{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9980866629150253,
  "eval_steps": 500,
  "global_step": 1480,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0013505908835115363,
      "grad_norm": 2.148791476511835,
      "learning_rate": 6.756756756756757e-08,
      "loss": 2.1725,
      "step": 1
    },
    {
      "epoch": 0.006752954417557681,
      "grad_norm": 2.0693435263055955,
      "learning_rate": 3.378378378378379e-07,
      "loss": 2.1697,
      "step": 5
    },
    {
      "epoch": 0.013505908835115363,
      "grad_norm": 1.698535954485606,
      "learning_rate": 6.756756756756758e-07,
      "loss": 2.0659,
      "step": 10
    },
    {
      "epoch": 0.020258863252673044,
      "grad_norm": 0.9452543971000898,
      "learning_rate": 1.0135135135135136e-06,
      "loss": 1.8928,
      "step": 15
    },
    {
      "epoch": 0.027011817670230726,
      "grad_norm": 0.8038988537572226,
      "learning_rate": 1.3513513513513515e-06,
      "loss": 1.5709,
      "step": 20
    },
    {
      "epoch": 0.03376477208778841,
      "grad_norm": 0.471508019359059,
      "learning_rate": 1.6891891891891894e-06,
      "loss": 1.1256,
      "step": 25
    },
    {
      "epoch": 0.04051772650534609,
      "grad_norm": 0.252352654342694,
      "learning_rate": 2.0270270270270273e-06,
      "loss": 0.9097,
      "step": 30
    },
    {
      "epoch": 0.04727068092290377,
      "grad_norm": 0.24974204597161465,
      "learning_rate": 2.364864864864865e-06,
      "loss": 0.7799,
      "step": 35
    },
    {
      "epoch": 0.05402363534046145,
      "grad_norm": 0.20795816160607852,
      "learning_rate": 2.702702702702703e-06,
      "loss": 0.7137,
      "step": 40
    },
    {
      "epoch": 0.06077658975801913,
      "grad_norm": 0.17844205306863006,
      "learning_rate": 3.040540540540541e-06,
      "loss": 0.5982,
      "step": 45
    },
    {
      "epoch": 0.06752954417557681,
      "grad_norm": 0.14449057063114276,
      "learning_rate": 3.3783783783783788e-06,
      "loss": 0.5388,
      "step": 50
    },
    {
      "epoch": 0.0742824985931345,
      "grad_norm": 0.12180944441742095,
      "learning_rate": 3.7162162162162162e-06,
      "loss": 0.5621,
      "step": 55
    },
    {
      "epoch": 0.08103545301069218,
      "grad_norm": 0.11672624417203623,
      "learning_rate": 4.0540540540540545e-06,
      "loss": 0.4885,
      "step": 60
    },
    {
      "epoch": 0.08778840742824986,
      "grad_norm": 0.12365779514235553,
      "learning_rate": 4.391891891891892e-06,
      "loss": 0.4725,
      "step": 65
    },
    {
      "epoch": 0.09454136184580754,
      "grad_norm": 0.13848942929552646,
      "learning_rate": 4.72972972972973e-06,
      "loss": 0.4737,
      "step": 70
    },
    {
      "epoch": 0.10129431626336523,
      "grad_norm": 0.13947287893715526,
      "learning_rate": 5.067567567567568e-06,
      "loss": 0.4686,
      "step": 75
    },
    {
      "epoch": 0.1080472706809229,
      "grad_norm": 0.10498499488363304,
      "learning_rate": 5.405405405405406e-06,
      "loss": 0.4607,
      "step": 80
    },
    {
      "epoch": 0.11480022509848059,
      "grad_norm": 0.1263112338839764,
      "learning_rate": 5.743243243243244e-06,
      "loss": 0.471,
      "step": 85
    },
    {
      "epoch": 0.12155317951603826,
      "grad_norm": 0.1082136353357591,
      "learning_rate": 6.081081081081082e-06,
      "loss": 0.432,
      "step": 90
    },
    {
      "epoch": 0.12830613393359594,
      "grad_norm": 0.08923160665434038,
      "learning_rate": 6.41891891891892e-06,
      "loss": 0.4433,
      "step": 95
    },
    {
      "epoch": 0.13505908835115363,
      "grad_norm": 0.11807106476588053,
      "learning_rate": 6.7567567567567575e-06,
      "loss": 0.4398,
      "step": 100
    },
    {
      "epoch": 0.14181204276871132,
      "grad_norm": 0.09395401671145927,
      "learning_rate": 7.0945945945945946e-06,
      "loss": 0.3899,
      "step": 105
    },
    {
      "epoch": 0.148564997186269,
      "grad_norm": 0.09428015614908754,
      "learning_rate": 7.4324324324324324e-06,
      "loss": 0.3983,
      "step": 110
    },
    {
      "epoch": 0.15531795160382666,
      "grad_norm": 0.08605902096557952,
      "learning_rate": 7.77027027027027e-06,
      "loss": 0.415,
      "step": 115
    },
    {
      "epoch": 0.16207090602138435,
      "grad_norm": 0.08713323888522238,
      "learning_rate": 8.108108108108109e-06,
      "loss": 0.4075,
      "step": 120
    },
    {
      "epoch": 0.16882386043894204,
      "grad_norm": 0.07858143188282206,
      "learning_rate": 8.445945945945948e-06,
      "loss": 0.4089,
      "step": 125
    },
    {
      "epoch": 0.17557681485649973,
      "grad_norm": 0.07427563021554226,
      "learning_rate": 8.783783783783785e-06,
      "loss": 0.4291,
      "step": 130
    },
    {
      "epoch": 0.1823297692740574,
      "grad_norm": 0.09144835191366023,
      "learning_rate": 9.121621621621622e-06,
      "loss": 0.3895,
      "step": 135
    },
    {
      "epoch": 0.18908272369161508,
      "grad_norm": 0.07052586919720526,
      "learning_rate": 9.45945945945946e-06,
      "loss": 0.4029,
      "step": 140
    },
    {
      "epoch": 0.19583567810917277,
      "grad_norm": 0.0769561636008079,
      "learning_rate": 9.797297297297298e-06,
      "loss": 0.4148,
      "step": 145
    },
    {
      "epoch": 0.20258863252673046,
      "grad_norm": 0.07216832297932871,
      "learning_rate": 9.999944372378571e-06,
      "loss": 0.3801,
      "step": 150
    },
    {
      "epoch": 0.20934158694428812,
      "grad_norm": 0.06883498336890326,
      "learning_rate": 9.999318575852451e-06,
      "loss": 0.3728,
      "step": 155
    },
    {
      "epoch": 0.2160945413618458,
      "grad_norm": 0.07007134671747867,
      "learning_rate": 9.99799753559161e-06,
      "loss": 0.3831,
      "step": 160
    },
    {
      "epoch": 0.2228474957794035,
      "grad_norm": 0.0638944841323381,
      "learning_rate": 9.995981435310078e-06,
      "loss": 0.3852,
      "step": 165
    },
    {
      "epoch": 0.22960045019696118,
      "grad_norm": 0.0732472641559478,
      "learning_rate": 9.993270555382283e-06,
      "loss": 0.4058,
      "step": 170
    },
    {
      "epoch": 0.23635340461451884,
      "grad_norm": 0.06688107743191332,
      "learning_rate": 9.989865272804064e-06,
      "loss": 0.3981,
      "step": 175
    },
    {
      "epoch": 0.24310635903207653,
      "grad_norm": 0.0750542374720311,
      "learning_rate": 9.985766061140233e-06,
      "loss": 0.4034,
      "step": 180
    },
    {
      "epoch": 0.24985931344963422,
      "grad_norm": 0.06261496896671846,
      "learning_rate": 9.980973490458728e-06,
      "loss": 0.3611,
      "step": 185
    },
    {
      "epoch": 0.2566122678671919,
      "grad_norm": 0.07059993646852766,
      "learning_rate": 9.97548822725133e-06,
      "loss": 0.3603,
      "step": 190
    },
    {
      "epoch": 0.26336522228474957,
      "grad_norm": 0.06560938196431165,
      "learning_rate": 9.969311034340977e-06,
      "loss": 0.3853,
      "step": 195
    },
    {
      "epoch": 0.27011817670230726,
      "grad_norm": 0.05548643801780881,
      "learning_rate": 9.962442770775675e-06,
      "loss": 0.3659,
      "step": 200
    },
    {
      "epoch": 0.27687113111986494,
      "grad_norm": 0.06003482909108544,
      "learning_rate": 9.954884391709043e-06,
      "loss": 0.3907,
      "step": 205
    },
    {
      "epoch": 0.28362408553742263,
      "grad_norm": 0.04978856362198666,
      "learning_rate": 9.946636948267468e-06,
      "loss": 0.349,
      "step": 210
    },
    {
      "epoch": 0.2903770399549803,
      "grad_norm": 0.056553641270721766,
      "learning_rate": 9.937701587403941e-06,
      "loss": 0.4041,
      "step": 215
    },
    {
      "epoch": 0.297129994372538,
      "grad_norm": 0.06048923977199324,
      "learning_rate": 9.928079551738542e-06,
      "loss": 0.3704,
      "step": 220
    },
    {
      "epoch": 0.30388294879009564,
      "grad_norm": 0.055403005803656744,
      "learning_rate": 9.91777217938564e-06,
      "loss": 0.3765,
      "step": 225
    },
    {
      "epoch": 0.31063590320765333,
      "grad_norm": 0.05878900335636033,
      "learning_rate": 9.906780903767799e-06,
      "loss": 0.3927,
      "step": 230
    },
    {
      "epoch": 0.317388857625211,
      "grad_norm": 0.05009538620005328,
      "learning_rate": 9.895107253416434e-06,
      "loss": 0.3826,
      "step": 235
    },
    {
      "epoch": 0.3241418120427687,
      "grad_norm": 0.06193069860962348,
      "learning_rate": 9.882752851759247e-06,
      "loss": 0.3694,
      "step": 240
    },
    {
      "epoch": 0.3308947664603264,
      "grad_norm": 0.05658978100208689,
      "learning_rate": 9.869719416894462e-06,
      "loss": 0.3729,
      "step": 245
    },
    {
      "epoch": 0.3376477208778841,
      "grad_norm": 0.060376802387617515,
      "learning_rate": 9.856008761351882e-06,
      "loss": 0.373,
      "step": 250
    },
    {
      "epoch": 0.34440067529544177,
      "grad_norm": 0.047418407201082834,
      "learning_rate": 9.841622791840839e-06,
      "loss": 0.3343,
      "step": 255
    },
    {
      "epoch": 0.35115362971299946,
      "grad_norm": 0.062039985277230605,
      "learning_rate": 9.826563508985017e-06,
      "loss": 0.3897,
      "step": 260
    },
    {
      "epoch": 0.3579065841305571,
      "grad_norm": 0.04965955489714088,
      "learning_rate": 9.810833007044247e-06,
      "loss": 0.3925,
      "step": 265
    },
    {
      "epoch": 0.3646595385481148,
      "grad_norm": 0.05316606675382464,
      "learning_rate": 9.794433473623249e-06,
      "loss": 0.377,
      "step": 270
    },
    {
      "epoch": 0.37141249296567247,
      "grad_norm": 0.0467476855297103,
      "learning_rate": 9.777367189367412e-06,
      "loss": 0.3779,
      "step": 275
    },
    {
      "epoch": 0.37816544738323016,
      "grad_norm": 0.04978059836724344,
      "learning_rate": 9.759636527645633e-06,
      "loss": 0.3521,
      "step": 280
    },
    {
      "epoch": 0.38491840180078785,
      "grad_norm": 0.05999211649054394,
      "learning_rate": 9.74124395422025e-06,
      "loss": 0.3725,
      "step": 285
    },
    {
      "epoch": 0.39167135621834553,
      "grad_norm": 0.04365453179095536,
      "learning_rate": 9.722192026904145e-06,
      "loss": 0.3402,
      "step": 290
    },
    {
      "epoch": 0.3984243106359032,
      "grad_norm": 0.04616864773915026,
      "learning_rate": 9.702483395205023e-06,
      "loss": 0.3655,
      "step": 295
    },
    {
      "epoch": 0.4051772650534609,
      "grad_norm": 0.04782080283525874,
      "learning_rate": 9.682120799956961e-06,
      "loss": 0.3263,
      "step": 300
    },
    {
      "epoch": 0.41193021947101854,
      "grad_norm": 0.0498656978185221,
      "learning_rate": 9.661107072939244e-06,
      "loss": 0.3387,
      "step": 305
    },
    {
      "epoch": 0.41868317388857623,
      "grad_norm": 0.04142545069347494,
      "learning_rate": 9.639445136482549e-06,
      "loss": 0.3388,
      "step": 310
    },
    {
      "epoch": 0.4254361283061339,
      "grad_norm": 0.04183998822986712,
      "learning_rate": 9.61713800306255e-06,
      "loss": 0.3615,
      "step": 315
    },
    {
      "epoch": 0.4321890827236916,
      "grad_norm": 0.045933316046675596,
      "learning_rate": 9.594188774880981e-06,
      "loss": 0.3607,
      "step": 320
    },
    {
      "epoch": 0.4389420371412493,
      "grad_norm": 0.04805726955974,
      "learning_rate": 9.570600643434217e-06,
      "loss": 0.3371,
      "step": 325
    },
    {
      "epoch": 0.445694991558807,
      "grad_norm": 0.04569635226134762,
      "learning_rate": 9.546376889069442e-06,
      "loss": 0.3582,
      "step": 330
    },
    {
      "epoch": 0.4524479459763647,
      "grad_norm": 0.050347567122321094,
      "learning_rate": 9.521520880528453e-06,
      "loss": 0.3831,
      "step": 335
    },
    {
      "epoch": 0.45920090039392236,
      "grad_norm": 0.051704382901289696,
      "learning_rate": 9.496036074479184e-06,
      "loss": 0.3604,
      "step": 340
    },
    {
      "epoch": 0.46595385481148,
      "grad_norm": 0.04717995332220621,
      "learning_rate": 9.46992601503499e-06,
      "loss": 0.3498,
      "step": 345
    },
    {
      "epoch": 0.4727068092290377,
      "grad_norm": 0.04394756827213909,
      "learning_rate": 9.44319433326178e-06,
      "loss": 0.3711,
      "step": 350
    },
    {
      "epoch": 0.47945976364659537,
      "grad_norm": 0.05455789140928886,
      "learning_rate": 9.415844746673047e-06,
      "loss": 0.3395,
      "step": 355
    },
    {
      "epoch": 0.48621271806415306,
      "grad_norm": 0.0486039405429815,
      "learning_rate": 9.387881058712888e-06,
      "loss": 0.3596,
      "step": 360
    },
    {
      "epoch": 0.49296567248171075,
      "grad_norm": 0.049127971484592976,
      "learning_rate": 9.359307158227067e-06,
      "loss": 0.3759,
      "step": 365
    },
    {
      "epoch": 0.49971862689926844,
      "grad_norm": 0.05804448606366045,
      "learning_rate": 9.330127018922195e-06,
      "loss": 0.3741,
      "step": 370
    },
    {
      "epoch": 0.5064715813168261,
      "grad_norm": 0.05121055195790688,
      "learning_rate": 9.300344698813124e-06,
      "loss": 0.3978,
      "step": 375
    },
    {
      "epoch": 0.5132245357343838,
      "grad_norm": 0.05072015885221285,
      "learning_rate": 9.269964339658605e-06,
      "loss": 0.349,
      "step": 380
    },
    {
      "epoch": 0.5199774901519415,
      "grad_norm": 0.0446955462817659,
      "learning_rate": 9.238990166385304e-06,
      "loss": 0.3783,
      "step": 385
    },
    {
      "epoch": 0.5267304445694991,
      "grad_norm": 0.047869039864268624,
      "learning_rate": 9.207426486500252e-06,
      "loss": 0.3587,
      "step": 390
    },
    {
      "epoch": 0.5334833989870569,
      "grad_norm": 0.04183937339521655,
      "learning_rate": 9.175277689491804e-06,
      "loss": 0.3385,
      "step": 395
    },
    {
      "epoch": 0.5402363534046145,
      "grad_norm": 0.04804994911136886,
      "learning_rate": 9.142548246219212e-06,
      "loss": 0.3696,
      "step": 400
    },
    {
      "epoch": 0.5469893078221723,
      "grad_norm": 0.056116289553118946,
      "learning_rate": 9.109242708290864e-06,
      "loss": 0.3427,
      "step": 405
    },
    {
      "epoch": 0.5537422622397299,
      "grad_norm": 0.04610990276019498,
      "learning_rate": 9.075365707431311e-06,
      "loss": 0.3684,
      "step": 410
    },
    {
      "epoch": 0.5604952166572875,
      "grad_norm": 0.04711791241060553,
      "learning_rate": 9.040921954837139e-06,
      "loss": 0.3262,
      "step": 415
    },
    {
      "epoch": 0.5672481710748453,
      "grad_norm": 0.0436305028584867,
      "learning_rate": 9.005916240521788e-06,
      "loss": 0.3466,
      "step": 420
    },
    {
      "epoch": 0.5740011254924029,
      "grad_norm": 0.043193879987331325,
      "learning_rate": 8.97035343264943e-06,
      "loss": 0.3213,
      "step": 425
    },
    {
      "epoch": 0.5807540799099606,
      "grad_norm": 0.04807145092887773,
      "learning_rate": 8.93423847685795e-06,
      "loss": 0.3553,
      "step": 430
    },
    {
      "epoch": 0.5875070343275183,
      "grad_norm": 0.04707322735993806,
      "learning_rate": 8.89757639557118e-06,
      "loss": 0.3611,
      "step": 435
    },
    {
      "epoch": 0.594259988745076,
      "grad_norm": 0.04694876715909799,
      "learning_rate": 8.860372287300432e-06,
      "loss": 0.3703,
      "step": 440
    },
    {
      "epoch": 0.6010129431626337,
      "grad_norm": 0.051713244981409834,
      "learning_rate": 8.822631325935463e-06,
      "loss": 0.3763,
      "step": 445
    },
    {
      "epoch": 0.6077658975801913,
      "grad_norm": 0.07337197692235135,
      "learning_rate": 8.78435876002496e-06,
      "loss": 0.3867,
      "step": 450
    },
    {
      "epoch": 0.614518851997749,
      "grad_norm": 0.03984879157392941,
      "learning_rate": 8.745559912046625e-06,
      "loss": 0.3292,
      "step": 455
    },
    {
      "epoch": 0.6212718064153067,
      "grad_norm": 0.042340122733838116,
      "learning_rate": 8.706240177667003e-06,
      "loss": 0.3476,
      "step": 460
    },
    {
      "epoch": 0.6280247608328644,
      "grad_norm": 0.04565714030234626,
      "learning_rate": 8.666405024991105e-06,
      "loss": 0.3361,
      "step": 465
    },
    {
      "epoch": 0.634777715250422,
      "grad_norm": 0.04169016093573778,
      "learning_rate": 8.626059993801988e-06,
      "loss": 0.339,
      "step": 470
    },
    {
      "epoch": 0.6415306696679798,
      "grad_norm": 0.04325992397568874,
      "learning_rate": 8.585210694790333e-06,
      "loss": 0.3366,
      "step": 475
    },
    {
      "epoch": 0.6482836240855374,
      "grad_norm": 0.0415410087146755,
      "learning_rate": 8.543862808774193e-06,
      "loss": 0.3489,
      "step": 480
    },
    {
      "epoch": 0.6550365785030952,
      "grad_norm": 0.04286421536314813,
      "learning_rate": 8.502022085908963e-06,
      "loss": 0.3471,
      "step": 485
    },
    {
      "epoch": 0.6617895329206528,
      "grad_norm": 0.04783002216381401,
      "learning_rate": 8.459694344887732e-06,
      "loss": 0.4098,
      "step": 490
    },
    {
      "epoch": 0.6685424873382104,
      "grad_norm": 0.04187887777828664,
      "learning_rate": 8.416885472132077e-06,
      "loss": 0.3462,
      "step": 495
    },
    {
      "epoch": 0.6752954417557682,
      "grad_norm": 0.05373723455097553,
      "learning_rate": 8.373601420973464e-06,
      "loss": 0.344,
      "step": 500
    },
    {
      "epoch": 0.6820483961733258,
      "grad_norm": 0.04300150710853549,
      "learning_rate": 8.329848210825322e-06,
      "loss": 0.3647,
      "step": 505
    },
    {
      "epoch": 0.6888013505908835,
      "grad_norm": 0.043097229603458374,
      "learning_rate": 8.285631926345943e-06,
      "loss": 0.3693,
      "step": 510
    },
    {
      "epoch": 0.6955543050084412,
      "grad_norm": 0.040768739194901454,
      "learning_rate": 8.240958716592304e-06,
      "loss": 0.3645,
      "step": 515
    },
    {
      "epoch": 0.7023072594259989,
      "grad_norm": 0.046324281574150986,
      "learning_rate": 8.195834794164925e-06,
      "loss": 0.3501,
      "step": 520
    },
    {
      "epoch": 0.7090602138435566,
      "grad_norm": 0.04048770768699556,
      "learning_rate": 8.150266434343904e-06,
      "loss": 0.3296,
      "step": 525
    },
    {
      "epoch": 0.7158131682611142,
      "grad_norm": 0.04249502701153697,
      "learning_rate": 8.104259974216219e-06,
      "loss": 0.3489,
      "step": 530
    },
    {
      "epoch": 0.7225661226786719,
      "grad_norm": 0.03877822496803113,
      "learning_rate": 8.057821811794457e-06,
      "loss": 0.3673,
      "step": 535
    },
    {
      "epoch": 0.7293190770962296,
      "grad_norm": 0.048769655901584476,
      "learning_rate": 8.010958405127048e-06,
      "loss": 0.3313,
      "step": 540
    },
    {
      "epoch": 0.7360720315137873,
      "grad_norm": 0.07133171500587031,
      "learning_rate": 7.963676271400158e-06,
      "loss": 0.3472,
      "step": 545
    },
    {
      "epoch": 0.7428249859313449,
      "grad_norm": 0.04064256549350507,
      "learning_rate": 7.915981986031367e-06,
      "loss": 0.3478,
      "step": 550
    },
    {
      "epoch": 0.7495779403489027,
      "grad_norm": 0.04549548056733051,
      "learning_rate": 7.86788218175523e-06,
      "loss": 0.3342,
      "step": 555
    },
    {
      "epoch": 0.7563308947664603,
      "grad_norm": 0.03599813075890018,
      "learning_rate": 7.819383547700889e-06,
      "loss": 0.3354,
      "step": 560
    },
    {
      "epoch": 0.7630838491840181,
      "grad_norm": 0.04316680817634258,
      "learning_rate": 7.770492828461824e-06,
      "loss": 0.3622,
      "step": 565
    },
    {
      "epoch": 0.7698368036015757,
      "grad_norm": 0.04908046724152037,
      "learning_rate": 7.721216823157896e-06,
      "loss": 0.3559,
      "step": 570
    },
    {
      "epoch": 0.7765897580191333,
      "grad_norm": 0.040354310023923715,
      "learning_rate": 7.671562384489819e-06,
      "loss": 0.3629,
      "step": 575
    },
    {
      "epoch": 0.7833427124366911,
      "grad_norm": 0.043579845331084145,
      "learning_rate": 7.621536417786159e-06,
      "loss": 0.3404,
      "step": 580
    },
    {
      "epoch": 0.7900956668542487,
      "grad_norm": 0.04040670293206675,
      "learning_rate": 7.571145880043036e-06,
      "loss": 0.3672,
      "step": 585
    },
    {
      "epoch": 0.7968486212718064,
      "grad_norm": 0.04244690517322693,
      "learning_rate": 7.520397778956623e-06,
      "loss": 0.3499,
      "step": 590
    },
    {
      "epoch": 0.8036015756893641,
      "grad_norm": 0.05433806018620802,
      "learning_rate": 7.469299171948608e-06,
      "loss": 0.3704,
      "step": 595
    },
    {
      "epoch": 0.8103545301069218,
      "grad_norm": 0.04428365952002865,
      "learning_rate": 7.417857165184723e-06,
      "loss": 0.3474,
      "step": 600
    },
    {
      "epoch": 0.8171074845244795,
      "grad_norm": 0.03948151196873183,
      "learning_rate": 7.366078912586523e-06,
      "loss": 0.3426,
      "step": 605
    },
    {
      "epoch": 0.8238604389420371,
      "grad_norm": 0.03804562874118621,
      "learning_rate": 7.313971614836496e-06,
      "loss": 0.3418,
      "step": 610
    },
    {
      "epoch": 0.8306133933595948,
      "grad_norm": 0.04327583528687119,
      "learning_rate": 7.261542518376677e-06,
      "loss": 0.3408,
      "step": 615
    },
    {
      "epoch": 0.8373663477771525,
      "grad_norm": 0.038581581579252,
      "learning_rate": 7.208798914400916e-06,
      "loss": 0.3429,
      "step": 620
    },
    {
      "epoch": 0.8441193021947102,
      "grad_norm": 0.040274824329860945,
      "learning_rate": 7.155748137840892e-06,
      "loss": 0.385,
      "step": 625
    },
    {
      "epoch": 0.8508722566122678,
      "grad_norm": 0.043944187715932226,
      "learning_rate": 7.102397566346073e-06,
      "loss": 0.3366,
      "step": 630
    },
    {
      "epoch": 0.8576252110298256,
      "grad_norm": 0.04145675205661903,
      "learning_rate": 7.048754619257716e-06,
      "loss": 0.3755,
      "step": 635
    },
    {
      "epoch": 0.8643781654473832,
      "grad_norm": 0.041045163518307656,
      "learning_rate": 6.994826756577082e-06,
      "loss": 0.3722,
      "step": 640
    },
    {
      "epoch": 0.871131119864941,
      "grad_norm": 0.05061812503051895,
      "learning_rate": 6.940621477927988e-06,
      "loss": 0.3687,
      "step": 645
    },
    {
      "epoch": 0.8778840742824986,
      "grad_norm": 0.03703919699696153,
      "learning_rate": 6.88614632151385e-06,
      "loss": 0.3454,
      "step": 650
    },
    {
      "epoch": 0.8846370287000562,
      "grad_norm": 0.039635652764389524,
      "learning_rate": 6.831408863069364e-06,
      "loss": 0.365,
      "step": 655
    },
    {
      "epoch": 0.891389983117614,
      "grad_norm": 0.0428338033091146,
      "learning_rate": 6.7764167148069695e-06,
      "loss": 0.3251,
      "step": 660
    },
    {
      "epoch": 0.8981429375351716,
      "grad_norm": 0.0393255629299852,
      "learning_rate": 6.721177524358226e-06,
      "loss": 0.3519,
      "step": 665
    },
    {
      "epoch": 0.9048958919527293,
      "grad_norm": 0.03605877433184958,
      "learning_rate": 6.665698973710289e-06,
      "loss": 0.3586,
      "step": 670
    },
    {
      "epoch": 0.911648846370287,
      "grad_norm": 0.0418094941760349,
      "learning_rate": 6.609988778137582e-06,
      "loss": 0.333,
      "step": 675
    },
    {
      "epoch": 0.9184018007878447,
      "grad_norm": 0.04338138841178037,
      "learning_rate": 6.554054685128857e-06,
      "loss": 0.334,
      "step": 680
    },
    {
      "epoch": 0.9251547552054024,
      "grad_norm": 0.04225445092421321,
      "learning_rate": 6.497904473309766e-06,
      "loss": 0.3519,
      "step": 685
    },
    {
      "epoch": 0.93190770962296,
      "grad_norm": 0.04826057110998316,
      "learning_rate": 6.44154595136111e-06,
      "loss": 0.3598,
      "step": 690
    },
    {
      "epoch": 0.9386606640405177,
      "grad_norm": 0.04381442124287324,
      "learning_rate": 6.384986956932897e-06,
      "loss": 0.3339,
      "step": 695
    },
    {
      "epoch": 0.9454136184580754,
      "grad_norm": 0.0414974847898415,
      "learning_rate": 6.328235355554382e-06,
      "loss": 0.3645,
      "step": 700
    },
    {
      "epoch": 0.9521665728756331,
      "grad_norm": 0.04276356905517851,
      "learning_rate": 6.271299039540228e-06,
      "loss": 0.3645,
      "step": 705
    },
    {
      "epoch": 0.9589195272931907,
      "grad_norm": 0.046676300677672276,
      "learning_rate": 6.214185926892936e-06,
      "loss": 0.3551,
      "step": 710
    },
    {
      "epoch": 0.9656724817107485,
      "grad_norm": 0.04006191125365821,
      "learning_rate": 6.156903960201709e-06,
      "loss": 0.3395,
      "step": 715
    },
    {
      "epoch": 0.9724254361283061,
      "grad_norm": 0.047147919741739285,
      "learning_rate": 6.099461105537889e-06,
      "loss": 0.3321,
      "step": 720
    },
    {
      "epoch": 0.9791783905458639,
      "grad_norm": 0.041770572631853936,
      "learning_rate": 6.041865351347146e-06,
      "loss": 0.3421,
      "step": 725
    },
    {
      "epoch": 0.9859313449634215,
      "grad_norm": 0.04239922397171486,
      "learning_rate": 5.984124707338528e-06,
      "loss": 0.3373,
      "step": 730
    },
    {
      "epoch": 0.9926842993809791,
      "grad_norm": 0.046194589474773985,
      "learning_rate": 5.926247203370583e-06,
      "loss": 0.3405,
      "step": 735
    },
    {
      "epoch": 0.9994372537985369,
      "grad_norm": 0.03944962664459983,
      "learning_rate": 5.8682408883346535e-06,
      "loss": 0.2909,
      "step": 740
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.3324998915195465,
      "eval_runtime": 18.1309,
      "eval_samples_per_second": 19.8,
      "eval_steps_per_second": 4.964,
      "step": 741
    },
    {
      "epoch": 1.005402363534046,
      "grad_norm": 0.040121304388969325,
      "learning_rate": 5.810113829035544e-06,
      "loss": 0.355,
      "step": 745
    },
    {
      "epoch": 1.0121553179516039,
      "grad_norm": 0.047689669773125845,
      "learning_rate": 5.751874109069685e-06,
      "loss": 0.3611,
      "step": 750
    },
    {
      "epoch": 1.0189082723691616,
      "grad_norm": 0.03912739831058518,
      "learning_rate": 5.693529827700967e-06,
      "loss": 0.2899,
      "step": 755
    },
    {
      "epoch": 1.0256612267867191,
      "grad_norm": 0.04377909856345864,
      "learning_rate": 5.635089098734394e-06,
      "loss": 0.3197,
      "step": 760
    },
    {
      "epoch": 1.0324141812042769,
      "grad_norm": 0.037393171944725745,
      "learning_rate": 5.576560049387717e-06,
      "loss": 0.339,
      "step": 765
    },
    {
      "epoch": 1.0391671356218346,
      "grad_norm": 0.042256918616508245,
      "learning_rate": 5.517950819161196e-06,
      "loss": 0.328,
      "step": 770
    },
    {
      "epoch": 1.0459200900393921,
      "grad_norm": 0.040529502097932234,
      "learning_rate": 5.459269558705667e-06,
      "loss": 0.3259,
      "step": 775
    },
    {
      "epoch": 1.0526730444569499,
      "grad_norm": 0.04379220562588465,
      "learning_rate": 5.400524428689035e-06,
      "loss": 0.3464,
      "step": 780
    },
    {
      "epoch": 1.0594259988745076,
      "grad_norm": 0.04725991662965075,
      "learning_rate": 5.341723598661409e-06,
      "loss": 0.3267,
      "step": 785
    },
    {
      "epoch": 1.0661789532920654,
      "grad_norm": 0.036832603953267457,
      "learning_rate": 5.282875245918963e-06,
      "loss": 0.3261,
      "step": 790
    },
    {
      "epoch": 1.0729319077096229,
      "grad_norm": 0.04175828917214628,
      "learning_rate": 5.2239875543667465e-06,
      "loss": 0.3346,
      "step": 795
    },
    {
      "epoch": 1.0796848621271806,
      "grad_norm": 0.03618010757790516,
      "learning_rate": 5.165068713380568e-06,
      "loss": 0.3182,
      "step": 800
    },
    {
      "epoch": 1.0864378165447384,
      "grad_norm": 0.04097386666815235,
      "learning_rate": 5.106126916668118e-06,
      "loss": 0.3192,
      "step": 805
    },
    {
      "epoch": 1.093190770962296,
      "grad_norm": 0.040300039979862916,
      "learning_rate": 5.047170361129484e-06,
      "loss": 0.3138,
      "step": 810
    },
    {
      "epoch": 1.0999437253798536,
      "grad_norm": 0.0413703774570022,
      "learning_rate": 4.988207245717232e-06,
      "loss": 0.3558,
      "step": 815
    },
    {
      "epoch": 1.1066966797974114,
      "grad_norm": 0.03938201981503368,
      "learning_rate": 4.929245770296191e-06,
      "loss": 0.3436,
      "step": 820
    },
    {
      "epoch": 1.1134496342149691,
      "grad_norm": 0.043903225983181984,
      "learning_rate": 4.870294134503123e-06,
      "loss": 0.3301,
      "step": 825
    },
    {
      "epoch": 1.1202025886325266,
      "grad_norm": 0.04208713272188743,
      "learning_rate": 4.811360536606416e-06,
      "loss": 0.3108,
      "step": 830
    },
    {
      "epoch": 1.1269555430500844,
      "grad_norm": 0.03660393810864938,
      "learning_rate": 4.752453172365966e-06,
      "loss": 0.333,
      "step": 835
    },
    {
      "epoch": 1.1337084974676421,
      "grad_norm": 0.04407131833701405,
      "learning_rate": 4.69358023389342e-06,
      "loss": 0.3605,
      "step": 840
    },
    {
      "epoch": 1.1404614518851997,
      "grad_norm": 0.04616999836259198,
      "learning_rate": 4.634749908512907e-06,
      "loss": 0.3439,
      "step": 845
    },
    {
      "epoch": 1.1472144063027574,
      "grad_norm": 0.04325365689860756,
      "learning_rate": 4.575970377622456e-06,
      "loss": 0.3049,
      "step": 850
    },
    {
      "epoch": 1.1539673607203151,
      "grad_norm": 0.03846831221970018,
      "learning_rate": 4.517249815556219e-06,
      "loss": 0.3511,
      "step": 855
    },
    {
      "epoch": 1.1607203151378729,
      "grad_norm": 0.04569268985609919,
      "learning_rate": 4.458596388447691e-06,
      "loss": 0.3209,
      "step": 860
    },
    {
      "epoch": 1.1674732695554304,
      "grad_norm": 0.0409252349150715,
      "learning_rate": 4.400018253094065e-06,
      "loss": 0.308,
      "step": 865
    },
    {
      "epoch": 1.1742262239729881,
      "grad_norm": 0.04133167625196089,
      "learning_rate": 4.341523555821882e-06,
      "loss": 0.3377,
      "step": 870
    },
    {
      "epoch": 1.180979178390546,
      "grad_norm": 0.043097502447341854,
      "learning_rate": 4.283120431354137e-06,
      "loss": 0.3117,
      "step": 875
    },
    {
      "epoch": 1.1877321328081036,
      "grad_norm": 0.042543275375795706,
      "learning_rate": 4.224817001679011e-06,
      "loss": 0.3344,
      "step": 880
    },
    {
      "epoch": 1.1944850872256612,
      "grad_norm": 0.04304997898343135,
      "learning_rate": 4.1666213749203545e-06,
      "loss": 0.3237,
      "step": 885
    },
    {
      "epoch": 1.201238041643219,
      "grad_norm": 0.04100222680465783,
      "learning_rate": 4.1085416442101205e-06,
      "loss": 0.3641,
      "step": 890
    },
    {
      "epoch": 1.2079909960607766,
      "grad_norm": 0.04523396323757555,
      "learning_rate": 4.050585886562858e-06,
      "loss": 0.3494,
      "step": 895
    },
    {
      "epoch": 1.2147439504783342,
      "grad_norm": 0.042513327919157136,
      "learning_rate": 3.992762161752474e-06,
      "loss": 0.343,
      "step": 900
    },
    {
      "epoch": 1.221496904895892,
      "grad_norm": 0.046594930027348866,
      "learning_rate": 3.935078511191368e-06,
      "loss": 0.3235,
      "step": 905
    },
    {
      "epoch": 1.2282498593134497,
      "grad_norm": 0.03906041128591151,
      "learning_rate": 3.877542956812137e-06,
      "loss": 0.3221,
      "step": 910
    },
    {
      "epoch": 1.2350028137310074,
      "grad_norm": 0.044515080037505254,
      "learning_rate": 3.820163499951984e-06,
      "loss": 0.3579,
      "step": 915
    },
    {
      "epoch": 1.241755768148565,
      "grad_norm": 0.04205841651552162,
      "learning_rate": 3.7629481202399886e-06,
      "loss": 0.3375,
      "step": 920
    },
    {
      "epoch": 1.2485087225661227,
      "grad_norm": 0.04152789361719496,
      "learning_rate": 3.705904774487396e-06,
      "loss": 0.2971,
      "step": 925
    },
    {
      "epoch": 1.2552616769836804,
      "grad_norm": 0.05137094607363216,
      "learning_rate": 3.64904139558109e-06,
      "loss": 0.3343,
      "step": 930
    },
    {
      "epoch": 1.2620146314012382,
      "grad_norm": 0.04499346319735512,
      "learning_rate": 3.5923658913803726e-06,
      "loss": 0.3274,
      "step": 935
    },
    {
      "epoch": 1.2687675858187957,
      "grad_norm": 0.04194049880241552,
      "learning_rate": 3.5358861436172487e-06,
      "loss": 0.3436,
      "step": 940
    },
    {
      "epoch": 1.2755205402363534,
      "grad_norm": 0.04739038652189242,
      "learning_rate": 3.47961000680032e-06,
      "loss": 0.3395,
      "step": 945
    },
    {
      "epoch": 1.2822734946539112,
      "grad_norm": 0.04407332378962217,
      "learning_rate": 3.4235453071224882e-06,
      "loss": 0.3283,
      "step": 950
    },
    {
      "epoch": 1.2890264490714687,
      "grad_norm": 0.046768041779961654,
      "learning_rate": 3.3676998413725726e-06,
      "loss": 0.3457,
      "step": 955
    },
    {
      "epoch": 1.2957794034890264,
      "grad_norm": 0.04405892619650361,
      "learning_rate": 3.3120813758510385e-06,
      "loss": 0.3446,
      "step": 960
    },
    {
      "epoch": 1.3025323579065842,
      "grad_norm": 0.04917585118916133,
      "learning_rate": 3.2566976452899507e-06,
      "loss": 0.3267,
      "step": 965
    },
    {
      "epoch": 1.3092853123241417,
      "grad_norm": 0.04327329891699119,
      "learning_rate": 3.2015563517773214e-06,
      "loss": 0.3382,
      "step": 970
    },
    {
      "epoch": 1.3160382667416994,
      "grad_norm": 0.04751650625794217,
      "learning_rate": 3.1466651636860025e-06,
      "loss": 0.3325,
      "step": 975
    },
    {
      "epoch": 1.3227912211592572,
      "grad_norm": 0.05680333767716324,
      "learning_rate": 3.0920317146072577e-06,
      "loss": 0.3207,
      "step": 980
    },
    {
      "epoch": 1.329544175576815,
      "grad_norm": 0.0497656718123479,
      "learning_rate": 3.0376636022891813e-06,
      "loss": 0.329,
      "step": 985
    },
    {
      "epoch": 1.3362971299943727,
      "grad_norm": 0.048529124206900136,
      "learning_rate": 2.983568387580094e-06,
      "loss": 0.3322,
      "step": 990
    },
    {
      "epoch": 1.3430500844119302,
      "grad_norm": 0.04475425551771912,
      "learning_rate": 2.9297535933770732e-06,
      "loss": 0.3278,
      "step": 995
    },
    {
      "epoch": 1.349803038829488,
      "grad_norm": 0.043952297509005415,
      "learning_rate": 2.8762267035797607e-06,
      "loss": 0.2961,
      "step": 1000
    },
    {
      "epoch": 1.3565559932470457,
      "grad_norm": 0.04672219098205047,
      "learning_rate": 2.822995162049599e-06,
      "loss": 0.3618,
      "step": 1005
    },
    {
      "epoch": 1.3633089476646032,
      "grad_norm": 0.04260321337685024,
      "learning_rate": 2.7700663715746213e-06,
      "loss": 0.3382,
      "step": 1010
    },
    {
      "epoch": 1.370061902082161,
      "grad_norm": 0.048008697667558865,
      "learning_rate": 2.7174476928399685e-06,
      "loss": 0.3466,
      "step": 1015
    },
    {
      "epoch": 1.3768148564997187,
      "grad_norm": 0.04892067375449484,
      "learning_rate": 2.66514644340426e-06,
      "loss": 0.3425,
      "step": 1020
    },
    {
      "epoch": 1.3835678109172762,
      "grad_norm": 0.04689882121063575,
      "learning_rate": 2.613169896681949e-06,
      "loss": 0.2877,
      "step": 1025
    },
    {
      "epoch": 1.390320765334834,
      "grad_norm": 0.054157177491435074,
      "learning_rate": 2.5615252809318287e-06,
      "loss": 0.3306,
      "step": 1030
    },
    {
      "epoch": 1.3970737197523917,
      "grad_norm": 0.048672772164928886,
      "learning_rate": 2.5102197782518145e-06,
      "loss": 0.3216,
      "step": 1035
    },
    {
      "epoch": 1.4038266741699492,
      "grad_norm": 0.06752970884761282,
      "learning_rate": 2.4592605235801544e-06,
      "loss": 0.3276,
      "step": 1040
    },
    {
      "epoch": 1.410579628587507,
      "grad_norm": 0.04710574868399225,
      "learning_rate": 2.4086546037031734e-06,
      "loss": 0.3173,
      "step": 1045
    },
    {
      "epoch": 1.4173325830050647,
      "grad_norm": 0.045184769590470804,
      "learning_rate": 2.3584090562697427e-06,
      "loss": 0.3164,
      "step": 1050
    },
    {
      "epoch": 1.4240855374226224,
      "grad_norm": 0.043485542556231396,
      "learning_rate": 2.3085308688125695e-06,
      "loss": 0.3027,
      "step": 1055
    },
    {
      "epoch": 1.4308384918401802,
      "grad_norm": 0.05260603355819636,
      "learning_rate": 2.2590269777764516e-06,
      "loss": 0.3004,
      "step": 1060
    },
    {
      "epoch": 1.4375914462577377,
      "grad_norm": 0.051099932583330926,
      "learning_rate": 2.2099042675536437e-06,
      "loss": 0.3368,
      "step": 1065
    },
    {
      "epoch": 1.4443444006752955,
      "grad_norm": 0.045328406677686464,
      "learning_rate": 2.161169569526461e-06,
      "loss": 0.3056,
      "step": 1070
    },
    {
      "epoch": 1.4510973550928532,
      "grad_norm": 0.05676490778567245,
      "learning_rate": 2.1128296611172593e-06,
      "loss": 0.3052,
      "step": 1075
    },
    {
      "epoch": 1.4578503095104107,
      "grad_norm": 0.044797358674842205,
      "learning_rate": 2.0648912648459072e-06,
      "loss": 0.3447,
      "step": 1080
    },
    {
      "epoch": 1.4646032639279685,
      "grad_norm": 0.049289994897252776,
      "learning_rate": 2.0173610473949048e-06,
      "loss": 0.3375,
      "step": 1085
    },
    {
      "epoch": 1.4713562183455262,
      "grad_norm": 0.04706434397429944,
      "learning_rate": 1.9702456186822595e-06,
      "loss": 0.3464,
      "step": 1090
    },
    {
      "epoch": 1.4781091727630837,
      "grad_norm": 0.05030965485900401,
      "learning_rate": 1.9235515309422685e-06,
      "loss": 0.3119,
      "step": 1095
    },
    {
      "epoch": 1.4848621271806415,
      "grad_norm": 0.04413503044032754,
      "learning_rate": 1.8772852778143064e-06,
      "loss": 0.3371,
      "step": 1100
    },
    {
      "epoch": 1.4916150815981992,
      "grad_norm": 0.06233556882470439,
      "learning_rate": 1.831453293439771e-06,
      "loss": 0.3253,
      "step": 1105
    },
    {
      "epoch": 1.498368036015757,
      "grad_norm": 0.05282578176334435,
      "learning_rate": 1.7860619515673034e-06,
      "loss": 0.3032,
      "step": 1110
    },
    {
      "epoch": 1.5051209904333147,
      "grad_norm": 0.05220647521633855,
      "learning_rate": 1.7411175646664103e-06,
      "loss": 0.3161,
      "step": 1115
    },
    {
      "epoch": 1.5118739448508722,
      "grad_norm": 0.047487304887968124,
      "learning_rate": 1.6966263830495939e-06,
      "loss": 0.3104,
      "step": 1120
    },
    {
      "epoch": 1.51862689926843,
      "grad_norm": 0.05098969364978786,
      "learning_rate": 1.6525945940031407e-06,
      "loss": 0.3692,
      "step": 1125
    },
    {
      "epoch": 1.5253798536859877,
      "grad_norm": 0.04810126110395747,
      "learning_rate": 1.6090283209266682e-06,
      "loss": 0.3383,
      "step": 1130
    },
    {
      "epoch": 1.5321328081035452,
      "grad_norm": 0.06656533903224827,
      "learning_rate": 1.5659336224815642e-06,
      "loss": 0.3413,
      "step": 1135
    },
    {
      "epoch": 1.538885762521103,
      "grad_norm": 0.057091145465475585,
      "learning_rate": 1.5233164917484117e-06,
      "loss": 0.3247,
      "step": 1140
    },
    {
      "epoch": 1.5456387169386607,
      "grad_norm": 0.05098329100385167,
      "learning_rate": 1.4811828553935498e-06,
      "loss": 0.3284,
      "step": 1145
    },
    {
      "epoch": 1.5523916713562183,
      "grad_norm": 0.04852633539753427,
      "learning_rate": 1.439538572844873e-06,
      "loss": 0.307,
      "step": 1150
    },
    {
      "epoch": 1.559144625773776,
      "grad_norm": 0.05935932625720792,
      "learning_rate": 1.3983894354769616e-06,
      "loss": 0.3247,
      "step": 1155
    },
    {
      "epoch": 1.5658975801913337,
      "grad_norm": 0.06306295104638487,
      "learning_rate": 1.3577411658056965e-06,
      "loss": 0.3286,
      "step": 1160
    },
    {
      "epoch": 1.5726505346088913,
      "grad_norm": 0.06348638606124773,
      "learning_rate": 1.3175994166924394e-06,
      "loss": 0.3247,
      "step": 1165
    },
    {
      "epoch": 1.5794034890264492,
      "grad_norm": 0.051811719215914646,
      "learning_rate": 1.2779697705579058e-06,
      "loss": 0.3246,
      "step": 1170
    },
    {
      "epoch": 1.5861564434440067,
      "grad_norm": 0.06830322751813911,
      "learning_rate": 1.2388577386058248e-06,
      "loss": 0.3326,
      "step": 1175
    },
    {
      "epoch": 1.5929093978615643,
      "grad_norm": 0.04922857249304829,
      "learning_rate": 1.2002687600565138e-06,
      "loss": 0.3061,
      "step": 1180
    },
    {
      "epoch": 1.5996623522791222,
      "grad_norm": 0.047108708806732846,
      "learning_rate": 1.1622082013904535e-06,
      "loss": 0.3118,
      "step": 1185
    },
    {
      "epoch": 1.6064153066966798,
      "grad_norm": 0.059687932246599115,
      "learning_rate": 1.1246813556019925e-06,
      "loss": 0.3372,
      "step": 1190
    },
    {
      "epoch": 1.6131682611142375,
      "grad_norm": 0.05955615759336213,
      "learning_rate": 1.0876934414632523e-06,
      "loss": 0.2956,
      "step": 1195
    },
    {
      "epoch": 1.6199212155317952,
      "grad_norm": 0.06542544159973171,
      "learning_rate": 1.0512496027983715e-06,
      "loss": 0.3118,
      "step": 1200
    },
    {
      "epoch": 1.6266741699493528,
      "grad_norm": 0.05198203822638065,
      "learning_rate": 1.0153549077681617e-06,
      "loss": 0.346,
      "step": 1205
    },
    {
      "epoch": 1.6334271243669105,
      "grad_norm": 0.06101598768362396,
      "learning_rate": 9.80014348165298e-07,
      "loss": 0.2952,
      "step": 1210
    },
    {
      "epoch": 1.6401800787844683,
      "grad_norm": 0.05552589226769097,
      "learning_rate": 9.452328387201104e-07,
      "loss": 0.2903,
      "step": 1215
    },
    {
      "epoch": 1.6469330332020258,
      "grad_norm": 0.06472026824610987,
      "learning_rate": 9.110152164171127e-07,
      "loss": 0.3081,
      "step": 1220
    },
    {
      "epoch": 1.6536859876195835,
      "grad_norm": 0.07431179400574185,
      "learning_rate": 8.773662398223276e-07,
      "loss": 0.3095,
      "step": 1225
    },
    {
      "epoch": 1.6604389420371413,
      "grad_norm": 0.06752269915658678,
      "learning_rate": 8.44290588421533e-07,
      "loss": 0.3187,
      "step": 1230
    },
    {
      "epoch": 1.6671918964546988,
      "grad_norm": 0.06578494925550707,
      "learning_rate": 8.117928619694848e-07,
      "loss": 0.3036,
      "step": 1235
    },
    {
      "epoch": 1.6739448508722568,
      "grad_norm": 0.06180051968265024,
      "learning_rate": 7.798775798502484e-07,
      "loss": 0.3023,
      "step": 1240
    },
    {
      "epoch": 1.6806978052898143,
      "grad_norm": 0.07560845712778133,
      "learning_rate": 7.485491804486972e-07,
      "loss": 0.2951,
      "step": 1245
    },
    {
      "epoch": 1.687450759707372,
      "grad_norm": 0.07194010404747912,
      "learning_rate": 7.178120205332717e-07,
      "loss": 0.3372,
      "step": 1250
    },
    {
      "epoch": 1.6942037141249298,
      "grad_norm": 0.06976609157173444,
      "learning_rate": 6.876703746500984e-07,
      "loss": 0.2973,
      "step": 1255
    },
    {
      "epoch": 1.7009566685424873,
      "grad_norm": 0.07659553993307627,
      "learning_rate": 6.581284345285371e-07,
      "loss": 0.284,
      "step": 1260
    },
    {
      "epoch": 1.707709622960045,
      "grad_norm": 0.05961865699755965,
      "learning_rate": 6.291903084982481e-07,
      "loss": 0.3153,
      "step": 1265
    },
    {
      "epoch": 1.7144625773776028,
      "grad_norm": 0.0812482207068001,
      "learning_rate": 6.008600209178539e-07,
      "loss": 0.342,
      "step": 1270
    },
    {
      "epoch": 1.7212155317951603,
      "grad_norm": 0.06939464438812673,
      "learning_rate": 5.7314151161528e-07,
      "loss": 0.3287,
      "step": 1275
    },
    {
      "epoch": 1.727968486212718,
      "grad_norm": 0.06615067862128174,
      "learning_rate": 5.460386353398583e-07,
      "loss": 0.3036,
      "step": 1280
    },
    {
      "epoch": 1.7347214406302758,
      "grad_norm": 0.08487351817752944,
      "learning_rate": 5.195551612262478e-07,
      "loss": 0.3059,
      "step": 1285
    },
    {
      "epoch": 1.7414743950478333,
      "grad_norm": 0.0660084040820942,
      "learning_rate": 4.936947722702762e-07,
      "loss": 0.3004,
      "step": 1290
    },
    {
      "epoch": 1.7482273494653913,
      "grad_norm": 0.07500490253176417,
      "learning_rate": 4.6846106481675035e-07,
      "loss": 0.3163,
      "step": 1295
    },
    {
      "epoch": 1.7549803038829488,
      "grad_norm": 0.07341914462851476,
      "learning_rate": 4.43857548059321e-07,
      "loss": 0.3028,
      "step": 1300
    },
    {
      "epoch": 1.7617332583005063,
      "grad_norm": 0.05936547346570651,
      "learning_rate": 4.198876435524718e-07,
      "loss": 0.3071,
      "step": 1305
    },
    {
      "epoch": 1.7684862127180643,
      "grad_norm": 0.06361743408706552,
      "learning_rate": 3.9655468473568436e-07,
      "loss": 0.3151,
      "step": 1310
    },
    {
      "epoch": 1.7752391671356218,
      "grad_norm": 0.07650389023350329,
      "learning_rate": 3.7386191646987094e-07,
      "loss": 0.333,
      "step": 1315
    },
    {
      "epoch": 1.7819921215531795,
      "grad_norm": 0.06885824668728105,
      "learning_rate": 3.51812494586114e-07,
      "loss": 0.2898,
      "step": 1320
    },
    {
      "epoch": 1.7887450759707373,
      "grad_norm": 0.0645999509118337,
      "learning_rate": 3.3040948544679817e-07,
      "loss": 0.3091,
      "step": 1325
    },
    {
      "epoch": 1.7954980303882948,
      "grad_norm": 0.07732866321608588,
      "learning_rate": 3.096558655191706e-07,
      "loss": 0.3148,
      "step": 1330
    },
    {
      "epoch": 1.8022509848058526,
      "grad_norm": 0.0733331982280887,
      "learning_rate": 2.895545209614176e-07,
      "loss": 0.3101,
      "step": 1335
    },
    {
      "epoch": 1.8090039392234103,
      "grad_norm": 0.06997792694817997,
      "learning_rate": 2.701082472212879e-07,
      "loss": 0.3302,
      "step": 1340
    },
    {
      "epoch": 1.8157568936409678,
      "grad_norm": 0.07837574138979145,
      "learning_rate": 2.5131974864734063e-07,
      "loss": 0.3425,
      "step": 1345
    },
    {
      "epoch": 1.8225098480585256,
      "grad_norm": 0.07239388554199513,
      "learning_rate": 2.331916381128535e-07,
      "loss": 0.3054,
      "step": 1350
    },
    {
      "epoch": 1.8292628024760833,
      "grad_norm": 0.0674988666661417,
      "learning_rate": 2.157264366524603e-07,
      "loss": 0.28,
      "step": 1355
    },
    {
      "epoch": 1.8360157568936408,
      "grad_norm": 0.07181164697688498,
      "learning_rate": 1.989265731115525e-07,
      "loss": 0.3307,
      "step": 1360
    },
    {
      "epoch": 1.8427687113111988,
      "grad_norm": 0.07548202922443094,
      "learning_rate": 1.827943838085111e-07,
      "loss": 0.306,
      "step": 1365
    },
    {
      "epoch": 1.8495216657287563,
      "grad_norm": 0.05982075375718766,
      "learning_rate": 1.6733211220979316e-07,
      "loss": 0.2981,
      "step": 1370
    },
    {
      "epoch": 1.856274620146314,
      "grad_norm": 0.08355451454560252,
      "learning_rate": 1.5254190861794415e-07,
      "loss": 0.3083,
      "step": 1375
    },
    {
      "epoch": 1.8630275745638718,
      "grad_norm": 0.07033662219869818,
      "learning_rate": 1.3842582987255494e-07,
      "loss": 0.293,
      "step": 1380
    },
    {
      "epoch": 1.8697805289814293,
      "grad_norm": 0.06615537508166416,
      "learning_rate": 1.2498583906422779e-07,
      "loss": 0.3072,
      "step": 1385
    },
    {
      "epoch": 1.876533483398987,
      "grad_norm": 0.059237664909188886,
      "learning_rate": 1.1222380526156929e-07,
      "loss": 0.313,
      "step": 1390
    },
    {
      "epoch": 1.8832864378165448,
      "grad_norm": 0.08185957993390121,
      "learning_rate": 1.0014150325126315e-07,
      "loss": 0.3071,
      "step": 1395
    },
    {
      "epoch": 1.8900393922341023,
      "grad_norm": 0.06113218790264024,
      "learning_rate": 8.874061329125939e-08,
      "loss": 0.3116,
      "step": 1400
    },
    {
      "epoch": 1.89679234665166,
      "grad_norm": 0.08593985138096456,
      "learning_rate": 7.802272087709951e-08,
      "loss": 0.3106,
      "step": 1405
    },
    {
      "epoch": 1.9035453010692178,
      "grad_norm": 0.0753339474248732,
      "learning_rate": 6.798931652142738e-08,
      "loss": 0.3118,
      "step": 1410
    },
    {
      "epoch": 1.9102982554867753,
      "grad_norm": 0.06406340322576629,
      "learning_rate": 5.864179554670724e-08,
      "loss": 0.3384,
      "step": 1415
    },
    {
      "epoch": 1.9170512099043333,
      "grad_norm": 0.06739406393286214,
      "learning_rate": 4.998145789118114e-08,
      "loss": 0.2755,
      "step": 1420
    },
    {
      "epoch": 1.9238041643218908,
      "grad_norm": 0.07101405306951455,
      "learning_rate": 4.2009507928084116e-08,
      "loss": 0.2802,
      "step": 1425
    },
    {
      "epoch": 1.9305571187394484,
      "grad_norm": 0.07509273619426268,
      "learning_rate": 3.4727054298161475e-08,
      "loss": 0.2778,
      "step": 1430
    },
    {
      "epoch": 1.9373100731570063,
      "grad_norm": 0.0719074660592959,
      "learning_rate": 2.8135109755487723e-08,
      "loss": 0.3238,
      "step": 1435
    },
    {
      "epoch": 1.9440630275745638,
      "grad_norm": 0.07042162359075328,
      "learning_rate": 2.223459102662695e-08,
      "loss": 0.3235,
      "step": 1440
    },
    {
      "epoch": 1.9508159819921216,
      "grad_norm": 0.07906944364159181,
      "learning_rate": 1.7026318683147082e-08,
      "loss": 0.3115,
      "step": 1445
    },
    {
      "epoch": 1.9575689364096793,
      "grad_norm": 0.08613249297162148,
      "learning_rate": 1.2511017027501682e-08,
      "loss": 0.3032,
      "step": 1450
    },
    {
      "epoch": 1.9643218908272368,
      "grad_norm": 0.07379471104805907,
      "learning_rate": 8.689313992306104e-09,
      "loss": 0.303,
      "step": 1455
    },
    {
      "epoch": 1.9710748452447946,
      "grad_norm": 0.07085802534414035,
      "learning_rate": 5.561741053010661e-09,
      "loss": 0.3059,
      "step": 1460
    },
    {
      "epoch": 1.9778277996623523,
      "grad_norm": 0.07455102650353135,
      "learning_rate": 3.1287331539903155e-09,
      "loss": 0.3165,
      "step": 1465
    },
    {
      "epoch": 1.9845807540799099,
      "grad_norm": 0.07699115693937987,
      "learning_rate": 1.3906286480563913e-09,
      "loss": 0.3108,
      "step": 1470
    },
    {
      "epoch": 1.9913337084974676,
      "grad_norm": 0.06672257472875329,
      "learning_rate": 3.4766924940476954e-10,
      "loss": 0.3134,
      "step": 1475
    },
    {
      "epoch": 1.9980866629150253,
      "grad_norm": 0.06196469052553269,
      "learning_rate": 0.0,
      "loss": 0.3054,
      "step": 1480
    },
    {
      "epoch": 1.9980866629150253,
      "eval_loss": 0.3086739182472229,
      "eval_runtime": 17.7246,
      "eval_samples_per_second": 20.254,
      "eval_steps_per_second": 5.078,
      "step": 1480
    },
    {
      "epoch": 1.9980866629150253,
      "step": 1480,
      "total_flos": 3.6719988787579453e+18,
      "train_loss": 0.37467364362768224,
      "train_runtime": 13260.7751,
      "train_samples_per_second": 5.36,
      "train_steps_per_second": 0.112
    }
  ],
  "logging_steps": 5,
  "max_steps": 1480,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.6719988787579453e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}