rate": 3.4592884720021654e-07,
      "loss": 1.9869,
      "num_input_tokens_seen": 465221984,
      "step": 456900
    },
    {
      "epoch": 8.997657065228092,
      "grad_norm": 1.79349946975708,
      "learning_rate": 3.4573271250655345e-07,
      "loss": 1.9616,
      "num_input_tokens_seen": 465323832,
      "step": 457000
    },
    {
      "epoch": 8.999625917977594,
      "grad_norm": 1.846758484840393,
      "learning_rate": 3.4553660404474136e-07,
      "loss": 1.9666,
      "num_input_tokens_seen": 465425696,
      "step": 457100
    },
    {
      "epoch": 9.001594770727097,
      "grad_norm": 3.0008890628814697,
      "learning_rate": 3.453405218481271e-07,
      "loss": 2.0245,
      "num_input_tokens_seen": 465527952,
      "step": 457200
    },
    {
      "epoch": 9.0035636234766,
      "grad_norm": 1.8662006855010986,
      "learning_rate": 3.451444659500524e-07,
      "loss": 1.9865,
      "num_input_tokens_seen": 465630352,
      "step": 457300
    },
    {
      "epoch": 9.005532476226103,
      "grad_norm": 1.9708071947097778,
      "learning_rate": 3.449484363838554e-07,
      "loss": 1.9751,
      "num_input_tokens_seen": 465732208,
      "step": 457400
    },
    {
      "epoch": 9.007501328975605,
      "grad_norm": 1.8690742254257202,
      "learning_rate": 3.4475243318286885e-07,
      "loss": 1.9877,
      "num_input_tokens_seen": 465834608,
      "step": 457500
    },
    {
      "epoch": 9.009470181725108,
      "grad_norm": 1.891236424446106,
      "learning_rate": 3.4455645638042173e-07,
      "loss": 1.9947,
      "num_input_tokens_seen": 465935952,
      "step": 457600
    },
    {
      "epoch": 9.011439034474611,
      "grad_norm": 1.934617519378662,
      "learning_rate": 3.44360506009838e-07,
      "loss": 1.967,
      "num_input_tokens_seen": 466037480,
      "step": 457700
    },
    {
      "epoch": 9.013407887224114,
      "grad_norm": 2.242408037185669,
      "learning_rate": 3.4416458210443755e-07,
      "loss": 2.0061,
      "num_input_tokens_seen": 466138688,
      "step": 457800
    },
    {
      "epoch": 9.015376739973618,
      "grad_norm": 2.3857059478759766,
      "learning_rate": 3.439686846975357e-07,
      "loss": 1.9726,
      "num_input_tokens_seen": 466239656,
      "step": 457900
    },
    {
      "epoch": 9.017345592723121,
      "grad_norm": 1.8627309799194336,
      "learning_rate": 3.437728138224428e-07,
      "loss": 1.9786,
      "num_input_tokens_seen": 466340224,
      "step": 458000
    },
    {
      "epoch": 9.019314445472624,
      "grad_norm": 1.5528455972671509,
      "learning_rate": 3.435769695124653e-07,
      "loss": 1.9627,
      "num_input_tokens_seen": 466442104,
      "step": 458100
    },
    {
      "epoch": 9.021283298222126,
      "grad_norm": 2.029254674911499,
      "learning_rate": 3.433811518009048e-07,
      "loss": 1.9436,
      "num_input_tokens_seen": 466544504,
      "step": 458200
    },
    {
      "epoch": 9.02325215097163,
      "grad_norm": 2.0019540786743164,
      "learning_rate": 3.431853607210584e-07,
      "loss": 1.9877,
      "num_input_tokens_seen": 466646264,
      "step": 458300
    },
    {
      "epoch": 9.025221003721132,
      "grad_norm": 2.2268126010894775,
      "learning_rate": 3.429895963062188e-07,
      "loss": 1.9771,
      "num_input_tokens_seen": 466748064,
      "step": 458400
    },
    {
      "epoch": 9.027189856470635,
      "grad_norm": 2.0140953063964844,
      "learning_rate": 3.4279385858967414e-07,
      "loss": 1.991,
      "num_input_tokens_seen": 466850464,
      "step": 458500
    },
    {
      "epoch": 9.029158709220138,
      "grad_norm": 1.559033989906311,
      "learning_rate": 3.4259814760470774e-07,
      "loss": 2.0166,
      "num_input_tokens_seen": 466952864,
      "step": 458600
    },
    {
      "epoch": 9.03112756196964,
      "grad_norm": 2.0017471313476562,
      "learning_rate": 3.424024633845989e-07,
      "loss": 1.9757,
      "num_input_tokens_seen": 467054560,
      "step": 458700
    },
    {
      "epoch": 9.033096414719143,
      "grad_norm": 2.088620185852051,
      "learning_rate": 3.4220680596262186e-07,
      "loss": 2.0129,
      "num_input_tokens_seen": 467156424,
      "step": 458800
    },
    {
      "epoch": 9.035065267468646,
      "grad_norm": 1.869077205657959,
      "learning_rate": 3.4201117537204636e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 467258824,
      "step": 458900
    },
    {
      "epoch": 9.037034120218149,
      "grad_norm": 1.6668530702590942,
      "learning_rate": 3.4181557164613833e-07,
      "loss": 1.9674,
      "num_input_tokens_seen": 467360768,
      "step": 459000
    },
    {
      "epoch": 9.039002972967651,
      "grad_norm": 2.469242572784424,
      "learning_rate": 3.4161999481815785e-07,
      "loss": 1.9679,
      "num_input_tokens_seen": 467462536,
      "step": 459100
    },
    {
      "epoch": 9.040971825717154,
      "grad_norm": 4.868458271026611,
      "learning_rate": 3.414244449213617e-07,
      "loss": 1.9877,
      "num_input_tokens_seen": 467563664,
      "step": 459200
    },
    {
      "epoch": 9.042940678466657,
      "grad_norm": 1.9527435302734375,
      "learning_rate": 3.412289219890009e-07,
      "loss": 2.0316,
      "num_input_tokens_seen": 467663904,
      "step": 459300
    },
    {
      "epoch": 9.04490953121616,
      "grad_norm": 1.9218140840530396,
      "learning_rate": 3.4103342605432305e-07,
      "loss": 1.9859,
      "num_input_tokens_seen": 467765456,
      "step": 459400
    },
    {
      "epoch": 9.046878383965662,
      "grad_norm": 1.7027376890182495,
      "learning_rate": 3.408379571505701e-07,
      "loss": 1.9911,
      "num_input_tokens_seen": 467867856,
      "step": 459500
    },
    {
      "epoch": 9.048847236715167,
      "grad_norm": 2.346632480621338,
      "learning_rate": 3.4064251531098034e-07,
      "loss": 1.9657,
      "num_input_tokens_seen": 467968848,
      "step": 459600
    },
    {
      "epoch": 9.05081608946467,
      "grad_norm": 1.9321973323822021,
      "learning_rate": 3.4044710056878646e-07,
      "loss": 1.9945,
      "num_input_tokens_seen": 468071248,
      "step": 459700
    },
    {
      "epoch": 9.052784942214172,
      "grad_norm": 1.9816904067993164,
      "learning_rate": 3.402517129572175e-07,
      "loss": 2.0039,
      "num_input_tokens_seen": 468173648,
      "step": 459800
    },
    {
      "epoch": 9.054753794963675,
      "grad_norm": 2.163686513900757,
      "learning_rate": 3.400563525094975e-07,
      "loss": 2.0248,
      "num_input_tokens_seen": 468275312,
      "step": 459900
    },
    {
      "epoch": 9.056722647713178,
      "grad_norm": 1.7017338275909424,
      "learning_rate": 3.3986101925884556e-07,
      "loss": 1.9548,
      "num_input_tokens_seen": 468377216,
      "step": 460000
    },
    {
      "epoch": 9.05869150046268,
      "grad_norm": 1.8634601831436157,
      "learning_rate": 3.396657132384768e-07,
      "loss": 2.0347,
      "num_input_tokens_seen": 468479064,
      "step": 460100
    },
    {
      "epoch": 9.060660353212183,
      "grad_norm": 1.8698630332946777,
      "learning_rate": 3.3947043448160106e-07,
      "loss": 1.9563,
      "num_input_tokens_seen": 468580176,
      "step": 460200
    },
    {
      "epoch": 9.062629205961686,
      "grad_norm": 1.9434471130371094,
      "learning_rate": 3.392751830214241e-07,
      "loss": 1.9627,
      "num_input_tokens_seen": 468681344,
      "step": 460300
    },
    {
      "epoch": 9.064598058711189,
      "grad_norm": 3.5381059646606445,
      "learning_rate": 3.3907995889114647e-07,
      "loss": 1.9417,
      "num_input_tokens_seen": 468783744,
      "step": 460400
    },
    {
      "epoch": 9.066566911460692,
      "grad_norm": 1.76524817943573,
      "learning_rate": 3.388847621239648e-07,
      "loss": 1.9711,
      "num_input_tokens_seen": 468886144,
      "step": 460500
    },
    {
      "epoch": 9.068535764210194,
      "grad_norm": 2.0872108936309814,
      "learning_rate": 3.386895927530702e-07,
      "loss": 1.9846,
      "num_input_tokens_seen": 468988544,
      "step": 460600
    },
    {
      "epoch": 9.070504616959697,
      "grad_norm": 1.856087327003479,
      "learning_rate": 3.384944508116499e-07,
      "loss": 1.9648,
      "num_input_tokens_seen": 469090944,
      "step": 460700
    },
    {
      "epoch": 9.0724734697092,
      "grad_norm": 2.1265947818756104,
      "learning_rate": 3.38299336332886e-07,
      "loss": 1.9657,
      "num_input_tokens_seen": 469192808,
      "step": 460800
    },
    {
      "epoch": 9.074442322458703,
      "grad_norm": 1.7705391645431519,
      "learning_rate": 3.3810424934995605e-07,
      "loss": 1.9983,
      "num_input_tokens_seen": 469293984,
      "step": 460900
    },
    {
      "epoch": 9.076411175208206,
      "grad_norm": 2.031543493270874,
      "learning_rate": 3.379091898960331e-07,
      "loss": 1.9514,
      "num_input_tokens_seen": 469396384,
      "step": 461000
    },
    {
      "epoch": 9.078380027957708,
      "grad_norm": 1.767552375793457,
      "learning_rate": 3.3771415800428527e-07,
      "loss": 1.9298,
      "num_input_tokens_seen": 469498784,
      "step": 461100
    },
    {
      "epoch": 9.080348880707213,
      "grad_norm": 2.2106571197509766,
      "learning_rate": 3.375191537078761e-07,
      "loss": 1.9784,
      "num_input_tokens_seen": 469600832,
      "step": 461200
    },
    {
      "epoch": 9.082317733456716,
      "grad_norm": 1.9160747528076172,
      "learning_rate": 3.3732417703996435e-07,
      "loss": 2.0515,
      "num_input_tokens_seen": 469703232,
      "step": 461300
    },
    {
      "epoch": 9.084286586206218,
      "grad_norm": 2.072680950164795,
      "learning_rate": 3.3712922803370435e-07,
      "loss": 2.0081,
      "num_input_tokens_seen": 469804048,
      "step": 461400
    },
    {
      "epoch": 9.086255438955721,
      "grad_norm": 1.9819695949554443,
      "learning_rate": 3.369343067222452e-07,
      "loss": 2.025,
      "num_input_tokens_seen": 469906448,
      "step": 461500
    },
    {
      "epoch": 9.088224291705224,
      "grad_norm": 1.9884424209594727,
      "learning_rate": 3.3673941313873206e-07,
      "loss": 1.9767,
      "num_input_tokens_seen": 470008256,
      "step": 461600
    },
    {
      "epoch": 9.090193144454727,
      "grad_norm": 1.9748448133468628,
      "learning_rate": 3.3654454731630455e-07,
      "loss": 1.9599,
      "num_input_tokens_seen": 470109848,
      "step": 461700
    },
    {
      "epoch": 9.09216199720423,
      "grad_norm": 2.1398630142211914,
      "learning_rate": 3.36349709288098e-07,
      "loss": 1.9951,
      "num_input_tokens_seen": 470212248,
      "step": 461800
    },
    {
      "epoch": 9.094130849953732,
      "grad_norm": 2.4477145671844482,
      "learning_rate": 3.361548990872435e-07,
      "loss": 1.937,
      "num_input_tokens_seen": 470313776,
      "step": 461900
    },
    {
      "epoch": 9.096099702703235,
      "grad_norm": 2.2335894107818604,
      "learning_rate": 3.359601167468661e-07,
      "loss": 2.0437,
      "num_input_tokens_seen": 470415696,
      "step": 462000
    },
    {
      "epoch": 9.098068555452738,
      "grad_norm": 2.278855085372925,
      "learning_rate": 3.3576536230008747e-07,
      "loss": 2.0122,
      "num_input_tokens_seen": 470517224,
      "step": 462100
    },
    {
      "epoch": 9.10003740820224,
      "grad_norm": 1.8019264936447144,
      "learning_rate": 3.3557063578002365e-07,
      "loss": 1.9621,
      "num_input_tokens_seen": 470619624,
      "step": 462200
    },
    {
      "epoch": 9.102006260951743,
      "grad_norm": 1.8344491720199585,
      "learning_rate": 3.353759372197865e-07,
      "loss": 1.9552,
      "num_input_tokens_seen": 470722024,
      "step": 462300
    },
    {
      "epoch": 9.103975113701246,
      "grad_norm": 1.8435543775558472,
      "learning_rate": 3.351812666524826e-07,
      "loss": 1.9675,
      "num_input_tokens_seen": 470823176,
      "step": 462400
    },
    {
      "epoch": 9.105943966450749,
      "grad_norm": 1.8911259174346924,
      "learning_rate": 3.349866241112142e-07,
      "loss": 2.0105,
      "num_input_tokens_seen": 470924648,
      "step": 462500
    },
    {
      "epoch": 9.107912819200251,
      "grad_norm": 2.0813772678375244,
      "learning_rate": 3.347920096290785e-07,
      "loss": 2.0098,
      "num_input_tokens_seen": 471027048,
      "step": 462600
    },
    {
      "epoch": 9.109881671949754,
      "grad_norm": 2.155921459197998,
      "learning_rate": 3.345974232391683e-07,
      "loss": 1.9851,
      "num_input_tokens_seen": 471128008,
      "step": 462700
    },
    {
      "epoch": 9.111850524699257,
      "grad_norm": 2.0911948680877686,
      "learning_rate": 3.344028649745711e-07,
      "loss": 2.0088,
      "num_input_tokens_seen": 471229656,
      "step": 462800
    },
    {
      "epoch": 9.113819377448761,
      "grad_norm": 1.9464905261993408,
      "learning_rate": 3.3420833486837e-07,
      "loss": 2.0201,
      "num_input_tokens_seen": 471332056,
      "step": 462900
    },
    {
      "epoch": 9.115788230198264,
      "grad_norm": 1.860300064086914,
      "learning_rate": 3.3401383295364326e-07,
      "loss": 2.0,
      "num_input_tokens_seen": 471434456,
      "step": 463000
    },
    {
      "epoch": 9.117757082947767,
      "grad_norm": 1.978712558746338,
      "learning_rate": 3.3381935926346416e-07,
      "loss": 1.9544,
      "num_input_tokens_seen": 471536216,
      "step": 463100
    },
    {
      "epoch": 9.11972593569727,
      "grad_norm": 1.672761082649231,
      "learning_rate": 3.3362491383090144e-07,
      "loss": 1.9781,
      "num_input_tokens_seen": 471638144,
      "step": 463200
    },
    {
      "epoch": 9.121694788446773,
      "grad_norm": 2.275247097015381,
      "learning_rate": 3.334304966890188e-07,
      "loss": 1.9692,
      "num_input_tokens_seen": 471740544,
      "step": 463300
    },
    {
      "epoch": 9.123663641196275,
      "grad_norm": 1.9788198471069336,
      "learning_rate": 3.3323610787087544e-07,
      "loss": 2.0102,
      "num_input_tokens_seen": 471842408,
      "step": 463400
    },
    {
      "epoch": 9.125632493945778,
      "grad_norm": 2.262812376022339,
      "learning_rate": 3.330417474095252e-07,
      "loss": 2.0388,
      "num_input_tokens_seen": 471943608,
      "step": 463500
    },
    {
      "epoch": 9.12760134669528,
      "grad_norm": 2.114628791809082,
      "learning_rate": 3.328474153380177e-07,
      "loss": 2.0586,
      "num_input_tokens_seen": 472044560,
      "step": 463600
    },
    {
      "epoch": 9.129570199444784,
      "grad_norm": 1.8643527030944824,
      "learning_rate": 3.326531116893974e-07,
      "loss": 2.0018,
      "num_input_tokens_seen": 472145456,
      "step": 463700
    },
    {
      "epoch": 9.131539052194286,
      "grad_norm": 2.0743651390075684,
      "learning_rate": 3.324588364967038e-07,
      "loss": 2.0114,
      "num_input_tokens_seen": 472246504,
      "step": 463800
    },
    {
      "epoch": 9.133507904943789,
      "grad_norm": 1.91729736328125,
      "learning_rate": 3.3226458979297224e-07,
      "loss": 2.0308,
      "num_input_tokens_seen": 472348344,
      "step": 463900
    },
    {
      "epoch": 9.135476757693292,
      "grad_norm": 2.067162275314331,
      "learning_rate": 3.320703716112322e-07,
      "loss": 1.9563,
      "num_input_tokens_seen": 472450744,
      "step": 464000
    },
    {
      "epoch": 9.137445610442795,
      "grad_norm": 1.8478366136550903,
      "learning_rate": 3.3187618198450935e-07,
      "loss": 1.9685,
      "num_input_tokens_seen": 472552648,
      "step": 464100
    },
    {
      "epoch": 9.139414463192297,
      "grad_norm": 1.784961462020874,
      "learning_rate": 3.3168202094582344e-07,
      "loss": 2.0024,
      "num_input_tokens_seen": 472654536,
      "step": 464200
    },
    {
      "epoch": 9.1413833159418,
      "grad_norm": 1.885314702987671,
      "learning_rate": 3.3148788852819056e-07,
      "loss": 1.9917,
      "num_input_tokens_seen": 472755744,
      "step": 464300
    },
    {
      "epoch": 9.143352168691303,
      "grad_norm": 1.6667174100875854,
      "learning_rate": 3.3129378476462057e-07,
      "loss": 1.9891,
      "num_input_tokens_seen": 472858144,
      "step": 464400
    },
    {
      "epoch": 9.145321021440806,
      "grad_norm": 1.8386411666870117,
      "learning_rate": 3.310997096881198e-07,
      "loss": 1.9967,
      "num_input_tokens_seen": 472959768,
      "step": 464500
    },
    {
      "epoch": 9.14728987419031,
      "grad_norm": 1.8697896003723145,
      "learning_rate": 3.309056633316886e-07,
      "loss": 1.9958,
      "num_input_tokens_seen": 473062168,
      "step": 464600
    },
    {
      "epoch": 9.149258726939813,
      "grad_norm": 2.16806697845459,
      "learning_rate": 3.3071164572832346e-07,
      "loss": 2.0571,
      "num_input_tokens_seen": 473163344,
      "step": 464700
    },
    {
      "epoch": 9.151227579689316,
      "grad_norm": 1.8884488344192505,
      "learning_rate": 3.3051765691101474e-07,
      "loss": 1.9225,
      "num_input_tokens_seen": 473265112,
      "step": 464800
    },
    {
      "epoch": 9.153196432438818,
      "grad_norm": 1.9401971101760864,
      "learning_rate": 3.303236969127492e-07,
      "loss": 1.9586,
      "num_input_tokens_seen": 473367512,
      "step": 464900
    },
    {
      "epoch": 9.155165285188321,
      "grad_norm": 1.9680757522583008,
      "learning_rate": 3.3012976576650785e-07,
      "loss": 1.9973,
      "num_input_tokens_seen": 473469912,
      "step": 465000
    },
    {
      "epoch": 9.157134137937824,
      "grad_norm": 2.009850025177002,
      "learning_rate": 3.299358635052671e-07,
      "loss": 1.9918,
      "num_input_tokens_seen": 473571536,
      "step": 465100
    },
    {
      "epoch": 9.159102990687327,
      "grad_norm": 1.7671650648117065,
      "learning_rate": 3.2974199016199844e-07,
      "loss": 1.9498,
      "num_input_tokens_seen": 473673936,
      "step": 465200
    },
    {
      "epoch": 9.16107184343683,
      "grad_norm": 2.0762922763824463,
      "learning_rate": 3.2954814576966816e-07,
      "loss": 1.9651,
      "num_input_tokens_seen": 473775760,
      "step": 465300
    },
    {
      "epoch": 9.163040696186332,
      "grad_norm": 1.9856797456741333,
      "learning_rate": 3.293543303612382e-07,
      "loss": 1.9336,
      "num_input_tokens_seen": 473878160,
      "step": 465400
    },
    {
      "epoch": 9.165009548935835,
      "grad_norm": 1.8416602611541748,
      "learning_rate": 3.29160543969665e-07,
      "loss": 1.9652,
      "num_input_tokens_seen": 473979912,
      "step": 465500
    },
    {
      "epoch": 9.166978401685338,
      "grad_norm": 1.8863821029663086,
      "learning_rate": 3.289667866279004e-07,
      "loss": 1.9712,
      "num_input_tokens_seen": 474082312,
      "step": 465600
    },
    {
      "epoch": 9.16894725443484,
      "grad_norm": 1.877996802330017,
      "learning_rate": 3.2877305836889113e-07,
      "loss": 2.044,
      "num_input_tokens_seen": 474183328,
      "step": 465700
    },
    {
      "epoch": 9.170916107184343,
      "grad_norm": 1.847738265991211,
      "learning_rate": 3.285793592255791e-07,
      "loss": 1.9763,
      "num_input_tokens_seen": 474285552,
      "step": 465800
    },
    {
      "epoch": 9.172884959933846,
      "grad_norm": 2.134746551513672,
      "learning_rate": 3.2838568923090115e-07,
      "loss": 2.0322,
      "num_input_tokens_seen": 474387680,
      "step": 465900
    },
    {
      "epoch": 9.174853812683349,
      "grad_norm": 2.0179078578948975,
      "learning_rate": 3.2819204841778936e-07,
      "loss": 2.047,
      "num_input_tokens_seen": 474489416,
      "step": 466000
    },
    {
      "epoch": 9.176822665432852,
      "grad_norm": 1.8857274055480957,
      "learning_rate": 3.2799843681917066e-07,
      "loss": 1.9924,
      "num_input_tokens_seen": 474591816,
      "step": 466100
    },
    {
      "epoch": 9.178791518182354,
      "grad_norm": 1.9603630304336548,
      "learning_rate": 3.27804854467967e-07,
      "loss": 1.9742,
      "num_input_tokens_seen": 474692704,
      "step": 466200
    },
    {
      "epoch": 9.180760370931859,
      "grad_norm": 1.8099164962768555,
      "learning_rate": 3.276113013970955e-07,
      "loss": 1.9653,
      "num_input_tokens_seen": 474793976,
      "step": 466300
    },
    {
      "epoch": 9.182729223681362,
      "grad_norm": 2.3027429580688477,
      "learning_rate": 3.274177776394682e-07,
      "loss": 1.9498,
      "num_input_tokens_seen": 474895736,
      "step": 466400
    },
    {
      "epoch": 9.184698076430864,
      "grad_norm": 1.7471723556518555,
      "learning_rate": 3.272242832279922e-07,
      "loss": 1.9733,
      "num_input_tokens_seen": 474997872,
      "step": 466500
    },
    {
      "epoch": 9.186666929180367,
      "grad_norm": 1.8851635456085205,
      "learning_rate": 3.2703081819556957e-07,
      "loss": 1.988,
      "num_input_tokens_seen": 475099320,
      "step": 466600
    },
    {
      "epoch": 9.18863578192987,
      "grad_norm": 1.6516727209091187,
      "learning_rate": 3.268373825750974e-07,
      "loss": 1.9975,
      "num_input_tokens_seen": 475201616,
      "step": 466700
    },
    {
      "epoch": 9.190604634679373,
      "grad_norm": 1.8464986085891724,
      "learning_rate": 3.266439763994678e-07,
      "loss": 1.984,
      "num_input_tokens_seen": 475303720,
      "step": 466800
    },
    {
      "epoch": 9.192573487428875,
      "grad_norm": 1.9440335035324097,
      "learning_rate": 3.2645059970156776e-07,
      "loss": 2.0096,
      "num_input_tokens_seen": 475406120,
      "step": 466900
    },
    {
      "epoch": 9.194542340178378,
      "grad_norm": 1.971192479133606,
      "learning_rate": 3.262572525142797e-07,
      "loss": 1.9945,
      "num_input_tokens_seen": 475507832,
      "step": 467000
    },
    {
      "epoch": 9.19651119292788,
      "grad_norm": 2.179316759109497,
      "learning_rate": 3.2606393487048017e-07,
      "loss": 1.9935,
      "num_input_tokens_seen": 475608792,
      "step": 467100
    },
    {
      "epoch": 9.198480045677384,
      "grad_norm": 3.94113826751709,
      "learning_rate": 3.2587064680304177e-07,
      "loss": 1.9534,
      "num_input_tokens_seen": 475710400,
      "step": 467200
    },
    {
      "epoch": 9.200448898426886,
      "grad_norm": 1.9328970909118652,
      "learning_rate": 3.2567738834483086e-07,
      "loss": 1.9876,
      "num_input_tokens_seen": 475812800,
      "step": 467300
    },
    {
      "epoch": 9.20241775117639,
      "grad_norm": 1.877854347229004,
      "learning_rate": 3.254841595287101e-07,
      "loss": 1.9952,
      "num_input_tokens_seen": 475914248,
      "step": 467400
    },
    {
      "epoch": 9.204386603925892,
      "grad_norm": 1.911928415298462,
      "learning_rate": 3.252909603875358e-07,
      "loss": 2.0325,
      "num_input_tokens_seen": 476015624,
      "step": 467500
    },
    {
      "epoch": 9.206355456675395,
      "grad_norm": 1.9240180253982544,
      "learning_rate": 3.250977909541604e-07,
      "loss": 2.0211,
      "num_input_tokens_seen": 476117536,
      "step": 467600
    },
    {
      "epoch": 9.208324309424897,
      "grad_norm": 2.091864585876465,
      "learning_rate": 3.249046512614302e-07,
      "loss": 1.9722,
      "num_input_tokens_seen": 476219936,
      "step": 467700
    },
    {
      "epoch": 9.2102931621744,
      "grad_norm": 1.8455979824066162,
      "learning_rate": 3.2471154134218747e-07,
      "loss": 1.9958,
      "num_input_tokens_seen": 476321784,
      "step": 467800
    },
    {
      "epoch": 9.212262014923905,
      "grad_norm": 1.8556621074676514,
      "learning_rate": 3.245184612292684e-07,
      "loss": 1.9936,
      "num_input_tokens_seen": 476423248,
      "step": 467900
    },
    {
      "epoch": 9.214230867673407,
      "grad_norm": 1.7449880838394165,
      "learning_rate": 3.2432541095550496e-07,
      "loss": 1.9763,
      "num_input_tokens_seen": 476524256,
      "step": 468000
    },
    {
      "epoch": 9.21619972042291,
      "grad_norm": 2.0576798915863037,
      "learning_rate": 3.241323905537239e-07,
      "loss": 1.9921,
      "num_input_tokens_seen": 476625112,
      "step": 468100
    },
    {
      "epoch": 9.218168573172413,
      "grad_norm": 2.1071770191192627,
      "learning_rate": 3.2393940005674625e-07,
      "loss": 1.9985,
      "num_input_tokens_seen": 476725296,
      "step": 468200
    },
    {
      "epoch": 9.220137425921916,
      "grad_norm": 2.01064395904541,
      "learning_rate": 3.237464394973889e-07,
      "loss": 1.9309,
      "num_input_tokens_seen": 476827104,
      "step": 468300
    },
    {
      "epoch": 9.222106278671419,
      "grad_norm": 2.1058905124664307,
      "learning_rate": 3.2355350890846266e-07,
      "loss": 1.9663,
      "num_input_tokens_seen": 476929072,
      "step": 468400
    },
    {
      "epoch": 9.224075131420921,
      "grad_norm": 2.111330032348633,
      "learning_rate": 3.2336060832277423e-07,
      "loss": 1.9992,
      "num_input_tokens_seen": 477031472,
      "step": 468500
    },
    {
      "epoch": 9.226043984170424,
      "grad_norm": 1.7522377967834473,
      "learning_rate": 3.2316773777312445e-07,
      "loss": 1.9978,
      "num_input_tokens_seen": 477132880,
      "step": 468600
    },
    {
      "epoch": 9.228012836919927,
      "grad_norm": 1.8095093965530396,
      "learning_rate": 3.2297489729230944e-07,
      "loss": 1.9728,
      "num_input_tokens_seen": 477235280,
      "step": 468700
    },
    {
      "epoch": 9.22998168966943,
      "grad_norm": 1.850618600845337,
      "learning_rate": 3.2278208691311987e-07,
      "loss": 2.003,
      "num_input_tokens_seen": 477337232,
      "step": 468800
    },
    {
      "epoch": 9.231950542418932,
      "grad_norm": 2.0030128955841064,
      "learning_rate": 3.225893066683417e-07,
      "loss": 2.0079,
      "num_input_tokens_seen": 477438184,
      "step": 468900
    },
    {
      "epoch": 9.233919395168435,
      "grad_norm": 2.2313177585601807,
      "learning_rate": 3.223965565907558e-07,
      "loss": 1.9683,
      "num_input_tokens_seen": 477540120,
      "step": 469000
    },
    {
      "epoch": 9.235888247917938,
      "grad_norm": 2.0350351333618164,
      "learning_rate": 3.222038367131373e-07,
      "loss": 1.9713,
      "num_input_tokens_seen": 477641864,
      "step": 469100
    },
    {
      "epoch": 9.23785710066744,
      "grad_norm": 2.18227219581604,
      "learning_rate": 3.2201114706825687e-07,
      "loss": 2.0003,
      "num_input_tokens_seen": 477743704,
      "step": 469200
    },
    {
      "epoch": 9.239825953416943,
      "grad_norm": 1.923969030380249,
      "learning_rate": 3.2181848768887953e-07,
      "loss": 1.9419,
      "num_input_tokens_seen": 477846104,
      "step": 469300
    },
    {
      "epoch": 9.241794806166446,
      "grad_norm": 2.0169677734375,
      "learning_rate": 3.216258586077657e-07,
      "loss": 2.0125,
      "num_input_tokens_seen": 477947016,
      "step": 469400
    },
    {
      "epoch": 9.243763658915949,
      "grad_norm": 1.9119294881820679,
      "learning_rate": 3.2143325985766994e-07,
      "loss": 2.0128,
      "num_input_tokens_seen": 478048688,
      "step": 469500
    },
    {
      "epoch": 9.245732511665453,
      "grad_norm": 1.8002443313598633,
      "learning_rate": 3.2124069147134234e-07,
      "loss": 1.9549,
      "num_input_tokens_seen": 478150456,
      "step": 469600
    },
    {
      "epoch": 9.247701364414956,
      "grad_norm": 2.031088352203369,
      "learning_rate": 3.2104815348152743e-07,
      "loss": 2.0572,
      "num_input_tokens_seen": 478251496,
      "step": 469700
    },
    {
      "epoch": 9.249670217164459,
      "grad_norm": 1.9940271377563477,
      "learning_rate": 3.208556459209647e-07,
      "loss": 1.9898,
      "num_input_tokens_seen": 478353896,
      "step": 469800
    },
    {
      "epoch": 9.251639069913962,
      "grad_norm": 1.775673508644104,
      "learning_rate": 3.2066316882238833e-07,
      "loss": 1.9776,
      "num_input_tokens_seen": 478455536,
      "step": 469900
    },
    {
      "epoch": 9.253607922663464,
      "grad_norm": 1.9610161781311035,
      "learning_rate": 3.2047072221852734e-07,
      "loss": 1.9509,
      "num_input_tokens_seen": 478557312,
      "step": 470000
    },
    {
      "epoch": 9.255576775412967,
      "grad_norm": 2.22320818901062,
      "learning_rate": 3.2027830614210614e-07,
      "loss": 1.9946,
      "num_input_tokens_seen": 478659128,
      "step": 470100
    },
    {
      "epoch": 9.25754562816247,
      "grad_norm": 1.7737853527069092,
      "learning_rate": 3.2008592062584287e-07,
      "loss": 2.0302,
      "num_input_tokens_seen": 478761528,
      "step": 470200
    },
    {
      "epoch": 9.259514480911973,
      "grad_norm": 1.810951590538025,
      "learning_rate": 3.198935657024515e-07,
      "loss": 2.0035,
      "num_input_tokens_seen": 478863928,
      "step": 470300
    },
    {
      "epoch": 9.261483333661475,
      "grad_norm": 1.684280514717102,
      "learning_rate": 3.1970124140464017e-07,
      "loss": 1.9601,
      "num_input_tokens_seen": 478966328,
      "step": 470400
    },
    {
      "epoch": 9.263452186410978,
      "grad_norm": 1.8167426586151123,
      "learning_rate": 3.195089477651121e-07,
      "loss": 1.9842,
      "num_input_tokens_seen": 479068728,
      "step": 470500
    },
    {
      "epoch": 9.265421039160481,
      "grad_norm": 1.8957006931304932,
      "learning_rate": 3.19316684816565e-07,
      "loss": 1.9632,
      "num_input_tokens_seen": 479171032,
      "step": 470600
    },
    {
      "epoch": 9.267389891909984,
      "grad_norm": 1.9806638956069946,
      "learning_rate": 3.191244525916918e-07,
      "loss": 1.9479,
      "num_input_tokens_seen": 479273432,
      "step": 470700
    },
    {
      "epoch": 9.269358744659487,
      "grad_norm": 1.9928994178771973,
      "learning_rate": 3.1893225112317974e-07,
      "loss": 1.9992,
      "num_input_tokens_seen": 479374640,
      "step": 470800
    },
    {
      "epoch": 9.27132759740899,
      "grad_norm": 1.9113658666610718,
      "learning_rate": 3.187400804437113e-07,
      "loss": 1.9912,
      "num_input_tokens_seen": 479475344,
      "step": 470900
    },
    {
      "epoch": 9.273296450158492,
      "grad_norm": 1.908582091331482,
      "learning_rate": 3.1854794058596325e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 479577744,
      "step": 471000
    },
    {
      "epoch": 9.275265302907995,
      "grad_norm": 2.080770969390869,
      "learning_rate": 3.183558315826075e-07,
      "loss": 2.017,
      "num_input_tokens_seen": 479678816,
      "step": 471100
    },
    {
      "epoch": 9.277234155657498,
      "grad_norm": 1.9410804510116577,
      "learning_rate": 3.1816375346631064e-07,
      "loss": 1.954,
      "num_input_tokens_seen": 479781216,
      "step": 471200
    },
    {
      "epoch": 9.279203008407002,
      "grad_norm": 2.1430890560150146,
      "learning_rate": 3.179717062697337e-07,
      "loss": 1.9689,
      "num_input_tokens_seen": 479881016,
      "step": 471300
    },
    {
      "epoch": 9.281171861156505,
      "grad_norm": 1.9866336584091187,
      "learning_rate": 3.1777969002553287e-07,
      "loss": 1.9697,
      "num_input_tokens_seen": 479983416,
      "step": 471400
    },
    {
      "epoch": 9.283140713906008,
      "grad_norm": 1.7233686447143555,
      "learning_rate": 3.175877047663588e-07,
      "loss": 1.9627,
      "num_input_tokens_seen": 480085136,
      "step": 471500
    },
    {
      "epoch": 9.28510956665551,
      "grad_norm": 1.7687311172485352,
      "learning_rate": 3.1739575052485704e-07,
      "loss": 1.9896,
      "num_input_tokens_seen": 480187536,
      "step": 471600
    },
    {
      "epoch": 9.287078419405013,
      "grad_norm": 1.7554092407226562,
      "learning_rate": 3.172038273336677e-07,
      "loss": 1.999,
      "num_input_tokens_seen": 480289248,
      "step": 471700
    },
    {
      "epoch": 9.289047272154516,
      "grad_norm": 2.0111732482910156,
      "learning_rate": 3.1701193522542584e-07,
      "loss": 2.0432,
      "num_input_tokens_seen": 480390720,
      "step": 471800
    },
    {
      "epoch": 9.291016124904019,
      "grad_norm": 2.094632625579834,
      "learning_rate": 3.168200742327609e-07,
      "loss": 2.0247,
      "num_input_tokens_seen": 480489912,
      "step": 471900
    },
    {
      "epoch": 9.292984977653521,
      "grad_norm": 1.6701654195785522,
      "learning_rate": 3.166282443882972e-07,
      "loss": 1.9814,
      "num_input_tokens_seen": 480592312,
      "step": 472000
    },
    {
      "epoch": 9.294953830403024,
      "grad_norm": 2.229541778564453,
      "learning_rate": 3.164364457246542e-07,
      "loss": 1.9446,
      "num_input_tokens_seen": 480694712,
      "step": 472100
    },
    {
      "epoch": 9.296922683152527,
      "grad_norm": 1.8854907751083374,
      "learning_rate": 3.1624467827444514e-07,
      "loss": 1.9776,
      "num_input_tokens_seen": 480795456,
      "step": 472200
    },
    {
      "epoch": 9.29889153590203,
      "grad_norm": 1.9025156497955322,
      "learning_rate": 3.1605294207027896e-07,
      "loss": 1.9725,
      "num_input_tokens_seen": 480897112,
      "step": 472300
    },
    {
      "epoch": 9.300860388651532,
      "grad_norm": 1.6696339845657349,
      "learning_rate": 3.158612371447581e-07,
      "loss": 1.9986,
      "num_input_tokens_seen": 480999512,
      "step": 472400
    },
    {
      "epoch": 9.302829241401035,
      "grad_norm": 1.939001202583313,
      "learning_rate": 3.1566956353048114e-07,
      "loss": 1.9936,
      "num_input_tokens_seen": 481100288,
      "step": 472500
    },
    {
      "epoch": 9.304798094150538,
      "grad_norm": 5.727812767028809,
      "learning_rate": 3.154779212600399e-07,
      "loss": 2.0239,
      "num_input_tokens_seen": 481202688,
      "step": 472600
    },
    {
      "epoch": 9.30676694690004,
      "grad_norm": 1.8076066970825195,
      "learning_rate": 3.152863103660221e-07,
      "loss": 1.908,
      "num_input_tokens_seen": 481305088,
      "step": 472700
    },
    {
      "epoch": 9.308735799649543,
      "grad_norm": 2.0467591285705566,
      "learning_rate": 3.15094730881009e-07,
      "loss": 1.9849,
      "num_input_tokens_seen": 481405152,
      "step": 472800
    },
    {
      "epoch": 9.310704652399046,
      "grad_norm": 2.243654489517212,
      "learning_rate": 3.1490318283757757e-07,
      "loss": 1.9627,
      "num_input_tokens_seen": 481507552,
      "step": 472900
    },
    {
      "epoch": 9.31267350514855,
      "grad_norm": 1.7883583307266235,
      "learning_rate": 3.147116662682985e-07,
      "loss": 2.0307,
      "num_input_tokens_seen": 481609952,
      "step": 473000
    },
    {
      "epoch": 9.314642357898054,
      "grad_norm": 1.8222451210021973,
      "learning_rate": 3.1452018120573785e-07,
      "loss": 1.9673,
      "num_input_tokens_seen": 481711112,
      "step": 473100
    },
    {
      "epoch": 9.316611210647556,
      "grad_norm": 2.2083327770233154,
      "learning_rate": 3.143287276824561e-07,
      "loss": 1.9647,
      "num_input_tokens_seen": 481813512,
      "step": 473200
    },
    {
      "epoch": 9.318580063397059,
      "grad_norm": 1.7571516036987305,
      "learning_rate": 3.1413730573100807e-07,
      "loss": 2.039,
      "num_input_tokens_seen": 481914184,
      "step": 473300
    },
    {
      "epoch": 9.320548916146562,
      "grad_norm": 1.7453532218933105,
      "learning_rate": 3.1394591538394366e-07,
      "loss": 1.9752,
      "num_input_tokens_seen": 482016584,
      "step": 473400
    },
    {
      "epoch": 9.322517768896065,
      "grad_norm": 2.0322978496551514,
      "learning_rate": 3.1375455667380695e-07,
      "loss": 1.9844,
      "num_input_tokens_seen": 482118984,
      "step": 473500
    },
    {
      "epoch": 9.324486621645567,
      "grad_norm": 2.997708797454834,
      "learning_rate": 3.135632296331372e-07,
      "loss": 1.9718,
      "num_input_tokens_seen": 482221384,
      "step": 473600
    },
    {
      "epoch": 9.32645547439507,
      "grad_norm": 2.3402109146118164,
      "learning_rate": 3.133719342944676e-07,
      "loss": 2.0285,
      "num_input_tokens_seen": 482323344,
      "step": 473700
    },
    {
      "epoch": 9.328424327144573,
      "grad_norm": 2.0022799968719482,
      "learning_rate": 3.131806706903266e-07,
      "loss": 1.9986,
      "num_input_tokens_seen": 482424064,
      "step": 473800
    },
    {
      "epoch": 9.330393179894076,
      "grad_norm": 1.9793131351470947,
      "learning_rate": 3.129894388532367e-07,
      "loss": 1.9741,
      "num_input_tokens_seen": 482525984,
      "step": 473900
    },
    {
      "epoch": 9.332362032643578,
      "grad_norm": 1.8297178745269775,
      "learning_rate": 3.1279823881571545e-07,
      "loss": 1.989,
      "num_input_tokens_seen": 482627552,
      "step": 474000
    },
    {
      "epoch": 9.334330885393081,
      "grad_norm": 1.928903579711914,
      "learning_rate": 3.1260707061027485e-07,
      "loss": 1.9663,
      "num_input_tokens_seen": 482729952,
      "step": 474100
    },
    {
      "epoch": 9.336299738142584,
      "grad_norm": 1.9665343761444092,
      "learning_rate": 3.1241593426942123e-07,
      "loss": 1.9881,
      "num_input_tokens_seen": 482831120,
      "step": 474200
    },
    {
      "epoch": 9.338268590892087,
      "grad_norm": 1.8264588117599487,
      "learning_rate": 3.12224829825656e-07,
      "loss": 2.0132,
      "num_input_tokens_seen": 482933520,
      "step": 474300
    },
    {
      "epoch": 9.34023744364159,
      "grad_norm": 1.992934226989746,
      "learning_rate": 3.1203375731147455e-07,
      "loss": 1.9447,
      "num_input_tokens_seen": 483035288,
      "step": 474400
    },
    {
      "epoch": 9.342206296391092,
      "grad_norm": 1.948421597480774,
      "learning_rate": 3.1184271675936734e-07,
      "loss": 1.9601,
      "num_input_tokens_seen": 483136136,
      "step": 474500
    },
    {
      "epoch": 9.344175149140597,
      "grad_norm": 2.632755756378174,
      "learning_rate": 3.1165170820181906e-07,
      "loss": 1.9267,
      "num_input_tokens_seen": 483238536,
      "step": 474600
    },
    {
      "epoch": 9.3461440018901,
      "grad_norm": 1.4781384468078613,
      "learning_rate": 3.1146073167130936e-07,
      "loss": 2.0041,
      "num_input_tokens_seen": 483340120,
      "step": 474700
    },
    {
      "epoch": 9.348112854639602,
      "grad_norm": 1.8639442920684814,
      "learning_rate": 3.1126978720031194e-07,
      "loss": 2.0101,
      "num_input_tokens_seen": 483442240,
      "step": 474800
    },
    {
      "epoch": 9.350081707389105,
      "grad_norm": 1.8763703107833862,
      "learning_rate": 3.1107887482129537e-07,
      "loss": 2.0322,
      "num_input_tokens_seen": 483544640,
      "step": 474900
    },
    {
      "epoch": 9.352050560138608,
      "grad_norm": 1.7841733694076538,
      "learning_rate": 3.1088799456672267e-07,
      "loss": 1.9665,
      "num_input_tokens_seen": 483645528,
      "step": 475000
    },
    {
      "epoch": 9.35401941288811,
      "grad_norm": 1.8835313320159912,
      "learning_rate": 3.106971464690514e-07,
      "loss": 1.9719,
      "num_input_tokens_seen": 483747928,
      "step": 475100
    },
    {
      "epoch": 9.355988265637613,
      "grad_norm": 1.7878150939941406,
      "learning_rate": 3.105063305607339e-07,
      "loss": 1.9826,
      "num_input_tokens_seen": 483850328,
      "step": 475200
    },
    {
      "epoch": 9.357957118387116,
      "grad_norm": 1.845474362373352,
      "learning_rate": 3.1031554687421644e-07,
      "loss": 2.0131,
      "num_input_tokens_seen": 483951776,
      "step": 475300
    },
    {
      "epoch": 9.359925971136619,
      "grad_norm": 1.8708844184875488,
      "learning_rate": 3.101247954419406e-07,
      "loss": 2.0197,
      "num_input_tokens_seen": 484053648,
      "step": 475400
    },
    {
      "epoch": 9.361894823886121,
      "grad_norm": 1.9564247131347656,
      "learning_rate": 3.099340762963415e-07,
      "loss": 1.9762,
      "num_input_tokens_seen": 484154704,
      "step": 475500
    },
    {
      "epoch": 9.363863676635624,
      "grad_norm": 1.9573136568069458,
      "learning_rate": 3.0974338946985e-07,
      "loss": 1.9717,
      "num_input_tokens_seen": 484256280,
      "step": 475600
    },
    {
      "epoch": 9.365832529385127,
      "grad_norm": 2.103881597518921,
      "learning_rate": 3.095527349948901e-07,
      "loss": 1.9731,
      "num_input_tokens_seen": 484358008,
      "step": 475700
    },
    {
      "epoch": 9.36780138213463,
      "grad_norm": 1.5944429636001587,
      "learning_rate": 3.093621129038816e-07,
      "loss": 1.9819,
      "num_input_tokens_seen": 484458936,
      "step": 475800
    },
    {
      "epoch": 9.369770234884133,
      "grad_norm": 2.0732192993164062,
      "learning_rate": 3.091715232292376e-07,
      "loss": 1.9712,
      "num_input_tokens_seen": 484557976,
      "step": 475900
    },
    {
      "epoch": 9.371739087633635,
      "grad_norm": 1.8265455961227417,
      "learning_rate": 3.0898096600336676e-07,
      "loss": 2.0142,
      "num_input_tokens_seen": 484660376,
      "step": 476000
    },
    {
      "epoch": 9.373707940383138,
      "grad_norm": 1.9798468351364136,
      "learning_rate": 3.0879044125867133e-07,
      "loss": 1.949,
      "num_input_tokens_seen": 484762776,
      "step": 476100
    },
    {
      "epoch": 9.37567679313264,
      "grad_norm": 1.9368246793746948,
      "learning_rate": 3.085999490275487e-07,
      "loss": 1.9707,
      "num_input_tokens_seen": 484865176,
      "step": 476200
    },
    {
      "epoch": 9.377645645882145,
      "grad_norm": 2.264223098754883,
      "learning_rate": 3.0840948934239046e-07,
      "loss": 1.9919,
      "num_input_tokens_seen": 484967576,
      "step": 476300
    },
    {
      "epoch": 9.379614498631648,
      "grad_norm": 2.0374276638031006,
      "learning_rate": 3.0821906223558255e-07,
      "loss": 1.9463,
      "num_input_tokens_seen": 485069312,
      "step": 476400
    },
    {
      "epoch": 9.38158335138115,
      "grad_norm": 6.350859642028809,
      "learning_rate": 3.0802866773950563e-07,
      "loss": 1.9481,
      "num_input_tokens_seen": 485171712,
      "step": 476500
    },
    {
      "epoch": 9.383552204130654,
      "grad_norm": 2.699000835418701,
      "learning_rate": 3.0783830588653446e-07,
      "loss": 2.0063,
      "num_input_tokens_seen": 485274112,
      "step": 476600
    },
    {
      "epoch": 9.385521056880156,
      "grad_norm": 1.6473956108093262,
      "learning_rate": 3.076479767090388e-07,
      "loss": 1.9552,
      "num_input_tokens_seen": 485376512,
      "step": 476700
    },
    {
      "epoch": 9.38748990962966,
      "grad_norm": 1.934122085571289,
      "learning_rate": 3.074576802393822e-07,
      "loss": 1.964,
      "num_input_tokens_seen": 485478288,
      "step": 476800
    },
    {
      "epoch": 9.389458762379162,
      "grad_norm": 1.996249794960022,
      "learning_rate": 3.0726741650992317e-07,
      "loss": 1.9986,
      "num_input_tokens_seen": 485579840,
      "step": 476900
    },
    {
      "epoch": 9.391427615128665,
      "grad_norm": 1.9151747226715088,
      "learning_rate": 3.0707718555301434e-07,
      "loss": 1.9481,
      "num_input_tokens_seen": 485682240,
      "step": 477000
    },
    {
      "epoch": 9.393396467878167,
      "grad_norm": 1.7812050580978394,
      "learning_rate": 3.068869874010028e-07,
      "loss": 1.967,
      "num_input_tokens_seen": 485784424,
      "step": 477100
    },
    {
      "epoch": 9.39536532062767,
      "grad_norm": 1.9762742519378662,
      "learning_rate": 3.0669682208623047e-07,
      "loss": 1.9881,
      "num_input_tokens_seen": 485885896,
      "step": 477200
    },
    {
      "epoch": 9.397334173377173,
      "grad_norm": 1.9740040302276611,
      "learning_rate": 3.06506689641033e-07,
      "loss": 1.9643,
      "num_input_tokens_seen": 485987680,
      "step": 477300
    },
    {
      "epoch": 9.399303026126676,
      "grad_norm": 5.866965293884277,
      "learning_rate": 3.0631659009774106e-07,
      "loss": 2.0009,
      "num_input_tokens_seen": 486089264,
      "step": 477400
    },
    {
      "epoch": 9.401271878876178,
      "grad_norm": 1.9798721075057983,
      "learning_rate": 3.061265234886793e-07,
      "loss": 1.9808,
      "num_input_tokens_seen": 486191664,
      "step": 477500
    },
    {
      "epoch": 9.403240731625681,
      "grad_norm": 1.6924023628234863,
      "learning_rate": 3.059364898461671e-07,
      "loss": 1.9908,
      "num_input_tokens_seen": 486293832,
      "step": 477600
    },
    {
      "epoch": 9.405209584375184,
      "grad_norm": 1.7568439245224,
      "learning_rate": 3.0574648920251787e-07,
      "loss": 1.9616,
      "num_input_tokens_seen": 486395472,
      "step": 477700
    },
    {
      "epoch": 9.407178437124687,
      "grad_norm": 1.8347095251083374,
      "learning_rate": 3.055565215900399e-07,
      "loss": 1.9431,
      "num_input_tokens_seen": 486497872,
      "step": 477800
    },
    {
      "epoch": 9.40914728987419,
      "grad_norm": 1.718002438545227,
      "learning_rate": 3.053665870410353e-07,
      "loss": 2.0107,
      "num_input_tokens_seen": 486600272,
      "step": 477900
    },
    {
      "epoch": 9.411116142623694,
      "grad_norm": 2.002347230911255,
      "learning_rate": 3.051766855878011e-07,
      "loss": 1.9571,
      "num_input_tokens_seen": 486702672,
      "step": 478000
    },
    {
      "epoch": 9.413084995373197,
      "grad_norm": 1.8163652420043945,
      "learning_rate": 3.049868172626281e-07,
      "loss": 1.9536,
      "num_input_tokens_seen": 486804256,
      "step": 478100
    },
    {
      "epoch": 9.4150538481227,
      "grad_norm": 2.0375304222106934,
      "learning_rate": 3.047969820978019e-07,
      "loss": 1.9842,
      "num_input_tokens_seen": 486906656,
      "step": 478200
    },
    {
      "epoch": 9.417022700872202,
      "grad_norm": 2.047271966934204,
      "learning_rate": 3.046071801256029e-07,
      "loss": 2.025,
      "num_input_tokens_seen": 487008176,
      "step": 478300
    },
    {
      "epoch": 9.418991553621705,
      "grad_norm": 2.1189892292022705,
      "learning_rate": 3.044174113783045e-07,
      "loss": 1.9808,
      "num_input_tokens_seen": 487109512,
      "step": 478400
    },
    {
      "epoch": 9.420960406371208,
      "grad_norm": 2.0342628955841064,
      "learning_rate": 3.04227675888176e-07,
      "loss": 1.9879,
      "num_input_tokens_seen": 487211912,
      "step": 478500
    },
    {
      "epoch": 9.42292925912071,
      "grad_norm": 1.9001010656356812,
      "learning_rate": 3.0403797368747975e-07,
      "loss": 1.9978,
      "num_input_tokens_seen": 487314312,
      "step": 478600
    },
    {
      "epoch": 9.424898111870213,
      "grad_norm": 2.133366107940674,
      "learning_rate": 3.038483048084734e-07,
      "loss": 1.9807,
      "num_input_tokens_seen": 487416280,
      "step": 478700
    },
    {
      "epoch": 9.426866964619716,
      "grad_norm": 1.8415337800979614,
      "learning_rate": 3.0365866928340833e-07,
      "loss": 1.9728,
      "num_input_tokens_seen": 487517608,
      "step": 478800
    },
    {
      "epoch": 9.428835817369219,
      "grad_norm": 1.873547911643982,
      "learning_rate": 3.0346906714453056e-07,
      "loss": 2.0177,
      "num_input_tokens_seen": 487619256,
      "step": 478900
    },
    {
      "epoch": 9.430804670118722,
      "grad_norm": 1.9932811260223389,
      "learning_rate": 3.032794984240803e-07,
      "loss": 1.9777,
      "num_input_tokens_seen": 487721656,
      "step": 479000
    },
    {
      "epoch": 9.432773522868224,
      "grad_norm": 2.0202314853668213,
      "learning_rate": 3.0308996315429205e-07,
      "loss": 1.9758,
      "num_input_tokens_seen": 487824056,
      "step": 479100
    },
    {
      "epoch": 9.434742375617727,
      "grad_norm": 1.7601984739303589,
      "learning_rate": 3.029004613673949e-07,
      "loss": 1.9564,
      "num_input_tokens_seen": 487925704,
      "step": 479200
    },
    {
      "epoch": 9.43671122836723,
      "grad_norm": 1.6400821208953857,
      "learning_rate": 3.0271099309561164e-07,
      "loss": 1.9816,
      "num_input_tokens_seen": 488028104,
      "step": 479300
    },
    {
      "epoch": 9.438680081116733,
      "grad_norm": 2.037163257598877,
      "learning_rate": 3.0252155837116023e-07,
      "loss": 1.9441,
      "num_input_tokens_seen": 488129640,
      "step": 479400
    },
    {
      "epoch": 9.440648933866235,
      "grad_norm": 1.7901649475097656,
      "learning_rate": 3.023321572262521e-07,
      "loss": 1.9221,
      "num_input_tokens_seen": 488232040,
      "step": 479500
    },
    {
      "epoch": 9.442617786615738,
      "grad_norm": 1.7820320129394531,
      "learning_rate": 3.021427896930935e-07,
      "loss": 1.9781,
      "num_input_tokens_seen": 488334256,
      "step": 479600
    },
    {
      "epoch": 9.444586639365243,
      "grad_norm": 1.6406290531158447,
      "learning_rate": 3.019534558038846e-07,
      "loss": 2.0058,
      "num_input_tokens_seen": 488436656,
      "step": 479700
    },
    {
      "epoch": 9.446555492114745,
      "grad_norm": 1.8044099807739258,
      "learning_rate": 3.0176415559082026e-07,
      "loss": 1.9889,
      "num_input_tokens_seen": 488539056,
      "step": 479800
    },
    {
      "epoch": 9.448524344864248,
      "grad_norm": 1.8847109079360962,
      "learning_rate": 3.0157488908608917e-07,
      "loss": 2.0041,
      "num_input_tokens_seen": 488640640,
      "step": 479900
    },
    {
      "epoch": 9.450493197613751,
      "grad_norm": 3.0067076683044434,
      "learning_rate": 3.0138565632187463e-07,
      "loss": 2.0161,
      "num_input_tokens_seen": 488741480,
      "step": 480000
    },
    {
      "epoch": 9.452462050363254,
      "grad_norm": 2.0110690593719482,
      "learning_rate": 3.0119645733035406e-07,
      "loss": 1.959,
      "num_input_tokens_seen": 488843120,
      "step": 480100
    },
    {
      "epoch": 9.454430903112756,
      "grad_norm": 1.813224196434021,
      "learning_rate": 3.01007292143699e-07,
      "loss": 1.9811,
      "num_input_tokens_seen": 488944896,
      "step": 480200
    },
    {
      "epoch": 9.45639975586226,
      "grad_norm": 1.855810523033142,
      "learning_rate": 3.008181607940757e-07,
      "loss": 1.9601,
      "num_input_tokens_seen": 489047064,
      "step": 480300
    },
    {
      "epoch": 9.458368608611762,
      "grad_norm": 2.0644779205322266,
      "learning_rate": 3.00629063313644e-07,
      "loss": 2.0049,
      "num_input_tokens_seen": 489148944,
      "step": 480400
    },
    {
      "epoch": 9.460337461361265,
      "grad_norm": 2.047935962677002,
      "learning_rate": 3.004399997345588e-07,
      "loss": 1.975,
      "num_input_tokens_seen": 489251344,
      "step": 480500
    },
    {
      "epoch": 9.462306314110768,
      "grad_norm": 1.9466698169708252,
      "learning_rate": 3.0025097008896814e-07,
      "loss": 1.9407,
      "num_input_tokens_seen": 489353744,
      "step": 480600
    },
    {
      "epoch": 9.46427516686027,
      "grad_norm": 1.7519962787628174,
      "learning_rate": 3.000619744090156e-07,
      "loss": 2.0201,
      "num_input_tokens_seen": 489453912,
      "step": 480700
    },
    {
      "epoch": 9.466244019609773,
      "grad_norm": 2.454188108444214,
      "learning_rate": 2.998730127268376e-07,
      "loss": 1.9823,
      "num_input_tokens_seen": 489554936,
      "step": 480800
    },
    {
      "epoch": 9.468212872359276,
      "grad_norm": 1.8989869356155396,
      "learning_rate": 2.996840850745662e-07,
      "loss": 1.9776,
      "num_input_tokens_seen": 489656792,
      "step": 480900
    },
    {
      "epoch": 9.470181725108779,
      "grad_norm": 1.9428150653839111,
      "learning_rate": 2.994951914843262e-07,
      "loss": 1.9719,
      "num_input_tokens_seen": 489759192,
      "step": 481000
    },
    {
      "epoch": 9.472150577858281,
      "grad_norm": 2.1792547702789307,
      "learning_rate": 2.993063319882382e-07,
      "loss": 2.0045,
      "num_input_tokens_seen": 489860464,
      "step": 481100
    },
    {
      "epoch": 9.474119430607784,
      "grad_norm": 1.7961978912353516,
      "learning_rate": 2.9911750661841526e-07,
      "loss": 1.9754,
      "num_input_tokens_seen": 489962864,
      "step": 481200
    },
    {
      "epoch": 9.476088283357289,
      "grad_norm": 1.9791311025619507,
      "learning_rate": 2.989287154069662e-07,
      "loss": 1.9865,
      "num_input_tokens_seen": 490065056,
      "step": 481300
    },
    {
      "epoch": 9.478057136106791,
      "grad_norm": 1.8737519979476929,
      "learning_rate": 2.987399583859932e-07,
      "loss": 1.9861,
      "num_input_tokens_seen": 490167456,
      "step": 481400
    },
    {
      "epoch": 9.480025988856294,
      "grad_norm": 2.049654483795166,
      "learning_rate": 2.9855123558759275e-07,
      "loss": 1.9892,
      "num_input_tokens_seen": 490269040,
      "step": 481500
    },
    {
      "epoch": 9.481994841605797,
      "grad_norm": 2.0935235023498535,
      "learning_rate": 2.9836254704385567e-07,
      "loss": 1.9572,
      "num_input_tokens_seen": 490371440,
      "step": 481600
    },
    {
      "epoch": 9.4839636943553,
      "grad_norm": 1.6533153057098389,
      "learning_rate": 2.981738927868667e-07,
      "loss": 1.9963,
      "num_input_tokens_seen": 490473840,
      "step": 481700
    },
    {
      "epoch": 9.485932547104802,
      "grad_norm": 1.9395042657852173,
      "learning_rate": 2.9798527284870503e-07,
      "loss": 2.0024,
      "num_input_tokens_seen": 490573944,
      "step": 481800
    },
    {
      "epoch": 9.487901399854305,
      "grad_norm": 2.0553762912750244,
      "learning_rate": 2.977966872614437e-07,
      "loss": 2.039,
      "num_input_tokens_seen": 490674256,
      "step": 481900
    },
    {
      "epoch": 9.489870252603808,
      "grad_norm": 2.0024337768554688,
      "learning_rate": 2.976081360571504e-07,
      "loss": 1.9972,
      "num_input_tokens_seen": 490776656,
      "step": 482000
    },
    {
      "epoch": 9.49183910535331,
      "grad_norm": 1.929008960723877,
      "learning_rate": 2.9741961926788637e-07,
      "loss": 2.0485,
      "num_input_tokens_seen": 490878264,
      "step": 482100
    },
    {
      "epoch": 9.493807958102813,
      "grad_norm": 1.934262752532959,
      "learning_rate": 2.9723113692570755e-07,
      "loss": 1.9917,
      "num_input_tokens_seen": 490980664,
      "step": 482200
    },
    {
      "epoch": 9.495776810852316,
      "grad_norm": 1.8070461750030518,
      "learning_rate": 2.970426890626637e-07,
      "loss": 1.9525,
      "num_input_tokens_seen": 491083064,
      "step": 482300
    },
    {
      "epoch": 9.497745663601819,
      "grad_norm": 1.7284494638442993,
      "learning_rate": 2.9685427571079856e-07,
      "loss": 2.036,
      "num_input_tokens_seen": 491183808,
      "step": 482400
    },
    {
      "epoch": 9.499714516351322,
      "grad_norm": 1.7930779457092285,
      "learning_rate": 2.9666589690215066e-07,
      "loss": 1.969,
      "num_input_tokens_seen": 491286208,
      "step": 482500
    },
    {
      "epoch": 9.501683369100824,
      "grad_norm": 1.8933864831924438,
      "learning_rate": 2.9647755266875187e-07,
      "loss": 2.0328,
      "num_input_tokens_seen": 491387224,
      "step": 482600
    },
    {
      "epoch": 9.503652221850327,
      "grad_norm": 1.849589228630066,
      "learning_rate": 2.962892430426287e-07,
      "loss": 2.0247,
      "num_input_tokens_seen": 491488816,
      "step": 482700
    },
    {
      "epoch": 9.50562107459983,
      "grad_norm": 2.3360965251922607,
      "learning_rate": 2.9610096805580155e-07,
      "loss": 1.9656,
      "num_input_tokens_seen": 491590696,
      "step": 482800
    },
    {
      "epoch": 9.507589927349333,
      "grad_norm": 1.8918428421020508,
      "learning_rate": 2.9591272774028504e-07,
      "loss": 1.9105,
      "num_input_tokens_seen": 491693096,
      "step": 482900
    },
    {
      "epoch": 9.509558780098835,
      "grad_norm": 2.1820144653320312,
      "learning_rate": 2.9572452212808774e-07,
      "loss": 1.9887,
      "num_input_tokens_seen": 491794912,
      "step": 483000
    },
    {
      "epoch": 9.51152763284834,
      "grad_norm": 1.9347286224365234,
      "learning_rate": 2.9553635125121267e-07,
      "loss": 1.9926,
      "num_input_tokens_seen": 491897312,
      "step": 483100
    },
    {
      "epoch": 9.513496485597843,
      "grad_norm": 2.362489700317383,
      "learning_rate": 2.953482151416564e-07,
      "loss": 2.0144,
      "num_input_tokens_seen": 491999712,
      "step": 483200
    },
    {
      "epoch": 9.515465338347346,
      "grad_norm": 1.9346020221710205,
      "learning_rate": 2.9516011383140994e-07,
      "loss": 1.9935,
      "num_input_tokens_seen": 492102112,
      "step": 483300
    },
    {
      "epoch": 9.517434191096848,
      "grad_norm": 1.8565385341644287,
      "learning_rate": 2.9497204735245874e-07,
      "loss": 1.9682,
      "num_input_tokens_seen": 492204512,
      "step": 483400
    },
    {
      "epoch": 9.519403043846351,
      "grad_norm": 1.8313584327697754,
      "learning_rate": 2.9478401573678124e-07,
      "loss": 1.9826,
      "num_input_tokens_seen": 492306088,
      "step": 483500
    },
    {
      "epoch": 9.521371896595854,
      "grad_norm": 1.8167219161987305,
      "learning_rate": 2.9459601901635135e-07,
      "loss": 1.9787,
      "num_input_tokens_seen": 492408488,
      "step": 483600
    },
    {
      "epoch": 9.523340749345357,
      "grad_norm": 2.275207281112671,
      "learning_rate": 2.944080572231357e-07,
      "loss": 2.0177,
      "num_input_tokens_seen": 492510224,
      "step": 483700
    },
    {
      "epoch": 9.52530960209486,
      "grad_norm": 2.0315277576446533,
      "learning_rate": 2.942201303890961e-07,
      "loss": 1.9909,
      "num_input_tokens_seen": 492612624,
      "step": 483800
    },
    {
      "epoch": 9.527278454844362,
      "grad_norm": 1.9449855089187622,
      "learning_rate": 2.940322385461874e-07,
      "loss": 1.9944,
      "num_input_tokens_seen": 492714232,
      "step": 483900
    },
    {
      "epoch": 9.529247307593865,
      "grad_norm": 1.933132290840149,
      "learning_rate": 2.938443817263597e-07,
      "loss": 1.982,
      "num_input_tokens_seen": 492816632,
      "step": 484000
    },
    {
      "epoch": 9.531216160343368,
      "grad_norm": 2.044750928878784,
      "learning_rate": 2.9365655996155573e-07,
      "loss": 1.9994,
      "num_input_tokens_seen": 492919024,
      "step": 484100
    },
    {
      "epoch": 9.53318501309287,
      "grad_norm": 1.9562097787857056,
      "learning_rate": 2.934687732837135e-07,
      "loss": 1.9182,
      "num_input_tokens_seen": 493021424,
      "step": 484200
    },
    {
      "epoch": 9.535153865842373,
      "grad_norm": 1.870526909828186,
      "learning_rate": 2.9328102172476447e-07,
      "loss": 1.9992,
      "num_input_tokens_seen": 493122904,
      "step": 484300
    },
    {
      "epoch": 9.537122718591876,
      "grad_norm": 2.134713649749756,
      "learning_rate": 2.930933053166341e-07,
      "loss": 2.0263,
      "num_input_tokens_seen": 493224592,
      "step": 484400
    },
    {
      "epoch": 9.539091571341379,
      "grad_norm": 2.2230257987976074,
      "learning_rate": 2.9290562409124207e-07,
      "loss": 1.9926,
      "num_input_tokens_seen": 493326136,
      "step": 484500
    },
    {
      "epoch": 9.541060424090883,
      "grad_norm": 2.0468411445617676,
      "learning_rate": 2.927179780805019e-07,
      "loss": 1.975,
      "num_input_tokens_seen": 493428536,
      "step": 484600
    },
    {
      "epoch": 9.543029276840386,
      "grad_norm": 2.568345785140991,
      "learning_rate": 2.925303673163213e-07,
      "loss": 1.9885,
      "num_input_tokens_seen": 493530936,
      "step": 484700
    },
    {
      "epoch": 9.544998129589889,
      "grad_norm": 2.144352912902832,
      "learning_rate": 2.923427918306018e-07,
      "loss": 1.9855,
      "num_input_tokens_seen": 493632416,
      "step": 484800
    },
    {
      "epoch": 9.546966982339391,
      "grad_norm": 1.7890158891677856,
      "learning_rate": 2.9215525165523914e-07,
      "loss": 1.9792,
      "num_input_tokens_seen": 493734264,
      "step": 484900
    },
    {
      "epoch": 9.548935835088894,
      "grad_norm": 1.8331222534179688,
      "learning_rate": 2.919677468221229e-07,
      "loss": 1.989,
      "num_input_tokens_seen": 493836040,
      "step": 485000
    },
    {
      "epoch": 9.550904687838397,
      "grad_norm": 2.1179535388946533,
      "learning_rate": 2.917802773631367e-07,
      "loss": 1.9415,
      "num_input_tokens_seen": 493937768,
      "step": 485100
    },
    {
      "epoch": 9.5528735405879,
      "grad_norm": 1.8272769451141357,
      "learning_rate": 2.9159284331015797e-07,
      "loss": 2.0194,
      "num_input_tokens_seen": 494040168,
      "step": 485200
    },
    {
      "epoch": 9.554842393337402,
      "grad_norm": 1.80509352684021,
      "learning_rate": 2.914054446950586e-07,
      "loss": 1.9741,
      "num_input_tokens_seen": 494141928,
      "step": 485300
    },
    {
      "epoch": 9.556811246086905,
      "grad_norm": 1.732566237449646,
      "learning_rate": 2.912180815497039e-07,
      "loss": 2.0084,
      "num_input_tokens_seen": 494243072,
      "step": 485400
    },
    {
      "epoch": 9.558780098836408,
      "grad_norm": 1.8130834102630615,
      "learning_rate": 2.9103075390595347e-07,
      "loss": 2.0252,
      "num_input_tokens_seen": 494343600,
      "step": 485500
    },
    {
      "epoch": 9.56074895158591,
      "grad_norm": 2.0498244762420654,
      "learning_rate": 2.9084346179566107e-07,
      "loss": 1.9835,
      "num_input_tokens_seen": 494446000,
      "step": 485600
    },
    {
      "epoch": 9.562717804335414,
      "grad_norm": 2.3239004611968994,
      "learning_rate": 2.906562052506736e-07,
      "loss": 2.0066,
      "num_input_tokens_seen": 494548400,
      "step": 485700
    },
    {
      "epoch": 9.564686657084916,
      "grad_norm": 2.157916784286499,
      "learning_rate": 2.9046898430283286e-07,
      "loss": 2.0113,
      "num_input_tokens_seen": 494650040,
      "step": 485800
    },
    {
      "epoch": 9.566655509834419,
      "grad_norm": 2.1191813945770264,
      "learning_rate": 2.9028179898397397e-07,
      "loss": 1.9713,
      "num_input_tokens_seen": 494752440,
      "step": 485900
    },
    {
      "epoch": 9.568624362583922,
      "grad_norm": 1.954754114151001,
      "learning_rate": 2.9009464932592653e-07,
      "loss": 2.033,
      "num_input_tokens_seen": 494854840,
      "step": 486000
    },
    {
      "epoch": 9.570593215333425,
      "grad_norm": 2.3735392093658447,
      "learning_rate": 2.8990753536051335e-07,
      "loss": 2.0372,
      "num_input_tokens_seen": 494955728,
      "step": 486100
    },
    {
      "epoch": 9.572562068082927,
      "grad_norm": 1.8865255117416382,
      "learning_rate": 2.8972045711955186e-07,
      "loss": 1.9987,
      "num_input_tokens_seen": 495058128,
      "step": 486200
    },
    {
      "epoch": 9.57453092083243,
      "grad_norm": 1.7551848888397217,
      "learning_rate": 2.89533414634853e-07,
      "loss": 1.9405,
      "num_input_tokens_seen": 495160024,
      "step": 486300
    },
    {
      "epoch": 9.576499773581935,
      "grad_norm": 1.8886034488677979,
      "learning_rate": 2.8934640793822184e-07,
      "loss": 1.9612,
      "num_input_tokens_seen": 495262424,
      "step": 486400
    },
    {
      "epoch": 9.578468626331437,
      "grad_norm": 1.8521831035614014,
      "learning_rate": 2.891594370614575e-07,
      "loss": 2.0034,
      "num_input_tokens_seen": 495364824,
      "step": 486500
    },
    {
      "epoch": 9.58043747908094,
      "grad_norm": 1.8598312139511108,
      "learning_rate": 2.8897250203635227e-07,
      "loss": 2.0004,
      "num_input_tokens_seen": 495466600,
      "step": 486600
    },
    {
      "epoch": 9.582406331830443,
      "grad_norm": 1.9452985525131226,
      "learning_rate": 2.8878560289469323e-07,
      "loss": 1.9654,
      "num_input_tokens_seen": 495569000,
      "step": 486700
    },
    {
      "epoch": 9.584375184579946,
      "grad_norm": 2.159031629562378,
      "learning_rate": 2.88598739668261e-07,
      "loss": 1.9751,
      "num_input_tokens_seen": 495670232,
      "step": 486800
    },
    {
      "epoch": 9.586344037329448,
      "grad_norm": 1.8398820161819458,
      "learning_rate": 2.8841191238883024e-07,
      "loss": 1.9639,
      "num_input_tokens_seen": 495771984,
      "step": 486900
    },
    {
      "epoch": 9.588312890078951,
      "grad_norm": 1.9648598432540894,
      "learning_rate": 2.8822512108816895e-07,
      "loss": 1.9311,
      "num_input_tokens_seen": 495874384,
      "step": 487000
    },
    {
      "epoch": 9.590281742828454,
      "grad_norm": 1.8710896968841553,
      "learning_rate": 2.880383657980396e-07,
      "loss": 1.9724,
      "num_input_tokens_seen": 495976248,
      "step": 487100
    },
    {
      "epoch": 9.592250595577957,
      "grad_norm": 1.960744023323059,
      "learning_rate": 2.8785164655019837e-07,
      "loss": 1.9517,
      "num_input_tokens_seen": 496078368,
      "step": 487200
    },
    {
      "epoch": 9.59421944832746,
      "grad_norm": 1.7784029245376587,
      "learning_rate": 2.876649633763953e-07,
      "loss": 1.9817,
      "num_input_tokens_seen": 496180008,
      "step": 487300
    },
    {
      "epoch": 9.596188301076962,
      "grad_norm": 1.8506587743759155,
      "learning_rate": 2.874783163083746e-07,
      "loss": 2.005,
      "num_input_tokens_seen": 496281200,
      "step": 487400
    },
    {
      "epoch": 9.598157153826465,
      "grad_norm": 1.8617209196090698,
      "learning_rate": 2.872917053778732e-07,
      "loss": 1.9451,
      "num_input_tokens_seen": 496383600,
      "step": 487500
    },
    {
      "epoch": 9.600126006575968,
      "grad_norm": 1.8235269784927368,
      "learning_rate": 2.871051306166237e-07,
      "loss": 2.006,
      "num_input_tokens_seen": 496485952,
      "step": 487600
    },
    {
      "epoch": 9.60209485932547,
      "grad_norm": 2.2079827785491943,
      "learning_rate": 2.869185920563509e-07,
      "loss": 2.0255,
      "num_input_tokens_seen": 496588352,
      "step": 487700
    },
    {
      "epoch": 9.604063712074973,
      "grad_norm": 2.095198154449463,
      "learning_rate": 2.867320897287745e-07,
      "loss": 1.979,
      "num_input_tokens_seen": 496689888,
      "step": 487800
    },
    {
      "epoch": 9.606032564824476,
      "grad_norm": 1.7441695928573608,
      "learning_rate": 2.8654562366560695e-07,
      "loss": 1.9682,
      "num_input_tokens_seen": 496792288,
      "step": 487900
    },
    {
      "epoch": 9.60800141757398,
      "grad_norm": 1.9506421089172363,
      "learning_rate": 2.863591938985563e-07,
      "loss": 1.9592,
      "num_input_tokens_seen": 496893816,
      "step": 488000
    },
    {
      "epoch": 9.609970270323483,
      "grad_norm": 1.9684830904006958,
      "learning_rate": 2.8617280045932254e-07,
      "loss": 1.9533,
      "num_input_tokens_seen": 496996216,
      "step": 488100
    },
    {
      "epoch": 9.611939123072986,
      "grad_norm": 1.8145133256912231,
      "learning_rate": 2.859864433796008e-07,
      "loss": 1.9905,
      "num_input_tokens_seen": 497098616,
      "step": 488200
    },
    {
      "epoch": 9.613907975822489,
      "grad_norm": 1.9941847324371338,
      "learning_rate": 2.858001226910788e-07,
      "loss": 1.9753,
      "num_input_tokens_seen": 497200288,
      "step": 488300
    },
    {
      "epoch": 9.615876828571992,
      "grad_norm": 1.7515360116958618,
      "learning_rate": 2.8561383842543953e-07,
      "loss": 1.9834,
      "num_input_tokens_seen": 497302144,
      "step": 488400
    },
    {
      "epoch": 9.617845681321494,
      "grad_norm": 2.0508158206939697,
      "learning_rate": 2.85427590614359e-07,
      "loss": 2.0187,
      "num_input_tokens_seen": 497403072,
      "step": 488500
    },
    {
      "epoch": 9.619814534070997,
      "grad_norm": 2.0787603855133057,
      "learning_rate": 2.8524137928950676e-07,
      "loss": 1.9858,
      "num_input_tokens_seen": 497505472,
      "step": 488600
    },
    {
      "epoch": 9.6217833868205,
      "grad_norm": 1.88705575466156,
      "learning_rate": 2.8505520448254647e-07,
      "loss": 1.9786,
      "num_input_tokens_seen": 497607288,
      "step": 488700
    },
    {
      "epoch": 9.623752239570003,
      "grad_norm": 1.6387436389923096,
      "learning_rate": 2.848690662251359e-07,
      "loss": 2.017,
      "num_input_tokens_seen": 497708920,
      "step": 488800
    },
    {
      "epoch": 9.625721092319505,
      "grad_norm": 2.3381474018096924,
      "learning_rate": 2.8468296454892625e-07,
      "loss": 2.0175,
      "num_input_tokens_seen": 497811320,
      "step": 488900
    },
    {
      "epoch": 9.627689945069008,
      "grad_norm": 1.9387003183364868,
      "learning_rate": 2.8449689948556223e-07,
      "loss": 1.9561,
      "num_input_tokens_seen": 497913720,
      "step": 489000
    },
    {
      "epoch": 9.62965879781851,
      "grad_norm": 3.1934409141540527,
      "learning_rate": 2.843108710666828e-07,
      "loss": 1.9508,
      "num_input_tokens_seen": 498015576,
      "step": 489100
    },
    {
      "epoch": 9.631627650568014,
      "grad_norm": 2.009428024291992,
      "learning_rate": 2.841248793239205e-07,
      "loss": 2.0133,
      "num_input_tokens_seen": 498117384,
      "step": 489200
    },
    {
      "epoch": 9.633596503317516,
      "grad_norm": 2.0938000679016113,
      "learning_rate": 2.8393892428890176e-07,
      "loss": 1.9905,
      "num_input_tokens_seen": 498219784,
      "step": 489300
    },
    {
      "epoch": 9.63556535606702,
      "grad_norm": 1.8964952230453491,
      "learning_rate": 2.837530059932468e-07,
      "loss": 1.9921,
      "num_input_tokens_seen": 498321480,
      "step": 489400
    },
    {
      "epoch": 9.637534208816522,
      "grad_norm": 2.0135340690612793,
      "learning_rate": 2.8356712446856903e-07,
      "loss": 1.9707,
      "num_input_tokens_seen": 498423232,
      "step": 489500
    },
    {
      "epoch": 9.639503061566025,
      "grad_norm": 1.9531112909317017,
      "learning_rate": 2.8338127974647623e-07,
      "loss": 1.9575,
      "num_input_tokens_seen": 498525632,
      "step": 489600
    },
    {
      "epoch": 9.641471914315527,
      "grad_norm": 1.8988596200942993,
      "learning_rate": 2.8319547185856983e-07,
      "loss": 1.9986,
      "num_input_tokens_seen": 498627744,
      "step": 489700
    },
    {
      "epoch": 9.643440767065032,
      "grad_norm": 2.2402443885803223,
      "learning_rate": 2.8300970083644505e-07,
      "loss": 1.9953,
      "num_input_tokens_seen": 498729304,
      "step": 489800
    },
    {
      "epoch": 9.645409619814535,
      "grad_norm": 2.510845422744751,
      "learning_rate": 2.828239667116902e-07,
      "loss": 1.9903,
      "num_input_tokens_seen": 498831704,
      "step": 489900
    },
    {
      "epoch": 9.647378472564037,
      "grad_norm": 1.8515900373458862,
      "learning_rate": 2.82638269515888e-07,
      "loss": 1.9601,
      "num_input_tokens_seen": 498933656,
      "step": 490000
    },
    {
      "epoch": 9.64934732531354,
      "grad_norm": 1.7519137859344482,
      "learning_rate": 2.8245260928061485e-07,
      "loss": 1.9976,
      "num_input_tokens_seen": 499036056,
      "step": 490100
    },
    {
      "epoch": 9.651316178063043,
      "grad_norm": 2.0999104976654053,
      "learning_rate": 2.8226698603744084e-07,
      "loss": 1.9393,
      "num_input_tokens_seen": 499137696,
      "step": 490200
    },
    {
      "epoch": 9.653285030812546,
      "grad_norm": 1.8033839464187622,
      "learning_rate": 2.820813998179291e-07,
      "loss": 2.0059,
      "num_input_tokens_seen": 499240096,
      "step": 490300
    },
    {
      "epoch": 9.655253883562049,
      "grad_norm": 2.0442311763763428,
      "learning_rate": 2.818958506536372e-07,
      "loss": 2.0026,
      "num_input_tokens_seen": 499342496,
      "step": 490400
    },
    {
      "epoch": 9.657222736311551,
      "grad_norm": 1.8606821298599243,
      "learning_rate": 2.817103385761167e-07,
      "loss": 2.002,
      "num_input_tokens_seen": 499444128,
      "step": 490500
    },
    {
      "epoch": 9.659191589061054,
      "grad_norm": 1.8472729921340942,
      "learning_rate": 2.815248636169118e-07,
      "loss": 1.9577,
      "num_input_tokens_seen": 499546528,
      "step": 490600
    },
    {
      "epoch": 9.661160441810557,
      "grad_norm": 2.1541848182678223,
      "learning_rate": 2.813394258075613e-07,
      "loss": 1.9521,
      "num_input_tokens_seen": 499648928,
      "step": 490700
    },
    {
      "epoch": 9.66312929456006,
      "grad_norm": 2.046351432800293,
      "learning_rate": 2.811540251795968e-07,
      "loss": 1.9455,
      "num_input_tokens_seen": 499751328,
      "step": 490800
    },
    {
      "epoch": 9.665098147309562,
      "grad_norm": 1.9514741897583008,
      "learning_rate": 2.809686617645449e-07,
      "loss": 1.9988,
      "num_input_tokens_seen": 499853024,
      "step": 490900
    },
    {
      "epoch": 9.667067000059065,
      "grad_norm": 1.676711916923523,
      "learning_rate": 2.8078333559392445e-07,
      "loss": 2.0047,
      "num_input_tokens_seen": 499954528,
      "step": 491000
    },
    {
      "epoch": 9.669035852808568,
      "grad_norm": 1.87375807762146,
      "learning_rate": 2.805980466992488e-07,
      "loss": 1.9715,
      "num_input_tokens_seen": 500056312,
      "step": 491100
    },
    {
      "epoch": 9.67100470555807,
      "grad_norm": 1.7538889646530151,
      "learning_rate": 2.804127951120248e-07,
      "loss": 1.9975,
      "num_input_tokens_seen": 500158144,
      "step": 491200
    },
    {
      "epoch": 9.672973558307575,
      "grad_norm": 1.7303611040115356,
      "learning_rate": 2.802275808637529e-07,
      "loss": 1.9748,
      "num_input_tokens_seen": 500260544,
      "step": 491300
    },
    {
      "epoch": 9.674942411057078,
      "grad_norm": 1.888983130455017,
      "learning_rate": 2.800424039859275e-07,
      "loss": 2.0078,
      "num_input_tokens_seen": 500361696,
      "step": 491400
    },
    {
      "epoch": 9.67691126380658,
      "grad_norm": 1.9586635828018188,
      "learning_rate": 2.7985726451003575e-07,
      "loss": 2.0308,
      "num_input_tokens_seen": 500462624,
      "step": 491500
    },
    {
      "epoch": 9.678880116556083,
      "grad_norm": 1.8989180326461792,
      "learning_rate": 2.796721624675595e-07,
      "loss": 1.976,
      "num_input_tokens_seen": 500565024,
      "step": 491600
    },
    {
      "epoch": 9.680848969305586,
      "grad_norm": 1.726993203163147,
      "learning_rate": 2.7948709788997373e-07,
      "loss": 1.9558,
      "num_input_tokens_seen": 500666680,
      "step": 491700
    },
    {
      "epoch": 9.682817822055089,
      "grad_norm": 1.9682456254959106,
      "learning_rate": 2.7930207080874723e-07,
      "loss": 2.0421,
      "num_input_tokens_seen": 500769080,
      "step": 491800
    },
    {
      "epoch": 9.684786674804592,
      "grad_norm": 2.2453274726867676,
      "learning_rate": 2.791170812553419e-07,
      "loss": 1.9356,
      "num_input_tokens_seen": 500871480,
      "step": 491900
    },
    {
      "epoch": 9.686755527554094,
      "grad_norm": 2.1051816940307617,
      "learning_rate": 2.789321292612139e-07,
      "loss": 2.0088,
      "num_input_tokens_seen": 500972744,
      "step": 492000
    },
    {
      "epoch": 9.688724380303597,
      "grad_norm": 2.17006516456604,
      "learning_rate": 2.7874721485781285e-07,
      "loss": 1.9819,
      "num_input_tokens_seen": 501074712,
      "step": 492100
    },
    {
      "epoch": 9.6906932330531,
      "grad_norm": 2.107874870300293,
      "learning_rate": 2.785623380765818e-07,
      "loss": 1.9853,
      "num_input_tokens_seen": 501176480,
      "step": 492200
    },
    {
      "epoch": 9.692662085802603,
      "grad_norm": 2.049481153488159,
      "learning_rate": 2.783774989489574e-07,
      "loss": 1.9698,
      "num_input_tokens_seen": 501278880,
      "step": 492300
    },
    {
      "epoch": 9.694630938552105,
      "grad_norm": 1.8123736381530762,
      "learning_rate": 2.7819269750637003e-07,
      "loss": 2.0016,
      "num_input_tokens_seen": 501380888,
      "step": 492400
    },
    {
      "epoch": 9.696599791301608,
      "grad_norm": 1.734336495399475,
      "learning_rate": 2.780079337802436e-07,
      "loss": 1.9658,
      "num_input_tokens_seen": 501483288,
      "step": 492500
    },
    {
      "epoch": 9.698568644051111,
      "grad_norm": 2.018087387084961,
      "learning_rate": 2.7782320780199575e-07,
      "loss": 1.9907,
      "num_input_tokens_seen": 501585688,
      "step": 492600
    },
    {
      "epoch": 9.700537496800614,
      "grad_norm": 2.8398308753967285,
      "learning_rate": 2.7763851960303766e-07,
      "loss": 2.0221,
      "num_input_tokens_seen": 501687240,
      "step": 492700
    },
    {
      "epoch": 9.702506349550116,
      "grad_norm": 2.0368359088897705,
      "learning_rate": 2.774538692147733e-07,
      "loss": 1.9818,
      "num_input_tokens_seen": 501788744,
      "step": 492800
    },
    {
      "epoch": 9.70447520229962,
      "grad_norm": 1.855392336845398,
      "learning_rate": 2.7726925666860196e-07,
      "loss": 1.9906,
      "num_input_tokens_seen": 501891144,
      "step": 492900
    },
    {
      "epoch": 9.706444055049122,
      "grad_norm": 1.8716174364089966,
      "learning_rate": 2.7708468199591474e-07,
      "loss": 1.996,
      "num_input_tokens_seen": 501991904,
      "step": 493000
    },
    {
      "epoch": 9.708412907798627,
      "grad_norm": 2.46576189994812,
      "learning_rate": 2.7690014522809734e-07,
      "loss": 1.9955,
      "num_input_tokens_seen": 502094304,
      "step": 493100
    },
    {
      "epoch": 9.71038176054813,
      "grad_norm": 2.07723069190979,
      "learning_rate": 2.7671564639652814e-07,
      "loss": 1.9854,
      "num_input_tokens_seen": 502196704,
      "step": 493200
    },
    {
      "epoch": 9.712350613297632,
      "grad_norm": 4.248500347137451,
      "learning_rate": 2.7653118553258034e-07,
      "loss": 2.0202,
      "num_input_tokens_seen": 502299104,
      "step": 493300
    },
    {
      "epoch": 9.714319466047135,
      "grad_norm": 2.7718570232391357,
      "learning_rate": 2.763467626676196e-07,
      "loss": 2.0088,
      "num_input_tokens_seen": 502400896,
      "step": 493400
    },
    {
      "epoch": 9.716288318796638,
      "grad_norm": 2.031266212463379,
      "learning_rate": 2.7616237783300534e-07,
      "loss": 1.9739,
      "num_input_tokens_seen": 502502600,
      "step": 493500
    },
    {
      "epoch": 9.71825717154614,
      "grad_norm": 2.1321840286254883,
      "learning_rate": 2.759780310600908e-07,
      "loss": 1.934,
      "num_input_tokens_seen": 502604488,
      "step": 493600
    },
    {
      "epoch": 9.720226024295643,
      "grad_norm": 1.8724679946899414,
      "learning_rate": 2.757937223802226e-07,
      "loss": 1.9673,
      "num_input_tokens_seen": 502705680,
      "step": 493700
    },
    {
      "epoch": 9.722194877045146,
      "grad_norm": 1.9844439029693604,
      "learning_rate": 2.7560945182474114e-07,
      "loss": 2.0069,
      "num_input_tokens_seen": 502808032,
      "step": 493800
    },
    {
      "epoch": 9.724163729794649,
      "grad_norm": 2.011350393295288,
      "learning_rate": 2.7542521942497956e-07,
      "loss": 1.955,
      "num_input_tokens_seen": 502908816,
      "step": 493900
    },
    {
      "epoch": 9.726132582544151,
      "grad_norm": 2.1120424270629883,
      "learning_rate": 2.752410252122652e-07,
      "loss": 1.9777,
      "num_input_tokens_seen": 503011216,
      "step": 494000
    },
    {
      "epoch": 9.728101435293654,
      "grad_norm": 2.089890956878662,
      "learning_rate": 2.750568692179189e-07,
      "loss": 1.9646,
      "num_input_tokens_seen": 503113120,
      "step": 494100
    },
    {
      "epoch": 9.730070288043157,
      "grad_norm": 2.004549503326416,
      "learning_rate": 2.748727514732549e-07,
      "loss": 2.0183,
      "num_input_tokens_seen": 503213472,
      "step": 494200
    },
    {
      "epoch": 9.73203914079266,
      "grad_norm": 2.12077260017395,
      "learning_rate": 2.7468867200958045e-07,
      "loss": 1.9554,
      "num_input_tokens_seen": 503315872,
      "step": 494300
    },
    {
      "epoch": 9.734007993542162,
      "grad_norm": 1.6632493734359741,
      "learning_rate": 2.7450463085819704e-07,
      "loss": 1.9637,
      "num_input_tokens_seen": 503416696,
      "step": 494400
    },
    {
      "epoch": 9.735976846291665,
      "grad_norm": 1.9577703475952148,
      "learning_rate": 2.743206280503992e-07,
      "loss": 1.9732,
      "num_input_tokens_seen": 503519096,
      "step": 494500
    },
    {
      "epoch": 9.737945699041168,
      "grad_norm": 1.8225675821304321,
      "learning_rate": 2.7413666361747506e-07,
      "loss": 1.9506,
      "num_input_tokens_seen": 503621496,
      "step": 494600
    },
    {
      "epoch": 9.739914551790672,
      "grad_norm": 2.062499523162842,
      "learning_rate": 2.7395273759070645e-07,
      "loss": 1.9915,
      "num_input_tokens_seen": 503723896,
      "step": 494700
    },
    {
      "epoch": 9.741883404540175,
      "grad_norm": 2.1235992908477783,
      "learning_rate": 2.7376885000136807e-07,
      "loss": 2.0217,
      "num_input_tokens_seen": 503825136,
      "step": 494800
    },
    {
      "epoch": 9.743852257289678,
      "grad_norm": 2.125648021697998,
      "learning_rate": 2.735850008807287e-07,
      "loss": 1.9994,
      "num_input_tokens_seen": 503927536,
      "step": 494900
    },
    {
      "epoch": 9.74582111003918,
      "grad_norm": 1.895817756652832,
      "learning_rate": 2.7340119026005015e-07,
      "loss": 2.062,
      "num_input_tokens_seen": 504028856,
      "step": 495000
    },
    {
      "epoch": 9.747789962788683,
      "grad_norm": 2.0151543617248535,
      "learning_rate": 2.732174181705882e-07,
      "loss": 1.9727,
      "num_input_tokens_seen": 504131168,
      "step": 495100
    },
    {
      "epoch": 9.749758815538186,
      "grad_norm": 2.3339152336120605,
      "learning_rate": 2.7303368464359133e-07,
      "loss": 1.9456,
      "num_input_tokens_seen": 504233568,
      "step": 495200
    },
    {
      "epoch": 9.751727668287689,
      "grad_norm": 1.84795081615448,
      "learning_rate": 2.7284998971030205e-07,
      "loss": 1.9769,
      "num_input_tokens_seen": 504335152,
      "step": 495300
    },
    {
      "epoch": 9.753696521037192,
      "grad_norm": 1.887708306312561,
      "learning_rate": 2.7266633340195615e-07,
      "loss": 1.9694,
      "num_input_tokens_seen": 504436984,
      "step": 495400
    },
    {
      "epoch": 9.755665373786695,
      "grad_norm": 1.8877062797546387,
      "learning_rate": 2.724827157497829e-07,
      "loss": 1.9582,
      "num_input_tokens_seen": 504538504,
      "step": 495500
    },
    {
      "epoch": 9.757634226536197,
      "grad_norm": 1.9336410760879517,
      "learning_rate": 2.7229913678500495e-07,
      "loss": 2.0069,
      "num_input_tokens_seen": 504640904,
      "step": 495600
    },
    {
      "epoch": 9.7596030792857,
      "grad_norm": 1.885657548904419,
      "learning_rate": 2.72115596538838e-07,
      "loss": 1.9436,
      "num_input_tokens_seen": 504743304,
      "step": 495700
    },
    {
      "epoch": 9.761571932035203,
      "grad_norm": 1.6186270713806152,
      "learning_rate": 2.7193209504249205e-07,
      "loss": 1.9954,
      "num_input_tokens_seen": 504845016,
      "step": 495800
    },
    {
      "epoch": 9.763540784784706,
      "grad_norm": 1.859894871711731,
      "learning_rate": 2.717486323271696e-07,
      "loss": 1.96,
      "num_input_tokens_seen": 504946704,
      "step": 495900
    },
    {
      "epoch": 9.765509637534208,
      "grad_norm": 1.8920289278030396,
      "learning_rate": 2.715652084240673e-07,
      "loss": 2.009,
      "num_input_tokens_seen": 505047872,
      "step": 496000
    },
    {
      "epoch": 9.767478490283711,
      "grad_norm": 2.127593994140625,
      "learning_rate": 2.7138182336437407e-07,
      "loss": 1.961,
      "num_input_tokens_seen": 505147912,
      "step": 496100
    },
    {
      "epoch": 9.769447343033214,
      "grad_norm": 1.780908226966858,
      "learning_rate": 2.711984771792741e-07,
      "loss": 1.9419,
      "num_input_tokens_seen": 505250312,
      "step": 496200
    },
    {
      "epoch": 9.771416195782717,
      "grad_norm": 1.4779224395751953,
      "learning_rate": 2.71015169899943e-07,
      "loss": 1.9676,
      "num_input_tokens_seen": 505351496,
      "step": 496300
    },
    {
      "epoch": 9.77338504853222,
      "grad_norm": 1.8564016819000244,
      "learning_rate": 2.7083190155755087e-07,
      "loss": 2.0155,
      "num_input_tokens_seen": 505453896,
      "step": 496400
    },
    {
      "epoch": 9.775353901281724,
      "grad_norm": 1.9791815280914307,
      "learning_rate": 2.7064867218326104e-07,
      "loss": 1.9633,
      "num_input_tokens_seen": 505556296,
      "step": 496500
    },
    {
      "epoch": 9.777322754031227,
      "grad_norm": 1.8004722595214844,
      "learning_rate": 2.7046548180823003e-07,
      "loss": 1.9416,
      "num_input_tokens_seen": 505658696,
      "step": 496600
    },
    {
      "epoch": 9.77929160678073,
      "grad_norm": 1.880846381187439,
      "learning_rate": 2.702823304636081e-07,
      "loss": 2.0206,
      "num_input_tokens_seen": 505760160,
      "step": 496700
    },
    {
      "epoch": 9.781260459530232,
      "grad_norm": 1.8842183351516724,
      "learning_rate": 2.700992181805382e-07,
      "loss": 2.0011,
      "num_input_tokens_seen": 505862048,
      "step": 496800
    },
    {
      "epoch": 9.783229312279735,
      "grad_norm": 2.29910945892334,
      "learning_rate": 2.69916144990157e-07,
      "loss": 1.9832,
      "num_input_tokens_seen": 505964448,
      "step": 496900
    },
    {
      "epoch": 9.785198165029238,
      "grad_norm": 1.9658163785934448,
      "learning_rate": 2.6973311092359485e-07,
      "loss": 1.9563,
      "num_input_tokens_seen": 506066032,
      "step": 497000
    },
    {
      "epoch": 9.78716701777874,
      "grad_norm": 1.6949522495269775,
      "learning_rate": 2.695501160119752e-07,
      "loss": 1.9488,
      "num_input_tokens_seen": 506167392,
      "step": 497100
    },
    {
      "epoch": 9.789135870528243,
      "grad_norm": 1.6766189336776733,
      "learning_rate": 2.6936716028641445e-07,
      "loss": 2.0115,
      "num_input_tokens_seen": 506269792,
      "step": 497200
    },
    {
      "epoch": 9.791104723277746,
      "grad_norm": 1.628969669342041,
      "learning_rate": 2.691842437780228e-07,
      "loss": 1.9597,
      "num_input_tokens_seen": 506371384,
      "step": 497300
    },
    {
      "epoch": 9.793073576027249,
      "grad_norm": 1.7124541997909546,
      "learning_rate": 2.690013665179037e-07,
      "loss": 2.0004,
      "num_input_tokens_seen": 506473040,
      "step": 497400
    },
    {
      "epoch": 9.795042428776751,
      "grad_norm": 2.026693820953369,
      "learning_rate": 2.6881852853715394e-07,
      "loss": 1.9905,
      "num_input_tokens_seen": 506573416,
      "step": 497500
    },
    {
      "epoch": 9.797011281526254,
      "grad_norm": 1.9993776082992554,
      "learning_rate": 2.6863572986686365e-07,
      "loss": 2.0227,
      "num_input_tokens_seen": 506675376,
      "step": 497600
    },
    {
      "epoch": 9.798980134275757,
      "grad_norm": 1.9662845134735107,
      "learning_rate": 2.68452970538116e-07,
      "loss": 1.9679,
      "num_input_tokens_seen": 506776608,
      "step": 497700
    },
    {
      "epoch": 9.80094898702526,
      "grad_norm": 1.9802907705307007,
      "learning_rate": 2.6827025058198767e-07,
      "loss": 1.9893,
      "num_input_tokens_seen": 506877784,
      "step": 497800
    },
    {
      "epoch": 9.802917839774762,
      "grad_norm": 1.615738868713379,
      "learning_rate": 2.6808757002954873e-07,
      "loss": 1.9751,
      "num_input_tokens_seen": 506980184,
      "step": 497900
    },
    {
      "epoch": 9.804886692524267,
      "grad_norm": 1.8563772439956665,
      "learning_rate": 2.6790492891186267e-07,
      "loss": 1.9837,
      "num_input_tokens_seen": 507081840,
      "step": 498000
    },
    {
      "epoch": 9.80685554527377,
      "grad_norm": 2.0212411880493164,
      "learning_rate": 2.677223272599857e-07,
      "loss": 1.9597,
      "num_input_tokens_seen": 507183472,
      "step": 498100
    },
    {
      "epoch": 9.808824398023273,
      "grad_norm": 2.157810926437378,
      "learning_rate": 2.675397651049678e-07,
      "loss": 1.9883,
      "num_input_tokens_seen": 507285232,
      "step": 498200
    },
    {
      "epoch": 9.810793250772775,
      "grad_norm": 1.9870649576187134,
      "learning_rate": 2.6735724247785224e-07,
      "loss": 1.9834,
      "num_input_tokens_seen": 507387632,
      "step": 498300
    },
    {
      "epoch": 9.812762103522278,
      "grad_norm": 1.8975505828857422,
      "learning_rate": 2.671747594096756e-07,
      "loss": 1.9949,
      "num_input_tokens_seen": 507488704,
      "step": 498400
    },
    {
      "epoch": 9.81473095627178,
      "grad_norm": 1.7841147184371948,
      "learning_rate": 2.6699231593146717e-07,
      "loss": 1.9388,
      "num_input_tokens_seen": 507590536,
      "step": 498500
    },
    {
      "epoch": 9.816699809021284,
      "grad_norm": 1.925437092781067,
      "learning_rate": 2.668099120742499e-07,
      "loss": 1.9734,
      "num_input_tokens_seen": 507692360,
      "step": 498600
    },
    {
      "epoch": 9.818668661770786,
      "grad_norm": 2.075392723083496,
      "learning_rate": 2.666275478690407e-07,
      "loss": 2.0219,
      "num_input_tokens_seen": 507794112,
      "step": 498700
    },
    {
      "epoch": 9.820637514520289,
      "grad_norm": 1.7077875137329102,
      "learning_rate": 2.6644522334684844e-07,
      "loss": 2.0062,
      "num_input_tokens_seen": 507895664,
      "step": 498800
    },
    {
      "epoch": 9.822606367269792,
      "grad_norm": 1.8576103448867798,
      "learning_rate": 2.662629385386762e-07,
      "loss": 1.963,
      "num_input_tokens_seen": 507996664,
      "step": 498900
    },
    {
      "epoch": 9.824575220019295,
      "grad_norm": 1.8298090696334839,
      "learning_rate": 2.660806934755195e-07,
      "loss": 1.9652,
      "num_input_tokens_seen": 508099064,
      "step": 499000
    },
    {
      "epoch": 9.826544072768797,
      "grad_norm": 2.252462387084961,
      "learning_rate": 2.658984881883683e-07,
      "loss": 2.0077,
      "num_input_tokens_seen": 508200608,
      "step": 499100
    },
    {
      "epoch": 9.8285129255183,
      "grad_norm": 3.3195366859436035,
      "learning_rate": 2.6571632270820443e-07,
      "loss": 1.9899,
      "num_input_tokens_seen": 508301488,
      "step": 499200
    },
    {
      "epoch": 9.830481778267803,
      "grad_norm": 1.695478916168213,
      "learning_rate": 2.655341970660041e-07,
      "loss": 1.9432,
      "num_input_tokens_seen": 508403888,
      "step": 499300
    },
    {
      "epoch": 9.832450631017306,
      "grad_norm": 1.786743402481079,
      "learning_rate": 2.653521112927356e-07,
      "loss": 1.9801,
      "num_input_tokens_seen": 508506288,
      "step": 499400
    },
    {
      "epoch": 9.834419483766808,
      "grad_norm": 2.0381052494049072,
      "learning_rate": 2.651700654193616e-07,
      "loss": 1.9688,
      "num_input_tokens_seen": 508608688,
      "step": 499500
    },
    {
      "epoch": 9.836388336516311,
      "grad_norm": 1.5066784620285034,
      "learning_rate": 2.6498805947683756e-07,
      "loss": 2.0125,
      "num_input_tokens_seen": 508711088,
      "step": 499600
    },
    {
      "epoch": 9.838357189265814,
      "grad_norm": 1.9943665266036987,
      "learning_rate": 2.648060934961116e-07,
      "loss": 1.9359,
      "num_input_tokens_seen": 508813288,
      "step": 499700
    },
    {
      "epoch": 9.840326042015318,
      "grad_norm": 1.9675488471984863,
      "learning_rate": 2.6462416750812577e-07,
      "loss": 2.0064,
      "num_input_tokens_seen": 508915152,
      "step": 499800
    },
    {
      "epoch": 9.842294894764821,
      "grad_norm": 2.042598009109497,
      "learning_rate": 2.6444228154381497e-07,
      "loss": 1.9775,
      "num_input_tokens_seen": 509017552,
      "step": 499900
    },
    {
      "epoch": 9.844263747514324,
      "grad_norm": 2.2934234142303467,
      "learning_rate": 2.642604356341076e-07,
      "loss": 1.9749,
      "num_input_tokens_seen": 509118664,
      "step": 500000
    },
    {
      "epoch": 9.846232600263827,
      "grad_norm": 2.0251336097717285,
      "learning_rate": 2.6407862980992467e-07,
      "loss": 2.0087,
      "num_input_tokens_seen": 509219320,
      "step": 500100
    },
    {
      "epoch": 9.84820145301333,
      "grad_norm": 1.9895894527435303,
      "learning_rate": 2.638968641021808e-07,
      "loss": 2.0134,
      "num_input_tokens_seen": 509321720,
      "step": 500200
    },
    {
      "epoch": 9.850170305762832,
      "grad_norm": 3.142385482788086,
      "learning_rate": 2.637151385417839e-07,
      "loss": 1.9761,
      "num_input_tokens_seen": 509424120,
      "step": 500300
    },
    {
      "epoch": 9.852139158512335,
      "grad_norm": 1.9831715822219849,
      "learning_rate": 2.635334531596349e-07,
      "loss": 1.9859,
      "num_input_tokens_seen": 509525688,
      "step": 500400
    },
    {
      "epoch": 9.854108011261838,
      "grad_norm": 1.9075015783309937,
      "learning_rate": 2.633518079866276e-07,
      "loss": 1.9598,
      "num_input_tokens_seen": 509628088,
      "step": 500500
    },
    {
      "epoch": 9.85607686401134,
      "grad_norm": 1.8580267429351807,
      "learning_rate": 2.6317020305364936e-07,
      "loss": 2.0112,
      "num_input_tokens_seen": 509728448,
      "step": 500600
    },
    {
      "epoch": 9.858045716760843,
      "grad_norm": 1.93879234790802,
      "learning_rate": 2.6298863839158056e-07,
      "loss": 1.9739,
      "num_input_tokens_seen": 509830848,
      "step": 500700
    },
    {
      "epoch": 9.860014569510346,
      "grad_norm": 2.9762351512908936,
      "learning_rate": 2.6280711403129477e-07,
      "loss": 1.9801,
      "num_input_tokens_seen": 509933248,
      "step": 500800
    },
    {
      "epoch": 9.861983422259849,
      "grad_norm": 1.956985354423523,
      "learning_rate": 2.6262563000365886e-07,
      "loss": 2.0005,
      "num_input_tokens_seen": 510034840,
      "step": 500900
    },
    {
      "epoch": 9.863952275009352,
      "grad_norm": 2.363708257675171,
      "learning_rate": 2.6244418633953226e-07,
      "loss": 1.9879,
      "num_input_tokens_seen": 510136696,
      "step": 501000
    },
    {
      "epoch": 9.865921127758854,
      "grad_norm": 2.0405540466308594,
      "learning_rate": 2.6226278306976815e-07,
      "loss": 1.9808,
      "num_input_tokens_seen": 510238304,
      "step": 501100
    },
    {
      "epoch": 9.867889980508357,
      "grad_norm": 2.091843366622925,
      "learning_rate": 2.620814202252125e-07,
      "loss": 2.0128,
      "num_input_tokens_seen": 510340704,
      "step": 501200
    },
    {
      "epoch": 9.86985883325786,
      "grad_norm": 1.728461742401123,
      "learning_rate": 2.61900097836705e-07,
      "loss": 2.0451,
      "num_input_tokens_seen": 510441816,
      "step": 501300
    },
    {
      "epoch": 9.871827686007364,
      "grad_norm": 1.9842027425765991,
      "learning_rate": 2.617188159350772e-07,
      "loss": 2.0257,
      "num_input_tokens_seen": 510543376,
      "step": 501400
    },
    {
      "epoch": 9.873796538756867,
      "grad_norm": 1.8193837404251099,
      "learning_rate": 2.615375745511551e-07,
      "loss": 1.9786,
      "num_input_tokens_seen": 510645064,
      "step": 501500
    },
    {
      "epoch": 9.87576539150637,
      "grad_norm": 1.8437957763671875,
      "learning_rate": 2.6135637371575743e-07,
      "loss": 2.0041,
      "num_input_tokens_seen": 510746392,
      "step": 501600
    },
    {
      "epoch": 9.877734244255873,
      "grad_norm": 2.14461612701416,
      "learning_rate": 2.611752134596954e-07,
      "loss": 1.9877,
      "num_input_tokens_seen": 510848192,
      "step": 501700
    },
    {
      "epoch": 9.879703097005375,
      "grad_norm": 1.79839026927948,
      "learning_rate": 2.609940938137739e-07,
      "loss": 1.9885,
      "num_input_tokens_seen": 510950592,
      "step": 501800
    },
    {
      "epoch": 9.881671949754878,
      "grad_norm": 2.0120837688446045,
      "learning_rate": 2.60813014808791e-07,
      "loss": 1.9937,
      "num_input_tokens_seen": 511052992,
      "step": 501900
    },
    {
      "epoch": 9.883640802504381,
      "grad_norm": 2.0450146198272705,
      "learning_rate": 2.6063197647553763e-07,
      "loss": 2.0191,
      "num_input_tokens_seen": 511152664,
      "step": 502000
    },
    {
      "epoch": 9.885609655253884,
      "grad_norm": 2.009990692138672,
      "learning_rate": 2.604509788447976e-07,
      "loss": 1.9992,
      "num_input_tokens_seen": 511253880,
      "step": 502100
    },
    {
      "epoch": 9.887578508003386,
      "grad_norm": 2.1507298946380615,
      "learning_rate": 2.6027002194734813e-07,
      "loss": 2.0121,
      "num_input_tokens_seen": 511354808,
      "step": 502200
    },
    {
      "epoch": 9.88954736075289,
      "grad_norm": 1.8799818754196167,
      "learning_rate": 2.6008910581395946e-07,
      "loss": 1.9703,
      "num_input_tokens_seen": 511456360,
      "step": 502300
    },
    {
      "epoch": 9.891516213502392,
      "grad_norm": 2.1749606132507324,
      "learning_rate": 2.599082304753951e-07,
      "loss": 1.9316,
      "num_input_tokens_seen": 511557536,
      "step": 502400
    },
    {
      "epoch": 9.893485066251895,
      "grad_norm": 2.0596413612365723,
      "learning_rate": 2.597273959624109e-07,
      "loss": 2.0277,
      "num_input_tokens_seen": 511658456,
      "step": 502500
    },
    {
      "epoch": 9.895453919001397,
      "grad_norm": 1.997752070426941,
      "learning_rate": 2.595466023057564e-07,
      "loss": 2.0189,
      "num_input_tokens_seen": 511760856,
      "step": 502600
    },
    {
      "epoch": 9.8974227717509,
      "grad_norm": 8.40396499633789,
      "learning_rate": 2.593658495361742e-07,
      "loss": 1.9804,
      "num_input_tokens_seen": 511861480,
      "step": 502700
    },
    {
      "epoch": 9.899391624500403,
      "grad_norm": 1.9299832582473755,
      "learning_rate": 2.591851376843996e-07,
      "loss": 1.973,
      "num_input_tokens_seen": 511963880,
      "step": 502800
    },
    {
      "epoch": 9.901360477249906,
      "grad_norm": 1.8149622678756714,
      "learning_rate": 2.5900446678116145e-07,
      "loss": 2.0083,
      "num_input_tokens_seen": 512065416,
      "step": 502900
    },
    {
      "epoch": 9.903329329999409,
      "grad_norm": 1.9619429111480713,
      "learning_rate": 2.588238368571809e-07,
      "loss": 2.0159,
      "num_input_tokens_seen": 512166376,
      "step": 503000
    },
    {
      "epoch": 9.905298182748911,
      "grad_norm": 2.074518918991089,
      "learning_rate": 2.5864324794317263e-07,
      "loss": 1.9427,
      "num_input_tokens_seen": 512268224,
      "step": 503100
    },
    {
      "epoch": 9.907267035498416,
      "grad_norm": 1.9608818292617798,
      "learning_rate": 2.584627000698444e-07,
      "loss": 2.0108,
      "num_input_tokens_seen": 512370176,
      "step": 503200
    },
    {
      "epoch": 9.909235888247919,
      "grad_norm": 1.947589635848999,
      "learning_rate": 2.58282193267897e-07,
      "loss": 2.0182,
      "num_input_tokens_seen": 512470304,
      "step": 503300
    },
    {
      "epoch": 9.911204740997421,
      "grad_norm": 1.9365743398666382,
      "learning_rate": 2.581017275680237e-07,
      "loss": 2.0415,
      "num_input_tokens_seen": 512572704,
      "step": 503400
    },
    {
      "epoch": 9.913173593746924,
      "grad_norm": 1.9300899505615234,
      "learning_rate": 2.579213030009114e-07,
      "loss": 2.0169,
      "num_input_tokens_seen": 512675104,
      "step": 503500
    },
    {
      "epoch": 9.915142446496427,
      "grad_norm": 2.0466909408569336,
      "learning_rate": 2.577409195972397e-07,
      "loss": 1.995,
      "num_input_tokens_seen": 512776680,
      "step": 503600
    },
    {
      "epoch": 9.91711129924593,
      "grad_norm": 1.806604027748108,
      "learning_rate": 2.5756057738768133e-07,
      "loss": 2.0216,
      "num_input_tokens_seen": 512878600,
      "step": 503700
    },
    {
      "epoch": 9.919080151995432,
      "grad_norm": 1.7431915998458862,
      "learning_rate": 2.573802764029021e-07,
      "loss": 2.0301,
      "num_input_tokens_seen": 512979456,
      "step": 503800
    },
    {
      "epoch": 9.921049004744935,
      "grad_norm": 2.1425445079803467,
      "learning_rate": 2.572000166735602e-07,
      "loss": 1.9495,
      "num_input_tokens_seen": 513081248,
      "step": 503900
    },
    {
      "epoch": 9.923017857494438,
      "grad_norm": 1.9135055541992188,
      "learning_rate": 2.5701979823030796e-07,
      "loss": 1.9724,
      "num_input_tokens_seen": 513183648,
      "step": 504000
    },
    {
      "epoch": 9.92498671024394,
      "grad_norm": 1.9988261461257935,
      "learning_rate": 2.5683962110378935e-07,
      "loss": 1.9874,
      "num_input_tokens_seen": 513286048,
      "step": 504100
    },
    {
      "epoch": 9.926955562993443,
      "grad_norm": 1.8992489576339722,
      "learning_rate": 2.5665948532464254e-07,
      "loss": 1.9934,
      "num_input_tokens_seen": 513387744,
      "step": 504200
    },
    {
      "epoch": 9.928924415742946,
      "grad_norm": 1.9562246799468994,
      "learning_rate": 2.5647939092349746e-07,
      "loss": 1.9569,
      "num_input_tokens_seen": 513490144,
      "step": 504300
    },
    {
      "epoch": 9.930893268492449,
      "grad_norm": 1.9197726249694824,
      "learning_rate": 2.562993379309784e-07,
      "loss": 1.9535,
      "num_input_tokens_seen": 513592184,
      "step": 504400
    },
    {
      "epoch": 9.932862121241952,
      "grad_norm": 2.091519594192505,
      "learning_rate": 2.561193263777012e-07,
      "loss": 1.9927,
      "num_input_tokens_seen": 513693624,
      "step": 504500
    },
    {
      "epoch": 9.934830973991454,
      "grad_norm": 1.6819130182266235,
      "learning_rate": 2.5593935629427555e-07,
      "loss": 2.032,
      "num_input_tokens_seen": 513794608,
      "step": 504600
    },
    {
      "epoch": 9.936799826740959,
      "grad_norm": 2.0900344848632812,
      "learning_rate": 2.557594277113039e-07,
      "loss": 1.9867,
      "num_input_tokens_seen": 513896696,
      "step": 504700
    },
    {
      "epoch": 9.938768679490462,
      "grad_norm": 1.9267830848693848,
      "learning_rate": 2.555795406593816e-07,
      "loss": 1.9819,
      "num_input_tokens_seen": 513998472,
      "step": 504800
    },
    {
      "epoch": 9.940737532239964,
      "grad_norm": 1.9433521032333374,
      "learning_rate": 2.55399695169097e-07,
      "loss": 2.0163,
      "num_input_tokens_seen": 514100400,
      "step": 504900
    },
    {
      "epoch": 9.942706384989467,
      "grad_norm": 1.8832790851593018,
      "learning_rate": 2.55219891271031e-07,
      "loss": 1.9759,
      "num_input_tokens_seen": 514202208,
      "step": 505000
    },
    {
      "epoch": 9.94467523773897,
      "grad_norm": 1.907757043838501,
      "learning_rate": 2.550401289957579e-07,
      "loss": 1.9478,
      "num_input_tokens_seen": 514304608,
      "step": 505100
    },
    {
      "epoch": 9.946644090488473,
      "grad_norm": 1.7344521284103394,
      "learning_rate": 2.548604083738448e-07,
      "loss": 1.9466,
      "num_input_tokens_seen": 514404648,
      "step": 505200
    },
    {
      "epoch": 9.948612943237976,
      "grad_norm": 1.7976120710372925,
      "learning_rate": 2.546807294358518e-07,
      "loss": 1.9911,
      "num_input_tokens_seen": 514506368,
      "step": 505300
    },
    {
      "epoch": 9.950581795987478,
      "grad_norm": 2.101142406463623,
      "learning_rate": 2.5450109221233157e-07,
      "loss": 1.9833,
      "num_input_tokens_seen": 514608768,
      "step": 505400
    },
    {
      "epoch": 9.952550648736981,
      "grad_norm": 2.016080856323242,
      "learning_rate": 2.543214967338299e-07,
      "loss": 2.0226,
      "num_input_tokens_seen": 514710648,
      "step": 505500
    },
    {
      "epoch": 9.954519501486484,
      "grad_norm": 3.0059051513671875,
      "learning_rate": 2.5414194303088555e-07,
      "loss": 2.0253,
      "num_input_tokens_seen": 514811992,
      "step": 505600
    },
    {
      "epoch": 9.956488354235987,
      "grad_norm": 2.1360247135162354,
      "learning_rate": 2.5396243113403016e-07,
      "loss": 2.0207,
      "num_input_tokens_seen": 514912672,
      "step": 505700
    },
    {
      "epoch": 9.95845720698549,
      "grad_norm": 1.8347407579421997,
      "learning_rate": 2.537829610737883e-07,
      "loss": 1.9898,
      "num_input_tokens_seen": 515013056,
      "step": 505800
    },
    {
      "epoch": 9.960426059734992,
      "grad_norm": 1.8011956214904785,
      "learning_rate": 2.5360353288067714e-07,
      "loss": 1.9628,
      "num_input_tokens_seen": 515114144,
      "step": 505900
    },
    {
      "epoch": 9.962394912484495,
      "grad_norm": 1.6846932172775269,
      "learning_rate": 2.534241465852069e-07,
      "loss": 2.02,
      "num_input_tokens_seen": 515216456,
      "step": 506000
    },
    {
      "epoch": 9.964363765233998,
      "grad_norm": 1.8579161167144775,
      "learning_rate": 2.532448022178809e-07,
      "loss": 2.0424,
      "num_input_tokens_seen": 515318376,
      "step": 506100
    },
    {
      "epoch": 9.9663326179835,
      "grad_norm": 1.856040358543396,
      "learning_rate": 2.5306549980919525e-07,
      "loss": 1.9999,
      "num_input_tokens_seen": 515420776,
      "step": 506200
    },
    {
      "epoch": 9.968301470733003,
      "grad_norm": 1.9598783254623413,
      "learning_rate": 2.5288623938963833e-07,
      "loss": 1.9778,
      "num_input_tokens_seen": 515523040,
      "step": 506300
    },
    {
      "epoch": 9.970270323482506,
      "grad_norm": 2.1430556774139404,
      "learning_rate": 2.5270702098969223e-07,
      "loss": 1.9611,
      "num_input_tokens_seen": 515624704,
      "step": 506400
    },
    {
      "epoch": 9.97223917623201,
      "grad_norm": 1.95353364944458,
      "learning_rate": 2.525278446398314e-07,
      "loss": 2.0287,
      "num_input_tokens_seen": 515727104,
      "step": 506500
    },
    {
      "epoch": 9.974208028981513,
      "grad_norm": 1.8301260471343994,
      "learning_rate": 2.523487103705233e-07,
      "loss": 1.9825,
      "num_input_tokens_seen": 515827264,
      "step": 506600
    },
    {
      "epoch": 9.976176881731016,
      "grad_norm": 1.9590789079666138,
      "learning_rate": 2.521696182122285e-07,
      "loss": 1.9909,
      "num_input_tokens_seen": 515929664,
      "step": 506700
    },
    {
      "epoch": 9.978145734480519,
      "grad_norm": 2.2142257690429688,
      "learning_rate": 2.5199056819539935e-07,
      "loss": 1.9831,
      "num_input_tokens_seen": 516032064,
      "step": 506800
    },
    {
      "epoch": 9.980114587230021,
      "grad_norm": 1.8652536869049072,
      "learning_rate": 2.5181156035048267e-07,
      "loss": 1.9951,
      "num_input_tokens_seen": 516133664,
      "step": 506900
    },
    {
      "epoch": 9.982083439979524,
      "grad_norm": 1.9279258251190186,
      "learning_rate": 2.5163259470791665e-07,
      "loss": 2.0282,
      "num_input_tokens_seen": 516233848,
      "step": 507000
    },
    {
      "epoch": 9.984052292729027,
      "grad_norm": 1.8755311965942383,
      "learning_rate": 2.514536712981332e-07,
      "loss": 1.9461,
      "num_input_tokens_seen": 516336248,
      "step": 507100
    },
    {
      "epoch": 9.98602114547853,
      "grad_norm": 1.9926308393478394,
      "learning_rate": 2.5127479015155633e-07,
      "loss": 1.9632,
      "num_input_tokens_seen": 516438648,
      "step": 507200
    },
    {
      "epoch": 9.987989998228032,
      "grad_norm": 1.844283938407898,
      "learning_rate": 2.5109595129860394e-07,
      "loss": 2.0052,
      "num_input_tokens_seen": 516540552,
      "step": 507300
    },
    {
      "epoch": 9.989958850977535,
      "grad_norm": 2.0283586978912354,
      "learning_rate": 2.5091715476968547e-07,
      "loss": 1.9831,
      "num_input_tokens_seen": 516641440,
      "step": 507400
    },
    {
      "epoch": 9.991927703727038,
      "grad_norm": 2.089303970336914,
      "learning_rate": 2.5073840059520407e-07,
      "loss": 1.9368,
      "num_input_tokens_seen": 516743064,
      "step": 507500
    },
    {
      "epoch": 9.99389655647654,
      "grad_norm": 1.672130823135376,
      "learning_rate": 2.50559688805555e-07,
      "loss": 1.9597,
      "num_input_tokens_seen": 516845464,
      "step": 507600
    },
    {
      "epoch": 9.995865409226043,
      "grad_norm": 3.2307937145233154,
      "learning_rate": 2.503810194311271e-07,
      "loss": 1.9873,
      "num_input_tokens_seen": 516946080,
      "step": 507700
    },
    {
      "epoch": 9.997834261975546,
      "grad_norm": 1.9267553091049194,
      "learning_rate": 2.502023925023017e-07,
      "loss": 1.9768,
      "num_input_tokens_seen": 517048480,
      "step": 507800
    },
    {
      "epoch": 9.999803114725049,
      "grad_norm": 1.6313378810882568,
      "learning_rate": 2.5002380804945233e-07,
      "loss": 1.9664,
      "num_input_tokens_seen": 517150880,
      "step": 507900
    },
    {
      "epoch": 10.001771967474552,
      "grad_norm": 1.7744789123535156,
      "learning_rate": 2.4984526610294596e-07,
      "loss": 2.0134,
      "num_input_tokens_seen": 517252488,
      "step": 508000
    },
    {
      "epoch": 10.003740820224056,
      "grad_norm": 1.9814518690109253,
      "learning_rate": 2.496667666931423e-07,
      "loss": 1.9949,
      "num_input_tokens_seen": 517352984,
      "step": 508100
    },
    {
      "epoch": 10.005709672973559,
      "grad_norm": 1.8977552652359009,
      "learning_rate": 2.494883098503937e-07,
      "loss": 1.9413,
      "num_input_tokens_seen": 517454640,
      "step": 508200
    },
    {
      "epoch": 10.007678525723062,
      "grad_norm": 2.129425287246704,
      "learning_rate": 2.493098956050449e-07,
      "loss": 1.9683,
      "num_input_tokens_seen": 517555064,
      "step": 508300
    },
    {
      "epoch": 10.009647378472565,
      "grad_norm": 1.9466496706008911,
      "learning_rate": 2.491315239874339e-07,
      "loss": 2.0346,
      "num_input_tokens_seen": 517656984,
      "step": 508400
    },
    {
      "epoch": 10.011616231222067,
      "grad_norm": 1.9415394067764282,
      "learning_rate": 2.4895319502789143e-07,
      "loss": 2.0483,
      "num_input_tokens_seen": 517759384,
      "step": 508500
    },
    {
      "epoch": 10.01358508397157,
      "grad_norm": 2.147742986679077,
      "learning_rate": 2.487749087567406e-07,
      "loss": 1.9784,
      "num_input_tokens_seen": 517861280,
      "step": 508600
    },
    {
      "epoch": 10.015553936721073,
      "grad_norm": 1.732603669166565,
      "learning_rate": 2.4859666520429785e-07,
      "loss": 1.9719,
      "num_input_tokens_seen": 517963208,
      "step": 508700
    },
    {
      "epoch": 10.017522789470576,
      "grad_norm": 2.3705923557281494,
      "learning_rate": 2.4841846440087156e-07,
      "loss": 2.0126,
      "num_input_tokens_seen": 518064736,
      "step": 508800
    },
    {
      "epoch": 10.019491642220078,
      "grad_norm": 1.7529703378677368,
      "learning_rate": 2.482403063767635e-07,
      "loss": 1.9861,
      "num_input_tokens_seen": 518167136,
      "step": 508900
    },
    {
      "epoch": 10.021460494969581,
      "grad_norm": 2.0923120975494385,
      "learning_rate": 2.4806219116226787e-07,
      "loss": 1.9833,
      "num_input_tokens_seen": 518268768,
      "step": 509000
    },
    {
      "epoch": 10.023429347719084,
      "grad_norm": 1.7131842374801636,
      "learning_rate": 2.4788411878767185e-07,
      "loss": 1.9851,
      "num_input_tokens_seen": 518370488,
      "step": 509100
    },
    {
      "epoch": 10.025398200468587,
      "grad_norm": 1.8414028882980347,
      "learning_rate": 2.477060892832548e-07,
      "loss": 2.0302,
      "num_input_tokens_seen": 518472800,
      "step": 509200
    },
    {
      "epoch": 10.02736705321809,
      "grad_norm": 1.709205150604248,
      "learning_rate": 2.475281026792893e-07,
      "loss": 2.0248,
      "num_input_tokens_seen": 518574760,
      "step": 509300
    },
    {
      "epoch": 10.029335905967592,
      "grad_norm": 2.290510654449463,
      "learning_rate": 2.4735015900604056e-07,
      "loss": 1.9267,
      "num_input_tokens_seen": 518676296,
      "step": 509400
    },
    {
      "epoch": 10.031304758717095,
      "grad_norm": 2.1205391883850098,
      "learning_rate": 2.471722582937665e-07,
      "loss": 1.9654,
      "num_input_tokens_seen": 518778696,
      "step": 509500
    },
    {
      "epoch": 10.033273611466598,
      "grad_norm": 2.234224557876587,
      "learning_rate": 2.469944005727171e-07,
      "loss": 1.9638,
      "num_input_tokens_seen": 518881096,
      "step": 509600
    },
    {
      "epoch": 10.0352424642161,
      "grad_norm": 2.0411813259124756,
      "learning_rate": 2.468165858731361e-07,
      "loss": 2.0003,
      "num_input_tokens_seen": 518982808,
      "step": 509700
    },
    {
      "epoch": 10.037211316965605,
      "grad_norm": 2.2885854244232178,
      "learning_rate": 2.466388142252594e-07,
      "loss": 1.9397,
      "num_input_tokens_seen": 519085208,
      "step": 509800
    },
    {
      "epoch": 10.039180169715108,
      "grad_norm": 1.8394403457641602,
      "learning_rate": 2.464610856593152e-07,
      "loss": 1.9773,
      "num_input_tokens_seen": 519187608,
      "step": 509900
    },
    {
      "epoch": 10.04114902246461,
      "grad_norm": 2.230210065841675,
      "learning_rate": 2.4628340020552503e-07,
      "loss": 2.0049,
      "num_input_tokens_seen": 519288824,
      "step": 510000
    },
    {
      "epoch": 10.043117875214113,
      "grad_norm": 2.0309078693389893,
      "learning_rate": 2.4610575789410266e-07,
      "loss": 1.9865,
      "num_input_tokens_seen": 519391224,
      "step": 510100
    },
    {
      "epoch": 10.045086727963616,
      "grad_norm": 2.0254244804382324,
      "learning_rate": 2.4592815875525483e-07,
      "loss": 2.0033,
      "num_input_tokens_seen": 519493624,
      "step": 510200
    },
    {
      "epoch": 10.047055580713119,
      "grad_norm": 2.09907865524292,
      "learning_rate": 2.457506028191806e-07,
      "loss": 1.9956,
      "num_input_tokens_seen": 519594240,
      "step": 510300
    },
    {
      "epoch": 10.049024433462622,
      "grad_norm": 1.7832585573196411,
      "learning_rate": 2.455730901160718e-07,
      "loss": 1.9841,
      "num_input_tokens_seen": 519695984,
      "step": 510400
    },
    {
      "epoch": 10.050993286212124,
      "grad_norm": 2.0728001594543457,
      "learning_rate": 2.453956206761132e-07,
      "loss": 1.9647,
      "num_input_tokens_seen": 519798384,
      "step": 510500
    },
    {
      "epoch": 10.052962138961627,
      "grad_norm": 2.0551185607910156,
      "learning_rate": 2.45218194529482e-07,
      "loss": 2.0005,
      "num_input_tokens_seen": 519900048,
      "step": 510600
    },
    {
      "epoch": 10.05493099171113,
      "grad_norm": 13.11880874633789,
      "learning_rate": 2.450408117063477e-07,
      "loss": 1.9848,
      "num_input_tokens_seen": 520001808,
      "step": 510700
    },
    {
      "epoch": 10.056899844460633,
      "grad_norm": 1.873674988746643,
      "learning_rate": 2.44863472236873e-07,
      "loss": 1.9745,
      "num_input_tokens_seen": 520102888,
      "step": 510800
    },
    {
      "epoch": 10.058868697210135,
      "grad_norm": 1.6636666059494019,
      "learning_rate": 2.4468617615121285e-07,
      "loss": 1.9473,
      "num_input_tokens_seen": 520205288,
      "step": 510900
    },
    {
      "epoch": 10.060837549959638,
      "grad_norm": 1.7565938234329224,
      "learning_rate": 2.44508923479515e-07,
      "loss": 1.9238,
      "num_input_tokens_seen": 520307688,
      "step": 511000
    },
    {
      "epoch": 10.06280640270914,
      "grad_norm": 1.8458462953567505,
      "learning_rate": 2.4433171425192013e-07,
      "loss": 2.0178,
      "num_input_tokens_seen": 520408568,
      "step": 511100
    },
    {
      "epoch": 10.064775255458644,
      "grad_norm": 3.9458611011505127,
      "learning_rate": 2.4415454849856053e-07,
      "loss": 2.0106,
      "num_input_tokens_seen": 520510488,
      "step": 511200
    },
    {
      "epoch": 10.066744108208146,
      "grad_norm": 1.7859681844711304,
      "learning_rate": 2.4397742624956214e-07,
      "loss": 2.0193,
      "num_input_tokens_seen": 520612128,
      "step": 511300
    },
    {
      "epoch": 10.06871296095765,
      "grad_norm": 1.888371467590332,
      "learning_rate": 2.4380034753504296e-07,
      "loss": 1.9404,
      "num_input_tokens_seen": 520712912,
      "step": 511400
    },
    {
      "epoch": 10.070681813707154,
      "grad_norm": 1.90412175655365,
      "learning_rate": 2.43623312385114e-07,
      "loss": 1.9989,
      "num_input_tokens_seen": 520813928,
      "step": 511500
    },
    {
      "epoch": 10.072650666456656,
      "grad_norm": 2.122999668121338,
      "learning_rate": 2.434463208298782e-07,
      "loss": 1.9649,
      "num_input_tokens_seen": 520914072,
      "step": 511600
    },
    {
      "epoch": 10.07461951920616,
      "grad_norm": 1.72954261302948,
      "learning_rate": 2.432693728994315e-07,
      "loss": 1.9807,
      "num_input_tokens_seen": 521015904,
      "step": 511700
    },
    {
      "epoch": 10.076588371955662,
      "grad_norm": 1.7599509954452515,
      "learning_rate": 2.4309246862386295e-07,
      "loss": 1.9625,
      "num_input_tokens_seen": 521118304,
      "step": 511800
    },
    {
      "epoch": 10.078557224705165,
      "grad_norm": 2.1503989696502686,
      "learning_rate": 2.4291560803325304e-07,
      "loss": 1.9813,
      "num_input_tokens_seen": 521218856,
      "step": 511900
    },
    {
      "epoch": 10.080526077454667,
      "grad_norm": 2.321514129638672,
      "learning_rate": 2.427387911576759e-07,
      "loss": 2.0091,
      "num_input_tokens_seen": 521321256,
      "step": 512000
    },
    {
      "epoch": 10.08249493020417,
      "grad_norm": 1.913076639175415,
      "learning_rate": 2.42562018027197e-07,
      "loss": 2.0034,
      "num_input_tokens_seen": 521422960,
      "step": 512100
    },
    {
      "epoch": 10.084463782953673,
      "grad_norm": 2.048002243041992,
      "learning_rate": 2.42385288671876e-07,
      "loss": 2.0196,
      "num_input_tokens_seen": 521524184,
      "step": 512200
    },
    {
      "epoch": 10.086432635703176,
      "grad_norm": 2.110062837600708,
      "learning_rate": 2.4220860312176364e-07,
      "loss": 1.965,
      "num_input_tokens_seen": 521625968,
      "step": 512300
    },
    {
      "epoch": 10.088401488452678,
      "grad_norm": 1.9550901651382446,
      "learning_rate": 2.420319614069043e-07,
      "loss": 2.0187,
      "num_input_tokens_seen": 521727608,
      "step": 512400
    },
    {
      "epoch": 10.090370341202181,
      "grad_norm": 1.8020589351654053,
      "learning_rate": 2.4185536355733363e-07,
      "loss": 1.9888,
      "num_input_tokens_seen": 521829152,
      "step": 512500
    },
    {
      "epoch": 10.092339193951684,
      "grad_norm": 1.8266799449920654,
      "learning_rate": 2.4167880960308153e-07,
      "loss": 1.9639,
      "num_input_tokens_seen": 521931552,
      "step": 512600
    },
    {
      "epoch": 10.094308046701187,
      "grad_norm": 1.8024038076400757,
      "learning_rate": 2.4150229957416887e-07,
      "loss": 1.9477,
      "num_input_tokens_seen": 522033448,
      "step": 512700
    },
    {
      "epoch": 10.09627689945069,
      "grad_norm": 1.9305915832519531,
      "learning_rate": 2.4132583350060984e-07,
      "loss": 1.9932,
      "num_input_tokens_seen": 522134376,
      "step": 512800
    },
    {
      "epoch": 10.098245752200192,
      "grad_norm": 1.8215798139572144,
      "learning_rate": 2.411494114124112e-07,
      "loss": 1.9581,
      "num_input_tokens_seen": 522236776,
      "step": 512900
    },
    {
      "epoch": 10.100214604949695,
      "grad_norm": 1.7347626686096191,
      "learning_rate": 2.409730333395718e-07,
      "loss": 1.9736,
      "num_input_tokens_seen": 522338552,
      "step": 513000
    },
    {
      "epoch": 10.102183457699198,
      "grad_norm": 1.7242077589035034,
      "learning_rate": 2.407966993120835e-07,
      "loss": 1.9951,
      "num_input_tokens_seen": 522440952,
      "step": 513100
    },
    {
      "epoch": 10.104152310448702,
      "grad_norm": 1.8284276723861694,
      "learning_rate": 2.406204093599302e-07,
      "loss": 1.9675,
      "num_input_tokens_seen": 522542616,
      "step": 513200
    },
    {
      "epoch": 10.106121163198205,
      "grad_norm": 2.1157314777374268,
      "learning_rate": 2.404441635130885e-07,
      "loss": 1.9681,
      "num_input_tokens_seen": 522645016,
      "step": 513300
    },
    {
      "epoch": 10.108090015947708,
      "grad_norm": 1.990410327911377,
      "learning_rate": 2.4026796180152757e-07,
      "loss": 1.9743,
      "num_input_tokens_seen": 522746672,
      "step": 513400
    },
    {
      "epoch": 10.11005886869721,
      "grad_norm": 2.300504446029663,
      "learning_rate": 2.400918042552093e-07,
      "loss": 1.9734,
      "num_input_tokens_seen": 522849072,
      "step": 513500
    },
    {
      "epoch": 10.112027721446713,
      "grad_norm": 2.0461835861206055,
      "learning_rate": 2.3991569090408737e-07,
      "loss": 2.0136,
      "num_input_tokens_seen": 522949760,
      "step": 513600
    },
    {
      "epoch": 10.113996574196216,
      "grad_norm": 1.8705706596374512,
      "learning_rate": 2.3973962177810845e-07,
      "loss": 1.9863,
      "num_input_tokens_seen": 523051608,
      "step": 513700
    },
    {
      "epoch": 10.115965426945719,
      "grad_norm": 1.970261812210083,
      "learning_rate": 2.3956359690721186e-07,
      "loss": 1.9778,
      "num_input_tokens_seen": 523153568,
      "step": 513800
    },
    {
      "epoch": 10.117934279695222,
      "grad_norm": 2.207791805267334,
      "learning_rate": 2.393876163213289e-07,
      "loss": 1.9951,
      "num_input_tokens_seen": 523255056,
      "step": 513900
    },
    {
      "epoch": 10.119903132444724,
      "grad_norm": 2.0701262950897217,
      "learning_rate": 2.3921168005038384e-07,
      "loss": 1.9755,
      "num_input_tokens_seen": 523357456,
      "step": 514000
    },
    {
      "epoch": 10.121871985194227,
      "grad_norm": 1.9004077911376953,
      "learning_rate": 2.390357881242929e-07,
      "loss": 1.9992,
      "num_input_tokens_seen": 523459856,
      "step": 514100
    },
    {
      "epoch": 10.12384083794373,
      "grad_norm": 2.1974658966064453,
      "learning_rate": 2.3885994057296506e-07,
      "loss": 1.9695,
      "num_input_tokens_seen": 523561464,
      "step": 514200
    },
    {
      "epoch": 10.125809690693233,
      "grad_norm": 2.118424892425537,
      "learning_rate": 2.386841374263017e-07,
      "loss": 1.9772,
      "num_input_tokens_seen": 523663864,
      "step": 514300
    },
    {
      "epoch": 10.127778543442735,
      "grad_norm": 1.9299614429473877,
      "learning_rate": 2.3850837871419703e-07,
      "loss": 1.9902,
      "num_input_tokens_seen": 523765712,
      "step": 514400
    },
    {
      "epoch": 10.129747396192238,
      "grad_norm": 1.9739350080490112,
      "learning_rate": 2.3833266446653682e-07,
      "loss": 2.0134,
      "num_input_tokens_seen": 523867632,
      "step": 514500
    },
    {
      "epoch": 10.131716248941741,
      "grad_norm": 1.9896650314331055,
      "learning_rate": 2.381569947132e-07,
      "loss": 1.9597,
      "num_input_tokens_seen": 523969264,
      "step": 514600
    },
    {
      "epoch": 10.133685101691244,
      "grad_norm": 1.8265329599380493,
      "learning_rate": 2.379813694840578e-07,
      "loss": 1.9922,
      "num_input_tokens_seen": 524069256,
      "step": 514700
    },
    {
      "epoch": 10.135653954440748,
      "grad_norm": 2.0936081409454346,
      "learning_rate": 2.3780578880897367e-07,
      "loss": 1.9293,
      "num_input_tokens_seen": 524170704,
      "step": 514800
    },
    {
      "epoch": 10.137622807190251,
      "grad_norm": 1.9497532844543457,
      "learning_rate": 2.3763025271780392e-07,
      "loss": 1.9511,
      "num_input_tokens_seen": 524272904,
      "step": 514900
    },
    {
      "epoch": 10.139591659939754,
      "grad_norm": 2.0265049934387207,
      "learning_rate": 2.3745476124039638e-07,
      "loss": 1.959,
      "num_input_tokens_seen": 524375304,
      "step": 515000
    },
    {
      "epoch": 10.141560512689257,
      "grad_norm": 1.735066294670105,
      "learning_rate": 2.3727931440659267e-07,
      "loss": 1.9715,
      "num_input_tokens_seen": 524477112,
      "step": 515100
    },
    {
      "epoch": 10.14352936543876,
      "grad_norm": 2.035606622695923,
      "learning_rate": 2.371039122462255e-07,
      "loss": 1.9913,
      "num_input_tokens_seen": 524579296,
      "step": 515200
    },
    {
      "epoch": 10.145498218188262,
      "grad_norm": 2.1595828533172607,
      "learning_rate": 2.3692855478912082e-07,
      "loss": 1.9816,
      "num_input_tokens_seen": 524681208,
      "step": 515300
    },
    {
      "epoch": 10.147467070937765,
      "grad_norm": 2.4264023303985596,
      "learning_rate": 2.3675324206509624e-07,
      "loss": 1.9869,
      "num_input_tokens_seen": 524781424,
      "step": 515400
    },
    {
      "epoch": 10.149435923687268,
      "grad_norm": 1.8765474557876587,
      "learning_rate": 2.3657797410396279e-07,
      "loss": 2.0115,
      "num_input_tokens_seen": 524883008,
      "step": 515500
    },
    {
      "epoch": 10.15140477643677,
      "grad_norm": 2.0481069087982178,
      "learning_rate": 2.3640275093552292e-07,
      "loss": 1.9893,
      "num_input_tokens_seen": 524985040,
      "step": 515600
    },
    {
      "epoch": 10.153373629186273,
      "grad_norm": 1.9797998666763306,
      "learning_rate": 2.3622757258957215e-07,
      "loss": 1.9814,
      "num_input_tokens_seen": 525086168,
      "step": 515700
    },
    {
      "epoch": 10.155342481935776,
      "grad_norm": 2.477825164794922,
      "learning_rate": 2.3605243909589756e-07,
      "loss": 2.0084,
      "num_input_tokens_seen": 525188568,
      "step": 515800
    },
    {
      "epoch": 10.157311334685279,
      "grad_norm": 1.7219065427780151,
      "learning_rate": 2.3587735048427964e-07,
      "loss": 2.0243,
      "num_input_tokens_seen": 525289304,
      "step": 515900
    },
    {
      "epoch": 10.159280187434781,
      "grad_norm": 1.96244215965271,
      "learning_rate": 2.357023067844907e-07,
      "loss": 2.0297,
      "num_input_tokens_seen": 525389840,
      "step": 516000
    },
    {
      "epoch": 10.161249040184284,
      "grad_norm": 1.9272196292877197,
      "learning_rate": 2.3552730802629518e-07,
      "loss": 1.985,
      "num_input_tokens_seen": 525492240,
      "step": 516100
    },
    {
      "epoch": 10.163217892933787,
      "grad_norm": 1.8310563564300537,
      "learning_rate": 2.3535235423945022e-07,
      "loss": 2.0207,
      "num_input_tokens_seen": 525594056,
      "step": 516200
    },
    {
      "epoch": 10.16518674568329,
      "grad_norm": 2.5422022342681885,
      "learning_rate": 2.3517744545370527e-07,
      "loss": 1.9616,
      "num_input_tokens_seen": 525696456,
      "step": 516300
    },
    {
      "epoch": 10.167155598432792,
      "grad_norm": 1.7177711725234985,
      "learning_rate": 2.3500258169880233e-07,
      "loss": 1.9646,
      "num_input_tokens_seen": 525798416,
      "step": 516400
    },
    {
      "epoch": 10.169124451182297,
      "grad_norm": 1.9400345087051392,
      "learning_rate": 2.348277630044751e-07,
      "loss": 1.9957,
      "num_input_tokens_seen": 525900816,
      "step": 516500
    },
    {
      "epoch": 10.1710933039318,
      "grad_norm": 2.4387450218200684,
      "learning_rate": 2.3465298940045014e-07,
      "loss": 1.9885,
      "num_input_tokens_seen": 526002544,
      "step": 516600
    },
    {
      "epoch": 10.173062156681302,
      "grad_norm": 5.567256450653076,
      "learning_rate": 2.3447826091644635e-07,
      "loss": 2.0208,
      "num_input_tokens_seen": 526103712,
      "step": 516700
    },
    {
      "epoch": 10.175031009430805,
      "grad_norm": 2.090766668319702,
      "learning_rate": 2.3430357758217478e-07,
      "loss": 2.0158,
      "num_input_tokens_seen": 526206112,
      "step": 516800
    },
    {
      "epoch": 10.176999862180308,
      "grad_norm": 1.9667357206344604,
      "learning_rate": 2.34128939427339e-07,
      "loss": 1.9853,
      "num_input_tokens_seen": 526307128,
      "step": 516900
    },
    {
      "epoch": 10.17896871492981,
      "grad_norm": 2.358248233795166,
      "learning_rate": 2.339543464816344e-07,
      "loss": 1.994,
      "num_input_tokens_seen": 526407776,
      "step": 517000
    },
    {
      "epoch": 10.180937567679313,
      "grad_norm": 2.1375815868377686,
      "learning_rate": 2.3377979877474935e-07,
      "loss": 1.9619,
      "num_input_tokens_seen": 526510176,
      "step": 517100
    },
    {
      "epoch": 10.182906420428816,
      "grad_norm": 4.232319355010986,
      "learning_rate": 2.3360529633636406e-07,
      "loss": 1.9753,
      "num_input_tokens_seen": 526612576,
      "step": 517200
    },
    {
      "epoch": 10.184875273178319,
      "grad_norm": 1.7390012741088867,
      "learning_rate": 2.334308391961514e-07,
      "loss": 2.0142,
      "num_input_tokens_seen": 526714120,
      "step": 517300
    },
    {
      "epoch": 10.186844125927822,
      "grad_norm": 2.1147119998931885,
      "learning_rate": 2.3325642738377604e-07,
      "loss": 1.9928,
      "num_input_tokens_seen": 526815200,
      "step": 517400
    },
    {
      "epoch": 10.188812978677324,
      "grad_norm": 1.6188441514968872,
      "learning_rate": 2.3308206092889537e-07,
      "loss": 1.9741,
      "num_input_tokens_seen": 526917600,
      "step": 517500
    },
    {
      "epoch": 10.190781831426827,
      "grad_norm": 2.1635310649871826,
      "learning_rate": 2.3290773986115896e-07,
      "loss": 1.9922,
      "num_input_tokens_seen": 527019424,
      "step": 517600
    },
    {
      "epoch": 10.19275068417633,
      "grad_norm": 2.92665433883667,
      "learning_rate": 2.3273346421020884e-07,
      "loss": 1.9675,
      "num_input_tokens_seen": 527119576,
      "step": 517700
    },
    {
      "epoch": 10.194719536925833,
      "grad_norm": 2.241255521774292,
      "learning_rate": 2.325592340056785e-07,
      "loss": 1.9674,
      "num_input_tokens_seen": 527220704,
      "step": 517800
    },
    {
      "epoch": 10.196688389675336,
      "grad_norm": 1.9111897945404053,
      "learning_rate": 2.3238504927719489e-07,
      "loss": 1.964,
      "num_input_tokens_seen": 527323104,
      "step": 517900
    },
    {
      "epoch": 10.198657242424838,
      "grad_norm": 1.8900012969970703,
      "learning_rate": 2.3221091005437671e-07,
      "loss": 1.9945,
      "num_input_tokens_seen": 527425504,
      "step": 518000
    },
    {
      "epoch": 10.200626095174341,
      "grad_norm": 2.0185890197753906,
      "learning_rate": 2.3203681636683443e-07,
      "loss": 1.9797,
      "num_input_tokens_seen": 527527904,
      "step": 518100
    },
    {
      "epoch": 10.202594947923846,
      "grad_norm": 1.8185033798217773,
      "learning_rate": 2.318627682441714e-07,
      "loss": 2.0198,
      "num_input_tokens_seen": 527629408,
      "step": 518200
    },
    {
      "epoch": 10.204563800673348,
      "grad_norm": 2.0610711574554443,
      "learning_rate": 2.3168876571598305e-07,
      "loss": 1.9578,
      "num_input_tokens_seen": 527731808,
      "step": 518300
    },
    {
      "epoch": 10.206532653422851,
      "grad_norm": 1.778607964515686,
      "learning_rate": 2.3151480881185724e-07,
      "loss": 1.9384,
      "num_input_tokens_seen": 527834208,
      "step": 518400
    },
    {
      "epoch": 10.208501506172354,
      "grad_norm": 2.1930012702941895,
      "learning_rate": 2.3134089756137355e-07,
      "loss": 1.9443,
      "num_input_tokens_seen": 527936608,
      "step": 518500
    },
    {
      "epoch": 10.210470358921857,
      "grad_norm": 2.0562498569488525,
      "learning_rate": 2.311670319941042e-07,
      "loss": 1.9886,
      "num_input_tokens_seen": 528038208,
      "step": 518600
    },
    {
      "epoch": 10.21243921167136,
      "grad_norm": 1.7914087772369385,
      "learning_rate": 2.3099321213961365e-07,
      "loss": 1.9796,
      "num_input_tokens_seen": 528138552,
      "step": 518700
    },
    {
      "epoch": 10.214408064420862,
      "grad_norm": 1.8735454082489014,
      "learning_rate": 2.3081943802745845e-07,
      "loss": 1.9795,
      "num_input_tokens_seen": 528240952,
      "step": 518800
    },
    {
      "epoch": 10.216376917170365,
      "grad_norm": 1.8071253299713135,
      "learning_rate": 2.306457096871876e-07,
      "loss": 1.9542,
      "num_input_tokens_seen": 528341680,
      "step": 518900
    },
    {
      "epoch": 10.218345769919868,
      "grad_norm": 1.6748353242874146,
      "learning_rate": 2.3047202714834185e-07,
      "loss": 1.9563,
      "num_input_tokens_seen": 528444080,
      "step": 519000
    },
    {
      "epoch": 10.22031462266937,
      "grad_norm": 2.3489108085632324,
      "learning_rate": 2.302983904404545e-07,
      "loss": 1.9612,
      "num_input_tokens_seen": 528545760,
      "step": 519100
    },
    {
      "epoch": 10.222283475418873,
      "grad_norm": 1.99221932888031,
      "learning_rate": 2.3012479959305114e-07,
      "loss": 1.9761,
      "num_input_tokens_seen": 528648160,
      "step": 519200
    },
    {
      "epoch": 10.224252328168376,
      "grad_norm": 2.010105609893799,
      "learning_rate": 2.2995125463564956e-07,
      "loss": 1.9773,
      "num_input_tokens_seen": 528750560,
      "step": 519300
    },
    {
      "epoch": 10.226221180917879,
      "grad_norm": 1.7972813844680786,
      "learning_rate": 2.2977775559775924e-07,
      "loss": 1.9713,
      "num_input_tokens_seen": 528852960,
      "step": 519400
    },
    {
      "epoch": 10.228190033667381,
      "grad_norm": 1.8483458757400513,
      "learning_rate": 2.2960430250888247e-07,
      "loss": 1.9711,
      "num_input_tokens_seen": 528955360,
      "step": 519500
    },
    {
      "epoch": 10.230158886416884,
      "grad_norm": 1.818363904953003,
      "learning_rate": 2.294308953985134e-07,
      "loss": 2.0199,
      "num_input_tokens_seen": 529055856,
      "step": 519600
    },
    {
      "epoch": 10.232127739166387,
      "grad_norm": 2.1480441093444824,
      "learning_rate": 2.2925753429613863e-07,
      "loss": 1.9829,
      "num_input_tokens_seen": 529158256,
      "step": 519700
    },
    {
      "epoch": 10.23409659191589,
      "grad_norm": 1.8674441576004028,
      "learning_rate": 2.2908421923123644e-07,
      "loss": 2.0033,
      "num_input_tokens_seen": 529260656,
      "step": 519800
    },
    {
      "epoch": 10.236065444665394,
      "grad_norm": 2.164078950881958,
      "learning_rate": 2.2891095023327762e-07,
      "loss": 1.9559,
      "num_input_tokens_seen": 529363056,
      "step": 519900
    },
    {
      "epoch": 10.238034297414897,
      "grad_norm": 1.9214270114898682,
      "learning_rate": 2.2873772733172559e-07,
      "loss": 1.9941,
      "num_input_tokens_seen": 529465456,
      "step": 520000
    },
    {
      "epoch": 10.2400031501644,
      "grad_norm": 1.9703806638717651,
      "learning_rate": 2.2856455055603485e-07,
      "loss": 2.0323,
      "num_input_tokens_seen": 529567856,
      "step": 520100
    },
    {
      "epoch": 10.241972002913903,
      "grad_norm": 2.055276393890381,
      "learning_rate": 2.2839141993565314e-07,
      "loss": 1.9791,
      "num_input_tokens_seen": 529670256,
      "step": 520200
    },
    {
      "epoch": 10.243940855663405,
      "grad_norm": 2.2363429069519043,
      "learning_rate": 2.2821833550001923e-07,
      "loss": 1.9815,
      "num_input_tokens_seen": 529772656,
      "step": 520300
    },
    {
      "epoch": 10.245909708412908,
      "grad_norm": 1.9203091859817505,
      "learning_rate": 2.2804529727856547e-07,
      "loss": 1.9928,
      "num_input_tokens_seen": 529875056,
      "step": 520400
    },
    {
      "epoch": 10.24787856116241,
      "grad_norm": 2.432593822479248,
      "learning_rate": 2.278723053007149e-07,
      "loss": 2.0213,
      "num_input_tokens_seen": 529975784,
      "step": 520500
    },
    {
      "epoch": 10.249847413911914,
      "grad_norm": 2.023099899291992,
      "learning_rate": 2.2769935959588377e-07,
      "loss": 1.9674,
      "num_input_tokens_seen": 530077432,
      "step": 520600
    },
    {
      "epoch": 10.251816266661416,
      "grad_norm": 2.149197816848755,
      "learning_rate": 2.2752646019347947e-07,
      "loss": 1.9796,
      "num_input_tokens_seen": 530179832,
      "step": 520700
    },
    {
      "epoch": 10.253785119410919,
      "grad_norm": 1.8262792825698853,
      "learning_rate": 2.273536071229029e-07,
      "loss": 2.0127,
      "num_input_tokens_seen": 530280344,
      "step": 520800
    },
    {
      "epoch": 10.255753972160422,
      "grad_norm": 2.2032392024993896,
      "learning_rate": 2.271808004135456e-07,
      "loss": 1.9719,
      "num_input_tokens_seen": 530382744,
      "step": 520900
    },
    {
      "epoch": 10.257722824909925,
      "grad_norm": 1.779129147529602,
      "learning_rate": 2.2700804009479212e-07,
      "loss": 1.9448,
      "num_input_tokens_seen": 530485144,
      "step": 521000
    },
    {
      "epoch": 10.259691677659427,
      "grad_norm": 1.8416393995285034,
      "learning_rate": 2.268353261960189e-07,
      "loss": 1.9488,
      "num_input_tokens_seen": 530587544,
      "step": 521100
    },
    {
      "epoch": 10.26166053040893,
      "grad_norm": 1.9500023126602173,
      "learning_rate": 2.266626587465945e-07,
      "loss": 2.0316,
      "num_input_tokens_seen": 530689944,
      "step": 521200
    },
    {
      "epoch": 10.263629383158433,
      "grad_norm": 1.6354200839996338,
      "learning_rate": 2.2649003777587978e-07,
      "loss": 1.9591,
      "num_input_tokens_seen": 530792344,
      "step": 521300
    },
    {
      "epoch": 10.265598235907936,
      "grad_norm": 1.780234694480896,
      "learning_rate": 2.2631746331322697e-07,
      "loss": 2.0049,
      "num_input_tokens_seen": 530893896,
      "step": 521400
    },
    {
      "epoch": 10.26756708865744,
      "grad_norm": 2.0246331691741943,
      "learning_rate": 2.2614493538798124e-07,
      "loss": 1.9504,
      "num_input_tokens_seen": 530996296,
      "step": 521500
    },
    {
      "epoch": 10.269535941406943,
      "grad_norm": 2.1935033798217773,
      "learning_rate": 2.2597245402947946e-07,
      "loss": 1.9635,
      "num_input_tokens_seen": 531098696,
      "step": 521600
    },
    {
      "epoch": 10.271504794156446,
      "grad_norm": 1.843631625175476,
      "learning_rate": 2.2580001926705073e-07,
      "loss": 1.949,
      "num_input_tokens_seen": 531201096,
      "step": 521700
    },
    {
      "epoch": 10.273473646905948,
      "grad_norm": 2.0420875549316406,
      "learning_rate": 2.2562763113001588e-07,
      "loss": 2.0177,
      "num_input_tokens_seen": 531302584,
      "step": 521800
    },
    {
      "epoch": 10.275442499655451,
      "grad_norm": 1.9334372282028198,
      "learning_rate": 2.2545528964768818e-07,
      "loss": 2.0019,
      "num_input_tokens_seen": 531404984,
      "step": 521900
    },
    {
      "epoch": 10.277411352404954,
      "grad_norm": 1.931643009185791,
      "learning_rate": 2.2528299484937285e-07,
      "loss": 2.0172,
      "num_input_tokens_seen": 531507384,
      "step": 522000
    },
    {
      "epoch": 10.279380205154457,
      "grad_norm": 2.018916368484497,
      "learning_rate": 2.251107467643672e-07,
      "loss": 2.0293,
      "num_input_tokens_seen": 531608608,
      "step": 522100
    },
    {
      "epoch": 10.28134905790396,
      "grad_norm": 1.982254981994629,
      "learning_rate": 2.2493854542196073e-07,
      "loss": 1.9613,
      "num_input_tokens_seen": 531710160,
      "step": 522200
    },
    {
      "epoch": 10.283317910653462,
      "grad_norm": 1.9437209367752075,
      "learning_rate": 2.247663908514344e-07,
      "loss": 1.949,
      "num_input_tokens_seen": 531811880,
      "step": 522300
    },
    {
      "epoch": 10.285286763402965,
      "grad_norm": 2.0421812534332275,
      "learning_rate": 2.245942830820619e-07,
      "loss": 1.9901,
      "num_input_tokens_seen": 531913512,
      "step": 522400
    },
    {
      "epoch": 10.287255616152468,
      "grad_norm": 1.7826402187347412,
      "learning_rate": 2.2442222214310869e-07,
      "loss": 2.029,
      "num_input_tokens_seen": 532014576,
      "step": 522500
    },
    {
      "epoch": 10.28922446890197,
      "grad_norm": 1.862229824066162,
      "learning_rate": 2.2425020806383248e-07,
      "loss": 1.9949,
      "num_input_tokens_seen": 532116232,
      "step": 522600
    },
    {
      "epoch": 10.291193321651473,
      "grad_norm": 1.9002376794815063,
      "learning_rate": 2.2407824087348237e-07,
      "loss": 2.0079,
      "num_input_tokens_seen": 532218536,
      "step": 522700
    },
    {
      "epoch": 10.293162174400976,
      "grad_norm": 2.0391688346862793,
      "learning_rate": 2.239063206013002e-07,
      "loss": 2.0133,
      "num_input_tokens_seen": 532319624,
      "step": 522800
    },
    {
      "epoch": 10.295131027150479,
      "grad_norm": 1.930080533027649,
      "learning_rate": 2.237344472765196e-07,
      "loss": 1.9766,
      "num_input_tokens_seen": 532421472,
      "step": 522900
    },
    {
      "epoch": 10.297099879899982,
      "grad_norm": 1.7099961042404175,
      "learning_rate": 2.2356262092836598e-07,
      "loss": 1.9889,
      "num_input_tokens_seen": 532523048,
      "step": 523000
    },
    {
      "epoch": 10.299068732649484,
      "grad_norm": 1.8478100299835205,
      "learning_rate": 2.2339084158605736e-07,
      "loss": 2.0042,
      "num_input_tokens_seen": 532623536,
      "step": 523100
    },
    {
      "epoch": 10.301037585398989,
      "grad_norm": 1.9965620040893555,
      "learning_rate": 2.2321910927880273e-07,
      "loss": 1.9936,
      "num_input_tokens_seen": 532725936,
      "step": 523200
    },
    {
      "epoch": 10.303006438148492,
      "grad_norm": 1.9363305568695068,
      "learning_rate": 2.230474240358044e-07,
      "loss": 1.9373,
      "num_input_tokens_seen": 532828336,
      "step": 523300
    },
    {
      "epoch": 10.304975290897994,
      "grad_norm": 2.0018746852874756,
      "learning_rate": 2.2287578588625555e-07,
      "loss": 1.9802,
      "num_input_tokens_seen": 532930736,
      "step": 523400
    },
    {
      "epoch": 10.306944143647497,
      "grad_norm": 1.9788929224014282,
      "learning_rate": 2.227041948593421e-07,
      "loss": 2.0174,
      "num_input_tokens_seen": 533033136,
      "step": 523500
    },
    {
      "epoch": 10.308912996397,
      "grad_norm": 1.749315857887268,
      "learning_rate": 2.2253265098424111e-07,
      "loss": 1.9514,
      "num_input_tokens_seen": 533134480,
      "step": 523600
    },
    {
      "epoch": 10.310881849146503,
      "grad_norm": 1.8094364404678345,
      "learning_rate": 2.2236115429012296e-07,
      "loss": 1.9811,
      "num_input_tokens_seen": 533236880,
      "step": 523700
    },
    {
      "epoch": 10.312850701896005,
      "grad_norm": 2.026655912399292,
      "learning_rate": 2.2218970480614862e-07,
      "loss": 1.9441,
      "num_input_tokens_seen": 533339280,
      "step": 523800
    },
    {
      "epoch": 10.314819554645508,
      "grad_norm": 1.825201153755188,
      "learning_rate": 2.220183025614718e-07,
      "loss": 2.0012,
      "num_input_tokens_seen": 533440728,
      "step": 523900
    },
    {
      "epoch": 10.316788407395011,
      "grad_norm": 1.8174968957901,
      "learning_rate": 2.2184694758523802e-07,
      "loss": 1.9657,
      "num_input_tokens_seen": 533542312,
      "step": 524000
    },
    {
      "epoch": 10.318757260144514,
      "grad_norm": 2.0007975101470947,
      "learning_rate": 2.2167563990658478e-07,
      "loss": 2.0017,
      "num_input_tokens_seen": 533644224,
      "step": 524100
    },
    {
      "epoch": 10.320726112894016,
      "grad_norm": 2.3326287269592285,
      "learning_rate": 2.2150437955464164e-07,
      "loss": 1.9784,
      "num_input_tokens_seen": 533746104,
      "step": 524200
    },
    {
      "epoch": 10.32269496564352,
      "grad_norm": 1.8556303977966309,
      "learning_rate": 2.2133316655852962e-07,
      "loss": 2.026,
      "num_input_tokens_seen": 533848504,
      "step": 524300
    },
    {
      "epoch": 10.324663818393022,
      "grad_norm": 1.7767612934112549,
      "learning_rate": 2.2116200094736226e-07,
      "loss": 1.9924,
      "num_input_tokens_seen": 533950312,
      "step": 524400
    },
    {
      "epoch": 10.326632671142525,
      "grad_norm": 1.9030683040618896,
      "learning_rate": 2.209908827502448e-07,
      "loss": 2.0138,
      "num_input_tokens_seen": 534052712,
      "step": 524500
    },
    {
      "epoch": 10.328601523892027,
      "grad_norm": 1.857700228691101,
      "learning_rate": 2.2081981199627464e-07,
      "loss": 1.9975,
      "num_input_tokens_seen": 534154120,
      "step": 524600
    },
    {
      "epoch": 10.33057037664153,
      "grad_norm": 1.9168223142623901,
      "learning_rate": 2.2064878871454056e-07,
      "loss": 1.9657,
      "num_input_tokens_seen": 534256520,
      "step": 524700
    },
    {
      "epoch": 10.332539229391035,
      "grad_norm": 1.792134165763855,
      "learning_rate": 2.2047781293412382e-07,
      "loss": 1.9913,
      "num_input_tokens_seen": 534358920,
      "step": 524800
    },
    {
      "epoch": 10.334508082140538,
      "grad_norm": 1.745578646659851,
      "learning_rate": 2.2030688468409736e-07,
      "loss": 2.0529,
      "num_input_tokens_seen": 534458480,
      "step": 524900
    },
    {
      "epoch": 10.33647693489004,
      "grad_norm": 1.7430298328399658,
      "learning_rate": 2.2013600399352617e-07,
      "loss": 1.9256,
      "num_input_tokens_seen": 534560408,
      "step": 525000
    },
    {
      "epoch": 10.338445787639543,
      "grad_norm": 2.0049538612365723,
      "learning_rate": 2.1996517089146716e-07,
      "loss": 1.9474,
      "num_input_tokens_seen": 534662808,
      "step": 525100
    },
    {
      "epoch": 10.340414640389046,
      "grad_norm": 2.332439661026001,
      "learning_rate": 2.1979438540696872e-07,
      "loss": 2.0159,
      "num_input_tokens_seen": 534764512,
      "step": 525200
    },
    {
      "epoch": 10.342383493138549,
      "grad_norm": 1.9742377996444702,
      "learning_rate": 2.196236475690717e-07,
      "loss": 1.9986,
      "num_input_tokens_seen": 534866272,
      "step": 525300
    },
    {
      "epoch": 10.344352345888051,
      "grad_norm": 1.7555639743804932,
      "learning_rate": 2.194529574068086e-07,
      "loss": 1.9787,
      "num_input_tokens_seen": 534968584,
      "step": 525400
    },
    {
      "epoch": 10.346321198637554,
      "grad_norm": 2.2423934936523438,
      "learning_rate": 2.19282314949204e-07,
      "loss": 2.0047,
      "num_input_tokens_seen": 535070224,
      "step": 525500
    },
    {
      "epoch": 10.348290051387057,
      "grad_norm": 2.0216166973114014,
      "learning_rate": 2.1911172022527397e-07,
      "loss": 2.0216,
      "num_input_tokens_seen": 535170960,
      "step": 525600
    },
    {
      "epoch": 10.35025890413656,
      "grad_norm": 2.083752155303955,
      "learning_rate": 2.1894117326402673e-07,
      "loss": 1.9949,
      "num_input_tokens_seen": 535273360,
      "step": 525700
    },
    {
      "epoch": 10.352227756886062,
      "grad_norm": 1.8411338329315186,
      "learning_rate": 2.1877067409446241e-07,
      "loss": 1.964,
      "num_input_tokens_seen": 535374984,
      "step": 525800
    },
    {
      "epoch": 10.354196609635565,
      "grad_norm": 1.942962884902954,
      "learning_rate": 2.1860022274557317e-07,
      "loss": 2.0191,
      "num_input_tokens_seen": 535477384,
      "step": 525900
    },
    {
      "epoch": 10.356165462385068,
      "grad_norm": 1.9197605848312378,
      "learning_rate": 2.1842981924634242e-07,
      "loss": 1.9617,
      "num_input_tokens_seen": 535579088,
      "step": 526000
    },
    {
      "epoch": 10.35813431513457,
      "grad_norm": 1.7973542213439941,
      "learning_rate": 2.182594636257459e-07,
      "loss": 1.9685,
      "num_input_tokens_seen": 535681488,
      "step": 526100
    },
    {
      "epoch": 10.360103167884073,
      "grad_norm": 1.6107827425003052,
      "learning_rate": 2.1808915591275155e-07,
      "loss": 2.0081,
      "num_input_tokens_seen": 535783448,
      "step": 526200
    },
    {
      "epoch": 10.362072020633576,
      "grad_norm": 1.6605485677719116,
      "learning_rate": 2.1791889613631836e-07,
      "loss": 1.9611,
      "num_input_tokens_seen": 535885848,
      "step": 526300
    },
    {
      "epoch": 10.364040873383079,
      "grad_norm": 1.8685499429702759,
      "learning_rate": 2.1774868432539773e-07,
      "loss": 1.9851,
      "num_input_tokens_seen": 535987608,
      "step": 526400
    },
    {
      "epoch": 10.366009726132582,
      "grad_norm": 1.9300727844238281,
      "learning_rate": 2.1757852050893265e-07,
      "loss": 2.0047,
      "num_input_tokens_seen": 536089304,
      "step": 526500
    },
    {
      "epoch": 10.367978578882086,
      "grad_norm": 2.131390333175659,
      "learning_rate": 2.1740840471585837e-07,
      "loss": 1.9521,
      "num_input_tokens_seen": 536191704,
      "step": 526600
    },
    {
      "epoch": 10.369947431631589,
      "grad_norm": 2.055431365966797,
      "learning_rate": 2.1723833697510113e-07,
      "loss": 1.9706,
      "num_input_tokens_seen": 536293504,
      "step": 526700
    },
    {
      "epoch": 10.371916284381092,
      "grad_norm": 2.128218650817871,
      "learning_rate": 2.170683173155798e-07,
      "loss": 1.9381,
      "num_input_tokens_seen": 536395376,
      "step": 526800
    },
    {
      "epoch": 10.373885137130594,
      "grad_norm": 1.9211279153823853,
      "learning_rate": 2.1689834576620474e-07,
      "loss": 2.0253,
      "num_input_tokens_seen": 536497776,
      "step": 526900
    },
    {
      "epoch": 10.375853989880097,
      "grad_norm": 1.6804500818252563,
      "learning_rate": 2.167284223558782e-07,
      "loss": 2.0039,
      "num_input_tokens_seen": 536599184,
      "step": 527000
    },
    {
      "epoch": 10.3778228426296,
      "grad_norm": 1.6547131538391113,
      "learning_rate": 2.1655854711349442e-07,
      "loss": 1.9946,
      "num_input_tokens_seen": 536701584,
      "step": 527100
    },
    {
      "epoch": 10.379791695379103,
      "grad_norm": 2.1227502822875977,
      "learning_rate": 2.163887200679388e-07,
      "loss": 1.9438,
      "num_input_tokens_seen": 536803128,
      "step": 527200
    },
    {
      "epoch": 10.381760548128605,
      "grad_norm": 1.9699610471725464,
      "learning_rate": 2.1621894124808932e-07,
      "loss": 1.9879,
      "num_input_tokens_seen": 536905528,
      "step": 527300
    },
    {
      "epoch": 10.383729400878108,
      "grad_norm": 1.9580391645431519,
      "learning_rate": 2.1604921068281528e-07,
      "loss": 1.9822,
      "num_input_tokens_seen": 537007928,
      "step": 527400
    },
    {
      "epoch": 10.385698253627611,
      "grad_norm": 1.9856661558151245,
      "learning_rate": 2.1587952840097823e-07,
      "loss": 1.964,
      "num_input_tokens_seen": 537109896,
      "step": 527500
    },
    {
      "epoch": 10.387667106377114,
      "grad_norm": 1.7396836280822754,
      "learning_rate": 2.1570989443143084e-07,
      "loss": 2.0012,
      "num_input_tokens_seen": 537211504,
      "step": 527600
    },
    {
      "epoch": 10.389635959126617,
      "grad_norm": 2.057490587234497,
      "learning_rate": 2.1554030880301803e-07,
      "loss": 2.0267,
      "num_input_tokens_seen": 537313448,
      "step": 527700
    },
    {
      "epoch": 10.39160481187612,
      "grad_norm": 1.6570557355880737,
      "learning_rate": 2.1537077154457656e-07,
      "loss": 2.0146,
      "num_input_tokens_seen": 537415304,
      "step": 527800
    },
    {
      "epoch": 10.393573664625622,
      "grad_norm": 2.0204179286956787,
      "learning_rate": 2.152012826849348e-07,
      "loss": 1.9446,
      "num_input_tokens_seen": 537516896,
      "step": 527900
    },
    {
      "epoch": 10.395542517375125,
      "grad_norm": 1.902204990386963,
      "learning_rate": 2.150318422529127e-07,
      "loss": 1.9952,
      "num_input_tokens_seen": 537618984,
      "step": 528000
    },
    {
      "epoch": 10.397511370124628,
      "grad_norm": 1.7257249355316162,
      "learning_rate": 2.1486245027732208e-07,
      "loss": 1.9506,
      "num_input_tokens_seen": 537721384,
      "step": 528100
    },
    {
      "epoch": 10.399480222874132,
      "grad_norm": 1.6369906663894653,
      "learning_rate": 2.1469310678696718e-07,
      "loss": 1.9586,
      "num_input_tokens_seen": 537823104,
      "step": 528200
    },
    {
      "epoch": 10.401449075623635,
      "grad_norm": 1.8127775192260742,
      "learning_rate": 2.1452381181064293e-07,
      "loss": 2.0144,
      "num_input_tokens_seen": 537925504,
      "step": 528300
    },
    {
      "epoch": 10.403417928373138,
      "grad_norm": 1.918745756149292,
      "learning_rate": 2.1435456537713681e-07,
      "loss": 1.9708,
      "num_input_tokens_seen": 538027088,
      "step": 528400
    },
    {
      "epoch": 10.40538678112264,
      "grad_norm": 1.9066112041473389,
      "learning_rate": 2.1418536751522726e-07,
      "loss": 1.9414,
      "num_input_tokens_seen": 538129488,
      "step": 528500
    },
    {
      "epoch": 10.407355633872143,
      "grad_norm": 1.8209941387176514,
      "learning_rate": 2.1401621825368565e-07,
      "loss": 1.9672,
      "num_input_tokens_seen": 538230440,
      "step": 528600
    },
    {
      "epoch": 10.409324486621646,
      "grad_norm": 1.7263050079345703,
      "learning_rate": 2.138471176212739e-07,
      "loss": 1.9203,
      "num_input_tokens_seen": 538332840,
      "step": 528700
    },
    {
      "epoch": 10.411293339371149,
      "grad_norm": 1.8965415954589844,
      "learning_rate": 2.136780656467464e-07,
      "loss": 1.9704,
      "num_input_tokens_seen": 538433896,
      "step": 528800
    },
    {
      "epoch": 10.413262192120651,
      "grad_norm": 1.8575900793075562,
      "learning_rate": 2.1350906235884852e-07,
      "loss": 1.9656,
      "num_input_tokens_seen": 538534640,
      "step": 528900
    },
    {
      "epoch": 10.415231044870154,
      "grad_norm": 2.3903491497039795,
      "learning_rate": 2.133401077863184e-07,
      "loss": 1.985,
      "num_input_tokens_seen": 538636176,
      "step": 529000
    },
    {
      "epoch": 10.417199897619657,
      "grad_norm": 1.6923577785491943,
      "learning_rate": 2.131712019578853e-07,
      "loss": 2.019,
      "num_input_tokens_seen": 538738576,
      "step": 529100
    },
    {
      "epoch": 10.41916875036916,
      "grad_norm": 4.005575180053711,
      "learning_rate": 2.1300234490226992e-07,
      "loss": 2.0357,
      "num_input_tokens_seen": 538839320,
      "step": 529200
    },
    {
      "epoch": 10.421137603118662,
      "grad_norm": 2.0925350189208984,
      "learning_rate": 2.1283353664818516e-07,
      "loss": 2.0347,
      "num_input_tokens_seen": 538941720,
      "step": 529300
    },
    {
      "epoch": 10.423106455868165,
      "grad_norm": 1.9640816450119019,
      "learning_rate": 2.126647772243354e-07,
      "loss": 2.0019,
      "num_input_tokens_seen": 539044120,
      "step": 529400
    },
    {
      "epoch": 10.425075308617668,
      "grad_norm": 1.8902928829193115,
      "learning_rate": 2.12496066659417e-07,
      "loss": 1.9818,
      "num_input_tokens_seen": 539146520,
      "step": 529500
    },
    {
      "epoch": 10.42704416136717,
      "grad_norm": 1.8575787544250488,
      "learning_rate": 2.1232740498211731e-07,
      "loss": 2.0276,
      "num_input_tokens_seen": 539248232,
      "step": 529600
    },
    {
      "epoch": 10.429013014116673,
      "grad_norm": 2.171523094177246,
      "learning_rate": 2.121587922211161e-07,
      "loss": 2.0075,
      "num_input_tokens_seen": 539350632,
      "step": 529700
    },
    {
      "epoch": 10.430981866866176,
      "grad_norm": 2.09633731842041,
      "learning_rate": 2.1199022840508452e-07,
      "loss": 1.9499,
      "num_input_tokens_seen": 539451976,
      "step": 529800
    },
    {
      "epoch": 10.43295071961568,
      "grad_norm": 2.0456607341766357,
      "learning_rate": 2.1182171356268552e-07,
      "loss": 1.9291,
      "num_input_tokens_seen": 539553840,
      "step": 529900
    },
    {
      "epoch": 10.434919572365184,
      "grad_norm": 1.7530344724655151,
      "learning_rate": 2.116532477225734e-07,
      "loss": 2.0332,
      "num_input_tokens_seen": 539655664,
      "step": 530000
    },
    {
      "epoch": 10.436888425114686,
      "grad_norm": 1.7530919313430786,
      "learning_rate": 2.1148483091339447e-07,
      "loss": 2.0249,
      "num_input_tokens_seen": 539757336,
      "step": 530100
    },
    {
      "epoch": 10.438857277864189,
      "grad_norm": 1.8370981216430664,
      "learning_rate": 2.1131646316378655e-07,
      "loss": 1.9842,
      "num_input_tokens_seen": 539858816,
      "step": 530200
    },
    {
      "epoch": 10.440826130613692,
      "grad_norm": 1.7831873893737793,
      "learning_rate": 2.111481445023793e-07,
      "loss": 1.9736,
      "num_input_tokens_seen": 539958856,
      "step": 530300
    },
    {
      "epoch": 10.442794983363195,
      "grad_norm": 1.8758455514907837,
      "learning_rate": 2.1097987495779385e-07,
      "loss": 1.9289,
      "num_input_tokens_seen": 540060736,
      "step": 530400
    },
    {
      "epoch": 10.444763836112697,
      "grad_norm": 1.8488799333572388,
      "learning_rate": 2.108116545586428e-07,
      "loss": 2.0148,
      "num_input_tokens_seen": 540162928,
      "step": 530500
    },
    {
      "epoch": 10.4467326888622,
      "grad_norm": 2.0831823348999023,
      "learning_rate": 2.1064348333353084e-07,
      "loss": 1.9506,
      "num_input_tokens_seen": 540265328,
      "step": 530600
    },
    {
      "epoch": 10.448701541611703,
      "grad_norm": 1.8798476457595825,
      "learning_rate": 2.1047536131105392e-07,
      "loss": 1.9451,
      "num_input_tokens_seen": 540367728,
      "step": 530700
    },
    {
      "epoch": 10.450670394361206,
      "grad_norm": 1.7741024494171143,
      "learning_rate": 2.1030728851980008e-07,
      "loss": 2.0824,
      "num_input_tokens_seen": 540469616,
      "step": 530800
    },
    {
      "epoch": 10.452639247110708,
      "grad_norm": 1.8567248582839966,
      "learning_rate": 2.1013926498834822e-07,
      "loss": 1.9891,
      "num_input_tokens_seen": 540570664,
      "step": 530900
    },
    {
      "epoch": 10.454608099860211,
      "grad_norm": 1.8395527601242065,
      "learning_rate": 2.0997129074526936e-07,
      "loss": 1.9399,
      "num_input_tokens_seen": 540672496,
      "step": 531000
    },
    {
      "epoch": 10.456576952609714,
      "grad_norm": 1.7833343744277954,
      "learning_rate": 2.0980336581912672e-07,
      "loss": 1.9804,
      "num_input_tokens_seen": 540774072,
      "step": 531100
    },
    {
      "epoch": 10.458545805359217,
      "grad_norm": 1.900362253189087,
      "learning_rate": 2.0963549023847383e-07,
      "loss": 1.9774,
      "num_input_tokens_seen": 540876472,
      "step": 531200
    },
    {
      "epoch": 10.46051465810872,
      "grad_norm": 2.158243179321289,
      "learning_rate": 2.09467664031857e-07,
      "loss": 2.0,
      "num_input_tokens_seen": 540978000,
      "step": 531300
    },
    {
      "epoch": 10.462483510858222,
      "grad_norm": 1.7789075374603271,
      "learning_rate": 2.092998872278131e-07,
      "loss": 1.968,
      "num_input_tokens_seen": 541080400,
      "step": 531400
    },
    {
      "epoch": 10.464452363607727,
      "grad_norm": 2.066131830215454,
      "learning_rate": 2.091321598548718e-07,
      "loss": 1.9518,
      "num_input_tokens_seen": 541181352,
      "step": 531500
    },
    {
      "epoch": 10.46642121635723,
      "grad_norm": 1.716525912284851,
      "learning_rate": 2.0896448194155332e-07,
      "loss": 1.97,
      "num_input_tokens_seen": 541283752,
      "step": 531600
    },
    {
      "epoch": 10.468390069106732,
      "grad_norm": 1.8623192310333252,
      "learning_rate": 2.0879685351637011e-07,
      "loss": 1.9683,
      "num_input_tokens_seen": 541384760,
      "step": 531700
    },
    {
      "epoch": 10.470358921856235,
      "grad_norm": 1.951213002204895,
      "learning_rate": 2.0862927460782547e-07,
      "loss": 1.9893,
      "num_input_tokens_seen": 541487160,
      "step": 531800
    },
    {
      "epoch": 10.472327774605738,
      "grad_norm": 2.0958969593048096,
      "learning_rate": 2.0846174524441557e-07,
      "loss": 1.9839,
      "num_input_tokens_seen": 541588088,
      "step": 531900
    },
    {
      "epoch": 10.47429662735524,
      "grad_norm": 1.9766016006469727,
      "learning_rate": 2.082942654546267e-07,
      "loss": 1.9573,
      "num_input_tokens_seen": 541690488,
      "step": 532000
    },
    {
      "epoch": 10.476265480104743,
      "grad_norm": 2.235281229019165,
      "learning_rate": 2.0812683526693765e-07,
      "loss": 1.9708,
      "num_input_tokens_seen": 541792056,
      "step": 532100
    },
    {
      "epoch": 10.478234332854246,
      "grad_norm": 1.8524541854858398,
      "learning_rate": 2.0795945470981853e-07,
      "loss": 2.0323,
      "num_input_tokens_seen": 541893760,
      "step": 532200
    },
    {
      "epoch": 10.480203185603749,
      "grad_norm": 1.9473456144332886,
      "learning_rate": 2.077921238117309e-07,
      "loss": 2.0381,
      "num_input_tokens_seen": 541995800,
      "step": 532300
    },
    {
      "epoch": 10.482172038353252,
      "grad_norm": 1.7756537199020386,
      "learning_rate": 2.0762484260112824e-07,
      "loss": 1.9719,
      "num_input_tokens_seen": 542098200,
      "step": 532400
    },
    {
      "epoch": 10.484140891102754,
      "grad_norm": 1.9330987930297852,
      "learning_rate": 2.0745761110645483e-07,
      "loss": 2.0419,
      "num_input_tokens_seen": 542199584,
      "step": 532500
    },
    {
      "epoch": 10.486109743852257,
      "grad_norm": 2.0194287300109863,
      "learning_rate": 2.0729042935614733e-07,
      "loss": 2.027,
      "num_input_tokens_seen": 542301064,
      "step": 532600
    },
    {
      "epoch": 10.48807859660176,
      "grad_norm": 2.013430118560791,
      "learning_rate": 2.0712329737863343e-07,
      "loss": 1.967,
      "num_input_tokens_seen": 542403464,
      "step": 532700
    },
    {
      "epoch": 10.490047449351263,
      "grad_norm": 1.8868472576141357,
      "learning_rate": 2.0695621520233275e-07,
      "loss": 1.9957,
      "num_input_tokens_seen": 542505864,
      "step": 532800
    },
    {
      "epoch": 10.492016302100765,
      "grad_norm": 1.6957166194915771,
      "learning_rate": 2.0678918285565589e-07,
      "loss": 2.0094,
      "num_input_tokens_seen": 542606008,
      "step": 532900
    },
    {
      "epoch": 10.493985154850268,
      "grad_norm": 3.040231227874756,
      "learning_rate": 2.0662220036700545e-07,
      "loss": 1.998,
      "num_input_tokens_seen": 542707056,
      "step": 533000
    },
    {
      "epoch": 10.49595400759977,
      "grad_norm": 2.069591999053955,
      "learning_rate": 2.0645526776477535e-07,
      "loss": 2.0218,
      "num_input_tokens_seen": 542808688,
      "step": 533100
    },
    {
      "epoch": 10.497922860349274,
      "grad_norm": 2.020448923110962,
      "learning_rate": 2.0628838507735107e-07,
      "loss": 2.0525,
      "num_input_tokens_seen": 542910160,
      "step": 533200
    },
    {
      "epoch": 10.499891713098778,
      "grad_norm": 2.019073724746704,
      "learning_rate": 2.061215523331098e-07,
      "loss": 2.0243,
      "num_input_tokens_seen": 543010928,
      "step": 533300
    },
    {
      "epoch": 10.50186056584828,
      "grad_norm": 2.085312604904175,
      "learning_rate": 2.0595476956041974e-07,
      "loss": 2.0143,
      "num_input_tokens_seen": 543112216,
      "step": 533400
    },
    {
      "epoch": 10.503829418597784,
      "grad_norm": 1.8515269756317139,
      "learning_rate": 2.05788036787641e-07,
      "loss": 1.985,
      "num_input_tokens_seen": 543213808,
      "step": 533500
    },
    {
      "epoch": 10.505798271347286,
      "grad_norm": 1.855665683746338,
      "learning_rate": 2.056213540431251e-07,
      "loss": 1.9961,
      "num_input_tokens_seen": 543315576,
      "step": 533600
    },
    {
      "epoch": 10.50776712409679,
      "grad_norm": 1.9936466217041016,
      "learning_rate": 2.0545472135521523e-07,
      "loss": 2.0058,
      "num_input_tokens_seen": 543417976,
      "step": 533700
    },
    {
      "epoch": 10.509735976846292,
      "grad_norm": 1.6375447511672974,
      "learning_rate": 2.052881387522455e-07,
      "loss": 2.017,
      "num_input_tokens_seen": 543519456,
      "step": 533800
    },
    {
      "epoch": 10.511704829595795,
      "grad_norm": 2.0896337032318115,
      "learning_rate": 2.0512160626254205e-07,
      "loss": 1.9796,
      "num_input_tokens_seen": 543621368,
      "step": 533900
    },
    {
      "epoch": 10.513673682345297,
      "grad_norm": 1.7808884382247925,
      "learning_rate": 2.0495512391442238e-07,
      "loss": 2.0095,
      "num_input_tokens_seen": 543723768,
      "step": 534000
    },
    {
      "epoch": 10.5156425350948,
      "grad_norm": 1.9391080141067505,
      "learning_rate": 2.0478869173619534e-07,
      "loss": 1.962,
      "num_input_tokens_seen": 543826168,
      "step": 534100
    },
    {
      "epoch": 10.517611387844303,
      "grad_norm": 2.5441837310791016,
      "learning_rate": 2.046223097561615e-07,
      "loss": 1.9686,
      "num_input_tokens_seen": 543928016,
      "step": 534200
    },
    {
      "epoch": 10.519580240593806,
      "grad_norm": 1.7621675729751587,
      "learning_rate": 2.044559780026122e-07,
      "loss": 2.0071,
      "num_input_tokens_seen": 544029240,
      "step": 534300
    },
    {
      "epoch": 10.521549093343308,
      "grad_norm": 2.0743985176086426,
      "learning_rate": 2.042896965038315e-07,
      "loss": 1.9813,
      "num_input_tokens_seen": 544130888,
      "step": 534400
    },
    {
      "epoch": 10.523517946092811,
      "grad_norm": 1.8300565481185913,
      "learning_rate": 2.0412346528809353e-07,
      "loss": 1.9375,
      "num_input_tokens_seen": 544233288,
      "step": 534500
    },
    {
      "epoch": 10.525486798842314,
      "grad_norm": 1.6652699708938599,
      "learning_rate": 2.0395728438366482e-07,
      "loss": 2.0245,
      "num_input_tokens_seen": 544332856,
      "step": 534600
    },
    {
      "epoch": 10.527455651591817,
      "grad_norm": 1.6961266994476318,
      "learning_rate": 2.0379115381880295e-07,
      "loss": 1.9959,
      "num_input_tokens_seen": 544434272,
      "step": 534700
    },
    {
      "epoch": 10.52942450434132,
      "grad_norm": 1.836819052696228,
      "learning_rate": 2.0362507362175718e-07,
      "loss": 2.0373,
      "num_input_tokens_seen": 544533528,
      "step": 534800
    },
    {
      "epoch": 10.531393357090824,
      "grad_norm": 1.9433895349502563,
      "learning_rate": 2.0345904382076772e-07,
      "loss": 2.0041,
      "num_input_tokens_seen": 544635120,
      "step": 534900
    },
    {
      "epoch": 10.533362209840327,
      "grad_norm": 1.8199739456176758,
      "learning_rate": 2.032930644440668e-07,
      "loss": 1.9818,
      "num_input_tokens_seen": 544737520,
      "step": 535000
    },
    {
      "epoch": 10.53533106258983,
      "grad_norm": 2.019948959350586,
      "learning_rate": 2.0312713551987764e-07,
      "loss": 1.9781,
      "num_input_tokens_seen": 544838688,
      "step": 535100
    },
    {
      "epoch": 10.537299915339332,
      "grad_norm": 1.9360240697860718,
      "learning_rate": 2.0296125707641515e-07,
      "loss": 2.001,
      "num_input_tokens_seen": 544941088,
      "step": 535200
    },
    {
      "epoch": 10.539268768088835,
      "grad_norm": 1.9009687900543213,
      "learning_rate": 2.0279542914188575e-07,
      "loss": 1.9559,
      "num_input_tokens_seen": 545042840,
      "step": 535300
    },
    {
      "epoch": 10.541237620838338,
      "grad_norm": 1.8150891065597534,
      "learning_rate": 2.026296517444867e-07,
      "loss": 2.0156,
      "num_input_tokens_seen": 545144968,
      "step": 535400
    },
    {
      "epoch": 10.54320647358784,
      "grad_norm": 2.0322012901306152,
      "learning_rate": 2.0246392491240726e-07,
      "loss": 1.9822,
      "num_input_tokens_seen": 545246592,
      "step": 535500
    },
    {
      "epoch": 10.545175326337343,
      "grad_norm": 1.6474072933197021,
      "learning_rate": 2.0229824867382784e-07,
      "loss": 2.0302,
      "num_input_tokens_seen": 545348992,
      "step": 535600
    },
    {
      "epoch": 10.547144179086846,
      "grad_norm": 1.8624540567398071,
      "learning_rate": 2.0213262305692046e-07,
      "loss": 1.9629,
      "num_input_tokens_seen": 545450464,
      "step": 535700
    },
    {
      "epoch": 10.549113031836349,
      "grad_norm": 1.7304649353027344,
      "learning_rate": 2.0196704808984805e-07,
      "loss": 1.9757,
      "num_input_tokens_seen": 545552864,
      "step": 535800
    },
    {
      "epoch": 10.551081884585852,
      "grad_norm": 1.6681280136108398,
      "learning_rate": 2.0180152380076532e-07,
      "loss": 2.0186,
      "num_input_tokens_seen": 545654192,
      "step": 535900
    },
    {
      "epoch": 10.553050737335354,
      "grad_norm": 1.9983810186386108,
      "learning_rate": 2.0163605021781843e-07,
      "loss": 2.0142,
      "num_input_tokens_seen": 545755120,
      "step": 536000
    },
    {
      "epoch": 10.555019590084857,
      "grad_norm": 2.0335702896118164,
      "learning_rate": 2.0147062736914468e-07,
      "loss": 1.9556,
      "num_input_tokens_seen": 545857024,
      "step": 536100
    },
    {
      "epoch": 10.55698844283436,
      "grad_norm": 1.5784399509429932,
      "learning_rate": 2.0130525528287297e-07,
      "loss": 2.0136,
      "num_input_tokens_seen": 545959424,
      "step": 536200
    },
    {
      "epoch": 10.558957295583863,
      "grad_norm": 2.0112950801849365,
      "learning_rate": 2.0113993398712291e-07,
      "loss": 1.9556,
      "num_input_tokens_seen": 546061824,
      "step": 536300
    },
    {
      "epoch": 10.560926148333365,
      "grad_norm": 8.489437103271484,
      "learning_rate": 2.0097466351000675e-07,
      "loss": 1.9847,
      "num_input_tokens_seen": 546163608,
      "step": 536400
    },
    {
      "epoch": 10.562895001082868,
      "grad_norm": 1.8609848022460938,
      "learning_rate": 2.0080944387962683e-07,
      "loss": 2.0082,
      "num_input_tokens_seen": 546265304,
      "step": 536500
    },
    {
      "epoch": 10.564863853832373,
      "grad_norm": 1.7672981023788452,
      "learning_rate": 2.0064427512407762e-07,
      "loss": 1.9857,
      "num_input_tokens_seen": 546366488,
      "step": 536600
    },
    {
      "epoch": 10.566832706581875,
      "grad_norm": 1.8648737668991089,
      "learning_rate": 2.0047915727144422e-07,
      "loss": 1.9564,
      "num_input_tokens_seen": 546466864,
      "step": 536700
    },
    {
      "epoch": 10.568801559331378,
      "grad_norm": 2.1185355186462402,
      "learning_rate": 2.0031409034980418e-07,
      "loss": 2.0156,
      "num_input_tokens_seen": 546569264,
      "step": 536800
    },
    {
      "epoch": 10.570770412080881,
      "grad_norm": 1.8192405700683594,
      "learning_rate": 2.0014907438722522e-07,
      "loss": 2.0054,
      "num_input_tokens_seen": 546671072,
      "step": 536900
    },
    {
      "epoch": 10.572739264830384,
      "grad_norm": 1.9865933656692505,
      "learning_rate": 1.999841094117673e-07,
      "loss": 2.0417,
      "num_input_tokens_seen": 546773472,
      "step": 537000
    },
    {
      "epoch": 10.574708117579886,
      "grad_norm": 1.863865613937378,
      "learning_rate": 1.9981919545148078e-07,
      "loss": 1.9848,
      "num_input_tokens_seen": 546875872,
      "step": 537100
    },
    {
      "epoch": 10.57667697032939,
      "grad_norm": 2.1517233848571777,
      "learning_rate": 1.996543325344084e-07,
      "loss": 1.9481,
      "num_input_tokens_seen": 546978272,
      "step": 537200
    },
    {
      "epoch": 10.578645823078892,
      "grad_norm": 15.011783599853516,
      "learning_rate": 1.994895206885837e-07,
      "loss": 2.0419,
      "num_input_tokens_seen": 547080048,
      "step": 537300
    },
    {
      "epoch": 10.580614675828395,
      "grad_norm": 5.901471138000488,
      "learning_rate": 1.9932475994203123e-07,
      "loss": 2.0005,
      "num_input_tokens_seen": 547180208,
      "step": 537400
    },
    {
      "epoch": 10.582583528577898,
      "grad_norm": 2.052987575531006,
      "learning_rate": 1.991600503227673e-07,
      "loss": 2.0015,
      "num_input_tokens_seen": 547281936,
      "step": 537500
    },
    {
      "epoch": 10.5845523813274,
      "grad_norm": 1.9954569339752197,
      "learning_rate": 1.9899539185879943e-07,
      "loss": 2.0246,
      "num_input_tokens_seen": 547383752,
      "step": 537600
    },
    {
      "epoch": 10.586521234076903,
      "grad_norm": 2.0651237964630127,
      "learning_rate": 1.988307845781265e-07,
      "loss": 2.0164,
      "num_input_tokens_seen": 547484800,
      "step": 537700
    },
    {
      "epoch": 10.588490086826406,
      "grad_norm": 1.91893470287323,
      "learning_rate": 1.9866622850873838e-07,
      "loss": 2.0221,
      "num_input_tokens_seen": 547585760,
      "step": 537800
    },
    {
      "epoch": 10.590458939575909,
      "grad_norm": 1.7563591003417969,
      "learning_rate": 1.985017236786165e-07,
      "loss": 1.9704,
      "num_input_tokens_seen": 547687624,
      "step": 537900
    },
    {
      "epoch": 10.592427792325411,
      "grad_norm": 1.728994607925415,
      "learning_rate": 1.983372701157336e-07,
      "loss": 2.0119,
      "num_input_tokens_seen": 547788776,
      "step": 538000
    },
    {
      "epoch": 10.594396645074914,
      "grad_norm": 1.9012501239776611,
      "learning_rate": 1.981728678480537e-07,
      "loss": 1.9796,
      "num_input_tokens_seen": 547889920,
      "step": 538100
    },
    {
      "epoch": 10.596365497824419,
      "grad_norm": 1.881460428237915,
      "learning_rate": 1.9800851690353171e-07,
      "loss": 1.9646,
      "num_input_tokens_seen": 547992320,
      "step": 538200
    },
    {
      "epoch": 10.598334350573921,
      "grad_norm": 2.05285906791687,
      "learning_rate": 1.9784421731011435e-07,
      "loss": 2.0002,
      "num_input_tokens_seen": 548094264,
      "step": 538300
    },
    {
      "epoch": 10.600303203323424,
      "grad_norm": 1.824878215789795,
      "learning_rate": 1.9767996909573931e-07,
      "loss": 1.9559,
      "num_input_tokens_seen": 548196064,
      "step": 538400
    },
    {
      "epoch": 10.602272056072927,
      "grad_norm": 1.8374119997024536,
      "learning_rate": 1.9751577228833565e-07,
      "loss": 2.0164,
      "num_input_tokens_seen": 548297648,
      "step": 538500
    },
    {
      "epoch": 10.60424090882243,
      "grad_norm": 21.481945037841797,
      "learning_rate": 1.973516269158238e-07,
      "loss": 2.036,
      "num_input_tokens_seen": 548399568,
      "step": 538600
    },
    {
      "epoch": 10.606209761571932,
      "grad_norm": 1.8211166858673096,
      "learning_rate": 1.9718753300611495e-07,
      "loss": 1.9437,
      "num_input_tokens_seen": 548501968,
      "step": 538700
    },
    {
      "epoch": 10.608178614321435,
      "grad_norm": 2.0096049308776855,
      "learning_rate": 1.9702349058711216e-07,
      "loss": 1.9995,
      "num_input_tokens_seen": 548603816,
      "step": 538800
    },
    {
      "epoch": 10.610147467070938,
      "grad_norm": 2.055584192276001,
      "learning_rate": 1.968594996867093e-07,
      "loss": 1.9478,
      "num_input_tokens_seen": 548705456,
      "step": 538900
    },
    {
      "epoch": 10.61211631982044,
      "grad_norm": 1.9225744009017944,
      "learning_rate": 1.9669556033279188e-07,
      "loss": 2.0025,
      "num_input_tokens_seen": 548807856,
      "step": 539000
    },
    {
      "epoch": 10.614085172569943,
      "grad_norm": 1.860952615737915,
      "learning_rate": 1.965316725532361e-07,
      "loss": 2.0287,
      "num_input_tokens_seen": 548910256,
      "step": 539100
    },
    {
      "epoch": 10.616054025319446,
      "grad_norm": 1.8782844543457031,
      "learning_rate": 1.9636783637590964e-07,
      "loss": 1.9812,
      "num_input_tokens_seen": 549011888,
      "step": 539200
    },
    {
      "epoch": 10.618022878068949,
      "grad_norm": 2.0483462810516357,
      "learning_rate": 1.96204051828672e-07,
      "loss": 1.9846,
      "num_input_tokens_seen": 549113040,
      "step": 539300
    },
    {
      "epoch": 10.619991730818452,
      "grad_norm": 2.2186572551727295,
      "learning_rate": 1.9604031893937284e-07,
      "loss": 1.9836,
      "num_input_tokens_seen": 549215440,
      "step": 539400
    },
    {
      "epoch": 10.621960583567954,
      "grad_norm": 1.8505703210830688,
      "learning_rate": 1.9587663773585395e-07,
      "loss": 1.9707,
      "num_input_tokens_seen": 549315832,
      "step": 539500
    },
    {
      "epoch": 10.623929436317457,
      "grad_norm": 1.854048728942871,
      "learning_rate": 1.9571300824594727e-07,
      "loss": 1.9586,
      "num_input_tokens_seen": 549418232,
      "step": 539600
    },
    {
      "epoch": 10.62589828906696,
      "grad_norm": 4.737663745880127,
      "learning_rate": 1.955494304974775e-07,
      "loss": 2.0373,
      "num_input_tokens_seen": 549520480,
      "step": 539700
    },
    {
      "epoch": 10.627867141816463,
      "grad_norm": 1.8604631423950195,
      "learning_rate": 1.9538590451825898e-07,
      "loss": 2.0142,
      "num_input_tokens_seen": 549622152,
      "step": 539800
    },
    {
      "epoch": 10.629835994565966,
      "grad_norm": 1.6579934358596802,
      "learning_rate": 1.952224303360984e-07,
      "loss": 1.9897,
      "num_input_tokens_seen": 549724552,
      "step": 539900
    },
    {
      "epoch": 10.63180484731547,
      "grad_norm": 2.0413172245025635,
      "learning_rate": 1.9505900797879255e-07,
      "loss": 1.998,
      "num_input_tokens_seen": 549825912,
      "step": 540000
    },
    {
      "epoch": 10.633773700064973,
      "grad_norm": 1.9077850580215454,
      "learning_rate": 1.9489563747413075e-07,
      "loss": 1.9597,
      "num_input_tokens_seen": 549928312,
      "step": 540100
    },
    {
      "epoch": 10.635742552814476,
      "grad_norm": 1.9252736568450928,
      "learning_rate": 1.9473231884989228e-07,
      "loss": 2.0268,
      "num_input_tokens_seen": 550030712,
      "step": 540200
    },
    {
      "epoch": 10.637711405563978,
      "grad_norm": 1.7537686824798584,
      "learning_rate": 1.9456905213384818e-07,
      "loss": 1.9472,
      "num_input_tokens_seen": 550133112,
      "step": 540300
    },
    {
      "epoch": 10.639680258313481,
      "grad_norm": 1.9000575542449951,
      "learning_rate": 1.944058373537607e-07,
      "loss": 1.9933,
      "num_input_tokens_seen": 550233992,
      "step": 540400
    },
    {
      "epoch": 10.641649111062984,
      "grad_norm": 2.1509549617767334,
      "learning_rate": 1.9424267453738307e-07,
      "loss": 2.0051,
      "num_input_tokens_seen": 550336392,
      "step": 540500
    },
    {
      "epoch": 10.643617963812487,
      "grad_norm": 1.662351369857788,
      "learning_rate": 1.9407956371245993e-07,
      "loss": 1.9517,
      "num_input_tokens_seen": 550438792,
      "step": 540600
    },
    {
      "epoch": 10.64558681656199,
      "grad_norm": 2.894723892211914,
      "learning_rate": 1.939165049067265e-07,
      "loss": 1.9642,
      "num_input_tokens_seen": 550540632,
      "step": 540700
    },
    {
      "epoch": 10.647555669311492,
      "grad_norm": 1.9298900365829468,
      "learning_rate": 1.9375349814790985e-07,
      "loss": 1.9798,
      "num_input_tokens_seen": 550643032,
      "step": 540800
    },
    {
      "epoch": 10.649524522060995,
      "grad_norm": 2.13678240776062,
      "learning_rate": 1.9359054346372784e-07,
      "loss": 2.0034,
      "num_input_tokens_seen": 550744768,
      "step": 540900
    },
    {
      "epoch": 10.651493374810498,
      "grad_norm": 1.8163970708847046,
      "learning_rate": 1.9342764088188967e-07,
      "loss": 1.9512,
      "num_input_tokens_seen": 550847168,
      "step": 541000
    },
    {
      "epoch": 10.65346222756,
      "grad_norm": 1.9950670003890991,
      "learning_rate": 1.9326479043009526e-07,
      "loss": 1.9809,
      "num_input_tokens_seen": 550949000,
      "step": 541100
    },
    {
      "epoch": 10.655431080309503,
      "grad_norm": 1.9394340515136719,
      "learning_rate": 1.9310199213603611e-07,
      "loss": 1.9153,
      "num_input_tokens_seen": 551051400,
      "step": 541200
    },
    {
      "epoch": 10.657399933059006,
      "grad_norm": 2.0396976470947266,
      "learning_rate": 1.929392460273947e-07,
      "loss": 1.9686,
      "num_input_tokens_seen": 551152200,
      "step": 541300
    },
    {
      "epoch": 10.659368785808509,
      "grad_norm": 1.7576956748962402,
      "learning_rate": 1.9277655213184462e-07,
      "loss": 1.9648,
      "num_input_tokens_seen": 551254600,
      "step": 541400
    },
    {
      "epoch": 10.661337638558011,
      "grad_norm": 2.1296796798706055,
      "learning_rate": 1.9261391047705083e-07,
      "loss": 1.9509,
      "num_input_tokens_seen": 551356448,
      "step": 541500
    },
    {
      "epoch": 10.663306491307516,
      "grad_norm": 1.871614933013916,
      "learning_rate": 1.9245132109066874e-07,
      "loss": 1.9598,
      "num_input_tokens_seen": 551458264,
      "step": 541600
    },
    {
      "epoch": 10.665275344057019,
      "grad_norm": 2.2402889728546143,
      "learning_rate": 1.9228878400034548e-07,
      "loss": 1.9813,
      "num_input_tokens_seen": 551560664,
      "step": 541700
    },
    {
      "epoch": 10.667244196806521,
      "grad_norm": 2.0696139335632324,
      "learning_rate": 1.921262992337192e-07,
      "loss": 1.9625,
      "num_input_tokens_seen": 551662648,
      "step": 541800
    },
    {
      "epoch": 10.669213049556024,
      "grad_norm": 1.936038613319397,
      "learning_rate": 1.9196386681841914e-07,
      "loss": 1.9948,
      "num_input_tokens_seen": 551763808,
      "step": 541900
    },
    {
      "epoch": 10.671181902305527,
      "grad_norm": 2.043972969055176,
      "learning_rate": 1.918014867820653e-07,
      "loss": 1.975,
      "num_input_tokens_seen": 551866208,
      "step": 542000
    },
    {
      "epoch": 10.67315075505503,
      "grad_norm": 2.106046676635742,
      "learning_rate": 1.9163915915226914e-07,
      "loss": 1.9942,
      "num_input_tokens_seen": 551968296,
      "step": 542100
    },
    {
      "epoch": 10.675119607804533,
      "grad_norm": 2.1447432041168213,
      "learning_rate": 1.9147688395663315e-07,
      "loss": 1.9886,
      "num_input_tokens_seen": 552069288,
      "step": 542200
    },
    {
      "epoch": 10.677088460554035,
      "grad_norm": 2.026850700378418,
      "learning_rate": 1.9131466122275085e-07,
      "loss": 1.9473,
      "num_input_tokens_seen": 552171088,
      "step": 542300
    },
    {
      "epoch": 10.679057313303538,
      "grad_norm": 2.0418670177459717,
      "learning_rate": 1.9115249097820695e-07,
      "loss": 1.9987,
      "num_input_tokens_seen": 552272080,
      "step": 542400
    },
    {
      "epoch": 10.68102616605304,
      "grad_norm": 1.7151198387145996,
      "learning_rate": 1.9099037325057672e-07,
      "loss": 1.9792,
      "num_input_tokens_seen": 552373664,
      "step": 542500
    },
    {
      "epoch": 10.682995018802544,
      "grad_norm": 2.2091479301452637,
      "learning_rate": 1.9082830806742755e-07,
      "loss": 2.0089,
      "num_input_tokens_seen": 552476064,
      "step": 542600
    },
    {
      "epoch": 10.684963871552046,
      "grad_norm": 2.1580698490142822,
      "learning_rate": 1.9066629545631675e-07,
      "loss": 1.971,
      "num_input_tokens_seen": 552577944,
      "step": 542700
    },
    {
      "epoch": 10.686932724301549,
      "grad_norm": 1.9133449792861938,
      "learning_rate": 1.9050433544479338e-07,
      "loss": 1.9845,
      "num_input_tokens_seen": 552680344,
      "step": 542800
    },
    {
      "epoch": 10.688901577051052,
      "grad_norm": 1.8320341110229492,
      "learning_rate": 1.903424280603973e-07,
      "loss": 1.9252,
      "num_input_tokens_seen": 552782264,
      "step": 542900
    },
    {
      "epoch": 10.690870429800555,
      "grad_norm": 2.1758272647857666,
      "learning_rate": 1.9018057333065979e-07,
      "loss": 2.0062,
      "num_input_tokens_seen": 552883224,
      "step": 543000
    },
    {
      "epoch": 10.692839282550057,
      "grad_norm": 1.948440670967102,
      "learning_rate": 1.9001877128310246e-07,
      "loss": 1.9744,
      "num_input_tokens_seen": 552984800,
      "step": 543100
    },
    {
      "epoch": 10.69480813529956,
      "grad_norm": 1.7925294637680054,
      "learning_rate": 1.8985702194523856e-07,
      "loss": 1.9985,
      "num_input_tokens_seen": 553087200,
      "step": 543200
    },
    {
      "epoch": 10.696776988049065,
      "grad_norm": 1.9080694913864136,
      "learning_rate": 1.8969532534457223e-07,
      "loss": 2.0328,
      "num_input_tokens_seen": 553188536,
      "step": 543300
    },
    {
      "epoch": 10.698745840798567,
      "grad_norm": 1.8982210159301758,
      "learning_rate": 1.8953368150859854e-07,
      "loss": 1.9933,
      "num_input_tokens_seen": 553289712,
      "step": 543400
    },
    {
      "epoch": 10.70071469354807,
      "grad_norm": 1.95067298412323,
      "learning_rate": 1.8937209046480395e-07,
      "loss": 1.9887,
      "num_input_tokens_seen": 553391664,
      "step": 543500
    },
    {
      "epoch": 10.702683546297573,
      "grad_norm": 1.7763067483901978,
      "learning_rate": 1.8921055224066522e-07,
      "loss": 1.9374,
      "num_input_tokens_seen": 553494064,
      "step": 543600
    },
    {
      "epoch": 10.704652399047076,
      "grad_norm": 2.1477227210998535,
      "learning_rate": 1.890490668636508e-07,
      "loss": 1.9607,
      "num_input_tokens_seen": 553596464,
      "step": 543700
    },
    {
      "epoch": 10.706621251796578,
      "grad_norm": 1.7865806818008423,
      "learning_rate": 1.8888763436121986e-07,
      "loss": 1.9777,
      "num_input_tokens_seen": 553697952,
      "step": 543800
    },
    {
      "epoch": 10.708590104546081,
      "grad_norm": 1.9819812774658203,
      "learning_rate": 1.8872625476082277e-07,
      "loss": 1.9756,
      "num_input_tokens_seen": 553800352,
      "step": 543900
    },
    {
      "epoch": 10.710558957295584,
      "grad_norm": 2.0499982833862305,
      "learning_rate": 1.8856492808990055e-07,
      "loss": 1.9781,
      "num_input_tokens_seen": 553901904,
      "step": 544000
    },
    {
      "epoch": 10.712527810045087,
      "grad_norm": 1.919305443763733,
      "learning_rate": 1.8840365437588552e-07,
      "loss": 2.0011,
      "num_input_tokens_seen": 554004304,
      "step": 544100
    },
    {
      "epoch": 10.71449666279459,
      "grad_norm": 1.708212971687317,
      "learning_rate": 1.8824243364620091e-07,
      "loss": 1.9838,
      "num_input_tokens_seen": 554106208,
      "step": 544200
    },
    {
      "epoch": 10.716465515544092,
      "grad_norm": 1.9649652242660522,
      "learning_rate": 1.88081265928261e-07,
      "loss": 1.985,
      "num_input_tokens_seen": 554208608,
      "step": 544300
    },
    {
      "epoch": 10.718434368293595,
      "grad_norm": 2.2554948329925537,
      "learning_rate": 1.8792015124947118e-07,
      "loss": 1.9327,
      "num_input_tokens_seen": 554311008,
      "step": 544400
    },
    {
      "epoch": 10.720403221043098,
      "grad_norm": 1.8740923404693604,
      "learning_rate": 1.8775908963722703e-07,
      "loss": 1.9693,
      "num_input_tokens_seen": 554412944,
      "step": 544500
    },
    {
      "epoch": 10.7223720737926,
      "grad_norm": 1.6839097738265991,
      "learning_rate": 1.875980811189165e-07,
      "loss": 1.9832,
      "num_input_tokens_seen": 554515344,
      "step": 544600
    },
    {
      "epoch": 10.724340926542103,
      "grad_norm": 2.1475319862365723,
      "learning_rate": 1.8743712572191722e-07,
      "loss": 1.924,
      "num_input_tokens_seen": 554617744,
      "step": 544700
    },
    {
      "epoch": 10.726309779291606,
      "grad_norm": 1.9010370969772339,
      "learning_rate": 1.8727622347359857e-07,
      "loss": 1.9734,
      "num_input_tokens_seen": 554718976,
      "step": 544800
    },
    {
      "epoch": 10.72827863204111,
      "grad_norm": 2.0754222869873047,
      "learning_rate": 1.8711537440132013e-07,
      "loss": 1.9793,
      "num_input_tokens_seen": 554821376,
      "step": 544900
    },
    {
      "epoch": 10.730247484790613,
      "grad_norm": 2.4671974182128906,
      "learning_rate": 1.8695457853243363e-07,
      "loss": 1.9605,
      "num_input_tokens_seen": 554923200,
      "step": 545000
    },
    {
      "epoch": 10.732216337540116,
      "grad_norm": 2.042914628982544,
      "learning_rate": 1.8679383589428056e-07,
      "loss": 1.9705,
      "num_input_tokens_seen": 555025600,
      "step": 545100
    },
    {
      "epoch": 10.734185190289619,
      "grad_norm": 1.9371339082717896,
      "learning_rate": 1.8663314651419416e-07,
      "loss": 2.0378,
      "num_input_tokens_seen": 555127408,
      "step": 545200
    },
    {
      "epoch": 10.736154043039122,
      "grad_norm": 2.2449545860290527,
      "learning_rate": 1.8647251041949775e-07,
      "loss": 1.996,
      "num_input_tokens_seen": 555228912,
      "step": 545300
    },
    {
      "epoch": 10.738122895788624,
      "grad_norm": 2.1624972820281982,
      "learning_rate": 1.8631192763750676e-07,
      "loss": 1.9941,
      "num_input_tokens_seen": 555331312,
      "step": 545400
    },
    {
      "epoch": 10.740091748538127,
      "grad_norm": 2.1800129413604736,
      "learning_rate": 1.8615139819552683e-07,
      "loss": 1.9384,
      "num_input_tokens_seen": 555432840,
      "step": 545500
    },
    {
      "epoch": 10.74206060128763,
      "grad_norm": 1.8301419019699097,
      "learning_rate": 1.8599092212085437e-07,
      "loss": 1.9991,
      "num_input_tokens_seen": 555535240,
      "step": 545600
    },
    {
      "epoch": 10.744029454037133,
      "grad_norm": 1.8693466186523438,
      "learning_rate": 1.858304994407771e-07,
      "loss": 1.996,
      "num_input_tokens_seen": 555636504,
      "step": 545700
    },
    {
      "epoch": 10.745998306786635,
      "grad_norm": 2.3593571186065674,
      "learning_rate": 1.856701301825736e-07,
      "loss": 2.0408,
      "num_input_tokens_seen": 555738904,
      "step": 545800
    },
    {
      "epoch": 10.747967159536138,
      "grad_norm": 1.9689140319824219,
      "learning_rate": 1.8550981437351343e-07,
      "loss": 1.9639,
      "num_input_tokens_seen": 555840704,
      "step": 545900
    },
    {
      "epoch": 10.74993601228564,
      "grad_norm": 2.083340883255005,
      "learning_rate": 1.8534955204085668e-07,
      "loss": 1.9826,
      "num_input_tokens_seen": 555942048,
      "step": 546000
    },
    {
      "epoch": 10.751904865035144,
      "grad_norm": 2.0336508750915527,
      "learning_rate": 1.8518934321185464e-07,
      "loss": 1.9606,
      "num_input_tokens_seen": 556043680,
      "step": 546100
    },
    {
      "epoch": 10.753873717784646,
      "grad_norm": 3.2070682048797607,
      "learning_rate": 1.850291879137496e-07,
      "loss": 1.9645,
      "num_input_tokens_seen": 556146080,
      "step": 546200
    },
    {
      "epoch": 10.75584257053415,
      "grad_norm": 1.8451169729232788,
      "learning_rate": 1.8486908617377456e-07,
      "loss": 1.9802,
      "num_input_tokens_seen": 556247888,
      "step": 546300
    },
    {
      "epoch": 10.757811423283652,
      "grad_norm": 1.7789429426193237,
      "learning_rate": 1.8470903801915366e-07,
      "loss": 1.9566,
      "num_input_tokens_seen": 556349440,
      "step": 546400
    },
    {
      "epoch": 10.759780276033155,
      "grad_norm": 1.8773837089538574,
      "learning_rate": 1.8454904347710138e-07,
      "loss": 1.9755,
      "num_input_tokens_seen": 556451840,
      "step": 546500
    },
    {
      "epoch": 10.761749128782657,
      "grad_norm": 1.8822336196899414,
      "learning_rate": 1.843891025748236e-07,
      "loss": 1.9436,
      "num_input_tokens_seen": 556554240,
      "step": 546600
    },
    {
      "epoch": 10.763717981532162,
      "grad_norm": 1.7115010023117065,
      "learning_rate": 1.8422921533951697e-07,
      "loss": 2.0207,
      "num_input_tokens_seen": 556655296,
      "step": 546700
    },
    {
      "epoch": 10.765686834281665,
      "grad_norm": 2.0768983364105225,
      "learning_rate": 1.8406938179836912e-07,
      "loss": 1.992,
      "num_input_tokens_seen": 556757696,
      "step": 546800
    },
    {
      "epoch": 10.767655687031167,
      "grad_norm": 1.9417039155960083,
      "learning_rate": 1.83909601978558e-07,
      "loss": 1.9803,
      "num_input_tokens_seen": 556859232,
      "step": 546900
    },
    {
      "epoch": 10.76962453978067,
      "grad_norm": 1.8766381740570068,
      "learning_rate": 1.8374987590725306e-07,
      "loss": 1.9494,
      "num_input_tokens_seen": 556961136,
      "step": 547000
    },
    {
      "epoch": 10.771593392530173,
      "grad_norm": 2.033201217651367,
      "learning_rate": 1.8359020361161437e-07,
      "loss": 1.9882,
      "num_input_tokens_seen": 557063536,
      "step": 547100
    },
    {
      "epoch": 10.773562245279676,
      "grad_norm": 2.4288578033447266,
      "learning_rate": 1.8343058511879296e-07,
      "loss": 2.014,
      "num_input_tokens_seen": 557164472,
      "step": 547200
    },
    {
      "epoch": 10.775531098029179,
      "grad_norm": 1.9378972053527832,
      "learning_rate": 1.832710204559303e-07,
      "loss": 1.9844,
      "num_input_tokens_seen": 557266872,
      "step": 547300
    },
    {
      "epoch": 10.777499950778681,
      "grad_norm": 1.8342008590698242,
      "learning_rate": 1.831115096501591e-07,
      "loss": 1.9789,
      "num_input_tokens_seen": 557367608,
      "step": 547400
    },
    {
      "epoch": 10.779468803528184,
      "grad_norm": 1.6611577272415161,
      "learning_rate": 1.829520527286032e-07,
      "loss": 2.0199,
      "num_input_tokens_seen": 557470008,
      "step": 547500
    },
    {
      "epoch": 10.781437656277687,
      "grad_norm": 1.9544681310653687,
      "learning_rate": 1.8279264971837644e-07,
      "loss": 2.0638,
      "num_input_tokens_seen": 557571088,
      "step": 547600
    },
    {
      "epoch": 10.78340650902719,
      "grad_norm": 1.902199625968933,
      "learning_rate": 1.8263330064658434e-07,
      "loss": 1.9636,
      "num_input_tokens_seen": 557673488,
      "step": 547700
    },
    {
      "epoch": 10.785375361776692,
      "grad_norm": 2.241950750350952,
      "learning_rate": 1.8247400554032233e-07,
      "loss": 1.9807,
      "num_input_tokens_seen": 557775312,
      "step": 547800
    },
    {
      "epoch": 10.787344214526195,
      "grad_norm": 1.654066801071167,
      "learning_rate": 1.8231476442667794e-07,
      "loss": 1.9661,
      "num_input_tokens_seen": 557876896,
      "step": 547900
    },
    {
      "epoch": 10.789313067275698,
      "grad_norm": 1.7208657264709473,
      "learning_rate": 1.8215557733272813e-07,
      "loss": 1.996,
      "num_input_tokens_seen": 557979296,
      "step": 548000
    },
    {
      "epoch": 10.7912819200252,
      "grad_norm": 1.879143476486206,
      "learning_rate": 1.8199644428554177e-07,
      "loss": 1.9259,
      "num_input_tokens_seen": 558081696,
      "step": 548100
    },
    {
      "epoch": 10.793250772774703,
      "grad_norm": 1.8063020706176758,
      "learning_rate": 1.8183736531217753e-07,
      "loss": 1.9632,
      "num_input_tokens_seen": 558184096,
      "step": 548200
    },
    {
      "epoch": 10.795219625524208,
      "grad_norm": 1.8183461427688599,
      "learning_rate": 1.8167834043968616e-07,
      "loss": 1.9789,
      "num_input_tokens_seen": 558285808,
      "step": 548300
    },
    {
      "epoch": 10.79718847827371,
      "grad_norm": 2.0113089084625244,
      "learning_rate": 1.81519369695108e-07,
      "loss": 1.9752,
      "num_input_tokens_seen": 558388136,
      "step": 548400
    },
    {
      "epoch": 10.799157331023213,
      "grad_norm": 1.9190913438796997,
      "learning_rate": 1.8136045310547483e-07,
      "loss": 1.9626,
      "num_input_tokens_seen": 558490088,
      "step": 548500
    },
    {
      "epoch": 10.801126183772716,
      "grad_norm": 2.0270869731903076,
      "learning_rate": 1.812015906978091e-07,
      "loss": 2.0121,
      "num_input_tokens_seen": 558592488,
      "step": 548600
    },
    {
      "epoch": 10.803095036522219,
      "grad_norm": 1.844152808189392,
      "learning_rate": 1.8104278249912398e-07,
      "loss": 1.9753,
      "num_input_tokens_seen": 558694888,
      "step": 548700
    },
    {
      "epoch": 10.805063889271722,
      "grad_norm": 1.7602601051330566,
      "learning_rate": 1.8088402853642365e-07,
      "loss": 1.9593,
      "num_input_tokens_seen": 558796840,
      "step": 548800
    },
    {
      "epoch": 10.807032742021224,
      "grad_norm": 2.2236902713775635,
      "learning_rate": 1.8072532883670256e-07,
      "loss": 1.9732,
      "num_input_tokens_seen": 558898864,
      "step": 548900
    },
    {
      "epoch": 10.809001594770727,
      "grad_norm": 2.0315370559692383,
      "learning_rate": 1.8056668342694637e-07,
      "loss": 1.9944,
      "num_input_tokens_seen": 558999960,
      "step": 549000
    },
    {
      "epoch": 10.81097044752023,
      "grad_norm": 1.8718507289886475,
      "learning_rate": 1.8040809233413147e-07,
      "loss": 1.9797,
      "num_input_tokens_seen": 559100584,
      "step": 549100
    },
    {
      "epoch": 10.812939300269733,
      "grad_norm": 2.095684289932251,
      "learning_rate": 1.8024955558522509e-07,
      "loss": 2.0122,
      "num_input_tokens_seen": 559202984,
      "step": 549200
    },
    {
      "epoch": 10.814908153019235,
      "grad_norm": 4.291043281555176,
      "learning_rate": 1.800910732071847e-07,
      "loss": 1.9312,
      "num_input_tokens_seen": 559304560,
      "step": 549300
    },
    {
      "epoch": 10.816877005768738,
      "grad_norm": 2.2442626953125,
      "learning_rate": 1.7993264522695905e-07,
      "loss": 1.986,
      "num_input_tokens_seen": 559405872,
      "step": 549400
    },
    {
      "epoch": 10.818845858518241,
      "grad_norm": 1.9597442150115967,
      "learning_rate": 1.7977427167148758e-07,
      "loss": 2.0036,
      "num_input_tokens_seen": 559508272,
      "step": 549500
    },
    {
      "epoch": 10.820814711267744,
      "grad_norm": 1.7878895998001099,
      "learning_rate": 1.7961595256770024e-07,
      "loss": 1.9619,
      "num_input_tokens_seen": 559610112,
      "step": 549600
    },
    {
      "epoch": 10.822783564017247,
      "grad_norm": 2.0321359634399414,
      "learning_rate": 1.794576879425181e-07,
      "loss": 2.0236,
      "num_input_tokens_seen": 559712512,
      "step": 549700
    },
    {
      "epoch": 10.82475241676675,
      "grad_norm": 1.8442186117172241,
      "learning_rate": 1.7929947782285239e-07,
      "loss": 1.9962,
      "num_input_tokens_seen": 559814912,
      "step": 549800
    },
    {
      "epoch": 10.826721269516252,
      "grad_norm": 2.162869691848755,
      "learning_rate": 1.7914132223560558e-07,
      "loss": 2.0282,
      "num_input_tokens_seen": 559917168,
      "step": 549900
    },
    {
      "epoch": 10.828690122265757,
      "grad_norm": 3.412938356399536,
      "learning_rate": 1.7898322120767063e-07,
      "loss": 2.0156,
      "num_input_tokens_seen": 560018000,
      "step": 550000
    },
    {
      "epoch": 10.83065897501526,
      "grad_norm": 1.93895423412323,
      "learning_rate": 1.788251747659315e-07,
      "loss": 2.0037,
      "num_input_tokens_seen": 560119568,
      "step": 550100
    },
    {
      "epoch": 10.832627827764762,
      "grad_norm": 2.056131601333618,
      "learning_rate": 1.7866718293726236e-07,
      "loss": 1.9807,
      "num_input_tokens_seen": 560221512,
      "step": 550200
    },
    {
      "epoch": 10.834596680514265,
      "grad_norm": 2.139589786529541,
      "learning_rate": 1.7850924574852854e-07,
      "loss": 2.0092,
      "num_input_tokens_seen": 560320424,
      "step": 550300
    },
    {
      "epoch": 10.836565533263768,
      "grad_norm": 2.064422369003296,
      "learning_rate": 1.7835136322658584e-07,
      "loss": 1.939,
      "num_input_tokens_seen": 560422480,
      "step": 550400
    },
    {
      "epoch": 10.83853438601327,
      "grad_norm": 1.7997618913650513,
      "learning_rate": 1.7819353539828097e-07,
      "loss": 1.9604,
      "num_input_tokens_seen": 560524240,
      "step": 550500
    },
    {
      "epoch": 10.840503238762773,
      "grad_norm": 1.8308550119400024,
      "learning_rate": 1.7803576229045125e-07,
      "loss": 1.9907,
      "num_input_tokens_seen": 560625744,
      "step": 550600
    },
    {
      "epoch": 10.842472091512276,
      "grad_norm": 1.782130241394043,
      "learning_rate": 1.7787804392992433e-07,
      "loss": 2.0138,
      "num_input_tokens_seen": 560727272,
      "step": 550700
    },
    {
      "epoch": 10.844440944261779,
      "grad_norm": 1.8118078708648682,
      "learning_rate": 1.7772038034351943e-07,
      "loss": 1.9414,
      "num_input_tokens_seen": 560829672,
      "step": 550800
    },
    {
      "epoch": 10.846409797011281,
      "grad_norm": 1.910395622253418,
      "learning_rate": 1.7756277155804545e-07,
      "loss": 1.9839,
      "num_input_tokens_seen": 560931520,
      "step": 550900
    },
    {
      "epoch": 10.848378649760784,
      "grad_norm": 1.6024221181869507,
      "learning_rate": 1.7740521760030264e-07,
      "loss": 1.9676,
      "num_input_tokens_seen": 561033416,
      "step": 551000
    },
    {
      "epoch": 10.850347502510287,
      "grad_norm": 1.9158639907836914,
      "learning_rate": 1.7724771849708164e-07,
      "loss": 1.9416,
      "num_input_tokens_seen": 561135816,
      "step": 551100
    },
    {
      "epoch": 10.85231635525979,
      "grad_norm": 2.0344066619873047,
      "learning_rate": 1.770902742751641e-07,
      "loss": 1.9874,
      "num_input_tokens_seen": 561237280,
      "step": 551200
    },
    {
      "epoch": 10.854285208009292,
      "grad_norm": 1.8660359382629395,
      "learning_rate": 1.769328849613217e-07,
      "loss": 2.0408,
      "num_input_tokens_seen": 561338968,
      "step": 551300
    },
    {
      "epoch": 10.856254060758795,
      "grad_norm": 2.2782697677612305,
      "learning_rate": 1.7677555058231731e-07,
      "loss": 1.9723,
      "num_input_tokens_seen": 561441368,
      "step": 551400
    },
    {
      "epoch": 10.858222913508298,
      "grad_norm": 1.8551559448242188,
      "learning_rate": 1.7661827116490435e-07,
      "loss": 1.9687,
      "num_input_tokens_seen": 561542992,
      "step": 551500
    },
    {
      "epoch": 10.860191766257802,
      "grad_norm": 1.6949559450149536,
      "learning_rate": 1.7646104673582685e-07,
      "loss": 1.9687,
      "num_input_tokens_seen": 561645392,
      "step": 551600
    },
    {
      "epoch": 10.862160619007305,
      "grad_norm": 1.8333609104156494,
      "learning_rate": 1.7630387732181973e-07,
      "loss": 1.9726,
      "num_input_tokens_seen": 561747792,
      "step": 551700
    },
    {
      "epoch": 10.864129471756808,
      "grad_norm": 1.9584308862686157,
      "learning_rate": 1.7614676294960785e-07,
      "loss": 1.9377,
      "num_input_tokens_seen": 561849584,
      "step": 551800
    },
    {
      "epoch": 10.86609832450631,
      "grad_norm": 1.767691731452942,
      "learning_rate": 1.7598970364590742e-07,
      "loss": 2.0214,
      "num_input_tokens_seen": 561951112,
      "step": 551900
    },
    {
      "epoch": 10.868067177255814,
      "grad_norm": 27.583389282226562,
      "learning_rate": 1.7583269943742513e-07,
      "loss": 2.0145,
      "num_input_tokens_seen": 562052280,
      "step": 552000
    },
    {
      "epoch": 10.870036030005316,
      "grad_norm": 1.7727301120758057,
      "learning_rate": 1.7567575035085824e-07,
      "loss": 2.0146,
      "num_input_tokens_seen": 562154680,
      "step": 552100
    },
    {
      "epoch": 10.872004882754819,
      "grad_norm": 1.8757035732269287,
      "learning_rate": 1.7551885641289437e-07,
      "loss": 1.9759,
      "num_input_tokens_seen": 562257080,
      "step": 552200
    },
    {
      "epoch": 10.873973735504322,
      "grad_norm": 1.8349326848983765,
      "learning_rate": 1.753620176502122e-07,
      "loss": 2.0018,
      "num_input_tokens_seen": 562359024,
      "step": 552300
    },
    {
      "epoch": 10.875942588253825,
      "grad_norm": 1.9872115850448608,
      "learning_rate": 1.752052340894808e-07,
      "loss": 2.0174,
      "num_input_tokens_seen": 562460808,
      "step": 552400
    },
    {
      "epoch": 10.877911441003327,
      "grad_norm": 1.9421464204788208,
      "learning_rate": 1.7504850575735984e-07,
      "loss": 1.9412,
      "num_input_tokens_seen": 562562624,
      "step": 552500
    },
    {
      "epoch": 10.87988029375283,
      "grad_norm": 1.8458161354064941,
      "learning_rate": 1.7489183268049994e-07,
      "loss": 2.0346,
      "num_input_tokens_seen": 562665024,
      "step": 552600
    },
    {
      "epoch": 10.881849146502333,
      "grad_norm": 2.0116019248962402,
      "learning_rate": 1.747352148855415e-07,
      "loss": 1.9846,
      "num_input_tokens_seen": 562767424,
      "step": 552700
    },
    {
      "epoch": 10.883817999251836,
      "grad_norm": 1.963964819908142,
      "learning_rate": 1.7457865239911663e-07,
      "loss": 1.9838,
      "num_input_tokens_seen": 562869016,
      "step": 552800
    },
    {
      "epoch": 10.885786852001338,
      "grad_norm": 35.831180572509766,
      "learning_rate": 1.7442214524784705e-07,
      "loss": 2.006,
      "num_input_tokens_seen": 562971416,
      "step": 552900
    },
    {
      "epoch": 10.887755704750841,
      "grad_norm": 2.1121113300323486,
      "learning_rate": 1.7426569345834573e-07,
      "loss": 1.9823,
      "num_input_tokens_seen": 563073816,
      "step": 553000
    },
    {
      "epoch": 10.889724557500344,
      "grad_norm": 21.630117416381836,
      "learning_rate": 1.741092970572156e-07,
      "loss": 2.0341,
      "num_input_tokens_seen": 563175296,
      "step": 553100
    },
    {
      "epoch": 10.891693410249847,
      "grad_norm": 1.8024646043777466,
      "learning_rate": 1.7395295607105116e-07,
      "loss": 2.0162,
      "num_input_tokens_seen": 563277328,
      "step": 553200
    },
    {
      "epoch": 10.89366226299935,
      "grad_norm": 2.0604710578918457,
      "learning_rate": 1.7379667052643632e-07,
      "loss": 1.9976,
      "num_input_tokens_seen": 563378568,
      "step": 553300
    },
    {
      "epoch": 10.895631115748854,
      "grad_norm": 2.136566638946533,
      "learning_rate": 1.736404404499463e-07,
      "loss": 1.9893,
      "num_input_tokens_seen": 563480968,
      "step": 553400
    },
    {
      "epoch": 10.897599968498357,
      "grad_norm": 1.924424171447754,
      "learning_rate": 1.734842658681468e-07,
      "loss": 2.036,
      "num_input_tokens_seen": 563581760,
      "step": 553500
    },
    {
      "epoch": 10.89956882124786,
      "grad_norm": 9.791131019592285,
      "learning_rate": 1.7332814680759396e-07,
      "loss": 1.9807,
      "num_input_tokens_seen": 563683200,
      "step": 553600
    },
    {
      "epoch": 10.901537673997362,
      "grad_norm": 1.996576189994812,
      "learning_rate": 1.7317208329483456e-07,
      "loss": 1.95,
      "num_input_tokens_seen": 563784976,
      "step": 553700
    },
    {
      "epoch": 10.903506526746865,
      "grad_norm": 1.685051679611206,
      "learning_rate": 1.7301607535640566e-07,
      "loss": 1.9791,
      "num_input_tokens_seen": 563886624,
      "step": 553800
    },
    {
      "epoch": 10.905475379496368,
      "grad_norm": 2.1437501907348633,
      "learning_rate": 1.7286012301883523e-07,
      "loss": 1.9887,
      "num_input_tokens_seen": 563989024,
      "step": 553900
    },
    {
      "epoch": 10.90744423224587,
      "grad_norm": 2.030673027038574,
      "learning_rate": 1.727042263086417e-07,
      "loss": 2.0081,
      "num_input_tokens_seen": 564089872,
      "step": 554000
    },
    {
      "epoch": 10.909413084995373,
      "grad_norm": 1.8840281963348389,
      "learning_rate": 1.7254838525233406e-07,
      "loss": 1.9486,
      "num_input_tokens_seen": 564192272,
      "step": 554100
    },
    {
      "epoch": 10.911381937744876,
      "grad_norm": 15.175052642822266,
      "learning_rate": 1.7239259987641152e-07,
      "loss": 2.0381,
      "num_input_tokens_seen": 564293720,
      "step": 554200
    },
    {
      "epoch": 10.913350790494379,
      "grad_norm": 1.842354416847229,
      "learning_rate": 1.7223687020736421e-07,
      "loss": 1.9525,
      "num_input_tokens_seen": 564396120,
      "step": 554300
    },
    {
      "epoch": 10.915319643243881,
      "grad_norm": 1.9844493865966797,
      "learning_rate": 1.7208119627167262e-07,
      "loss": 1.9901,
      "num_input_tokens_seen": 564497872,
      "step": 554400
    },
    {
      "epoch": 10.917288495993384,
      "grad_norm": 2.2026803493499756,
      "learning_rate": 1.7192557809580782e-07,
      "loss": 1.9713,
      "num_input_tokens_seen": 564600272,
      "step": 554500
    },
    {
      "epoch": 10.919257348742887,
      "grad_norm": 2.121382713317871,
      "learning_rate": 1.7177001570623152e-07,
      "loss": 1.9652,
      "num_input_tokens_seen": 564701904,
      "step": 554600
    },
    {
      "epoch": 10.92122620149239,
      "grad_norm": 1.7426812648773193,
      "learning_rate": 1.7161450912939546e-07,
      "loss": 1.9955,
      "num_input_tokens_seen": 564803376,
      "step": 554700
    },
    {
      "epoch": 10.923195054241893,
      "grad_norm": 1.8384121656417847,
      "learning_rate": 1.7145905839174241e-07,
      "loss": 1.9527,
      "num_input_tokens_seen": 564905776,
      "step": 554800
    },
    {
      "epoch": 10.925163906991397,
      "grad_norm": 2.107618808746338,
      "learning_rate": 1.7130366351970538e-07,
      "loss": 1.9534,
      "num_input_tokens_seen": 565008176,
      "step": 554900
    },
    {
      "epoch": 10.9271327597409,
      "grad_norm": 2.13130784034729,
      "learning_rate": 1.7114832453970822e-07,
      "loss": 1.9654,
      "num_input_tokens_seen": 565110576,
      "step": 555000
    },
    {
      "epoch": 10.929101612490403,
      "grad_norm": 1.6640058755874634,
      "learning_rate": 1.7099304147816467e-07,
      "loss": 1.9346,
      "num_input_tokens_seen": 565212792,
      "step": 555100
    },
    {
      "epoch": 10.931070465239905,
      "grad_norm": 1.8980098962783813,
      "learning_rate": 1.7083781436147943e-07,
      "loss": 1.969,
      "num_input_tokens_seen": 565315192,
      "step": 555200
    },
    {
      "epoch": 10.933039317989408,
      "grad_norm": 2.0015316009521484,
      "learning_rate": 1.7068264321604753e-07,
      "loss": 2.0155,
      "num_input_tokens_seen": 565415928,
      "step": 555300
    },
    {
      "epoch": 10.93500817073891,
      "grad_norm": 2.2221791744232178,
      "learning_rate": 1.7052752806825477e-07,
      "loss": 2.0495,
      "num_input_tokens_seen": 565517616,
      "step": 555400
    },
    {
      "epoch": 10.936977023488414,
      "grad_norm": 1.7417666912078857,
      "learning_rate": 1.7037246894447677e-07,
      "loss": 1.9904,
      "num_input_tokens_seen": 565619440,
      "step": 555500
    },
    {
      "epoch": 10.938945876237916,
      "grad_norm": 1.8724528551101685,
      "learning_rate": 1.702174658710801e-07,
      "loss": 1.9493,
      "num_input_tokens_seen": 565721296,
      "step": 555600
    },
    {
      "epoch": 10.94091472898742,
      "grad_norm": 1.9935745000839233,
      "learning_rate": 1.7006251887442213e-07,
      "loss": 1.9937,
      "num_input_tokens_seen": 565823608,
      "step": 555700
    },
    {
      "epoch": 10.942883581736922,
      "grad_norm": 1.8080589771270752,
      "learning_rate": 1.6990762798084984e-07,
      "loss": 2.0009,
      "num_input_tokens_seen": 565926008,
      "step": 555800
    },
    {
      "epoch": 10.944852434486425,
      "grad_norm": 2.0000579357147217,
      "learning_rate": 1.6975279321670148e-07,
      "loss": 1.9993,
      "num_input_tokens_seen": 566027696,
      "step": 555900
    },
    {
      "epoch": 10.946821287235927,
      "grad_norm": 1.9233328104019165,
      "learning_rate": 1.6959801460830487e-07,
      "loss": 2.0226,
      "num_input_tokens_seen": 566129400,
      "step": 556000
    },
    {
      "epoch": 10.94879013998543,
      "grad_norm": 1.6917973756790161,
      "learning_rate": 1.6944329218197943e-07,
      "loss": 1.9986,
      "num_input_tokens_seen": 566231272,
      "step": 556100
    },
    {
      "epoch": 10.950758992734933,
      "grad_norm": 1.933417558670044,
      "learning_rate": 1.6928862596403397e-07,
      "loss": 2.003,
      "num_input_tokens_seen": 566333672,
      "step": 556200
    },
    {
      "epoch": 10.952727845484436,
      "grad_norm": 1.990510106086731,
      "learning_rate": 1.6913401598076848e-07,
      "loss": 1.9752,
      "num_input_tokens_seen": 566436072,
      "step": 556300
    },
    {
      "epoch": 10.954696698233938,
      "grad_norm": 1.8237090110778809,
      "learning_rate": 1.6897946225847258e-07,
      "loss": 1.9981,
      "num_input_tokens_seen": 566537920,
      "step": 556400
    },
    {
      "epoch": 10.956665550983441,
      "grad_norm": 1.7059775590896606,
      "learning_rate": 1.6882496482342734e-07,
      "loss": 1.9692,
      "num_input_tokens_seen": 566640320,
      "step": 556500
    },
    {
      "epoch": 10.958634403732944,
      "grad_norm": 1.910999059677124,
      "learning_rate": 1.6867052370190376e-07,
      "loss": 1.9715,
      "num_input_tokens_seen": 566742720,
      "step": 556600
    },
    {
      "epoch": 10.960603256482448,
      "grad_norm": 1.8922381401062012,
      "learning_rate": 1.6851613892016286e-07,
      "loss": 1.9685,
      "num_input_tokens_seen": 566844328,
      "step": 556700
    },
    {
      "epoch": 10.962572109231951,
      "grad_norm": 2.10164475440979,
      "learning_rate": 1.683618105044567e-07,
      "loss": 1.9762,
      "num_input_tokens_seen": 566946600,
      "step": 556800
    },
    {
      "epoch": 10.964540961981454,
      "grad_norm": 1.8269226551055908,
      "learning_rate": 1.6820753848102748e-07,
      "loss": 1.9475,
      "num_input_tokens_seen": 567049000,
      "step": 556900
    },
    {
      "epoch": 10.966509814730957,
      "grad_norm": 1.7695835828781128,
      "learning_rate": 1.6805332287610807e-07,
      "loss": 2.0123,
      "num_input_tokens_seen": 567148664,
      "step": 557000
    },
    {
      "epoch": 10.96847866748046,
      "grad_norm": 1.875718593597412,
      "learning_rate": 1.678991637159211e-07,
      "loss": 2.0617,
      "num_input_tokens_seen": 567250544,
      "step": 557100
    },
    {
      "epoch": 10.970447520229962,
      "grad_norm": 1.7459222078323364,
      "learning_rate": 1.6774506102668035e-07,
      "loss": 1.9657,
      "num_input_tokens_seen": 567352448,
      "step": 557200
    },
    {
      "epoch": 10.972416372979465,
      "grad_norm": 2.0527901649475098,
      "learning_rate": 1.6759101483458954e-07,
      "loss": 1.9856,
      "num_input_tokens_seen": 567453400,
      "step": 557300
    },
    {
      "epoch": 10.974385225728968,
      "grad_norm": 1.7636523246765137,
      "learning_rate": 1.6743702516584323e-07,
      "loss": 2.0378,
      "num_input_tokens_seen": 567553656,
      "step": 557400
    },
    {
      "epoch": 10.97635407847847,
      "grad_norm": 1.9239298105239868,
      "learning_rate": 1.672830920466256e-07,
      "loss": 1.9628,
      "num_input_tokens_seen": 567656056,
      "step": 557500
    },
    {
      "epoch": 10.978322931227973,
      "grad_norm": 1.8326318264007568,
      "learning_rate": 1.6712921550311192e-07,
      "loss": 1.9569,
      "num_input_tokens_seen": 567758456,
      "step": 557600
    },
    {
      "epoch": 10.980291783977476,
      "grad_norm": 1.7801740169525146,
      "learning_rate": 1.6697539556146762e-07,
      "loss": 1.9886,
      "num_input_tokens_seen": 567860680,
      "step": 557700
    },
    {
      "epoch": 10.982260636726979,
      "grad_norm": 1.9315011501312256,
      "learning_rate": 1.668216322478484e-07,
      "loss": 2.0117,
      "num_input_tokens_seen": 567962464,
      "step": 557800
    },
    {
      "epoch": 10.984229489476482,
      "grad_norm": 2.050309419631958,
      "learning_rate": 1.6666792558840058e-07,
      "loss": 1.966,
      "num_input_tokens_seen": 568063992,
      "step": 557900
    },
    {
      "epoch": 10.986198342225984,
      "grad_norm": 1.7155494689941406,
      "learning_rate": 1.6651427560926053e-07,
      "loss": 1.9484,
      "num_input_tokens_seen": 568164104,
      "step": 558000
    },
    {
      "epoch": 10.988167194975487,
      "grad_norm": 3.1228291988372803,
      "learning_rate": 1.6636068233655504e-07,
      "loss": 1.9726,
      "num_input_tokens_seen": 568265776,
      "step": 558100
    },
    {
      "epoch": 10.99013604772499,
      "grad_norm": 1.9787405729293823,
      "learning_rate": 1.6620714579640155e-07,
      "loss": 1.9986,
      "num_input_tokens_seen": 568367792,
      "step": 558200
    },
    {
      "epoch": 10.992104900474494,
      "grad_norm": 1.7454941272735596,
      "learning_rate": 1.6605366601490767e-07,
      "loss": 1.9918,
      "num_input_tokens_seen": 568469384,
      "step": 558300
    },
    {
      "epoch": 10.994073753223997,
      "grad_norm": 2.203538179397583,
      "learning_rate": 1.659002430181712e-07,
      "loss": 2.0342,
      "num_input_tokens_seen": 568571048,
      "step": 558400
    },
    {
      "epoch": 10.9960426059735,
      "grad_norm": 2.1025588512420654,
      "learning_rate": 1.6574687683228016e-07,
      "loss": 2.0571,
      "num_input_tokens_seen": 568671840,
      "step": 558500
    },
    {
      "epoch": 10.998011458723003,
      "grad_norm": 18.16162109375,
      "learning_rate": 1.6559356748331394e-07,
      "loss": 1.9717,
      "num_input_tokens_seen": 568774240,
      "step": 558600
    },
    {
      "epoch": 10.999980311472505,
      "grad_norm": 1.8461054563522339,
      "learning_rate": 1.6544031499734085e-07,
      "loss": 2.0256,
      "num_input_tokens_seen": 568876208,
      "step": 558700
    },
    {
      "epoch": 11.001949164222008,
      "grad_norm": 1.747460126876831,
      "learning_rate": 1.6528711940042056e-07,
      "loss": 1.9573,
      "num_input_tokens_seen": 568978608,
      "step": 558800
    },
    {
      "epoch": 11.003918016971511,
      "grad_norm": 1.770186424255371,
      "learning_rate": 1.651339807186022e-07,
      "loss": 1.944,
      "num_input_tokens_seen": 569081008,
      "step": 558900
    },
    {
      "epoch": 11.005886869721014,
      "grad_norm": 1.6608457565307617,
      "learning_rate": 1.6498089897792634e-07,
      "loss": 2.0115,
      "num_input_tokens_seen": 569183408,
      "step": 559000
    },
    {
      "epoch": 11.007855722470516,
      "grad_norm": 1.6924635171890259,
      "learning_rate": 1.648278742044228e-07,
      "loss": 1.9466,
      "num_input_tokens_seen": 569284776,
      "step": 559100
    },
    {
      "epoch": 11.00982457522002,
      "grad_norm": 1.9252318143844604,
      "learning_rate": 1.646749064241124e-07,
      "loss": 1.9733,
      "num_input_tokens_seen": 569387176,
      "step": 559200
    },
    {
      "epoch": 11.011793427969522,
      "grad_norm": 1.762007474899292,
      "learning_rate": 1.6452199566300557e-07,
      "loss": 1.9397,
      "num_input_tokens_seen": 569489576,
      "step": 559300
    },
    {
      "epoch": 11.013762280719025,
      "grad_norm": 1.9775484800338745,
      "learning_rate": 1.6436914194710416e-07,
      "loss": 1.9486,
      "num_input_tokens_seen": 569591344,
      "step": 559400
    },
    {
      "epoch": 11.015731133468528,
      "grad_norm": 1.9309196472167969,
      "learning_rate": 1.6421634530239913e-07,
      "loss": 2.0292,
      "num_input_tokens_seen": 569693144,
      "step": 559500
    },
    {
      "epoch": 11.01769998621803,
      "grad_norm": 2.1214234828948975,
      "learning_rate": 1.6406360575487254e-07,
      "loss": 2.0163,
      "num_input_tokens_seen": 569793808,
      "step": 559600
    },
    {
      "epoch": 11.019668838967533,
      "grad_norm": 1.9098743200302124,
      "learning_rate": 1.6391092333049627e-07,
      "loss": 1.9609,
      "num_input_tokens_seen": 569894256,
      "step": 559700
    },
    {
      "epoch": 11.021637691717036,
      "grad_norm": 2.009169340133667,
      "learning_rate": 1.6375829805523285e-07,
      "loss": 2.0057,
      "num_input_tokens_seen": 569996112,
      "step": 559800
    },
    {
      "epoch": 11.023606544466539,
      "grad_norm": 2.2146265506744385,
      "learning_rate": 1.6360572995503503e-07,
      "loss": 1.9741,
      "num_input_tokens_seen": 570098512,
      "step": 559900
    },
    {
      "epoch": 11.025575397216043,
      "grad_norm": 2.114055633544922,
      "learning_rate": 1.634532190558453e-07,
      "loss": 1.9892,
      "num_input_tokens_seen": 570200912,
      "step": 560000
    },
    {
      "epoch": 11.027544249965546,
      "grad_norm": 1.9288376569747925,
      "learning_rate": 1.6330076538359717e-07,
      "loss": 2.0215,
      "num_input_tokens_seen": 570302456,
      "step": 560100
    },
    {
      "epoch": 11.029513102715049,
      "grad_norm": 1.7856731414794922,
      "learning_rate": 1.63148368964214e-07,
      "loss": 1.938,
      "num_input_tokens_seen": 570404128,
      "step": 560200
    },
    {
      "epoch": 11.031481955464551,
      "grad_norm": 2.039823293685913,
      "learning_rate": 1.6299602982360977e-07,
      "loss": 1.9825,
      "num_input_tokens_seen": 570504216,
      "step": 560300
    },
    {
      "epoch": 11.033450808214054,
      "grad_norm": 31.957164764404297,
      "learning_rate": 1.6284374798768802e-07,
      "loss": 2.0262,
      "num_input_tokens_seen": 570605856,
      "step": 560400
    },
    {
      "epoch": 11.035419660963557,
      "grad_norm": 1.9994714260101318,
      "learning_rate": 1.6269152348234328e-07,
      "loss": 1.9857,
      "num_input_tokens_seen": 570706544,
      "step": 560500
    },
    {
      "epoch": 11.03738851371306,
      "grad_norm": 2.319862127304077,
      "learning_rate": 1.6253935633345995e-07,
      "loss": 1.9742,
      "num_input_tokens_seen": 570808264,
      "step": 560600
    },
    {
      "epoch": 11.039357366462562,
      "grad_norm": 1.614006519317627,
      "learning_rate": 1.6238724656691277e-07,
      "loss": 1.9635,
      "num_input_tokens_seen": 570910664,
      "step": 560700
    },
    {
      "epoch": 11.041326219212065,
      "grad_norm": 1.8084412813186646,
      "learning_rate": 1.6223519420856697e-07,
      "loss": 1.9952,
      "num_input_tokens_seen": 571011392,
      "step": 560800
    },
    {
      "epoch": 11.043295071961568,
      "grad_norm": 1.8977243900299072,
      "learning_rate": 1.6208319928427732e-07,
      "loss": 1.9233,
      "num_input_tokens_seen": 571113056,
      "step": 560900
    },
    {
      "epoch": 11.04526392471107,
      "grad_norm": 1.9307817220687866,
      "learning_rate": 1.6193126181988958e-07,
      "loss": 2.0118,
      "num_input_tokens_seen": 571214160,
      "step": 561000
    },
    {
      "epoch": 11.047232777460573,
      "grad_norm": 2.055455207824707,
      "learning_rate": 1.6177938184123928e-07,
      "loss": 1.9679,
      "num_input_tokens_seen": 571315792,
      "step": 561100
    },
    {
      "epoch": 11.049201630210076,
      "grad_norm": 1.7375178337097168,
      "learning_rate": 1.6162755937415268e-07,
      "loss": 1.9897,
      "num_input_tokens_seen": 571416736,
      "step": 561200
    },
    {
      "epoch": 11.051170482959579,
      "grad_norm": 1.7575231790542603,
      "learning_rate": 1.6147579444444526e-07,
      "loss": 1.978,
      "num_input_tokens_seen": 571518296,
      "step": 561300
    },
    {
      "epoch": 11.053139335709082,
      "grad_norm": 1.8998470306396484,
      "learning_rate": 1.6132408707792411e-07,
      "loss": 2.0047,
      "num_input_tokens_seen": 571619800,
      "step": 561400
    },
    {
      "epoch": 11.055108188458584,
      "grad_norm": 2.5749189853668213,
      "learning_rate": 1.611724373003852e-07,
      "loss": 1.9374,
      "num_input_tokens_seen": 571721408,
      "step": 561500
    },
    {
      "epoch": 11.057077041208087,
      "grad_norm": 1.9108294248580933,
      "learning_rate": 1.610208451376156e-07,
      "loss": 1.9598,
      "num_input_tokens_seen": 571822328,
      "step": 561600
    },
    {
      "epoch": 11.059045893957592,
      "grad_norm": 2.0555667877197266,
      "learning_rate": 1.6086931061539217e-07,
      "loss": 1.9683,
      "num_input_tokens_seen": 571923528,
      "step": 561700
    },
    {
      "epoch": 11.061014746707095,
      "grad_norm": 2.043241024017334,
      "learning_rate": 1.607178337594821e-07,
      "loss": 1.9853,
      "num_input_tokens_seen": 572025928,
      "step": 561800
    },
    {
      "epoch": 11.062983599456597,
      "grad_norm": 1.7151275873184204,
      "learning_rate": 1.6056641459564292e-07,
      "loss": 1.9622,
      "num_input_tokens_seen": 572128328,
      "step": 561900
    },
    {
      "epoch": 11.0649524522061,
      "grad_norm": 1.8281558752059937,
      "learning_rate": 1.604150531496219e-07,
      "loss": 1.9895,
      "num_input_tokens_seen": 572230728,
      "step": 562000
    },
    {
      "epoch": 11.066921304955603,
      "grad_norm": 2.052183151245117,
      "learning_rate": 1.6026374944715682e-07,
      "loss": 1.9744,
      "num_input_tokens_seen": 572333128,
      "step": 562100
    },
    {
      "epoch": 11.068890157705106,
      "grad_norm": 2.437272548675537,
      "learning_rate": 1.6011250351397575e-07,
      "loss": 1.952,
      "num_input_tokens_seen": 572435528,
      "step": 562200
    },
    {
      "epoch": 11.070859010454608,
      "grad_norm": 1.8878471851348877,
      "learning_rate": 1.5996131537579688e-07,
      "loss": 1.9921,
      "num_input_tokens_seen": 572537136,
      "step": 562300
    },
    {
      "epoch": 11.072827863204111,
      "grad_norm": 1.868731141090393,
      "learning_rate": 1.598101850583281e-07,
      "loss": 1.965,
      "num_input_tokens_seen": 572639536,
      "step": 562400
    },
    {
      "epoch": 11.074796715953614,
      "grad_norm": 1.9930466413497925,
      "learning_rate": 1.596591125872681e-07,
      "loss": 2.0186,
      "num_input_tokens_seen": 572741464,
      "step": 562500
    },
    {
      "epoch": 11.076765568703117,
      "grad_norm": 2.220581293106079,
      "learning_rate": 1.595080979883054e-07,
      "loss": 2.0402,
      "num_input_tokens_seen": 572843864,
      "step": 562600
    },
    {
      "epoch": 11.07873442145262,
      "grad_norm": 1.8037691116333008,
      "learning_rate": 1.5935714128711886e-07,
      "loss": 1.9514,
      "num_input_tokens_seen": 572946264,
      "step": 562700
    },
    {
      "epoch": 11.080703274202122,
      "grad_norm": 1.759264588356018,
      "learning_rate": 1.5920624250937743e-07,
      "loss": 1.9852,
      "num_input_tokens_seen": 573048664,
      "step": 562800
    },
    {
      "epoch": 11.082672126951625,
      "grad_norm": 2.2461397647857666,
      "learning_rate": 1.5905540168073993e-07,
      "loss": 1.95,
      "num_input_tokens_seen": 573151064,
      "step": 562900
    },
    {
      "epoch": 11.084640979701128,
      "grad_norm": 1.8784043788909912,
      "learning_rate": 1.589046188268557e-07,
      "loss": 1.9567,
      "num_input_tokens_seen": 573253464,
      "step": 563000
    },
    {
      "epoch": 11.08660983245063,
      "grad_norm": 2.1749541759490967,
      "learning_rate": 1.5875389397336413e-07,
      "loss": 1.9779,
      "num_input_tokens_seen": 573355048,
      "step": 563100
    },
    {
      "epoch": 11.088578685200133,
      "grad_norm": 2.208847999572754,
      "learning_rate": 1.5860322714589487e-07,
      "loss": 1.9522,
      "num_input_tokens_seen": 573456920,
      "step": 563200
    },
    {
      "epoch": 11.090547537949636,
      "grad_norm": 1.80336332321167,
      "learning_rate": 1.5845261837006712e-07,
      "loss": 1.9963,
      "num_input_tokens_seen": 573559216,
      "step": 563300
    },
    {
      "epoch": 11.09251639069914,
      "grad_norm": 1.8268429040908813,
      "learning_rate": 1.5830206767149097e-07,
      "loss": 1.994,
      "num_input_tokens_seen": 573661616,
      "step": 563400
    },
    {
      "epoch": 11.094485243448643,
      "grad_norm": 1.6226472854614258,
      "learning_rate": 1.581515750757661e-07,
      "loss": 1.993,
      "num_input_tokens_seen": 573761264,
      "step": 563500
    },
    {
      "epoch": 11.096454096198146,
      "grad_norm": 1.9020010232925415,
      "learning_rate": 1.5800114060848268e-07,
      "loss": 1.9719,
      "num_input_tokens_seen": 573863528,
      "step": 563600
    },
    {
      "epoch": 11.098422948947649,
      "grad_norm": 2.0572657585144043,
      "learning_rate": 1.5785076429522089e-07,
      "loss": 1.9866,
      "num_input_tokens_seen": 573964744,
      "step": 563700
    },
    {
      "epoch": 11.100391801697151,
      "grad_norm": 2.0211639404296875,
      "learning_rate": 1.5770044616155048e-07,
      "loss": 1.9711,
      "num_input_tokens_seen": 574067144,
      "step": 563800
    },
    {
      "epoch": 11.102360654446654,
      "grad_norm": 1.8248523473739624,
      "learning_rate": 1.5755018623303247e-07,
      "loss": 1.9967,
      "num_input_tokens_seen": 574169232,
      "step": 563900
    },
    {
      "epoch": 11.104329507196157,
      "grad_norm": 1.644024133682251,
      "learning_rate": 1.5739998453521668e-07,
      "loss": 1.9236,
      "num_input_tokens_seen": 574271632,
      "step": 564000
    },
    {
      "epoch": 11.10629835994566,
      "grad_norm": 2.086744785308838,
      "learning_rate": 1.5724984109364414e-07,
      "loss": 1.9722,
      "num_input_tokens_seen": 574373296,
      "step": 564100
    },
    {
      "epoch": 11.108267212695162,
      "grad_norm": 2.0320801734924316,
      "learning_rate": 1.5709975593384489e-07,
      "loss": 1.9822,
      "num_input_tokens_seen": 574475408,
      "step": 564200
    },
    {
      "epoch": 11.110236065444665,
      "grad_norm": 2.0280330181121826,
      "learning_rate": 1.5694972908134034e-07,
      "loss": 2.032,
      "num_input_tokens_seen": 574576424,
      "step": 564300
    },
    {
      "epoch": 11.112204918194168,
      "grad_norm": 1.6259900331497192,
      "learning_rate": 1.5679976056164074e-07,
      "loss": 2.0213,
      "num_input_tokens_seen": 574678512,
      "step": 564400
    },
    {
      "epoch": 11.11417377094367,
      "grad_norm": 1.9069445133209229,
      "learning_rate": 1.5664985040024737e-07,
      "loss": 1.9384,
      "num_input_tokens_seen": 574779736,
      "step": 564500
    },
    {
      "epoch": 11.116142623693174,
      "grad_norm": 1.8923521041870117,
      "learning_rate": 1.564999986226506e-07,
      "loss": 2.0487,
      "num_input_tokens_seen": 574881512,
      "step": 564600
    },
    {
      "epoch": 11.118111476442676,
      "grad_norm": 2.391019582748413,
      "learning_rate": 1.5635020525433207e-07,
      "loss": 1.9697,
      "num_input_tokens_seen": 574983912,
      "step": 564700
    },
    {
      "epoch": 11.120080329192179,
      "grad_norm": 2.0226306915283203,
      "learning_rate": 1.5620047032076282e-07,
      "loss": 1.9658,
      "num_input_tokens_seen": 575086312,
      "step": 564800
    },
    {
      "epoch": 11.122049181941682,
      "grad_norm": 2.0029852390289307,
      "learning_rate": 1.5605079384740373e-07,
      "loss": 2.0302,
      "num_input_tokens_seen": 575188712,
      "step": 564900
    },
    {
      "epoch": 11.124018034691186,
      "grad_norm": 2.1440653800964355,
      "learning_rate": 1.5590117585970608e-07,
      "loss": 1.9845,
      "num_input_tokens_seen": 575291112,
      "step": 565000
    },
    {
      "epoch": 11.125986887440689,
      "grad_norm": 1.9851350784301758,
      "learning_rate": 1.5575161638311124e-07,
      "loss": 1.9315,
      "num_input_tokens_seen": 575393512,
      "step": 565100
    },
    {
      "epoch": 11.127955740190192,
      "grad_norm": 2.07045316696167,
      "learning_rate": 1.5560211544305063e-07,
      "loss": 1.9842,
      "num_input_tokens_seen": 575495392,
      "step": 565200
    },
    {
      "epoch": 11.129924592939695,
      "grad_norm": 2.118616819381714,
      "learning_rate": 1.5545267306494537e-07,
      "loss": 1.9941,
      "num_input_tokens_seen": 575597792,
      "step": 565300
    },
    {
      "epoch": 11.131893445689197,
      "grad_norm": 1.6898269653320312,
      "learning_rate": 1.5530328927420705e-07,
      "loss": 1.998,
      "num_input_tokens_seen": 575699280,
      "step": 565400
    },
    {
      "epoch": 11.1338622984387,
      "grad_norm": 1.656571865081787,
      "learning_rate": 1.5515396409623704e-07,
      "loss": 1.9904,
      "num_input_tokens_seen": 575799888,
      "step": 565500
    },
    {
      "epoch": 11.135831151188203,
      "grad_norm": 2.069849967956543,
      "learning_rate": 1.5500469755642702e-07,
      "loss": 1.9609,
      "num_input_tokens_seen": 575902288,
      "step": 565600
    },
    {
      "epoch": 11.137800003937706,
      "grad_norm": 2.019660234451294,
      "learning_rate": 1.548554896801582e-07,
      "loss": 1.9981,
      "num_input_tokens_seen": 576002808,
      "step": 565700
    },
    {
      "epoch": 11.139768856687208,
      "grad_norm": 1.7920044660568237,
      "learning_rate": 1.5470634049280228e-07,
      "loss": 2.0262,
      "num_input_tokens_seen": 576104288,
      "step": 565800
    },
    {
      "epoch": 11.141737709436711,
      "grad_norm": 2.0878922939300537,
      "learning_rate": 1.545572500197208e-07,
      "loss": 2.0091,
      "num_input_tokens_seen": 576206136,
      "step": 565900
    },
    {
      "epoch": 11.143706562186214,
      "grad_norm": 1.9078084230422974,
      "learning_rate": 1.5440821828626538e-07,
      "loss": 1.9671,
      "num_input_tokens_seen": 576308536,
      "step": 566000
    },
    {
      "epoch": 11.145675414935717,
      "grad_norm": 1.8267896175384521,
      "learning_rate": 1.542592453177778e-07,
      "loss": 1.98,
      "num_input_tokens_seen": 576410368,
      "step": 566100
    },
    {
      "epoch": 11.14764426768522,
      "grad_norm": 2.090945243835449,
      "learning_rate": 1.541103311395892e-07,
      "loss": 1.9989,
      "num_input_tokens_seen": 576511456,
      "step": 566200
    },
    {
      "epoch": 11.149613120434722,
      "grad_norm": 2.000983476638794,
      "learning_rate": 1.5396147577702145e-07,
      "loss": 1.9486,
      "num_input_tokens_seen": 576612360,
      "step": 566300
    },
    {
      "epoch": 11.151581973184225,
      "grad_norm": 1.9991554021835327,
      "learning_rate": 1.538126792553861e-07,
      "loss": 2.0102,
      "num_input_tokens_seen": 576713904,
      "step": 566400
    },
    {
      "epoch": 11.153550825933728,
      "grad_norm": 1.842217206954956,
      "learning_rate": 1.5366394159998498e-07,
      "loss": 1.9764,
      "num_input_tokens_seen": 576815784,
      "step": 566500
    },
    {
      "epoch": 11.15551967868323,
      "grad_norm": 1.7024178504943848,
      "learning_rate": 1.5351526283610922e-07,
      "loss": 1.9892,
      "num_input_tokens_seen": 576917360,
      "step": 566600
    },
    {
      "epoch": 11.157488531432735,
      "grad_norm": 1.959771990776062,
      "learning_rate": 1.533666429890405e-07,
      "loss": 1.9749,
      "num_input_tokens_seen": 577019760,
      "step": 566700
    },
    {
      "epoch": 11.159457384182238,
      "grad_norm": 2.245302200317383,
      "learning_rate": 1.5321808208405084e-07,
      "loss": 1.9923,
      "num_input_tokens_seen": 577121680,
      "step": 566800
    },
    {
      "epoch": 11.16142623693174,
      "grad_norm": 1.8373440504074097,
      "learning_rate": 1.5306958014640125e-07,
      "loss": 2.0039,
      "num_input_tokens_seen": 577224080,
      "step": 566900
    },
    {
      "epoch": 11.163395089681243,
      "grad_norm": 1.9024267196655273,
      "learning_rate": 1.529211372013436e-07,
      "loss": 1.9718,
      "num_input_tokens_seen": 577326480,
      "step": 567000
    },
    {
      "epoch": 11.165363942430746,
      "grad_norm": 1.8979979753494263,
      "learning_rate": 1.527727532741188e-07,
      "loss": 1.976,
      "num_input_tokens_seen": 577428200,
      "step": 567100
    },
    {
      "epoch": 11.167332795180249,
      "grad_norm": 1.8532240390777588,
      "learning_rate": 1.5262442838995903e-07,
      "loss": 1.9921,
      "num_input_tokens_seen": 577530600,
      "step": 567200
    },
    {
      "epoch": 11.169301647929752,
      "grad_norm": 2.6826493740081787,
      "learning_rate": 1.5247616257408508e-07,
      "loss": 1.9831,
      "num_input_tokens_seen": 577631408,
      "step": 567300
    },
    {
      "epoch": 11.171270500679254,
      "grad_norm": 1.8805820941925049,
      "learning_rate": 1.5232795585170876e-07,
      "loss": 1.9984,
      "num_input_tokens_seen": 577731176,
      "step": 567400
    },
    {
      "epoch": 11.173239353428757,
      "grad_norm": 2.616795539855957,
      "learning_rate": 1.5217980824803078e-07,
      "loss": 1.9783,
      "num_input_tokens_seen": 577833576,
      "step": 567500
    },
    {
      "epoch": 11.17520820617826,
      "grad_norm": 1.8681564331054688,
      "learning_rate": 1.5203171978824315e-07,
      "loss": 2.0714,
      "num_input_tokens_seen": 577935976,
      "step": 567600
    },
    {
      "epoch": 11.177177058927763,
      "grad_norm": 2.055142402648926,
      "learning_rate": 1.5188369049752643e-07,
      "loss": 1.951,
      "num_input_tokens_seen": 578038376,
      "step": 567700
    },
    {
      "epoch": 11.179145911677265,
      "grad_norm": 2.36122727394104,
      "learning_rate": 1.51735720401052e-07,
      "loss": 1.9677,
      "num_input_tokens_seen": 578140776,
      "step": 567800
    },
    {
      "epoch": 11.181114764426768,
      "grad_norm": 2.0014007091522217,
      "learning_rate": 1.5158780952398092e-07,
      "loss": 2.0331,
      "num_input_tokens_seen": 578243176,
      "step": 567900
    },
    {
      "epoch": 11.18308361717627,
      "grad_norm": 1.9691665172576904,
      "learning_rate": 1.5143995789146423e-07,
      "loss": 2.016,
      "num_input_tokens_seen": 578344376,
      "step": 568000
    },
    {
      "epoch": 11.185052469925774,
      "grad_norm": 2.918562412261963,
      "learning_rate": 1.5129216552864294e-07,
      "loss": 1.9612,
      "num_input_tokens_seen": 578446216,
      "step": 568100
    },
    {
      "epoch": 11.187021322675276,
      "grad_norm": 1.950661540031433,
      "learning_rate": 1.5114443246064762e-07,
      "loss": 1.9729,
      "num_input_tokens_seen": 578548616,
      "step": 568200
    },
    {
      "epoch": 11.18899017542478,
      "grad_norm": 1.8653844594955444,
      "learning_rate": 1.5099675871259914e-07,
      "loss": 1.9807,
      "num_input_tokens_seen": 578650312,
      "step": 568300
    },
    {
      "epoch": 11.190959028174284,
      "grad_norm": 1.8881200551986694,
      "learning_rate": 1.5084914430960827e-07,
      "loss": 1.978,
      "num_input_tokens_seen": 578752168,
      "step": 568400
    },
    {
      "epoch": 11.192927880923786,
      "grad_norm": 1.8521870374679565,
      "learning_rate": 1.5070158927677568e-07,
      "loss": 1.9887,
      "num_input_tokens_seen": 578853584,
      "step": 568500
    },
    {
      "epoch": 11.19489673367329,
      "grad_norm": 1.9622387886047363,
      "learning_rate": 1.5055409363919153e-07,
      "loss": 1.9914,
      "num_input_tokens_seen": 578955536,
      "step": 568600
    },
    {
      "epoch": 11.196865586422792,
      "grad_norm": 1.9085460901260376,
      "learning_rate": 1.5040665742193636e-07,
      "loss": 2.0241,
      "num_input_tokens_seen": 579057936,
      "step": 568700
    },
    {
      "epoch": 11.198834439172295,
      "grad_norm": 2.0087790489196777,
      "learning_rate": 1.5025928065008053e-07,
      "loss": 1.989,
      "num_input_tokens_seen": 579160336,
      "step": 568800
    },
    {
      "epoch": 11.200803291921797,
      "grad_norm": 2.1040682792663574,
      "learning_rate": 1.501119633486842e-07,
      "loss": 1.9584,
      "num_input_tokens_seen": 579262048,
      "step": 568900
    },
    {
      "epoch": 11.2027721446713,
      "grad_norm": 1.6661858558654785,
      "learning_rate": 1.4996470554279755e-07,
      "loss": 1.9962,
      "num_input_tokens_seen": 579364360,
      "step": 569000
    },
    {
      "epoch": 11.204740997420803,
      "grad_norm": 2.2188713550567627,
      "learning_rate": 1.4981750725746023e-07,
      "loss": 2.0264,
      "num_input_tokens_seen": 579465560,
      "step": 569100
    },
    {
      "epoch": 11.206709850170306,
      "grad_norm": 2.0795061588287354,
      "learning_rate": 1.496703685177022e-07,
      "loss": 1.9811,
      "num_input_tokens_seen": 579567960,
      "step": 569200
    },
    {
      "epoch": 11.208678702919809,
      "grad_norm": 1.8177028894424438,
      "learning_rate": 1.495232893485432e-07,
      "loss": 2.0287,
      "num_input_tokens_seen": 579670360,
      "step": 569300
    },
    {
      "epoch": 11.210647555669311,
      "grad_norm": 7.807389259338379,
      "learning_rate": 1.4937626977499295e-07,
      "loss": 1.982,
      "num_input_tokens_seen": 579771912,
      "step": 569400
    },
    {
      "epoch": 11.212616408418814,
      "grad_norm": 1.7488486766815186,
      "learning_rate": 1.4922930982205045e-07,
      "loss": 1.9384,
      "num_input_tokens_seen": 579874312,
      "step": 569500
    },
    {
      "epoch": 11.214585261168317,
      "grad_norm": 1.8570404052734375,
      "learning_rate": 1.490824095147055e-07,
      "loss": 2.005,
      "num_input_tokens_seen": 579976712,
      "step": 569600
    },
    {
      "epoch": 11.21655411391782,
      "grad_norm": 1.8749765157699585,
      "learning_rate": 1.4893556887793702e-07,
      "loss": 2.0216,
      "num_input_tokens_seen": 580077800,
      "step": 569700
    },
    {
      "epoch": 11.218522966667322,
      "grad_norm": 1.9788738489151,
      "learning_rate": 1.487887879367139e-07,
      "loss": 2.024,
      "num_input_tokens_seen": 580180200,
      "step": 569800
    },
    {
      "epoch": 11.220491819416825,
      "grad_norm": 2.037383794784546,
      "learning_rate": 1.4864206671599522e-07,
      "loss": 2.033,
      "num_input_tokens_seen": 580282600,
      "step": 569900
    },
    {
      "epoch": 11.222460672166328,
      "grad_norm": 1.9531891345977783,
      "learning_rate": 1.4849540524072956e-07,
      "loss": 2.0203,
      "num_input_tokens_seen": 580384776,
      "step": 570000
    },
    {
      "epoch": 11.224429524915832,
      "grad_norm": 2.8155910968780518,
      "learning_rate": 1.4834880353585573e-07,
      "loss": 1.9994,
      "num_input_tokens_seen": 580486544,
      "step": 570100
    },
    {
      "epoch": 11.226398377665335,
      "grad_norm": 1.8555750846862793,
      "learning_rate": 1.4820226162630168e-07,
      "loss": 1.9608,
      "num_input_tokens_seen": 580588184,
      "step": 570200
    },
    {
      "epoch": 11.228367230414838,
      "grad_norm": 4.470939636230469,
      "learning_rate": 1.4805577953698589e-07,
      "loss": 1.9514,
      "num_input_tokens_seen": 580690584,
      "step": 570300
    },
    {
      "epoch": 11.23033608316434,
      "grad_norm": 1.8449416160583496,
      "learning_rate": 1.4790935729281627e-07,
      "loss": 1.9708,
      "num_input_tokens_seen": 580792984,
      "step": 570400
    },
    {
      "epoch": 11.232304935913843,
      "grad_norm": 1.840404987335205,
      "learning_rate": 1.4776299491869104e-07,
      "loss": 2.0117,
      "num_input_tokens_seen": 580894040,
      "step": 570500
    },
    {
      "epoch": 11.234273788663346,
      "grad_norm": 1.9039394855499268,
      "learning_rate": 1.4761669243949738e-07,
      "loss": 1.9612,
      "num_input_tokens_seen": 580995568,
      "step": 570600
    },
    {
      "epoch": 11.236242641412849,
      "grad_norm": 1.957252860069275,
      "learning_rate": 1.474704498801131e-07,
      "loss": 1.9442,
      "num_input_tokens_seen": 581097968,
      "step": 570700
    },
    {
      "epoch": 11.238211494162352,
      "grad_norm": 1.939706802368164,
      "learning_rate": 1.4732426726540538e-07,
      "loss": 1.986,
      "num_input_tokens_seen": 581200368,
      "step": 570800
    },
    {
      "epoch": 11.240180346911854,
      "grad_norm": 1.6529273986816406,
      "learning_rate": 1.4717814462023142e-07,
      "loss": 1.9895,
      "num_input_tokens_seen": 581302768,
      "step": 570900
    },
    {
      "epoch": 11.242149199661357,
      "grad_norm": 1.947439432144165,
      "learning_rate": 1.4703208196943833e-07,
      "loss": 1.9941,
      "num_input_tokens_seen": 581404080,
      "step": 571000
    },
    {
      "epoch": 11.24411805241086,
      "grad_norm": 1.882500171661377,
      "learning_rate": 1.4688607933786247e-07,
      "loss": 1.9897,
      "num_input_tokens_seen": 581505192,
      "step": 571100
    },
    {
      "epoch": 11.246086905160363,
      "grad_norm": 2.2510392665863037,
      "learning_rate": 1.4674013675033047e-07,
      "loss": 2.0142,
      "num_input_tokens_seen": 581607592,
      "step": 571200
    },
    {
      "epoch": 11.248055757909865,
      "grad_norm": 1.9028221368789673,
      "learning_rate": 1.465942542316587e-07,
      "loss": 1.9901,
      "num_input_tokens_seen": 581708536,
      "step": 571300
    },
    {
      "epoch": 11.250024610659368,
      "grad_norm": 2.0776336193084717,
      "learning_rate": 1.464484318066534e-07,
      "loss": 2.0462,
      "num_input_tokens_seen": 581810296,
      "step": 571400
    },
    {
      "epoch": 11.251993463408871,
      "grad_norm": 5.531641006469727,
      "learning_rate": 1.4630266950011012e-07,
      "loss": 1.995,
      "num_input_tokens_seen": 581910408,
      "step": 571500
    },
    {
      "epoch": 11.253962316158374,
      "grad_norm": 1.7137070894241333,
      "learning_rate": 1.461569673368147e-07,
      "loss": 1.9868,
      "num_input_tokens_seen": 582012000,
      "step": 571600
    },
    {
      "epoch": 11.255931168907878,
      "grad_norm": 1.8212003707885742,
      "learning_rate": 1.460113253415425e-07,
      "loss": 1.9675,
      "num_input_tokens_seen": 582113920,
      "step": 571700
    },
    {
      "epoch": 11.257900021657381,
      "grad_norm": 2.0662999153137207,
      "learning_rate": 1.458657435390588e-07,
      "loss": 1.9635,
      "num_input_tokens_seen": 582216320,
      "step": 571800
    },
    {
      "epoch": 11.259868874406884,
      "grad_norm": 3.0008654594421387,
      "learning_rate": 1.457202219541186e-07,
      "loss": 1.9334,
      "num_input_tokens_seen": 582317976,
      "step": 571900
    },
    {
      "epoch": 11.261837727156387,
      "grad_norm": 1.8614851236343384,
      "learning_rate": 1.4557476061146624e-07,
      "loss": 1.9132,
      "num_input_tokens_seen": 582419872,
      "step": 572000
    },
    {
      "epoch": 11.26380657990589,
      "grad_norm": 2.2235891819000244,
      "learning_rate": 1.4542935953583673e-07,
      "loss": 1.9686,
      "num_input_tokens_seen": 582521016,
      "step": 572100
    },
    {
      "epoch": 11.265775432655392,
      "grad_norm": 2.6485838890075684,
      "learning_rate": 1.4528401875195395e-07,
      "loss": 1.9833,
      "num_input_tokens_seen": 582623416,
      "step": 572200
    },
    {
      "epoch": 11.267744285404895,
      "grad_norm": 1.9150540828704834,
      "learning_rate": 1.4513873828453215e-07,
      "loss": 1.9421,
      "num_input_tokens_seen": 582725088,
      "step": 572300
    },
    {
      "epoch": 11.269713138154398,
      "grad_norm": 2.054717779159546,
      "learning_rate": 1.449935181582745e-07,
      "loss": 1.9725,
      "num_input_tokens_seen": 582826864,
      "step": 572400
    },
    {
      "epoch": 11.2716819909039,
      "grad_norm": 2.01460337638855,
      "learning_rate": 1.4484835839787512e-07,
      "loss": 2.0038,
      "num_input_tokens_seen": 582929264,
      "step": 572500
    },
    {
      "epoch": 11.273650843653403,
      "grad_norm": 1.9440685510635376,
      "learning_rate": 1.4470325902801684e-07,
      "loss": 1.9878,
      "num_input_tokens_seen": 583030960,
      "step": 572600
    },
    {
      "epoch": 11.275619696402906,
      "grad_norm": 1.9856462478637695,
      "learning_rate": 1.4455822007337271e-07,
      "loss": 2.0348,
      "num_input_tokens_seen": 583131856,
      "step": 572700
    },
    {
      "epoch": 11.277588549152409,
      "grad_norm": 2.501225233078003,
      "learning_rate": 1.4441324155860506e-07,
      "loss": 1.9922,
      "num_input_tokens_seen": 583233440,
      "step": 572800
    },
    {
      "epoch": 11.279557401901911,
      "grad_norm": 2.163364887237549,
      "learning_rate": 1.4426832350836671e-07,
      "loss": 2.0571,
      "num_input_tokens_seen": 583335224,
      "step": 572900
    },
    {
      "epoch": 11.281526254651414,
      "grad_norm": 1.930635690689087,
      "learning_rate": 1.4412346594729967e-07,
      "loss": 1.9999,
      "num_input_tokens_seen": 583437624,
      "step": 573000
    },
    {
      "epoch": 11.283495107400917,
      "grad_norm": 1.994875192642212,
      "learning_rate": 1.439786689000355e-07,
      "loss": 1.9369,
      "num_input_tokens_seen": 583540024,
      "step": 573100
    },
    {
      "epoch": 11.28546396015042,
      "grad_norm": 2.055906057357788,
      "learning_rate": 1.4383393239119585e-07,
      "loss": 1.9874,
      "num_input_tokens_seen": 583641720,
      "step": 573200
    },
    {
      "epoch": 11.287432812899922,
      "grad_norm": 2.022674798965454,
      "learning_rate": 1.4368925644539199e-07,
      "loss": 1.9549,
      "num_input_tokens_seen": 583744120,
      "step": 573300
    },
    {
      "epoch": 11.289401665649427,
      "grad_norm": 1.8915902376174927,
      "learning_rate": 1.4354464108722497e-07,
      "loss": 1.9397,
      "num_input_tokens_seen": 583846520,
      "step": 573400
    },
    {
      "epoch": 11.29137051839893,
      "grad_norm": 2.067164659500122,
      "learning_rate": 1.4340008634128503e-07,
      "loss": 1.9893,
      "num_input_tokens_seen": 583948448,
      "step": 573500
    },
    {
      "epoch": 11.293339371148432,
      "grad_norm": 1.889007329940796,
      "learning_rate": 1.4325559223215278e-07,
      "loss": 1.9454,
      "num_input_tokens_seen": 584050848,
      "step": 573600
    },
    {
      "epoch": 11.295308223897935,
      "grad_norm": 1.9478282928466797,
      "learning_rate": 1.4311115878439812e-07,
      "loss": 2.02,
      "num_input_tokens_seen": 584153248,
      "step": 573700
    },
    {
      "epoch": 11.297277076647438,
      "grad_norm": 1.7752412557601929,
      "learning_rate": 1.429667860225807e-07,
      "loss": 1.997,
      "num_input_tokens_seen": 584255008,
      "step": 573800
    },
    {
      "epoch": 11.29924592939694,
      "grad_norm": 2.0746665000915527,
      "learning_rate": 1.428224739712502e-07,
      "loss": 1.9901,
      "num_input_tokens_seen": 584356096,
      "step": 573900
    },
    {
      "epoch": 11.301214782146443,
      "grad_norm": 2.286505699157715,
      "learning_rate": 1.426782226549453e-07,
      "loss": 1.9713,
      "num_input_tokens_seen": 584457408,
      "step": 574000
    },
    {
      "epoch": 11.303183634895946,
      "grad_norm": 1.9341868162155151,
      "learning_rate": 1.4253403209819476e-07,
      "loss": 1.9918,
      "num_input_tokens_seen": 584559808,
      "step": 574100
    },
    {
      "epoch": 11.305152487645449,
      "grad_norm": 1.6348973512649536,
      "learning_rate": 1.4238990232551713e-07,
      "loss": 1.9851,
      "num_input_tokens_seen": 584661440,
      "step": 574200
    },
    {
      "epoch": 11.307121340394952,
      "grad_norm": 2.006460428237915,
      "learning_rate": 1.4224583336142043e-07,
      "loss": 1.994,
      "num_input_tokens_seen": 584762864,
      "step": 574300
    },
    {
      "epoch": 11.309090193144455,
      "grad_norm": 1.9484412670135498,
      "learning_rate": 1.4210182523040226e-07,
      "loss": 2.0,
      "num_input_tokens_seen": 584865264,
      "step": 574400
    },
    {
      "epoch": 11.311059045893957,
      "grad_norm": 1.9613430500030518,
      "learning_rate": 1.4195787795695007e-07,
      "loss": 1.9854,
      "num_input_tokens_seen": 584966056,
      "step": 574500
    },
    {
      "epoch": 11.31302789864346,
      "grad_norm": 1.8905065059661865,
      "learning_rate": 1.4181399156554087e-07,
      "loss": 1.9642,
      "num_input_tokens_seen": 585065648,
      "step": 574600
    },
    {
      "epoch": 11.314996751392963,
      "grad_norm": 1.8770084381103516,
      "learning_rate": 1.4167016608064142e-07,
      "loss": 2.0091,
      "num_input_tokens_seen": 585166184,
      "step": 574700
    },
    {
      "epoch": 11.316965604142466,
      "grad_norm": 1.7596750259399414,
      "learning_rate": 1.4152640152670782e-07,
      "loss": 2.0012,
      "num_input_tokens_seen": 585268584,
      "step": 574800
    },
    {
      "epoch": 11.318934456891968,
      "grad_norm": 1.8129584789276123,
      "learning_rate": 1.4138269792818596e-07,
      "loss": 1.9249,
      "num_input_tokens_seen": 585370144,
      "step": 574900
    },
    {
      "epoch": 11.320903309641471,
      "grad_norm": 2.135826826095581,
      "learning_rate": 1.412390553095119e-07,
      "loss": 2.0058,
      "num_input_tokens_seen": 585471936,
      "step": 575000
    },
    {
      "epoch": 11.322872162390976,
      "grad_norm": 2.023653984069824,
      "learning_rate": 1.4109547369511044e-07,
      "loss": 2.0141,
      "num_input_tokens_seen": 585574336,
      "step": 575100
    },
    {
      "epoch": 11.324841015140478,
      "grad_norm": 2.0453250408172607,
      "learning_rate": 1.4095195310939668e-07,
      "loss": 1.9526,
      "num_input_tokens_seen": 585676200,
      "step": 575200
    },
    {
      "epoch": 11.326809867889981,
      "grad_norm": 1.8389403820037842,
      "learning_rate": 1.408084935767746e-07,
      "loss": 2.0132,
      "num_input_tokens_seen": 585777960,
      "step": 575300
    },
    {
      "epoch": 11.328778720639484,
      "grad_norm": 1.771575927734375,
      "learning_rate": 1.4066509512163898e-07,
      "loss": 1.9714,
      "num_input_tokens_seen": 585879760,
      "step": 575400
    },
    {
      "epoch": 11.330747573388987,
      "grad_norm": 1.759372353553772,
      "learning_rate": 1.40521757768373e-07,
      "loss": 1.9772,
      "num_input_tokens_seen": 585981528,
      "step": 575500
    },
    {
      "epoch": 11.33271642613849,
      "grad_norm": 1.9839931726455688,
      "learning_rate": 1.4037848154135028e-07,
      "loss": 1.9412,
      "num_input_tokens_seen": 586083232,
      "step": 575600
    },
    {
      "epoch": 11.334685278887992,
      "grad_norm": 1.7879927158355713,
      "learning_rate": 1.4023526646493328e-07,
      "loss": 1.9495,
      "num_input_tokens_seen": 586185632,
      "step": 575700
    },
    {
      "epoch": 11.336654131637495,
      "grad_norm": 1.7677305936813354,
      "learning_rate": 1.4009211256347492e-07,
      "loss": 1.9981,
      "num_input_tokens_seen": 586287736,
      "step": 575800
    },
    {
      "epoch": 11.338622984386998,
      "grad_norm": 2.170419692993164,
      "learning_rate": 1.399490198613174e-07,
      "loss": 1.9876,
      "num_input_tokens_seen": 586390136,
      "step": 575900
    },
    {
      "epoch": 11.3405918371365,
      "grad_norm": 1.8429392576217651,
      "learning_rate": 1.3980598838279212e-07,
      "loss": 1.9817,
      "num_input_tokens_seen": 586492536,
      "step": 576000
    },
    {
      "epoch": 11.342560689886003,
      "grad_norm": 2.134028673171997,
      "learning_rate": 1.3966301815222049e-07,
      "loss": 1.9608,
      "num_input_tokens_seen": 586594272,
      "step": 576100
    },
    {
      "epoch": 11.344529542635506,
      "grad_norm": 1.9560480117797852,
      "learning_rate": 1.3952010919391344e-07,
      "loss": 2.005,
      "num_input_tokens_seen": 586696672,
      "step": 576200
    },
    {
      "epoch": 11.346498395385009,
      "grad_norm": 1.8093209266662598,
      "learning_rate": 1.393772615321715e-07,
      "loss": 1.9469,
      "num_input_tokens_seen": 586799072,
      "step": 576300
    },
    {
      "epoch": 11.348467248134511,
      "grad_norm": 2.019213914871216,
      "learning_rate": 1.3923447519128438e-07,
      "loss": 2.0158,
      "num_input_tokens_seen": 586900176,
      "step": 576400
    },
    {
      "epoch": 11.350436100884014,
      "grad_norm": 1.845041036605835,
      "learning_rate": 1.3909175019553206e-07,
      "loss": 1.9645,
      "num_input_tokens_seen": 587002576,
      "step": 576500
    },
    {
      "epoch": 11.352404953633517,
      "grad_norm": 2.8705618381500244,
      "learning_rate": 1.3894908656918346e-07,
      "loss": 1.9708,
      "num_input_tokens_seen": 587104016,
      "step": 576600
    },
    {
      "epoch": 11.35437380638302,
      "grad_norm": 1.7919387817382812,
      "learning_rate": 1.3880648433649767e-07,
      "loss": 1.9898,
      "num_input_tokens_seen": 587205496,
      "step": 576700
    },
    {
      "epoch": 11.356342659132524,
      "grad_norm": 1.9043221473693848,
      "learning_rate": 1.3866394352172262e-07,
      "loss": 1.9464,
      "num_input_tokens_seen": 587307896,
      "step": 576800
    },
    {
      "epoch": 11.358311511882027,
      "grad_norm": 1.738218069076538,
      "learning_rate": 1.3852146414909626e-07,
      "loss": 1.9639,
      "num_input_tokens_seen": 587410296,
      "step": 576900
    },
    {
      "epoch": 11.36028036463153,
      "grad_norm": 2.0904452800750732,
      "learning_rate": 1.3837904624284613e-07,
      "loss": 1.9839,
      "num_input_tokens_seen": 587511824,
      "step": 577000
    },
    {
      "epoch": 11.362249217381033,
      "grad_norm": 1.8238475322723389,
      "learning_rate": 1.3823668982718912e-07,
      "loss": 1.923,
      "num_input_tokens_seen": 587614224,
      "step": 577100
    },
    {
      "epoch": 11.364218070130535,
      "grad_norm": 1.5709075927734375,
      "learning_rate": 1.380943949263319e-07,
      "loss": 1.999,
      "num_input_tokens_seen": 587716080,
      "step": 577200
    },
    {
      "epoch": 11.366186922880038,
      "grad_norm": 1.9707057476043701,
      "learning_rate": 1.379521615644702e-07,
      "loss": 1.9936,
      "num_input_tokens_seen": 587818480,
      "step": 577300
    },
    {
      "epoch": 11.36815577562954,
      "grad_norm": 2.1370668411254883,
      "learning_rate": 1.3780998976578984e-07,
      "loss": 1.9804,
      "num_input_tokens_seen": 587919928,
      "step": 577400
    },
    {
      "epoch": 11.370124628379044,
      "grad_norm": 3.453627347946167,
      "learning_rate": 1.376678795544658e-07,
      "loss": 1.981,
      "num_input_tokens_seen": 588022328,
      "step": 577500
    },
    {
      "epoch": 11.372093481128546,
      "grad_norm": 1.943253517150879,
      "learning_rate": 1.3752583095466296e-07,
      "loss": 2.0243,
      "num_input_tokens_seen": 588123872,
      "step": 577600
    },
    {
      "epoch": 11.374062333878049,
      "grad_norm": 1.6537712812423706,
      "learning_rate": 1.3738384399053504e-07,
      "loss": 1.987,
      "num_input_tokens_seen": 588226272,
      "step": 577700
    },
    {
      "epoch": 11.376031186627552,
      "grad_norm": 1.8856823444366455,
      "learning_rate": 1.3724191868622626e-07,
      "loss": 2.0012,
      "num_input_tokens_seen": 588327776,
      "step": 577800
    },
    {
      "epoch": 11.378000039377055,
      "grad_norm": 2.0332820415496826,
      "learning_rate": 1.3710005506586948e-07,
      "loss": 1.9756,
      "num_input_tokens_seen": 588429640,
      "step": 577900
    },
    {
      "epoch": 11.379968892126557,
      "grad_norm": 2.006032943725586,
      "learning_rate": 1.3695825315358745e-07,
      "loss": 1.966,
      "num_input_tokens_seen": 588532040,
      "step": 578000
    },
    {
      "epoch": 11.38193774487606,
      "grad_norm": 1.7524176836013794,
      "learning_rate": 1.368165129734924e-07,
      "loss": 1.9319,
      "num_input_tokens_seen": 588634440,
      "step": 578100
    },
    {
      "epoch": 11.383906597625563,
      "grad_norm": 2.0088958740234375,
      "learning_rate": 1.3667483454968614e-07,
      "loss": 1.9359,
      "num_input_tokens_seen": 588736840,
      "step": 578200
    },
    {
      "epoch": 11.385875450375066,
      "grad_norm": 1.8933031558990479,
      "learning_rate": 1.3653321790626e-07,
      "loss": 1.9927,
      "num_input_tokens_seen": 588838304,
      "step": 578300
    },
    {
      "epoch": 11.38784430312457,
      "grad_norm": 1.934854507446289,
      "learning_rate": 1.3639166306729432e-07,
      "loss": 1.9792,
      "num_input_tokens_seen": 588939856,
      "step": 578400
    },
    {
      "epoch": 11.389813155874073,
      "grad_norm": 1.7932626008987427,
      "learning_rate": 1.3625017005685958e-07,
      "loss": 2.031,
      "num_input_tokens_seen": 589040408,
      "step": 578500
    },
    {
      "epoch": 11.391782008623576,
      "grad_norm": 2.015106439590454,
      "learning_rate": 1.361087388990154e-07,
      "loss": 2.0541,
      "num_input_tokens_seen": 589141480,
      "step": 578600
    },
    {
      "epoch": 11.393750861373078,
      "grad_norm": 2.0154497623443604,
      "learning_rate": 1.3596736961781115e-07,
      "loss": 1.9962,
      "num_input_tokens_seen": 589243880,
      "step": 578700
    },
    {
      "epoch": 11.395719714122581,
      "grad_norm": 1.980150818824768,
      "learning_rate": 1.358260622372851e-07,
      "loss": 1.9847,
      "num_input_tokens_seen": 589345728,
      "step": 578800
    },
    {
      "epoch": 11.397688566872084,
      "grad_norm": 1.8823072910308838,
      "learning_rate": 1.356848167814656e-07,
      "loss": 2.0179,
      "num_input_tokens_seen": 589446256,
      "step": 578900
    },
    {
      "epoch": 11.399657419621587,
      "grad_norm": 2.0101981163024902,
      "learning_rate": 1.3554363327437023e-07,
      "loss": 1.9559,
      "num_input_tokens_seen": 589548656,
      "step": 579000
    },
    {
      "epoch": 11.40162627237109,
      "grad_norm": 4.090921878814697,
      "learning_rate": 1.3540251174000606e-07,
      "loss": 1.9844,
      "num_input_tokens_seen": 589649688,
      "step": 579100
    },
    {
      "epoch": 11.403595125120592,
      "grad_norm": 1.813073992729187,
      "learning_rate": 1.3526145220236973e-07,
      "loss": 2.0004,
      "num_input_tokens_seen": 589749616,
      "step": 579200
    },
    {
      "epoch": 11.405563977870095,
      "grad_norm": 2.2304625511169434,
      "learning_rate": 1.3512045468544703e-07,
      "loss": 2.0328,
      "num_input_tokens_seen": 589852016,
      "step": 579300
    },
    {
      "epoch": 11.407532830619598,
      "grad_norm": 2.0579187870025635,
      "learning_rate": 1.3497951921321337e-07,
      "loss": 2.0259,
      "num_input_tokens_seen": 589953456,
      "step": 579400
    },
    {
      "epoch": 11.4095016833691,
      "grad_norm": 2.0351145267486572,
      "learning_rate": 1.3483864580963384e-07,
      "loss": 2.0185,
      "num_input_tokens_seen": 590055088,
      "step": 579500
    },
    {
      "epoch": 11.411470536118603,
      "grad_norm": 2.1744699478149414,
      "learning_rate": 1.346978344986628e-07,
      "loss": 2.0111,
      "num_input_tokens_seen": 590157488,
      "step": 579600
    },
    {
      "epoch": 11.413439388868106,
      "grad_norm": 1.747322916984558,
      "learning_rate": 1.3455708530424375e-07,
      "loss": 1.9501,
      "num_input_tokens_seen": 590259344,
      "step": 579700
    },
    {
      "epoch": 11.415408241617609,
      "grad_norm": 2.6944944858551025,
      "learning_rate": 1.3441639825031003e-07,
      "loss": 2.0623,
      "num_input_tokens_seen": 590360504,
      "step": 579800
    },
    {
      "epoch": 11.417377094367112,
      "grad_norm": 2.2272725105285645,
      "learning_rate": 1.3427577336078431e-07,
      "loss": 1.9529,
      "num_input_tokens_seen": 590462464,
      "step": 579900
    },
    {
      "epoch": 11.419345947116614,
      "grad_norm": 2.331481695175171,
      "learning_rate": 1.3413521065957873e-07,
      "loss": 2.0399,
      "num_input_tokens_seen": 590564472,
      "step": 580000
    },
    {
      "epoch": 11.421314799866119,
      "grad_norm": 2.06408429145813,
      "learning_rate": 1.3399471017059483e-07,
      "loss": 2.0402,
      "num_input_tokens_seen": 590665696,
      "step": 580100
    },
    {
      "epoch": 11.423283652615622,
      "grad_norm": 2.0692265033721924,
      "learning_rate": 1.3385427191772315e-07,
      "loss": 2.0511,
      "num_input_tokens_seen": 590768096,
      "step": 580200
    },
    {
      "epoch": 11.425252505365124,
      "grad_norm": 1.6960629224777222,
      "learning_rate": 1.3371389592484461e-07,
      "loss": 1.9878,
      "num_input_tokens_seen": 590869752,
      "step": 580300
    },
    {
      "epoch": 11.427221358114627,
      "grad_norm": 2.9542627334594727,
      "learning_rate": 1.3357358221582854e-07,
      "loss": 1.9829,
      "num_input_tokens_seen": 590972152,
      "step": 580400
    },
    {
      "epoch": 11.42919021086413,
      "grad_norm": 1.9720667600631714,
      "learning_rate": 1.3343333081453434e-07,
      "loss": 1.9853,
      "num_input_tokens_seen": 591074552,
      "step": 580500
    },
    {
      "epoch": 11.431159063613633,
      "grad_norm": 1.8150570392608643,
      "learning_rate": 1.3329314174481022e-07,
      "loss": 1.9869,
      "num_input_tokens_seen": 591176088,
      "step": 580600
    },
    {
      "epoch": 11.433127916363135,
      "grad_norm": 1.9600450992584229,
      "learning_rate": 1.3315301503049464e-07,
      "loss": 1.9784,
      "num_input_tokens_seen": 591278480,
      "step": 580700
    },
    {
      "epoch": 11.435096769112638,
      "grad_norm": 1.910221815109253,
      "learning_rate": 1.3301295069541464e-07,
      "loss": 1.9742,
      "num_input_tokens_seen": 591380304,
      "step": 580800
    },
    {
      "epoch": 11.437065621862141,
      "grad_norm": 1.8686867952346802,
      "learning_rate": 1.32872948763387e-07,
      "loss": 1.9576,
      "num_input_tokens_seen": 591481872,
      "step": 580900
    },
    {
      "epoch": 11.439034474611644,
      "grad_norm": 1.9554502964019775,
      "learning_rate": 1.3273300925821795e-07,
      "loss": 1.982,
      "num_input_tokens_seen": 591583600,
      "step": 581000
    },
    {
      "epoch": 11.441003327361146,
      "grad_norm": 1.949562668800354,
      "learning_rate": 1.3259313220370298e-07,
      "loss": 2.0285,
      "num_input_tokens_seen": 591686000,
      "step": 581100
    },
    {
      "epoch": 11.44297218011065,
      "grad_norm": 2.226060628890991,
      "learning_rate": 1.3245331762362723e-07,
      "loss": 2.0074,
      "num_input_tokens_seen": 591788400,
      "step": 581200
    },
    {
      "epoch": 11.444941032860152,
      "grad_norm": 2.0835230350494385,
      "learning_rate": 1.323135655417646e-07,
      "loss": 2.0276,
      "num_input_tokens_seen": 591890496,
      "step": 581300
    },
    {
      "epoch": 11.446909885609655,
      "grad_norm": 2.0677332878112793,
      "learning_rate": 1.3217387598187895e-07,
      "loss": 2.017,
      "num_input_tokens_seen": 591991800,
      "step": 581400
    },
    {
      "epoch": 11.448878738359157,
      "grad_norm": 1.9477649927139282,
      "learning_rate": 1.320342489677233e-07,
      "loss": 1.9957,
      "num_input_tokens_seen": 592093008,
      "step": 581500
    },
    {
      "epoch": 11.45084759110866,
      "grad_norm": 1.973305583000183,
      "learning_rate": 1.318946845230402e-07,
      "loss": 1.9603,
      "num_input_tokens_seen": 592194024,
      "step": 581600
    },
    {
      "epoch": 11.452816443858163,
      "grad_norm": 1.8793494701385498,
      "learning_rate": 1.317551826715611e-07,
      "loss": 1.9894,
      "num_input_tokens_seen": 592295600,
      "step": 581700
    },
    {
      "epoch": 11.454785296607668,
      "grad_norm": 1.967320203781128,
      "learning_rate": 1.3161574343700727e-07,
      "loss": 1.9842,
      "num_input_tokens_seen": 592396872,
      "step": 581800
    },
    {
      "epoch": 11.45675414935717,
      "grad_norm": 2.103950023651123,
      "learning_rate": 1.3147636684308917e-07,
      "loss": 2.0282,
      "num_input_tokens_seen": 592498104,
      "step": 581900
    },
    {
      "epoch": 11.458723002106673,
      "grad_norm": 1.748271107673645,
      "learning_rate": 1.313370529135066e-07,
      "loss": 1.9633,
      "num_input_tokens_seen": 592597736,
      "step": 582000
    },
    {
      "epoch": 11.460691854856176,
      "grad_norm": 2.0629665851593018,
      "learning_rate": 1.3119780167194893e-07,
      "loss": 1.9863,
      "num_input_tokens_seen": 592700136,
      "step": 582100
    },
    {
      "epoch": 11.462660707605679,
      "grad_norm": 2.018004894256592,
      "learning_rate": 1.3105861314209426e-07,
      "loss": 1.9724,
      "num_input_tokens_seen": 592801664,
      "step": 582200
    },
    {
      "epoch": 11.464629560355181,
      "grad_norm": 2.0310120582580566,
      "learning_rate": 1.3091948734761065e-07,
      "loss": 2.0213,
      "num_input_tokens_seen": 592903792,
      "step": 582300
    },
    {
      "epoch": 11.466598413104684,
      "grad_norm": 6.117821216583252,
      "learning_rate": 1.3078042431215524e-07,
      "loss": 1.9787,
      "num_input_tokens_seen": 593005504,
      "step": 582400
    },
    {
      "epoch": 11.468567265854187,
      "grad_norm": 1.8119415044784546,
      "learning_rate": 1.3064142405937472e-07,
      "loss": 1.9885,
      "num_input_tokens_seen": 593106544,
      "step": 582500
    },
    {
      "epoch": 11.47053611860369,
      "grad_norm": 2.0592005252838135,
      "learning_rate": 1.3050248661290452e-07,
      "loss": 1.9836,
      "num_input_tokens_seen": 593208352,
      "step": 582600
    },
    {
      "epoch": 11.472504971353192,
      "grad_norm": 1.7434622049331665,
      "learning_rate": 1.3036361199637002e-07,
      "loss": 1.9239,
      "num_input_tokens_seen": 593310000,
      "step": 582700
    },
    {
      "epoch": 11.474473824102695,
      "grad_norm": 1.8289612531661987,
      "learning_rate": 1.3022480023338567e-07,
      "loss": 2.0132,
      "num_input_tokens_seen": 593411416,
      "step": 582800
    },
    {
      "epoch": 11.476442676852198,
      "grad_norm": 1.9406064748764038,
      "learning_rate": 1.300860513475553e-07,
      "loss": 1.9853,
      "num_input_tokens_seen": 593513064,
      "step": 582900
    },
    {
      "epoch": 11.4784115296017,
      "grad_norm": 2.0607354640960693,
      "learning_rate": 1.299473653624718e-07,
      "loss": 1.975,
      "num_input_tokens_seen": 593615464,
      "step": 583000
    },
    {
      "epoch": 11.480380382351203,
      "grad_norm": 1.8850548267364502,
      "learning_rate": 1.2980874230171751e-07,
      "loss": 1.9976,
      "num_input_tokens_seen": 593716144,
      "step": 583100
    },
    {
      "epoch": 11.482349235100706,
      "grad_norm": 2.0662503242492676,
      "learning_rate": 1.296701821888645e-07,
      "loss": 1.9811,
      "num_input_tokens_seen": 593818544,
      "step": 583200
    },
    {
      "epoch": 11.484318087850209,
      "grad_norm": 1.8697055578231812,
      "learning_rate": 1.295316850474734e-07,
      "loss": 1.9803,
      "num_input_tokens_seen": 593920944,
      "step": 583300
    },
    {
      "epoch": 11.486286940599712,
      "grad_norm": 1.941184401512146,
      "learning_rate": 1.293932509010947e-07,
      "loss": 1.9224,
      "num_input_tokens_seen": 594023344,
      "step": 583400
    },
    {
      "epoch": 11.488255793349216,
      "grad_norm": 2.114168405532837,
      "learning_rate": 1.2925487977326753e-07,
      "loss": 2.0006,
      "num_input_tokens_seen": 594124760,
      "step": 583500
    },
    {
      "epoch": 11.490224646098719,
      "grad_norm": 1.7260112762451172,
      "learning_rate": 1.2911657168752138e-07,
      "loss": 1.9692,
      "num_input_tokens_seen": 594226512,
      "step": 583600
    },
    {
      "epoch": 11.492193498848222,
      "grad_norm": 1.895065188407898,
      "learning_rate": 1.289783266673739e-07,
      "loss": 2.0289,
      "num_input_tokens_seen": 594328912,
      "step": 583700
    },
    {
      "epoch": 11.494162351597724,
      "grad_norm": 2.090946912765503,
      "learning_rate": 1.2884014473633276e-07,
      "loss": 1.9865,
      "num_input_tokens_seen": 594431312,
      "step": 583800
    },
    {
      "epoch": 11.496131204347227,
      "grad_norm": 1.916996955871582,
      "learning_rate": 1.287020259178942e-07,
      "loss": 1.9655,
      "num_input_tokens_seen": 594533712,
      "step": 583900
    },
    {
      "epoch": 11.49810005709673,
      "grad_norm": 1.820854902267456,
      "learning_rate": 1.285639702355446e-07,
      "loss": 1.969,
      "num_input_tokens_seen": 594636112,
      "step": 584000
    },
    {
      "epoch": 11.500068909846233,
      "grad_norm": 1.9672008752822876,
      "learning_rate": 1.284259777127592e-07,
      "loss": 1.9958,
      "num_input_tokens_seen": 594737096,
      "step": 584100
    },
    {
      "epoch": 11.502037762595736,
      "grad_norm": 2.0690979957580566,
      "learning_rate": 1.282880483730021e-07,
      "loss": 1.9662,
      "num_input_tokens_seen": 594839496,
      "step": 584200
    },
    {
      "epoch": 11.504006615345238,
      "grad_norm": 2.628361701965332,
      "learning_rate": 1.2815018223972728e-07,
      "loss": 1.9606,
      "num_input_tokens_seen": 594941448,
      "step": 584300
    },
    {
      "epoch": 11.505975468094741,
      "grad_norm": 2.009925365447998,
      "learning_rate": 1.280123793363776e-07,
      "loss": 2.028,
      "num_input_tokens_seen": 595043848,
      "step": 584400
    },
    {
      "epoch": 11.507944320844244,
      "grad_norm": 2.4728260040283203,
      "learning_rate": 1.2787463968638546e-07,
      "loss": 1.979,
      "num_input_tokens_seen": 595145456,
      "step": 584500
    },
    {
      "epoch": 11.509913173593747,
      "grad_norm": 1.9369709491729736,
      "learning_rate": 1.277369633131721e-07,
      "loss": 2.0121,
      "num_input_tokens_seen": 595246472,
      "step": 584600
    },
    {
      "epoch": 11.51188202634325,
      "grad_norm": 4.695319652557373,
      "learning_rate": 1.2759935024014822e-07,
      "loss": 2.0076,
      "num_input_tokens_seen": 595348120,
      "step": 584700
    },
    {
      "epoch": 11.513850879092752,
      "grad_norm": 1.899884819984436,
      "learning_rate": 1.2746180049071392e-07,
      "loss": 1.9597,
      "num_input_tokens_seen": 595450520,
      "step": 584800
    },
    {
      "epoch": 11.515819731842255,
      "grad_norm": 1.9921207427978516,
      "learning_rate": 1.273243140882585e-07,
      "loss": 1.9856,
      "num_input_tokens_seen": 595552144,
      "step": 584900
    },
    {
      "epoch": 11.517788584591758,
      "grad_norm": 1.755434513092041,
      "learning_rate": 1.2718689105616e-07,
      "loss": 1.9415,
      "num_input_tokens_seen": 595654544,
      "step": 585000
    },
    {
      "epoch": 11.519757437341262,
      "grad_norm": 3.659200668334961,
      "learning_rate": 3.94675918389431e-09,
      "loss": 1.9812,
      "num_input_tokens_seen": 595756448,
      "step": 585100
    },
    {
      "epoch": 11.521726290090765,
      "grad_norm": 2.1249501705169678,
      "learning_rate": 3.914506629807912e-09,
      "loss": 2.0162,
      "num_input_tokens_seen": 595858648,
      "step": 585200
    },
    {
      "epoch": 11.523695142840268,
      "grad_norm": 2.6715171337127686,
      "learning_rate": 3.882385881513572e-09,
      "loss": 1.9694,
      "num_input_tokens_seen": 595960224,
      "step": 585300
    },
    {
      "epoch": 11.52566399558977,
      "grad_norm": 1.8062368631362915,
      "learning_rate": 3.8503969475454645e-09,
      "loss": 1.9692,
      "num_input_tokens_seen": 596059680,
      "step": 585400
    },
    {
      "epoch": 11.527632848339273,
      "grad_norm": 1.8949953317642212,
      "learning_rate": 3.818539836402901e-09,
      "loss": 1.9634,
      "num_input_tokens_seen": 596160920,
      "step": 585500
    },
    {
      "epoch": 11.529601701088776,
      "grad_norm": 1.8594651222229004,
      "learning_rate": 3.786814556549889e-09,
      "loss": 2.0249,
      "num_input_tokens_seen": 596262168,
      "step": 585600
    },
    {
      "epoch": 11.531570553838279,
      "grad_norm": 1.5639607906341553,
      "learning_rate": 3.755221116415685e-09,
      "loss": 1.9759,
      "num_input_tokens_seen": 596364568,
      "step": 585700
    },
    {
      "epoch": 11.533539406587781,
      "grad_norm": 1.8152952194213867,
      "learning_rate": 3.7237595243944097e-09,
      "loss": 2.0019,
      "num_input_tokens_seen": 596466344,
      "step": 585800
    },
    {
      "epoch": 11.535508259337284,
      "grad_norm": 2.1235945224761963,
      "learning_rate": 3.6924297888450974e-09,
      "loss": 1.98,
      "num_input_tokens_seen": 596568744,
      "step": 585900
    },
    {
      "epoch": 11.537477112086787,
      "grad_norm": 1.9921948909759521,
      "learning_rate": 3.661231918091756e-09,
      "loss": 1.9993,
      "num_input_tokens_seen": 596670520,
      "step": 586000
    },
    {
      "epoch": 11.53944596483629,
      "grad_norm": 2.057854652404785,
      "learning_rate": 3.6301659204235335e-09,
      "loss": 1.9438,
      "num_input_tokens_seen": 596771432,
      "step": 586100
    },
    {
      "epoch": 11.541414817585792,
      "grad_norm": 2.0799715518951416,
      "learning_rate": 3.5992318040942714e-09,
      "loss": 2.0,
      "num_input_tokens_seen": 596873104,
      "step": 586200
    },
    {
      "epoch": 11.543383670335295,
      "grad_norm": 2.0648226737976074,
      "learning_rate": 3.5684295773230066e-09,
      "loss": 2.0493,
      "num_input_tokens_seen": 596974128,
      "step": 586300
    },
    {
      "epoch": 11.545352523084798,
      "grad_norm": 1.8234760761260986,
      "learning_rate": 3.5377592482935816e-09,
      "loss": 1.9846,
      "num_input_tokens_seen": 597076528,
      "step": 586400
    },
    {
      "epoch": 11.5473213758343,
      "grad_norm": 1.8485310077667236,
      "learning_rate": 3.5072208251548665e-09,
      "loss": 1.9597,
      "num_input_tokens_seen": 597178928,
      "step": 586500
    },
    {
      "epoch": 11.549290228583803,
      "grad_norm": 1.9488697052001953,
      "learning_rate": 3.4768143160206486e-09,
      "loss": 2.0429,
      "num_input_tokens_seen": 597280600,
      "step": 586600
    },
    {
      "epoch": 11.551259081333306,
      "grad_norm": 2.2161998748779297,
      "learning_rate": 3.4465397289696884e-09,
      "loss": 1.9481,
      "num_input_tokens_seen": 597383000,
      "step": 586700
    },
    {
      "epoch": 11.553227934082809,
      "grad_norm": 2.0033934116363525,
      "learning_rate": 3.416397072045718e-09,
      "loss": 1.9765,
      "num_input_tokens_seen": 597484064,
      "step": 586800
    },
    {
      "epoch": 11.555196786832314,
      "grad_norm": 1.6994107961654663,
      "learning_rate": 3.386386353257387e-09,
      "loss": 2.05,
      "num_input_tokens_seen": 597586464,
      "step": 586900
    },
    {
      "epoch": 11.557165639581816,
      "grad_norm": 1.991618037223816,
      "learning_rate": 3.3565075805782605e-09,
      "loss": 2.0157,
      "num_input_tokens_seen": 597686616,
      "step": 587000
    },
    {
      "epoch": 11.559134492331319,
      "grad_norm": 1.9513188600540161,
      "learning_rate": 3.326760761946934e-09,
      "loss": 2.0172,
      "num_input_tokens_seen": 597787712,
      "step": 587100
    },
    {
      "epoch": 11.561103345080822,
      "grad_norm": 1.7938748598098755,
      "learning_rate": 3.2971459052668627e-09,
      "loss": 1.9731,
      "num_input_tokens_seen": 597890112,
      "step": 587200
    },
    {
      "epoch": 11.563072197830325,
      "grad_norm": 2.0966484546661377,
      "learning_rate": 3.2676630184064193e-09,
      "loss": 2.021,
      "num_input_tokens_seen": 597991024,
      "step": 587300
    },
    {
      "epoch": 11.565041050579827,
      "grad_norm": 2.132781505584717,
      "learning_rate": 3.238312109199004e-09,
      "loss": 1.9729,
      "num_input_tokens_seen": 598093424,
      "step": 587400
    },
    {
      "epoch": 11.56700990332933,
      "grad_norm": 1.893375277519226,
      "learning_rate": 3.2090931854429347e-09,
      "loss": 2.0153,
      "num_input_tokens_seen": 598195824,
      "step": 587500
    },
    {
      "epoch": 11.568978756078833,
      "grad_norm": 2.136014223098755,
      "learning_rate": 3.1800062549013906e-09,
      "loss": 1.9696,
      "num_input_tokens_seen": 598296504,
      "step": 587600
    },
    {
      "epoch": 11.570947608828336,
      "grad_norm": 1.9575252532958984,
      "learning_rate": 3.151051325302523e-09,
      "loss": 1.9419,
      "num_input_tokens_seen": 598398328,
      "step": 587700
    },
    {
      "epoch": 11.572916461577838,
      "grad_norm": 1.770530343055725,
      "learning_rate": 3.122228404339511e-09,
      "loss": 2.0142,
      "num_input_tokens_seen": 598500728,
      "step": 587800
    },
    {
      "epoch": 11.574885314327341,
      "grad_norm": 2.0865249633789062,
      "learning_rate": 3.0935374996702292e-09,
      "loss": 2.0028,
      "num_input_tokens_seen": 598602488,
      "step": 587900
    },
    {
      "epoch": 11.576854167076844,
      "grad_norm": 2.0338327884674072,
      "learning_rate": 3.0649786189176908e-09,
      "loss": 1.98,
      "num_input_tokens_seen": 598703832,
      "step": 588000
    },
    {
      "epoch": 11.578823019826347,
      "grad_norm": 1.9522576332092285,
      "learning_rate": 3.0365517696697705e-09,
      "loss": 2.0216,
      "num_input_tokens_seen": 598805000,
      "step": 588100
    },
    {
      "epoch": 11.58079187257585,
      "grad_norm": 2.122528314590454,
      "learning_rate": 3.0082569594792052e-09,
      "loss": 1.9755,
      "num_input_tokens_seen": 598906680,
      "step": 588200
    },
    {
      "epoch": 11.582760725325352,
      "grad_norm": 1.6766784191131592,
      "learning_rate": 2.9800941958637027e-09,
      "loss": 2.0083,
      "num_input_tokens_seen": 599006296,
      "step": 588300
    },
    {
      "epoch": 11.584729578074857,
      "grad_norm": 2.1729745864868164,
      "learning_rate": 2.9520634863058336e-09,
      "loss": 2.0161,
      "num_input_tokens_seen": 599106608,
      "step": 588400
    },
    {
      "epoch": 11.58669843082436,
      "grad_norm": 2.253821611404419,
      "learning_rate": 2.924164838253196e-09,
      "loss": 2.0157,
      "num_input_tokens_seen": 599208192,
      "step": 588500
    },
    {
      "epoch": 11.588667283573862,
      "grad_norm": 1.7080379724502563,
      "learning_rate": 2.8963982591182488e-09,
      "loss": 1.9792,
      "num_input_tokens_seen": 599309664,
      "step": 588600
    },
    {
      "epoch": 11.590636136323365,
      "grad_norm": 1.9355628490447998,
      "learning_rate": 2.868763756278203e-09,
      "loss": 1.9482,
      "num_input_tokens_seen": 599411336,
      "step": 588700
    },
    {
      "epoch": 11.592604989072868,
      "grad_norm": 1.8168165683746338,
      "learning_rate": 2.8412613370754624e-09,
      "loss": 2.0295,
      "num_input_tokens_seen": 599513736,
      "step": 588800
    },
    {
      "epoch": 11.59457384182237,
      "grad_norm": 1.9750293493270874,
      "learning_rate": 2.8138910088170708e-09,
      "loss": 1.9778,
      "num_input_tokens_seen": 599616136,
      "step": 588900
    },
    {
      "epoch": 11.596542694571873,
      "grad_norm": 2.0568883419036865,
      "learning_rate": 2.786652778775267e-09,
      "loss": 2.0103,
      "num_input_tokens_seen": 599718536,
      "step": 589000
    },
    {
      "epoch": 11.598511547321376,
      "grad_norm": 1.8482158184051514,
      "learning_rate": 2.7595466541868175e-09,
      "loss": 1.9449,
      "num_input_tokens_seen": 599820168,
      "step": 589100
    },
    {
      "epoch": 11.600480400070879,
      "grad_norm": 2.0327420234680176,
      "learning_rate": 2.7325726422536833e-09,
      "loss": 2.011,
      "num_input_tokens_seen": 599922480,
      "step": 589200
    },
    {
      "epoch": 11.602449252820382,
      "grad_norm": 1.763657569885254,
      "learning_rate": 2.705730750142743e-09,
      "loss": 1.9304,
      "num_input_tokens_seen": 600023624,
      "step": 589300
    },
    {
      "epoch": 11.604418105569884,
      "grad_norm": 1.5697556734085083,
      "learning_rate": 2.6790209849854584e-09,
      "loss": 1.9709,
      "num_input_tokens_seen": 600125280,
      "step": 589400
    },
    {
      "epoch": 11.606386958319387,
      "grad_norm": 1.759230136871338,
      "learning_rate": 2.6524433538785418e-09,
      "loss": 1.9486,
      "num_input_tokens_seen": 600227680,
      "step": 589500
    },
    {
      "epoch": 11.60835581106889,
      "grad_norm": 1.822941780090332,
      "learning_rate": 2.6259978638834e-09,
      "loss": 1.9941,
      "num_input_tokens_seen": 600328576,
      "step": 589600
    },
    {
      "epoch": 11.610324663818393,
      "grad_norm": 1.6592353582382202,
      "learning_rate": 2.599684522026413e-09,
      "loss": 1.9774,
      "num_input_tokens_seen": 600430928,
      "step": 589700
    },
    {
      "epoch": 11.612293516567895,
      "grad_norm": 2.0996155738830566,
      "learning_rate": 2.5735033352987656e-09,
      "loss": 1.9954,
      "num_input_tokens_seen": 600532056,
      "step": 589800
    },
    {
      "epoch": 11.614262369317398,
      "grad_norm": 1.7824513912200928,
      "learning_rate": 2.547454310656616e-09,
      "loss": 2.0438,
      "num_input_tokens_seen": 600632248,
      "step": 589900
    },
    {
      "epoch": 11.6162312220669,
      "grad_norm": 1.7518781423568726,
      "learning_rate": 2.521537455020928e-09,
      "loss": 2.0343,
      "num_input_tokens_seen": 600733752,
      "step": 590000
    },
    {
      "epoch": 11.618200074816404,
      "grad_norm": 2.099236011505127,
      "learning_rate": 2.4957527752776376e-09,
      "loss": 2.0148,
      "num_input_tokens_seen": 600835576,
      "step": 590100
    },
    {
      "epoch": 11.620168927565908,
      "grad_norm": 1.8308309316635132,
      "learning_rate": 2.470100278277487e-09,
      "loss": 1.9398,
      "num_input_tokens_seen": 600937968,
      "step": 590200
    },
    {
      "epoch": 11.622137780315411,
      "grad_norm": 1.9775854349136353,
      "learning_rate": 2.4445799708361914e-09,
      "loss": 2.0115,
      "num_input_tokens_seen": 601038016,
      "step": 590300
    },
    {
      "epoch": 11.624106633064914,
      "grad_norm": 2.018115997314453,
      "learning_rate": 2.419191859734271e-09,
      "loss": 1.9977,
      "num_input_tokens_seen": 601140416,
      "step": 590400
    },
    {
      "epoch": 11.626075485814416,
      "grad_norm": 1.8581650257110596,
      "learning_rate": 2.393935951717052e-09,
      "loss": 1.979,
      "num_input_tokens_seen": 601241464,
      "step": 590500
    },
    {
      "epoch": 11.62804433856392,
      "grad_norm": 1.8789161443710327,
      "learning_rate": 2.368812253494945e-09,
      "loss": 2.0179,
      "num_input_tokens_seen": 601343864,
      "step": 590600
    },
    {
      "epoch": 11.630013191313422,
      "grad_norm": 2.103886842727661,
      "learning_rate": 2.343820771742999e-09,
      "loss": 1.9422,
      "num_input_tokens_seen": 601445440,
      "step": 590700
    },
    {
      "epoch": 11.631982044062925,
      "grad_norm": 2.323665142059326,
      "learning_rate": 2.318961513101292e-09,
      "loss": 1.971,
      "num_input_tokens_seen": 601547200,
      "step": 590800
    },
    {
      "epoch": 11.633950896812427,
      "grad_norm": 1.7970967292785645,
      "learning_rate": 2.294234484174762e-09,
      "loss": 1.99,
      "num_input_tokens_seen": 601649600,
      "step": 590900
    },
    {
      "epoch": 11.63591974956193,
      "grad_norm": 1.786669373512268,
      "learning_rate": 2.269639691533043e-09,
      "loss": 2.0106,
      "num_input_tokens_seen": 601751416,
      "step": 591000
    },
    {
      "epoch": 11.637888602311433,
      "grad_norm": 2.2270216941833496,
      "learning_rate": 2.2451771417109077e-09,
      "loss": 1.9897,
      "num_input_tokens_seen": 601853120,
      "step": 591100
    },
    {
      "epoch": 11.639857455060936,
      "grad_norm": 1.8334543704986572,
      "learning_rate": 2.220846841207824e-09,
      "loss": 2.0037,
      "num_input_tokens_seen": 601955520,
      "step": 591200
    },
    {
      "epoch": 11.641826307810438,
      "grad_norm": 1.9184553623199463,
      "learning_rate": 2.1966487964880652e-09,
      "loss": 1.9609,
      "num_input_tokens_seen": 602057920,
      "step": 591300
    },
    {
      "epoch": 11.643795160559941,
      "grad_norm": 2.007113218307495,
      "learning_rate": 2.172583013980933e-09,
      "loss": 2.0072,
      "num_input_tokens_seen": 602159032,
      "step": 591400
    },
    {
      "epoch": 11.645764013309444,
      "grad_norm": 2.028715133666992,
      "learning_rate": 2.1486495000804796e-09,
      "loss": 1.9598,
      "num_input_tokens_seen": 602261432,
      "step": 591500
    },
    {
      "epoch": 11.647732866058947,
      "grad_norm": 1.79324209690094,
      "learning_rate": 2.124848261145673e-09,
      "loss": 1.9845,
      "num_input_tokens_seen": 602363384,
      "step": 591600
    },
    {
      "epoch": 11.64970171880845,
      "grad_norm": 1.9027572870254517,
      "learning_rate": 2.101179303500289e-09,
      "loss": 1.9423,
      "num_input_tokens_seen": 602465536,
      "step": 591700
    },
    {
      "epoch": 11.651670571557954,
      "grad_norm": 1.9445005655288696,
      "learning_rate": 2.0776426334329077e-09,
      "loss": 1.9774,
      "num_input_tokens_seen": 602567936,
      "step": 591800
    },
    {
      "epoch": 11.653639424307457,
      "grad_norm": 1.8726835250854492,
      "learning_rate": 2.0542382571971384e-09,
      "loss": 2.0093,
      "num_input_tokens_seen": 602669528,
      "step": 591900
    },
    {
      "epoch": 11.65560827705696,
      "grad_norm": 1.7551376819610596,
      "learning_rate": 2.0309661810112287e-09,
      "loss": 1.9717,
      "num_input_tokens_seen": 602771928,
      "step": 592000
    },
    {
      "epoch": 11.657577129806462,
      "grad_norm": 1.9877421855926514,
      "learning_rate": 2.0078264110584e-09,
      "loss": 1.9794,
      "num_input_tokens_seen": 602874328,
      "step": 592100
    },
    {
      "epoch": 11.659545982555965,
      "grad_norm": 1.8482903242111206,
      "learning_rate": 1.9848189534867332e-09,
      "loss": 1.9593,
      "num_input_tokens_seen": 602976112,
      "step": 592200
    },
    {
      "epoch": 11.661514835305468,
      "grad_norm": 2.0622079372406006,
      "learning_rate": 1.9619438144091173e-09,
      "loss": 1.9501,
      "num_input_tokens_seen": 603078512,
      "step": 592300
    },
    {
      "epoch": 11.66348368805497,
      "grad_norm": 1.8774306774139404,
      "learning_rate": 1.9392009999032455e-09,
      "loss": 1.9638,
      "num_input_tokens_seen": 603180912,
      "step": 592400
    },
    {
      "epoch": 11.665452540804473,
      "grad_norm": 2.006685972213745,
      "learning_rate": 1.916590516011729e-09,
      "loss": 1.9802,
      "num_input_tokens_seen": 603282784,
      "step": 592500
    },
    {
      "epoch": 11.667421393553976,
      "grad_norm": 1.8034919500350952,
      "learning_rate": 1.8941123687419292e-09,
      "loss": 1.9731,
      "num_input_tokens_seen": 603385184,
      "step": 592600
    },
    {
      "epoch": 11.669390246303479,
      "grad_norm": 1.901477575302124,
      "learning_rate": 1.871766564066235e-09,
      "loss": 1.9767,
      "num_input_tokens_seen": 603487584,
      "step": 592700
    },
    {
      "epoch": 11.671359099052982,
      "grad_norm": 1.9936275482177734,
      "learning_rate": 1.8495531079216199e-09,
      "loss": 1.9927,
      "num_input_tokens_seen": 603589520,
      "step": 592800
    },
    {
      "epoch": 11.673327951802484,
      "grad_norm": 2.0637118816375732,
      "learning_rate": 1.827472006210029e-09,
      "loss": 1.9931,
      "num_input_tokens_seen": 603690792,
      "step": 592900
    },
    {
      "epoch": 11.675296804551987,
      "grad_norm": 1.935971736907959,
      "learning_rate": 1.8055232647982698e-09,
      "loss": 1.9789,
      "num_input_tokens_seen": 603793192,
      "step": 593000
    },
    {
      "epoch": 11.67726565730149,
      "grad_norm": 2.0571725368499756,
      "learning_rate": 1.7837068895178997e-09,
      "loss": 2.0209,
      "num_input_tokens_seen": 603894648,
      "step": 593100
    },
    {
      "epoch": 11.679234510050993,
      "grad_norm": 2.2053656578063965,
      "learning_rate": 1.7620228861653374e-09,
      "loss": 1.9245,
      "num_input_tokens_seen": 603997000,
      "step": 593200
    },
    {
      "epoch": 11.681203362800495,
      "grad_norm": 1.9412939548492432,
      "learning_rate": 1.740471260501919e-09,
      "loss": 2.0252,
      "num_input_tokens_seen": 604099400,
      "step": 593300
    },
    {
      "epoch": 11.683172215549998,
      "grad_norm": 2.307305097579956,
      "learning_rate": 1.719052018253675e-09,
      "loss": 2.0151,
      "num_input_tokens_seen": 604201128,
      "step": 593400
    },
    {
      "epoch": 11.685141068299501,
      "grad_norm": 1.7656645774841309,
      "learning_rate": 1.6977651651114976e-09,
      "loss": 1.9501,
      "num_input_tokens_seen": 604302984,
      "step": 593500
    },
    {
      "epoch": 11.687109921049005,
      "grad_norm": 2.1728756427764893,
      "learning_rate": 1.6766107067311964e-09,
      "loss": 2.0443,
      "num_input_tokens_seen": 604404160,
      "step": 593600
    },
    {
      "epoch": 11.689078773798508,
      "grad_norm": 2.0321216583251953,
      "learning_rate": 1.6555886487332749e-09,
      "loss": 1.9772,
      "num_input_tokens_seen": 604505368,
      "step": 593700
    },
    {
      "epoch": 11.691047626548011,
      "grad_norm": 1.8394025564193726,
      "learning_rate": 1.6346989967031545e-09,
      "loss": 1.993,
      "num_input_tokens_seen": 604607768,
      "step": 593800
    },
    {
      "epoch": 11.693016479297514,
      "grad_norm": 6.243096351623535,
      "learning_rate": 1.6139417561910063e-09,
      "loss": 2.0026,
      "num_input_tokens_seen": 604709664,
      "step": 593900
    },
    {
      "epoch": 11.694985332047017,
      "grad_norm": 1.8713412284851074,
      "learning_rate": 1.5933169327119189e-09,
      "loss": 2.0245,
      "num_input_tokens_seen": 604809960,
      "step": 594000
    },
    {
      "epoch": 11.69695418479652,
      "grad_norm": 2.0375311374664307,
      "learning_rate": 1.5728245317456756e-09,
      "loss": 2.0352,
      "num_input_tokens_seen": 604912360,
      "step": 594100
    },
    {
      "epoch": 11.698923037546022,
      "grad_norm": 2.0358030796051025,
      "learning_rate": 1.5524645587369768e-09,
      "loss": 1.9875,
      "num_input_tokens_seen": 605014080,
      "step": 594200
    },
    {
      "epoch": 11.700891890295525,
      "grad_norm": 2.111898899078369,
      "learning_rate": 1.5322370190952727e-09,
      "loss": 1.9973,
      "num_input_tokens_seen": 605116480,
      "step": 594300
    },
    {
      "epoch": 11.702860743045028,
      "grad_norm": 2.085745334625244,
      "learning_rate": 1.5121419181948204e-09,
      "loss": 2.0091,
      "num_input_tokens_seen": 605218016,
      "step": 594400
    },
    {
      "epoch": 11.70482959579453,
      "grad_norm": 2.052598237991333,
      "learning_rate": 1.492179261374793e-09,
      "loss": 2.0294,
      "num_input_tokens_seen": 605319816,
      "step": 594500
    },
    {
      "epoch": 11.706798448544033,
      "grad_norm": 1.9992872476577759,
      "learning_rate": 1.4723490539390592e-09,
      "loss": 1.9653,
      "num_input_tokens_seen": 605421400,
      "step": 594600
    },
    {
      "epoch": 11.708767301293536,
      "grad_norm": 1.8020381927490234,
      "learning_rate": 1.4526513011564046e-09,
      "loss": 1.9635,
      "num_input_tokens_seen": 605523800,
      "step": 594700
    },
    {
      "epoch": 11.710736154043039,
      "grad_norm": 1.9601142406463623,
      "learning_rate": 1.4330860082602535e-09,
      "loss": 1.9859,
      "num_input_tokens_seen": 605626200,
      "step": 594800
    },
    {
      "epoch": 11.712705006792541,
      "grad_norm": 1.6479361057281494,
      "learning_rate": 1.4136531804490036e-09,
      "loss": 1.9933,
      "num_input_tokens_seen": 605728600,
      "step": 594900
    },
    {
      "epoch": 11.714673859542044,
      "grad_norm": 2.1334123611450195,
      "learning_rate": 1.3943528228858026e-09,
      "loss": 1.9638,
      "num_input_tokens_seen": 605831000,
      "step": 595000
    },
    {
      "epoch": 11.716642712291549,
      "grad_norm": 1.9267786741256714,
      "learning_rate": 1.3751849406985483e-09,
      "loss": 2.022,
      "num_input_tokens_seen": 605932824,
      "step": 595100
    },
    {
      "epoch": 11.718611565041051,
      "grad_norm": 1.9408087730407715,
      "learning_rate": 1.3561495389800004e-09,
      "loss": 1.9692,
      "num_input_tokens_seen": 606035224,
      "step": 595200
    },
    {
      "epoch": 11.720580417790554,
      "grad_norm": 1.870067834854126,
      "learning_rate": 1.3372466227877243e-09,
      "loss": 1.9671,
      "num_input_tokens_seen": 606136760,
      "step": 595300
    },
    {
      "epoch": 11.722549270540057,
      "grad_norm": 1.7808547019958496,
      "learning_rate": 1.3184761971440917e-09,
      "loss": 1.984,
      "num_input_tokens_seen": 606239160,
      "step": 595400
    },
    {
      "epoch": 11.72451812328956,
      "grad_norm": 1.933311104774475,
      "learning_rate": 1.299838267036224e-09,
      "loss": 1.9937,
      "num_input_tokens_seen": 606338368,
      "step": 595500
    },
    {
      "epoch": 11.726486976039062,
      "grad_norm": 1.8346141576766968,
      "learning_rate": 1.2813328374161047e-09,
      "loss": 1.9848,
      "num_input_tokens_seen": 606440768,
      "step": 595600
    },
    {
      "epoch": 11.728455828788565,
      "grad_norm": 2.006956100463867,
      "learning_rate": 1.2629599132003566e-09,
      "loss": 1.9859,
      "num_input_tokens_seen": 606541504,
      "step": 595700
    },
    {
      "epoch": 11.730424681538068,
      "grad_norm": 1.6994996070861816,
      "learning_rate": 1.24471949927063e-09,
      "loss": 2.0021,
      "num_input_tokens_seen": 606642144,
      "step": 595800
    },
    {
      "epoch": 11.73239353428757,
      "grad_norm": 2.083919048309326,
      "learning_rate": 1.2266116004731598e-09,
      "loss": 1.9589,
      "num_input_tokens_seen": 606744544,
      "step": 595900
    },
    {
      "epoch": 11.734362387037073,
      "grad_norm": 2.079982042312622,
      "learning_rate": 1.2086362216191525e-09,
      "loss": 2.0034,
      "num_input_tokens_seen": 606846352,
      "step": 596000
    },
    {
      "epoch": 11.736331239786576,
      "grad_norm": 1.7823718786239624,
      "learning_rate": 1.1907933674845106e-09,
      "loss": 1.9454,
      "num_input_tokens_seen": 606947992,
      "step": 596100
    },
    {
      "epoch": 11.738300092536079,
      "grad_norm": 1.944311499595642,
      "learning_rate": 1.173083042809886e-09,
      "loss": 1.9692,
      "num_input_tokens_seen": 607050392,
      "step": 596200
    },
    {
      "epoch": 11.740268945285582,
      "grad_norm": 2.179417848587036,
      "learning_rate": 1.155505252300737e-09,
      "loss": 1.9869,
      "num_input_tokens_seen": 607152792,
      "step": 596300
    },
    {
      "epoch": 11.742237798035084,
      "grad_norm": 1.756500244140625,
      "learning_rate": 1.1380600006273833e-09,
      "loss": 1.9897,
      "num_input_tokens_seen": 607254336,
      "step": 596400
    },
    {
      "epoch": 11.744206650784587,
      "grad_norm": 2.1323633193969727,
      "learning_rate": 1.120747292424895e-09,
      "loss": 1.9508,
      "num_input_tokens_seen": 607356072,
      "step": 596500
    },
    {
      "epoch": 11.74617550353409,
      "grad_norm": 2.6009182929992676,
      "learning_rate": 1.103567132293093e-09,
      "loss": 2.0472,
      "num_input_tokens_seen": 607458472,
      "step": 596600
    },
    {
      "epoch": 11.748144356283593,
      "grad_norm": 1.9170985221862793,
      "learning_rate": 1.0865195247966585e-09,
      "loss": 2.0183,
      "num_input_tokens_seen": 607560872,
      "step": 596700
    },
    {
      "epoch": 11.750113209033096,
      "grad_norm": 1.8585846424102783,
      "learning_rate": 1.0696044744649135e-09,
      "loss": 1.9619,
      "num_input_tokens_seen": 607663272,
      "step": 596800
    },
    {
      "epoch": 11.7520820617826,
      "grad_norm": 2.1044414043426514,
      "learning_rate": 1.0528219857920407e-09,
      "loss": 1.9705,
      "num_input_tokens_seen": 607765672,
      "step": 596900
    },
    {
      "epoch": 11.754050914532103,
      "grad_norm": 1.7990201711654663,
      "learning_rate": 1.036172063237084e-09,
      "loss": 1.9639,
      "num_input_tokens_seen": 607868072,
      "step": 597000
    },
    {
      "epoch": 11.756019767281606,
      "grad_norm": 1.767758846282959,
      "learning_rate": 1.0196547112237275e-09,
      "loss": 2.018,
      "num_input_tokens_seen": 607970472,
      "step": 597100
    },
    {
      "epoch": 11.757988620031108,
      "grad_norm": 1.8967418670654297,
      "learning_rate": 1.0032699341405716e-09,
      "loss": 1.9725,
      "num_input_tokens_seen": 608071744,
      "step": 597200
    },
    {
      "epoch": 11.759957472780611,
      "grad_norm": 2.6396100521087646,
      "learning_rate": 9.870177363408005e-10,
      "loss": 1.972,
      "num_input_tokens_seen": 608174144,
      "step": 597300
    },
    {
      "epoch": 11.761926325530114,
      "grad_norm": 1.9669471979141235,
      "learning_rate": 9.70898122142627e-10,
      "loss": 1.9493,
      "num_input_tokens_seen": 608276544,
      "step": 597400
    },
    {
      "epoch": 11.763895178279617,
      "grad_norm": 1.7169256210327148,
      "learning_rate": 9.549110958287365e-10,
      "loss": 1.9629,
      "num_input_tokens_seen": 608378944,
      "step": 597500
    },
    {
      "epoch": 11.76586403102912,
      "grad_norm": 2.1224753856658936,
      "learning_rate": 9.390566616468976e-10,
      "loss": 2.0116,
      "num_input_tokens_seen": 608480472,
      "step": 597600
    },
    {
      "epoch": 11.767832883778622,
      "grad_norm": 1.8127522468566895,
      "learning_rate": 9.23334823809463e-10,
      "loss": 1.9794,
      "num_input_tokens_seen": 608582048,
      "step": 597700
    },
    {
      "epoch": 11.769801736528125,
      "grad_norm": 1.8179248571395874,
      "learning_rate": 9.077455864934803e-10,
      "loss": 2.0005,
      "num_input_tokens_seen": 608682832,
      "step": 597800
    },
    {
      "epoch": 11.771770589277628,
      "grad_norm": 1.9912502765655518,
      "learning_rate": 8.92288953841025e-10,
      "loss": 2.0201,
      "num_input_tokens_seen": 608785232,
      "step": 597900
    },
    {
      "epoch": 11.77373944202713,
      "grad_norm": 1.807316780090332,
      "learning_rate": 8.769649299587568e-10,
      "loss": 2.0112,
      "num_input_tokens_seen": 608887632,
      "step": 598000
    },
    {
      "epoch": 11.775708294776633,
      "grad_norm": 2.461653470993042,
      "learning_rate": 8.617735189181408e-10,
      "loss": 1.983,
      "num_input_tokens_seen": 608988488,
      "step": 598100
    },
    {
      "epoch": 11.777677147526136,
      "grad_norm": 2.1969923973083496,
      "learning_rate": 8.467147247553375e-10,
      "loss": 1.9361,
      "num_input_tokens_seen": 609090216,
      "step": 598200
    },
    {
      "epoch": 11.779646000275639,
      "grad_norm": 1.9241276979446411,
      "learning_rate": 8.317885514714241e-10,
      "loss": 1.95,
      "num_input_tokens_seen": 609192616,
      "step": 598300
    },
    {
      "epoch": 11.781614853025141,
      "grad_norm": 1.9649975299835205,
      "learning_rate": 8.169950030321726e-10,
      "loss": 1.9692,
      "num_input_tokens_seen": 609294416,
      "step": 598400
    },
    {
      "epoch": 11.783583705774646,
      "grad_norm": 2.0693068504333496,
      "learning_rate": 8.023340833679948e-10,
      "loss": 1.9561,
      "num_input_tokens_seen": 609396352,
      "step": 598500
    },
    {
      "epoch": 11.785552558524149,
      "grad_norm": 1.730277180671692,
      "learning_rate": 7.878057963743301e-10,
      "loss": 1.9669,
      "num_input_tokens_seen": 609498752,
      "step": 598600
    },
    {
      "epoch": 11.787521411273651,
      "grad_norm": 1.7669689655303955,
      "learning_rate": 7.734101459110909e-10,
      "loss": 2.0171,
      "num_input_tokens_seen": 609601152,
      "step": 598700
    },
    {
      "epoch": 11.789490264023154,
      "grad_norm": 2.171191453933716,
      "learning_rate": 7.59147135803162e-10,
      "loss": 1.9596,
      "num_input_tokens_seen": 609702952,
      "step": 598800
    },
    {
      "epoch": 11.791459116772657,
      "grad_norm": 1.9326591491699219,
      "learning_rate": 7.450167698400678e-10,
      "loss": 1.9839,
      "num_input_tokens_seen": 609804584,
      "step": 598900
    },
    {
      "epoch": 11.79342796952216,
      "grad_norm": 1.955987811088562,
      "learning_rate": 7.310190517761383e-10,
      "loss": 1.967,
      "num_input_tokens_seen": 609905832,
      "step": 599000
    },
    {
      "epoch": 11.795396822271663,
      "grad_norm": 1.9000886678695679,
      "learning_rate": 7.171539853304542e-10,
      "loss": 1.959,
      "num_input_tokens_seen": 610007656,
      "step": 599100
    },
    {
      "epoch": 11.797365675021165,
      "grad_norm": 2.1839611530303955,
      "learning_rate": 7.034215741868465e-10,
      "loss": 1.9685,
      "num_input_tokens_seen": 610109456,
      "step": 599200
    },
    {
      "epoch": 11.799334527770668,
      "grad_norm": 1.9134931564331055,
      "learning_rate": 6.898218219938967e-10,
      "loss": 1.9602,
      "num_input_tokens_seen": 610211088,
      "step": 599300
    },
    {
      "epoch": 11.80130338052017,
      "grad_norm": 1.989763617515564,
      "learning_rate": 6.763547323649921e-10,
      "loss": 1.9667,
      "num_input_tokens_seen": 610312904,
      "step": 599400
    },
    {
      "epoch": 11.803272233269674,
      "grad_norm": 2.162590742111206,
      "learning_rate": 6.630203088781594e-10,
      "loss": 1.9997,
      "num_input_tokens_seen": 610414856,
      "step": 599500
    },
    {
      "epoch": 11.805241086019176,
      "grad_norm": 2.458364248275757,
      "learning_rate": 6.49818555076287e-10,
      "loss": 1.9776,
      "num_input_tokens_seen": 610514968,
      "step": 599600
    },
    {
      "epoch": 11.807209938768679,
      "grad_norm": 1.8544301986694336,
      "learning_rate": 6.367494744669577e-10,
      "loss": 1.937,
      "num_input_tokens_seen": 610617368,
      "step": 599700
    },
    {
      "epoch": 11.809178791518182,
      "grad_norm": 1.9206244945526123,
      "learning_rate": 6.238130705225053e-10,
      "loss": 1.9884,
      "num_input_tokens_seen": 610719000,
      "step": 599800
    },
    {
      "epoch": 11.811147644267685,
      "grad_norm": 1.8682975769042969,
      "learning_rate": 6.110093466800137e-10,
      "loss": 1.9506,
      "num_input_tokens_seen": 610820752,
      "step": 599900
    },
    {
      "epoch": 11.813116497017187,
      "grad_norm": 1.937517523765564,
      "learning_rate": 5.983383063413728e-10,
      "loss": 1.9925,
      "num_input_tokens_seen": 610923152,
      "step": 600000
    },
    {
      "epoch": 11.81508534976669,
      "grad_norm": 1.77732515335083,
      "learning_rate": 5.857999528731672e-10,
      "loss": 2.0176,
      "num_input_tokens_seen": 611024096,
      "step": 600100
    },
    {
      "epoch": 11.817054202516193,
      "grad_norm": 1.8825489282608032,
      "learning_rate": 5.733942896066768e-10,
      "loss": 1.977,
      "num_input_tokens_seen": 611126496,
      "step": 600200
    },
    {
      "epoch": 11.819023055265697,
      "grad_norm": 2.0327060222625732,
      "learning_rate": 5.611213198380427e-10,
      "loss": 2.0058,
      "num_input_tokens_seen": 611228088,
      "step": 600300
    },
    {
      "epoch": 11.8209919080152,
      "grad_norm": 2.0941362380981445,
      "learning_rate": 5.489810468280454e-10,
      "loss": 1.948,
      "num_input_tokens_seen": 611329448,
      "step": 600400
    },
    {
      "epoch": 11.822960760764703,
      "grad_norm": 2.052814483642578,
      "learning_rate": 5.369734738023269e-10,
      "loss": 1.9882,
      "num_input_tokens_seen": 611431072,
      "step": 600500
    },
    {
      "epoch": 11.824929613514206,
      "grad_norm": 1.8776636123657227,
      "learning_rate": 5.250986039511129e-10,
      "loss": 1.9543,
      "num_input_tokens_seen": 611533040,
      "step": 600600
    },
    {
      "epoch": 11.826898466263708,
      "grad_norm": 1.789001226425171,
      "learning_rate": 5.13356440429491e-10,
      "loss": 2.0162,
      "num_input_tokens_seen": 611633952,
      "step": 600700
    },
    {
      "epoch": 11.828867319013211,
      "grad_norm": 1.8962733745574951,
      "learning_rate": 5.017469863572987e-10,
      "loss": 1.97,
      "num_input_tokens_seen": 611736264,
      "step": 600800
    },
    {
      "epoch": 11.830836171762714,
      "grad_norm": 2.056675910949707,
      "learning_rate": 4.902702448190133e-10,
      "loss": 1.9556,
      "num_input_tokens_seen": 611838664,
      "step": 600900
    },
    {
      "epoch": 11.832805024512217,
      "grad_norm": 2.016049861907959,
      "learning_rate": 4.789262188639176e-10,
      "loss": 1.9164,
      "num_input_tokens_seen": 611940568,
      "step": 601000
    },
    {
      "epoch": 11.83477387726172,
      "grad_norm": 1.9803416728973389,
      "learning_rate": 4.677149115059897e-10,
      "loss": 2.0214,
      "num_input_tokens_seen": 612042136,
      "step": 601100
    },
    {
      "epoch": 11.836742730011222,
      "grad_norm": 2.1229701042175293,
      "learning_rate": 4.5663632572412456e-10,
      "loss": 1.9946,
      "num_input_tokens_seen": 612144536,
      "step": 601200
    },
    {
      "epoch": 11.838711582760725,
      "grad_norm": 1.9645607471466064,
      "learning_rate": 4.456904644616344e-10,
      "loss": 1.9869,
      "num_input_tokens_seen": 612246720,
      "step": 601300
    },
    {
      "epoch": 11.840680435510228,
      "grad_norm": 2.0595479011535645,
      "learning_rate": 4.348773306268594e-10,
      "loss": 1.9422,
      "num_input_tokens_seen": 612347272,
      "step": 601400
    },
    {
      "epoch": 11.84264928825973,
      "grad_norm": 2.179272413253784,
      "learning_rate": 4.241969270927237e-10,
      "loss": 2.0274,
      "num_input_tokens_seen": 612448768,
      "step": 601500
    },
    {
      "epoch": 11.844618141009233,
      "grad_norm": 1.8377439975738525,
      "learning_rate": 4.1364925669695736e-10,
      "loss": 1.9705,
      "num_input_tokens_seen": 612551168,
      "step": 601600
    },
    {
      "epoch": 11.846586993758736,
      "grad_norm": 1.7907642126083374,
      "learning_rate": 4.0323432224187435e-10,
      "loss": 2.0292,
      "num_input_tokens_seen": 612652048,
      "step": 601700
    },
    {
      "epoch": 11.84855584650824,
      "grad_norm": 1.8527629375457764,
      "learning_rate": 3.9295212649476105e-10,
      "loss": 1.9715,
      "num_input_tokens_seen": 612753936,
      "step": 601800
    },
    {
      "epoch": 11.850524699257743,
      "grad_norm": 1.6705670356750488,
      "learning_rate": 3.8280267218748773e-10,
      "loss": 1.9632,
      "num_input_tokens_seen": 612856336,
      "step": 601900
    },
    {
      "epoch": 11.852493552007246,
      "grad_norm": 1.7484914064407349,
      "learning_rate": 3.727859620166196e-10,
      "loss": 1.9986,
      "num_input_tokens_seen": 612957960,
      "step": 602000
    },
    {
      "epoch": 11.854462404756749,
      "grad_norm": 1.9333302974700928,
      "learning_rate": 3.6290199864358326e-10,
      "loss": 1.9875,
      "num_input_tokens_seen": 613059136,
      "step": 602100
    },
    {
      "epoch": 11.856431257506252,
      "grad_norm": 1.8948204517364502,
      "learning_rate": 3.531507846943893e-10,
      "loss": 1.9834,
      "num_input_tokens_seen": 613160608,
      "step": 602200
    },
    {
      "epoch": 11.858400110255754,
      "grad_norm": 2.2163522243499756,
      "learning_rate": 3.435323227599096e-10,
      "loss": 1.9761,
      "num_input_tokens_seen": 613262392,
      "step": 602300
    },
    {
      "epoch": 11.860368963005257,
      "grad_norm": 2.157076597213745,
      "learning_rate": 3.340466153957111e-10,
      "loss": 1.9958,
      "num_input_tokens_seen": 613364792,
      "step": 602400
    },
    {
      "epoch": 11.86233781575476,
      "grad_norm": 1.933104157447815,
      "learning_rate": 3.24693665122e-10,
      "loss": 1.9794,
      "num_input_tokens_seen": 613466696,
      "step": 602500
    },
    {
      "epoch": 11.864306668504263,
      "grad_norm": 2.285022020339966,
      "learning_rate": 3.1547347442384406e-10,
      "loss": 1.9959,
      "num_input_tokens_seen": 613567792,
      "step": 602600
    },
    {
      "epoch": 11.866275521253765,
      "grad_norm": 1.8007138967514038,
      "learning_rate": 3.063860457508949e-10,
      "loss": 2.0291,
      "num_input_tokens_seen": 613669760,
      "step": 602700
    },
    {
      "epoch": 11.868244374003268,
      "grad_norm": 1.9165966510772705,
      "learning_rate": 2.9743138151766546e-10,
      "loss": 1.9732,
      "num_input_tokens_seen": 613771232,
      "step": 602800
    },
    {
      "epoch": 11.870213226752771,
      "grad_norm": 1.8733264207839966,
      "learning_rate": 2.886094841033637e-10,
      "loss": 1.9883,
      "num_input_tokens_seen": 613871376,
      "step": 602900
    },
    {
      "epoch": 11.872182079502274,
      "grad_norm": 1.8785486221313477,
      "learning_rate": 2.799203558517815e-10,
      "loss": 2.0485,
      "num_input_tokens_seen": 613972416,
      "step": 603000
    },
    {
      "epoch": 11.874150932251776,
      "grad_norm": 2.178788185119629,
      "learning_rate": 2.713639990716832e-10,
      "loss": 1.9821,
      "num_input_tokens_seen": 614073528,
      "step": 603100
    },
    {
      "epoch": 11.87611978500128,
      "grad_norm": 1.9850804805755615,
      "learning_rate": 2.629404160363058e-10,
      "loss": 1.926,
      "num_input_tokens_seen": 614175928,
      "step": 603200
    },
    {
      "epoch": 11.878088637750782,
      "grad_norm": 2.022782325744629,
      "learning_rate": 2.546496089838035e-10,
      "loss": 1.9753,
      "num_input_tokens_seen": 614278328,
      "step": 603300
    },
    {
      "epoch": 11.880057490500285,
      "grad_norm": 2.0160112380981445,
      "learning_rate": 2.464915801169143e-10,
      "loss": 1.9688,
      "num_input_tokens_seen": 614379912,
      "step": 603400
    },
    {
      "epoch": 11.882026343249787,
      "grad_norm": 2.0020434856414795,
      "learning_rate": 2.384663316032376e-10,
      "loss": 2.0239,
      "num_input_tokens_seen": 614481512,
      "step": 603500
    },
    {
      "epoch": 11.883995195999292,
      "grad_norm": 1.973625659942627,
      "learning_rate": 2.3057386557495674e-10,
      "loss": 1.9905,
      "num_input_tokens_seen": 614583072,
      "step": 603600
    },
    {
      "epoch": 11.885964048748795,
      "grad_norm": 2.12579607963562,
      "learning_rate": 2.2281418412906095e-10,
      "loss": 1.9786,
      "num_input_tokens_seen": 614684672,
      "step": 603700
    },
    {
      "epoch": 11.887932901498298,
      "grad_norm": 1.7405120134353638,
      "learning_rate": 2.151872893271789e-10,
      "loss": 1.9951,
      "num_input_tokens_seen": 614786264,
      "step": 603800
    },
    {
      "epoch": 11.8899017542478,
      "grad_norm": 1.7212185859680176,
      "learning_rate": 2.0769318319580065e-10,
      "loss": 1.9598,
      "num_input_tokens_seen": 614888128,
      "step": 603900
    },
    {
      "epoch": 11.891870606997303,
      "grad_norm": 2.0192983150482178,
      "learning_rate": 2.0033186772594468e-10,
      "loss": 2.0105,
      "num_input_tokens_seen": 614990072,
      "step": 604000
    },
    {
      "epoch": 11.893839459746806,
      "grad_norm": 2.044783115386963,
      "learning_rate": 1.9310334487354641e-10,
      "loss": 1.982,
      "num_input_tokens_seen": 615091848,
      "step": 604100
    },
    {
      "epoch": 11.895808312496309,
      "grad_norm": 1.9060918092727661,
      "learning_rate": 1.8600761655912512e-10,
      "loss": 1.9581,
      "num_input_tokens_seen": 615194248,
      "step": 604200
    },
    {
      "epoch": 11.897777165245811,
      "grad_norm": 2.227484703063965,
      "learning_rate": 1.7904468466795054e-10,
      "loss": 1.9834,
      "num_input_tokens_seen": 615295800,
      "step": 604300
    },
    {
      "epoch": 11.899746017995314,
      "grad_norm": 2.0130362510681152,
      "learning_rate": 1.722145510499873e-10,
      "loss": 2.0045,
      "num_input_tokens_seen": 615398200,
      "step": 604400
    },
    {
      "epoch": 11.901714870744817,
      "grad_norm": 1.6936615705490112,
      "learning_rate": 1.655172175200059e-10,
      "loss": 1.9594,
      "num_input_tokens_seen": 615500072,
      "step": 604500
    },
    {
      "epoch": 11.90368372349432,
      "grad_norm": 2.2261927127838135,
      "learning_rate": 1.589526858574164e-10,
      "loss": 1.9738,
      "num_input_tokens_seen": 615601776,
      "step": 604600
    },
    {
      "epoch": 11.905652576243822,
      "grad_norm": 2.0863723754882812,
      "learning_rate": 1.5252095780637907e-10,
      "loss": 2.0208,
      "num_input_tokens_seen": 615704176,
      "step": 604700
    },
    {
      "epoch": 11.907621428993325,
      "grad_norm": 1.7478595972061157,
      "learning_rate": 1.4622203507569375e-10,
      "loss": 1.9997,
      "num_input_tokens_seen": 615805168,
      "step": 604800
    },
    {
      "epoch": 11.909590281742828,
      "grad_norm": 1.6486343145370483,
      "learning_rate": 1.400559193389106e-10,
      "loss": 2.0192,
      "num_input_tokens_seen": 615907568,
      "step": 604900
    },
    {
      "epoch": 11.91155913449233,
      "grad_norm": 1.9903755187988281,
      "learning_rate": 1.340226122344412e-10,
      "loss": 2.0046,
      "num_input_tokens_seen": 616009968,
      "step": 605000
    },
    {
      "epoch": 11.913527987241833,
      "grad_norm": 1.8391920328140259,
      "learning_rate": 1.2812211536517016e-10,
      "loss": 1.9669,
      "num_input_tokens_seen": 616111552,
      "step": 605100
    },
    {
      "epoch": 11.915496839991338,
      "grad_norm": 1.8323636054992676,
      "learning_rate": 1.223544302988433e-10,
      "loss": 1.9956,
      "num_input_tokens_seen": 616213952,
      "step": 605200
    },
    {
      "epoch": 11.91746569274084,
      "grad_norm": 1.731201410293579,
      "learning_rate": 1.167195585679015e-10,
      "loss": 2.0026,
      "num_input_tokens_seen": 616316352,
      "step": 605300
    },
    {
      "epoch": 11.919434545490343,
      "grad_norm": 2.0433614253997803,
      "learning_rate": 1.1121750166948052e-10,
      "loss": 1.9802,
      "num_input_tokens_seen": 616417960,
      "step": 605400
    },
    {
      "epoch": 11.921403398239846,
      "grad_norm": 1.8426910638809204,
      "learning_rate": 1.05848261065411e-10,
      "loss": 1.9955,
      "num_input_tokens_seen": 616520360,
      "step": 605500
    },
    {
      "epoch": 11.923372250989349,
      "grad_norm": 1.9176217317581177,
      "learning_rate": 1.0061183818227403e-10,
      "loss": 2.008,
      "num_input_tokens_seen": 616622760,
      "step": 605600
    },
    {
      "epoch": 11.925341103738852,
      "grad_norm": 1.7819119691848755,
      "learning_rate": 9.550823441134559e-11,
      "loss": 2.0313,
      "num_input_tokens_seen": 616723584,
      "step": 605700
    },
    {
      "epoch": 11.927309956488354,
      "grad_norm": 1.9518394470214844,
      "learning_rate": 9.053745110859656e-11,
      "loss": 2.01,
      "num_input_tokens_seen": 616824488,
      "step": 605800
    },
    {
      "epoch": 11.929278809237857,
      "grad_norm": 1.9034873247146606,
      "learning_rate": 8.569948959474827e-11,
      "loss": 2.0303,
      "num_input_tokens_seen": 616926888,
      "step": 605900
    },
    {
      "epoch": 11.93124766198736,
      "grad_norm": 2.1896543502807617,
      "learning_rate": 8.099435115521691e-11,
      "loss": 1.968,
      "num_input_tokens_seen": 617028512,
      "step": 606000
    },
    {
      "epoch": 11.933216514736863,
      "grad_norm": 1.953952670097351,
      "learning_rate": 7.642203704000261e-11,
      "loss": 2.0075,
      "num_input_tokens_seen": 617130264,
      "step": 606100
    },
    {
      "epoch": 11.935185367486365,
      "grad_norm": 1.916153073310852,
      "learning_rate": 7.198254846407792e-11,
      "loss": 1.979,
      "num_input_tokens_seen": 617232320,
      "step": 606200
    },
    {
      "epoch": 11.937154220235868,
      "grad_norm": 1.7991586923599243,
      "learning_rate": 6.767588660694378e-11,
      "loss": 2.0086,
      "num_input_tokens_seen": 617334088,
      "step": 606300
    },
    {
      "epoch": 11.939123072985371,
      "grad_norm": 1.894468903541565,
      "learning_rate": 6.350205261274055e-11,
      "loss": 1.9897,
      "num_input_tokens_seen": 617436488,
      "step": 606400
    },
    {
      "epoch": 11.941091925734874,
      "grad_norm": 1.8279722929000854,
      "learning_rate": 5.946104759058102e-11,
      "loss": 2.0329,
      "num_input_tokens_seen": 617537816,
      "step": 606500
    },
    {
      "epoch": 11.943060778484377,
      "grad_norm": 1.7615768909454346,
      "learning_rate": 5.555287261399533e-11,
      "loss": 1.9813,
      "num_input_tokens_seen": 617640216,
      "step": 606600
    },
    {
      "epoch": 11.94502963123388,
      "grad_norm": 1.692265510559082,
      "learning_rate": 5.1777528721430595e-11,
      "loss": 1.9675,
      "num_input_tokens_seen": 617742616,
      "step": 606700
    },
    {
      "epoch": 11.946998483983382,
      "grad_norm": 1.910791039466858,
      "learning_rate": 4.8135016915917815e-11,
      "loss": 2.0206,
      "num_input_tokens_seen": 617844024,
      "step": 606800
    },
    {
      "epoch": 11.948967336732887,
      "grad_norm": 2.1531341075897217,
      "learning_rate": 4.462533816523839e-11,
      "loss": 1.9695,
      "num_input_tokens_seen": 617946424,
      "step": 606900
    },
    {
      "epoch": 11.95093618948239,
      "grad_norm": 2.0287556648254395,
      "learning_rate": 4.1248493401924154e-11,
      "loss": 1.9561,
      "num_input_tokens_seen": 618048824,
      "step": 607000
    },
    {
      "epoch": 11.952905042231892,
      "grad_norm": 1.763012409210205,
      "learning_rate": 3.8004483523201846e-11,
      "loss": 1.9766,
      "num_input_tokens_seen": 618151224,
      "step": 607100
    },
    {
      "epoch": 11.954873894981395,
      "grad_norm": 1.935534954071045,
      "learning_rate": 3.4893309390882084e-11,
      "loss": 2.033,
      "num_input_tokens_seen": 618253480,
      "step": 607200
    },
    {
      "epoch": 11.956842747730898,
      "grad_norm": 1.7291593551635742,
      "learning_rate": 3.191497183169245e-11,
      "loss": 1.9738,
      "num_input_tokens_seen": 618355880,
      "step": 607300
    },
    {
      "epoch": 11.9588116004804,
      "grad_norm": 1.8182733058929443,
      "learning_rate": 2.9069471636833375e-11,
      "loss": 2.0233,
      "num_input_tokens_seen": 618457680,
      "step": 607400
    },
    {
      "epoch": 11.960780453229903,
      "grad_norm": 2.1735570430755615,
      "learning_rate": 2.6356809562422254e-11,
      "loss": 1.9637,
      "num_input_tokens_seen": 618559264,
      "step": 607500
    },
    {
      "epoch": 11.962749305979406,
      "grad_norm": 1.901559829711914,
      "learning_rate": 2.3776986329104853e-11,
      "loss": 1.9714,
      "num_input_tokens_seen": 618661664,
      "step": 607600
    },
    {
      "epoch": 11.964718158728909,
      "grad_norm": 2.0196540355682373,
      "learning_rate": 2.133000262244389e-11,
      "loss": 1.9451,
      "num_input_tokens_seen": 618763152,
      "step": 607700
    },
    {
      "epoch": 11.966687011478411,
      "grad_norm": 2.054826498031616,
      "learning_rate": 1.901585909247494e-11,
      "loss": 2.0128,
      "num_input_tokens_seen": 618865552,
      "step": 607800
    },
    {
      "epoch": 11.968655864227914,
      "grad_norm": 1.9896193742752075,
      "learning_rate": 1.683455635403952e-11,
      "loss": 1.9903,
      "num_input_tokens_seen": 618967856,
      "step": 607900
    },
    {
      "epoch": 11.970624716977417,
      "grad_norm": 1.827958106994629,
      "learning_rate": 1.478609498678507e-11,
      "loss": 2.005,
      "num_input_tokens_seen": 619069488,
      "step": 608000
    },
    {
      "epoch": 11.97259356972692,
      "grad_norm": 1.9769021272659302,
      "learning_rate": 1.2870475534942915e-11,
      "loss": 1.9584,
      "num_input_tokens_seen": 619171888,
      "step": 608100
    },
    {
      "epoch": 11.974562422476422,
      "grad_norm": 2.0425643920898438,
      "learning_rate": 1.1087698507383781e-11,
      "loss": 1.9683,
      "num_input_tokens_seen": 619273856,
      "step": 608200
    },
    {
      "epoch": 11.976531275225925,
      "grad_norm": 1.987949013710022,
      "learning_rate": 9.43776437789534e-12,
      "loss": 1.9758,
      "num_input_tokens_seen": 619374808,
      "step": 608300
    },
    {
      "epoch": 11.978500127975428,
      "grad_norm": 1.9493474960327148,
      "learning_rate": 7.920673584793646e-12,
      "loss": 2.0262,
      "num_input_tokens_seen": 619476568,
      "step": 608400
    },
    {
      "epoch": 11.980468980724932,
      "grad_norm": 8.260712623596191,
      "learning_rate": 6.536426531145167e-12,
      "loss": 1.9921,
      "num_input_tokens_seen": 619578008,
      "step": 608500
    },
    {
      "epoch": 11.982437833474435,
      "grad_norm": 1.871995210647583,
      "learning_rate": 5.28502358476679e-12,
      "loss": 1.991,
      "num_input_tokens_seen": 619680408,
      "step": 608600
    },
    {
      "epoch": 11.984406686223938,
      "grad_norm": 1.9008432626724243,
      "learning_rate": 4.1664650781148e-12,
      "loss": 1.9481,
      "num_input_tokens_seen": 619782808,
      "step": 608700
    },
    {
      "epoch": 11.98637553897344,
      "grad_norm": 1.7159595489501953,
      "learning_rate": 3.1807513084514127e-12,
      "loss": 1.9444,
      "num_input_tokens_seen": 619885000,
      "step": 608800
    },
    {
      "epoch": 11.988344391722944,
      "grad_norm": 2.0267257690429688,
      "learning_rate": 2.3278825375672163e-12,
      "loss": 1.957,
      "num_input_tokens_seen": 619987400,
      "step": 608900
    },
    {
      "epoch": 11.990313244472446,
      "grad_norm": 2.0029587745666504,
      "learning_rate": 1.6078589921142417e-12,
      "loss": 2.0438,
      "num_input_tokens_seen": 620089800,
      "step": 609000
    },
    {
      "epoch": 11.992282097221949,
      "grad_norm": 1.907485842704773,
      "learning_rate": 1.0206808634394271e-12,
      "loss": 1.9305,
      "num_input_tokens_seen": 620192200,
      "step": 609100
    },
    {
      "epoch": 11.994250949971452,
      "grad_norm": 2.12469220161438,
      "learning_rate": 5.663483074735965e-13,
      "loss": 1.9854,
      "num_input_tokens_seen": 620294600,
      "step": 609200
    },
    {
      "epoch": 11.996219802720955,
      "grad_norm": 1.963167667388916,
      "learning_rate": 2.4486144495350357e-13,
      "loss": 1.9765,
      "num_input_tokens_seen": 620397000,
      "step": 609300
    },
    {
      "epoch": 11.998188655470457,
      "grad_norm": 1.826084852218628,
      "learning_rate": 5.622036136632147e-14,
      "loss": 1.9388,
      "num_input_tokens_seen": 620499400,
      "step": 609400
    },
    {
      "epoch": 12.0,
      "num_input_tokens_seen": 620593344,
      "step": 609492,
      "total_flos": 3.16711124803584e+17,
      "train_loss": 0.07982336108915004,
      "train_runtime": 2874.0318,
      "train_samples_per_second": 212.069,
      "train_steps_per_second": 212.069
    }
  ],
  "logging_steps": 100,
  "max_steps": 609492,
  "num_input_tokens_seen": 620593344,
  "num_train_epochs": 12,
  "save_steps": 5000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.16711124803584e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}