{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.12197420799929033,
  "eval_steps": 500,
  "global_step": 11000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 3.2580924034118652,
      "learning_rate": 2.2172949002217296e-06,
      "loss": 10.2933,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.4347386360168457,
      "learning_rate": 4.434589800443459e-06,
      "loss": 10.1894,
      "step": 100
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.3895885944366455,
      "learning_rate": 6.651884700665188e-06,
      "loss": 10.1424,
      "step": 150
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.129647731781006,
      "learning_rate": 8.869179600886918e-06,
      "loss": 10.0995,
      "step": 200
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.3564186096191406,
      "learning_rate": 1.1086474501108649e-05,
      "loss": 10.0479,
      "step": 250
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.830551028251648,
      "learning_rate": 1.3303769401330377e-05,
      "loss": 9.9971,
      "step": 300
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1173911094665527,
      "learning_rate": 1.5521064301552106e-05,
      "loss": 9.9201,
      "step": 350
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6636557579040527,
      "learning_rate": 1.7738359201773837e-05,
      "loss": 9.8562,
      "step": 400
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.4503839015960693,
      "learning_rate": 1.9955654101995567e-05,
      "loss": 9.7599,
      "step": 450
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.822424054145813,
      "learning_rate": 2.2172949002217298e-05,
      "loss": 9.6608,
      "step": 500
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6598998308181763,
      "learning_rate": 2.4390243902439026e-05,
      "loss": 9.55,
      "step": 550
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8471707105636597,
      "learning_rate": 2.6607538802660753e-05,
      "loss": 9.4606,
      "step": 600
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4833533763885498,
      "learning_rate": 2.8824833702882487e-05,
      "loss": 9.3283,
      "step": 650
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.688541054725647,
      "learning_rate": 3.104212860310421e-05,
      "loss": 9.2229,
      "step": 700
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6466543674468994,
      "learning_rate": 3.325942350332594e-05,
      "loss": 9.1093,
      "step": 750
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4169293642044067,
      "learning_rate": 3.547671840354767e-05,
      "loss": 8.9703,
      "step": 800
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7079193592071533,
      "learning_rate": 3.7694013303769404e-05,
      "loss": 8.8351,
      "step": 850
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5513204336166382,
      "learning_rate": 3.9911308203991135e-05,
      "loss": 8.7111,
      "step": 900
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.485573172569275,
      "learning_rate": 4.212860310421286e-05,
      "loss": 8.5627,
      "step": 950
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.511690616607666,
      "learning_rate": 4.4345898004434597e-05,
      "loss": 8.5042,
      "step": 1000
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1478614807128906,
      "learning_rate": 4.656319290465632e-05,
      "loss": 8.3287,
      "step": 1050
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4060652256011963,
      "learning_rate": 4.878048780487805e-05,
      "loss": 8.2341,
      "step": 1100
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3950035572052002,
      "learning_rate": 5.099778270509978e-05,
      "loss": 8.1277,
      "step": 1150
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5197688341140747,
      "learning_rate": 5.3215077605321506e-05,
      "loss": 8.0311,
      "step": 1200
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3406693935394287,
      "learning_rate": 5.543237250554324e-05,
      "loss": 7.9824,
      "step": 1250
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4520119428634644,
      "learning_rate": 5.7649667405764975e-05,
      "loss": 7.9948,
      "step": 1300
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.179124116897583,
      "learning_rate": 5.98669623059867e-05,
      "loss": 7.9144,
      "step": 1350
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4039533138275146,
      "learning_rate": 6.208425720620842e-05,
      "loss": 7.8768,
      "step": 1400
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5542700290679932,
      "learning_rate": 6.430155210643016e-05,
      "loss": 7.894,
      "step": 1450
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4150550365447998,
      "learning_rate": 6.651884700665188e-05,
      "loss": 7.8409,
      "step": 1500
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6647827625274658,
      "learning_rate": 6.873614190687362e-05,
      "loss": 7.91,
      "step": 1550
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7795697450637817,
      "learning_rate": 7.095343680709535e-05,
      "loss": 7.8256,
      "step": 1600
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.933110237121582,
      "learning_rate": 7.317073170731707e-05,
      "loss": 7.8463,
      "step": 1650
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1942570209503174,
      "learning_rate": 7.538802660753881e-05,
      "loss": 7.7827,
      "step": 1700
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6759297847747803,
      "learning_rate": 7.760532150776053e-05,
      "loss": 7.8,
      "step": 1750
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.093256950378418,
      "learning_rate": 7.982261640798227e-05,
      "loss": 7.7461,
      "step": 1800
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.567872166633606,
      "learning_rate": 8.2039911308204e-05,
      "loss": 7.7338,
      "step": 1850
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3017679452896118,
      "learning_rate": 8.425720620842572e-05,
      "loss": 7.804,
      "step": 1900
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7510960102081299,
      "learning_rate": 8.647450110864746e-05,
      "loss": 7.7405,
      "step": 1950
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7215120792388916,
      "learning_rate": 8.869179600886919e-05,
      "loss": 7.7429,
      "step": 2000
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6202715635299683,
      "learning_rate": 9.090909090909092e-05,
      "loss": 7.6588,
      "step": 2050
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5680756568908691,
      "learning_rate": 9.312638580931264e-05,
      "loss": 7.6224,
      "step": 2100
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.462240219116211,
      "learning_rate": 9.534368070953438e-05,
      "loss": 7.6851,
      "step": 2150
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2018320560455322,
      "learning_rate": 9.75609756097561e-05,
      "loss": 7.6443,
      "step": 2200
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9520208835601807,
      "learning_rate": 9.977827050997783e-05,
      "loss": 7.6456,
      "step": 2250
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.115421175956726,
      "learning_rate": 0.00010199556541019956,
      "loss": 7.5894,
      "step": 2300
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6002250909805298,
      "learning_rate": 0.0001042128603104213,
      "loss": 7.6017,
      "step": 2350
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6516796350479126,
      "learning_rate": 0.00010643015521064301,
      "loss": 7.4548,
      "step": 2400
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2168257236480713,
      "learning_rate": 0.00010864745011086475,
      "loss": 7.5867,
      "step": 2450
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5447593927383423,
      "learning_rate": 0.00011086474501108647,
      "loss": 7.5317,
      "step": 2500
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6840906143188477,
      "learning_rate": 0.00011308203991130821,
      "loss": 7.5127,
      "step": 2550
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2965503931045532,
      "learning_rate": 0.00011529933481152995,
      "loss": 7.4911,
      "step": 2600
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.643584966659546,
      "learning_rate": 0.00011751662971175166,
      "loss": 7.4416,
      "step": 2650
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5419111251831055,
      "learning_rate": 0.0001197339246119734,
      "loss": 7.4944,
      "step": 2700
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7774205207824707,
      "learning_rate": 0.00012195121951219512,
      "loss": 7.4244,
      "step": 2750
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1709322929382324,
      "learning_rate": 0.00012416851441241685,
      "loss": 7.371,
      "step": 2800
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5503411293029785,
      "learning_rate": 0.0001263858093126386,
      "loss": 7.3031,
      "step": 2850
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7744035720825195,
      "learning_rate": 0.00012860310421286032,
      "loss": 7.3338,
      "step": 2900
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2014000415802,
      "learning_rate": 0.00013082039911308205,
      "loss": 7.2962,
      "step": 2950
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6716220378875732,
      "learning_rate": 0.00013303769401330377,
      "loss": 7.3348,
      "step": 3000
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7045074701309204,
      "learning_rate": 0.0001352549889135255,
      "loss": 7.2864,
      "step": 3050
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8933771848678589,
      "learning_rate": 0.00013747228381374724,
      "loss": 7.2744,
      "step": 3100
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.298779249191284,
      "learning_rate": 0.00013968957871396897,
      "loss": 7.2472,
      "step": 3150
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3420922756195068,
      "learning_rate": 0.0001419068736141907,
      "loss": 7.3019,
      "step": 3200
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9339039325714111,
      "learning_rate": 0.00014412416851441242,
      "loss": 7.2982,
      "step": 3250
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.69667387008667,
      "learning_rate": 0.00014634146341463414,
      "loss": 7.2851,
      "step": 3300
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3124189376831055,
      "learning_rate": 0.0001485587583148559,
      "loss": 7.258,
      "step": 3350
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.975651741027832,
      "learning_rate": 0.00015077605321507762,
      "loss": 7.1275,
      "step": 3400
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9704022407531738,
      "learning_rate": 0.00015299334811529934,
      "loss": 7.1473,
      "step": 3450
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5047757625579834,
      "learning_rate": 0.00015521064301552106,
      "loss": 7.1096,
      "step": 3500
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5465894937515259,
      "learning_rate": 0.0001574279379157428,
      "loss": 7.1501,
      "step": 3550
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9557933807373047,
      "learning_rate": 0.00015964523281596454,
      "loss": 7.2033,
      "step": 3600
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.420116424560547,
      "learning_rate": 0.00016186252771618626,
      "loss": 7.1275,
      "step": 3650
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.114737033843994,
      "learning_rate": 0.000164079822616408,
      "loss": 7.0932,
      "step": 3700
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3085389137268066,
      "learning_rate": 0.00016629711751662974,
      "loss": 7.0311,
      "step": 3750
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5679140090942383,
      "learning_rate": 0.00016851441241685144,
      "loss": 6.9168,
      "step": 3800
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8611838817596436,
      "learning_rate": 0.0001707317073170732,
      "loss": 7.0085,
      "step": 3850
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8603994846343994,
      "learning_rate": 0.0001729490022172949,
      "loss": 6.9432,
      "step": 3900
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4244627952575684,
      "learning_rate": 0.00017516629711751663,
      "loss": 6.9333,
      "step": 3950
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.177870750427246,
      "learning_rate": 0.00017738359201773839,
      "loss": 6.9499,
      "step": 4000
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9320554733276367,
      "learning_rate": 0.00017960088691796008,
      "loss": 6.8204,
      "step": 4050
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5062849521636963,
      "learning_rate": 0.00018181818181818183,
      "loss": 6.9505,
      "step": 4100
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.9272422790527344,
      "learning_rate": 0.00018403547671840356,
      "loss": 6.8701,
      "step": 4150
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0309596061706543,
      "learning_rate": 0.00018625277161862528,
      "loss": 6.924,
      "step": 4200
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0265886783599854,
      "learning_rate": 0.00018847006651884703,
      "loss": 6.9223,
      "step": 4250
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5160486698150635,
      "learning_rate": 0.00019068736141906876,
      "loss": 6.8708,
      "step": 4300
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.613301992416382,
      "learning_rate": 0.00019290465631929045,
      "loss": 6.8937,
      "step": 4350
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3031229972839355,
      "learning_rate": 0.0001951219512195122,
      "loss": 6.8337,
      "step": 4400
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.54779052734375,
      "learning_rate": 0.00019733924611973393,
      "loss": 6.8334,
      "step": 4450
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.8277971744537354,
      "learning_rate": 0.00019955654101995565,
      "loss": 6.7925,
      "step": 4500
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0113885402679443,
      "learning_rate": 0.00019999989242739025,
      "loss": 6.8458,
      "step": 4550
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2395377159118652,
      "learning_rate": 0.00019999945541405976,
      "loss": 6.6251,
      "step": 4600
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.445993423461914,
      "learning_rate": 0.0001999986822381884,
      "loss": 6.8099,
      "step": 4650
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.077752590179443,
      "learning_rate": 0.0001999975729023753,
      "loss": 6.8053,
      "step": 4700
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.167569875717163,
      "learning_rate": 0.00019999612741034963,
      "loss": 6.7706,
      "step": 4750
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.893659234046936,
      "learning_rate": 0.00019999434576697066,
      "loss": 6.8245,
      "step": 4800
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.6101326942443848,
      "learning_rate": 0.00019999222797822762,
      "loss": 6.7407,
      "step": 4850
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2858726978302,
      "learning_rate": 0.00019998977405123974,
      "loss": 6.74,
      "step": 4900
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9325459003448486,
      "learning_rate": 0.0001999869839942563,
      "loss": 6.716,
      "step": 4950
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0043437480926514,
      "learning_rate": 0.00019998385781665643,
      "loss": 6.6003,
      "step": 5000
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.151523113250732,
      "learning_rate": 0.00019998039552894924,
      "loss": 6.6801,
      "step": 5050
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.8407771587371826,
      "learning_rate": 0.00019997659714277372,
      "loss": 6.608,
      "step": 5100
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.230713129043579,
      "learning_rate": 0.00019997246267089867,
      "loss": 6.6479,
      "step": 5150
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2546942234039307,
      "learning_rate": 0.0001999679921272227,
      "loss": 6.6548,
      "step": 5200
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.180986166000366,
      "learning_rate": 0.00019996318552677425,
      "loss": 6.6851,
      "step": 5250
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.341231346130371,
      "learning_rate": 0.00019995804288571134,
      "loss": 6.547,
      "step": 5300
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1117124557495117,
      "learning_rate": 0.00019995256422132172,
      "loss": 6.7072,
      "step": 5350
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0082530975341797,
      "learning_rate": 0.0001999467495520227,
      "loss": 6.5422,
      "step": 5400
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.409489870071411,
      "learning_rate": 0.0001999405988973611,
      "loss": 6.3716,
      "step": 5450
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.649052381515503,
      "learning_rate": 0.00019993411227801328,
      "loss": 6.6434,
      "step": 5500
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.081116199493408,
      "learning_rate": 0.00019992728971578492,
      "loss": 6.4624,
      "step": 5550
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1578280925750732,
      "learning_rate": 0.00019992013123361102,
      "loss": 6.5416,
      "step": 5600
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.7874557971954346,
      "learning_rate": 0.0001999126368555559,
      "loss": 6.4512,
      "step": 5650
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.7693099975585938,
      "learning_rate": 0.00019990480660681293,
      "loss": 6.5105,
      "step": 5700
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4338185787200928,
      "learning_rate": 0.00019989680712666593,
      "loss": 6.5092,
      "step": 5750
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.656937837600708,
      "learning_rate": 0.00019988831193270577,
      "loss": 6.4269,
      "step": 5800
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.857292652130127,
      "learning_rate": 0.00019987948094982952,
      "loss": 6.4387,
      "step": 5850
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.4963467121124268,
      "learning_rate": 0.00019987031420772385,
      "loss": 6.3851,
      "step": 5900
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.602522611618042,
      "learning_rate": 0.00019986081173720396,
      "loss": 6.3413,
      "step": 5950
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.6455273628234863,
      "learning_rate": 0.00019985097357021385,
      "loss": 6.2965,
      "step": 6000
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.5592167377471924,
      "learning_rate": 0.0001998407997398259,
      "loss": 6.4293,
      "step": 6050
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.6016533374786377,
      "learning_rate": 0.00019983029028024094,
      "loss": 6.2897,
      "step": 6100
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5536839962005615,
      "learning_rate": 0.000199819445226788,
      "loss": 6.3157,
      "step": 6150
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0514349937438965,
      "learning_rate": 0.00019980826461592427,
      "loss": 6.3847,
      "step": 6200
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.72495174407959,
      "learning_rate": 0.00019979674848523505,
      "loss": 6.3517,
      "step": 6250
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.4264872074127197,
      "learning_rate": 0.00019978489687343335,
      "loss": 6.2533,
      "step": 6300
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8361423015594482,
      "learning_rate": 0.0001997727098203602,
      "loss": 6.3654,
      "step": 6350
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.9690892696380615,
      "learning_rate": 0.00019976018736698404,
      "loss": 6.3968,
      "step": 6400
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.6132867336273193,
      "learning_rate": 0.0001997473295554009,
      "loss": 6.3444,
      "step": 6450
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.820697784423828,
      "learning_rate": 0.00019973413642883424,
      "loss": 6.2019,
      "step": 6500
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2316782474517822,
      "learning_rate": 0.00019972060803163458,
      "loss": 6.2049,
      "step": 6550
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.9528305530548096,
      "learning_rate": 0.00019970674440927957,
      "loss": 6.1718,
      "step": 6600
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.891073226928711,
      "learning_rate": 0.0001996925456083738,
      "loss": 6.2393,
      "step": 6650
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.813270092010498,
      "learning_rate": 0.00019967801167664853,
      "loss": 6.2116,
      "step": 6700
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2726826667785645,
      "learning_rate": 0.00019966314266296173,
      "loss": 6.1521,
      "step": 6750
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3895318508148193,
      "learning_rate": 0.00019964793861729772,
      "loss": 6.1072,
      "step": 6800
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.190431833267212,
      "learning_rate": 0.000199632399590767,
      "loss": 6.2009,
      "step": 6850
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.79266095161438,
      "learning_rate": 0.00019961652563560634,
      "loss": 6.028,
      "step": 6900
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.260039806365967,
      "learning_rate": 0.00019960031680517826,
      "loss": 6.0733,
      "step": 6950
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0739686489105225,
      "learning_rate": 0.0001995837731539711,
      "loss": 6.0521,
      "step": 7000
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0517771244049072,
      "learning_rate": 0.00019956689473759872,
      "loss": 6.0544,
      "step": 7050
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.9524648189544678,
      "learning_rate": 0.0001995496816128003,
      "loss": 6.1326,
      "step": 7100
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.498497486114502,
      "learning_rate": 0.00019953213383744033,
      "loss": 6.236,
      "step": 7150
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.157576084136963,
      "learning_rate": 0.00019951425147050807,
      "loss": 5.9898,
      "step": 7200
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.9297516345977783,
      "learning_rate": 0.00019949603457211775,
      "loss": 6.086,
      "step": 7250
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.3214786052703857,
      "learning_rate": 0.00019947748320350804,
      "loss": 5.9589,
      "step": 7300
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8847291469573975,
      "learning_rate": 0.00019945859742704201,
      "loss": 6.1931,
      "step": 7350
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.387896776199341,
      "learning_rate": 0.00019943937730620702,
      "loss": 6.0539,
      "step": 7400
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.1214797496795654,
      "learning_rate": 0.00019941982290561417,
      "loss": 6.0288,
      "step": 7450
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.7995123863220215,
      "learning_rate": 0.00019939993429099841,
      "loss": 6.0526,
      "step": 7500
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.788393974304199,
      "learning_rate": 0.00019937971152921818,
      "loss": 5.9799,
      "step": 7550
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.009220123291016,
      "learning_rate": 0.0001993591546882552,
      "loss": 6.1223,
      "step": 7600
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.5576276779174805,
      "learning_rate": 0.00019933826383721428,
      "loss": 5.989,
      "step": 7650
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.1287412643432617,
      "learning_rate": 0.00019931703904632294,
      "loss": 6.0542,
      "step": 7700
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.6518595218658447,
      "learning_rate": 0.00019929548038693146,
      "loss": 6.041,
      "step": 7750
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.268080472946167,
      "learning_rate": 0.0001992735879315123,
      "loss": 5.888,
      "step": 7800
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.6055593490600586,
      "learning_rate": 0.00019925136175366007,
      "loss": 5.913,
      "step": 7850
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.866463661193848,
      "learning_rate": 0.00019922880192809137,
      "loss": 5.9858,
      "step": 7900
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.44808292388916,
      "learning_rate": 0.00019920590853064423,
      "loss": 5.7686,
      "step": 7950
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.9507765769958496,
      "learning_rate": 0.00019918268163827808,
      "loss": 5.8557,
      "step": 8000
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.441870927810669,
      "learning_rate": 0.00019915912132907352,
      "loss": 5.8268,
      "step": 8050
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.838809013366699,
      "learning_rate": 0.00019913522768223182,
      "loss": 5.9833,
      "step": 8100
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.165487289428711,
      "learning_rate": 0.00019911100077807498,
      "loss": 5.7422,
      "step": 8150
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.5947463512420654,
      "learning_rate": 0.0001990864406980452,
      "loss": 5.7479,
      "step": 8200
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.130446434020996,
      "learning_rate": 0.00019906154752470472,
      "loss": 5.7767,
      "step": 8250
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.866550922393799,
      "learning_rate": 0.00019903632134173554,
      "loss": 5.7681,
      "step": 8300
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.2839725017547607,
      "learning_rate": 0.00019901076223393903,
      "loss": 5.6656,
      "step": 8350
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0762476921081543,
      "learning_rate": 0.0001989848702872359,
      "loss": 5.789,
      "step": 8400
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.7109107971191406,
      "learning_rate": 0.00019895864558866556,
      "loss": 5.773,
      "step": 8450
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.400998115539551,
      "learning_rate": 0.00019893208822638618,
      "loss": 5.7506,
      "step": 8500
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.3062849044799805,
      "learning_rate": 0.00019890519828967413,
      "loss": 5.7515,
      "step": 8550
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.109920501708984,
      "learning_rate": 0.00019887797586892373,
      "loss": 5.7972,
      "step": 8600
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.4838390350341797,
      "learning_rate": 0.00019885042105564717,
      "loss": 5.6753,
      "step": 8650
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.251760959625244,
      "learning_rate": 0.00019882253394247381,
      "loss": 5.6303,
      "step": 8700
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.042376518249512,
      "learning_rate": 0.00019879431462315025,
      "loss": 5.5753,
      "step": 8750
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.239652633666992,
      "learning_rate": 0.0001987657631925398,
      "loss": 5.5335,
      "step": 8800
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.15481424331665,
      "learning_rate": 0.00019873687974662215,
      "loss": 5.5396,
      "step": 8850
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.36835241317749,
      "learning_rate": 0.00019870766438249317,
      "loss": 5.6017,
      "step": 8900
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.165258407592773,
      "learning_rate": 0.00019867811719836452,
      "loss": 5.7228,
      "step": 8950
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.125988006591797,
      "learning_rate": 0.0001986482382935633,
      "loss": 5.5787,
      "step": 9000
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.177731037139893,
      "learning_rate": 0.0001986180277685317,
      "loss": 5.5829,
      "step": 9050
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.006561279296875,
      "learning_rate": 0.00019858748572482683,
      "loss": 5.5466,
      "step": 9100
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.33070182800293,
      "learning_rate": 0.00019855661226512007,
      "loss": 5.5544,
      "step": 9150
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.358560085296631,
      "learning_rate": 0.00019852540749319708,
      "loss": 5.4599,
      "step": 9200
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.536096096038818,
      "learning_rate": 0.00019849387151395708,
      "loss": 5.4983,
      "step": 9250
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.66163444519043,
      "learning_rate": 0.0001984620044334129,
      "loss": 5.4097,
      "step": 9300
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.4319233894348145,
      "learning_rate": 0.00019842980635869024,
      "loss": 5.4093,
      "step": 9350
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.98419713973999,
      "learning_rate": 0.0001983972773980276,
      "loss": 5.4056,
      "step": 9400
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.6354339122772217,
      "learning_rate": 0.0001983644176607757,
      "loss": 5.3171,
      "step": 9450
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.495342254638672,
      "learning_rate": 0.00019833122725739736,
      "loss": 5.4521,
      "step": 9500
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.5558671951293945,
      "learning_rate": 0.00019829770629946678,
      "loss": 5.5158,
      "step": 9550
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.7165732383728027,
      "learning_rate": 0.00019826385489966957,
      "loss": 5.301,
      "step": 9600
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.030915260314941,
      "learning_rate": 0.00019822967317180204,
      "loss": 5.3316,
      "step": 9650
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.385923385620117,
      "learning_rate": 0.00019819516123077094,
      "loss": 5.3844,
      "step": 9700
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.383516788482666,
      "learning_rate": 0.00019816101926755305,
      "loss": 5.2995,
      "step": 9750
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.446406364440918,
      "learning_rate": 0.00019812585384780055,
      "loss": 5.386,
      "step": 9800
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.345483303070068,
      "learning_rate": 0.00019809035856388805,
      "loss": 5.2815,
      "step": 9850
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.791261672973633,
      "learning_rate": 0.00019805453353513813,
      "loss": 5.3757,
      "step": 9900
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.622151851654053,
      "learning_rate": 0.00019801837888198172,
      "loss": 5.4405,
      "step": 9950
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.934606075286865,
      "learning_rate": 0.0001979818947259579,
      "loss": 5.139,
      "step": 10000
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.9659693241119385,
      "learning_rate": 0.0001979450811897134,
      "loss": 5.1726,
      "step": 10050
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.214992046356201,
      "learning_rate": 0.00019790793839700226,
      "loss": 5.2864,
      "step": 10100
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.5359601974487305,
      "learning_rate": 0.00019787046647268524,
      "loss": 5.1443,
      "step": 10150
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.26462984085083,
      "learning_rate": 0.00019783266554272962,
      "loss": 5.0597,
      "step": 10200
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.053945064544678,
      "learning_rate": 0.00019779453573420873,
      "loss": 5.2946,
      "step": 10250
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.082211494445801,
      "learning_rate": 0.00019775607717530127,
      "loss": 5.2075,
      "step": 10300
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.107390403747559,
      "learning_rate": 0.00019771728999529132,
      "loss": 5.1394,
      "step": 10350
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.58411169052124,
      "learning_rate": 0.00019767817432456752,
      "loss": 5.1064,
      "step": 10400
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.38965892791748,
      "learning_rate": 0.00019763952239228627,
      "loss": 5.0808,
      "step": 10450
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.885803699493408,
      "learning_rate": 0.00019759975669894338,
      "loss": 5.0664,
      "step": 10500
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.1605916023254395,
      "learning_rate": 0.00019755966290999167,
      "loss": 5.2469,
      "step": 10550
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.821887016296387,
      "learning_rate": 0.00019751924116021225,
      "loss": 5.2451,
      "step": 10600
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.865694761276245,
      "learning_rate": 0.00019747849158548858,
      "loss": 5.2334,
      "step": 10650
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.640681028366089,
      "learning_rate": 0.00019743741432280625,
      "loss": 5.1206,
      "step": 10700
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.04166316986084,
      "learning_rate": 0.00019739600951025236,
      "loss": 5.0059,
      "step": 10750
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.637605667114258,
      "learning_rate": 0.00019735427728701516,
      "loss": 5.0302,
      "step": 10800
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.08723783493042,
      "learning_rate": 0.0001973122177933835,
      "loss": 5.1551,
      "step": 10850
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.7944953441619873,
      "learning_rate": 0.00019726983117074643,
      "loss": 5.0665,
      "step": 10900
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.2847371101379395,
      "learning_rate": 0.00019722711756159266,
      "loss": 5.2212,
      "step": 10950
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.109150409698486,
      "learning_rate": 0.00019718407710951012,
      "loss": 5.2645,
      "step": 11000
    }
  ],
  "logging_steps": 50,
  "max_steps": 90183,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 3955495796736000.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}