{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 541.0,
  "eval_steps": 500,
  "global_step": 9738,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 5.56,
      "grad_norm": 0.40853407979011536,
      "learning_rate": 4.998998767795805e-05,
      "loss": 2.5454,
      "step": 100
    },
    {
      "epoch": 11.11,
      "grad_norm": 0.27508941292762756,
      "learning_rate": 4.995538747800402e-05,
      "loss": 0.3265,
      "step": 200
    },
    {
      "epoch": 16.67,
      "grad_norm": 0.19116361439228058,
      "learning_rate": 4.989610999768349e-05,
      "loss": 0.1042,
      "step": 300
    },
    {
      "epoch": 22.22,
      "grad_norm": 0.10925190895795822,
      "learning_rate": 4.9812213853878376e-05,
      "loss": 0.0633,
      "step": 400
    },
    {
      "epoch": 27.78,
      "grad_norm": 0.08680253475904465,
      "learning_rate": 4.970378200777949e-05,
      "loss": 0.0377,
      "step": 500
    },
    {
      "epoch": 33.33,
      "grad_norm": 0.029545623809099197,
      "learning_rate": 4.957092168284987e-05,
      "loss": 0.0181,
      "step": 600
    },
    {
      "epoch": 38.89,
      "grad_norm": 0.030413279309868813,
      "learning_rate": 4.941376425879624e-05,
      "loss": 0.008,
      "step": 700
    },
    {
      "epoch": 44.44,
      "grad_norm": 0.05130019411444664,
      "learning_rate": 4.923246514165339e-05,
      "loss": 0.0037,
      "step": 800
    },
    {
      "epoch": 50.0,
      "grad_norm": 0.005339763592928648,
      "learning_rate": 4.902720361011007e-05,
      "loss": 0.0021,
      "step": 900
    },
    {
      "epoch": 55.56,
      "grad_norm": 0.023293694481253624,
      "learning_rate": 4.8798182638228166e-05,
      "loss": 0.0014,
      "step": 1000
    },
    {
      "epoch": 61.11,
      "grad_norm": 0.014095759950578213,
      "learning_rate": 4.8545628694730624e-05,
      "loss": 0.001,
      "step": 1100
    },
    {
      "epoch": 66.67,
      "grad_norm": 0.00977824255824089,
      "learning_rate": 4.826979151905655e-05,
      "loss": 0.0007,
      "step": 1200
    },
    {
      "epoch": 72.22,
      "grad_norm": 0.006551814265549183,
      "learning_rate": 4.797094387440491e-05,
      "loss": 0.0006,
      "step": 1300
    },
    {
      "epoch": 77.78,
      "grad_norm": 0.008941611275076866,
      "learning_rate": 4.7649381278011e-05,
      "loss": 0.0005,
      "step": 1400
    },
    {
      "epoch": 83.33,
      "grad_norm": 0.0047577545046806335,
      "learning_rate": 4.73054217089226e-05,
      "loss": 0.0004,
      "step": 1500
    },
    {
      "epoch": 88.89,
      "grad_norm": 0.005813396070152521,
      "learning_rate": 4.693940529356444e-05,
      "loss": 0.0003,
      "step": 1600
    },
    {
      "epoch": 94.44,
      "grad_norm": 0.006802932824939489,
      "learning_rate": 4.655169396940229e-05,
      "loss": 0.0003,
      "step": 1700
    },
    {
      "epoch": 100.0,
      "grad_norm": 0.0018754027551040053,
      "learning_rate": 4.6142671127038905e-05,
      "loss": 0.0002,
      "step": 1800
    },
    {
      "epoch": 105.56,
      "grad_norm": 0.0058812773786485195,
      "learning_rate": 4.571274123109606e-05,
      "loss": 0.0002,
      "step": 1900
    },
    {
      "epoch": 111.11,
      "grad_norm": 0.003945046104490757,
      "learning_rate": 4.52623294202573e-05,
      "loss": 0.0002,
      "step": 2000
    },
    {
      "epoch": 116.67,
      "grad_norm": 0.0026905699633061886,
      "learning_rate": 4.479188108686714e-05,
      "loss": 0.0002,
      "step": 2100
    },
    {
      "epoch": 122.22,
      "grad_norm": 0.00241717416793108,
      "learning_rate": 4.4301861436502156e-05,
      "loss": 0.0001,
      "step": 2200
    },
    {
      "epoch": 127.78,
      "grad_norm": 0.002276218729093671,
      "learning_rate": 4.379275502794983e-05,
      "loss": 0.0001,
      "step": 2300
    },
    {
      "epoch": 133.33,
      "grad_norm": 0.0011824576649814844,
      "learning_rate": 4.326506529404972e-05,
      "loss": 0.0001,
      "step": 2400
    },
    {
      "epoch": 138.89,
      "grad_norm": 0.001206545508466661,
      "learning_rate": 4.271931404387096e-05,
      "loss": 0.0001,
      "step": 2500
    },
    {
      "epoch": 144.44,
      "grad_norm": 0.0025704980362206697,
      "learning_rate": 4.215604094671835e-05,
      "loss": 0.0001,
      "step": 2600
    },
    {
      "epoch": 150.0,
      "grad_norm": 0.0012343807611614466,
      "learning_rate": 4.157580299847717e-05,
      "loss": 0.0001,
      "step": 2700
    },
    {
      "epoch": 155.56,
      "grad_norm": 0.0010726838372647762,
      "learning_rate": 4.0979173970824626e-05,
      "loss": 0.0001,
      "step": 2800
    },
    {
      "epoch": 161.11,
      "grad_norm": 0.0008975835517048836,
      "learning_rate": 4.036674384385231e-05,
      "loss": 0.0001,
      "step": 2900
    },
    {
      "epoch": 166.67,
      "grad_norm": 0.0015915404073894024,
      "learning_rate": 3.973911822266099e-05,
      "loss": 0.0001,
      "step": 3000
    },
    {
      "epoch": 172.22,
      "grad_norm": 0.0013330359943211079,
      "learning_rate": 3.909691773850445e-05,
      "loss": 0.0001,
      "step": 3100
    },
    {
      "epoch": 177.78,
      "grad_norm": 0.0007300216238945723,
      "learning_rate": 3.844077743507468e-05,
      "loss": 0.0001,
      "step": 3200
    },
    {
      "epoch": 183.33,
      "grad_norm": 0.0006974161951802671,
      "learning_rate": 3.777134614053522e-05,
      "loss": 0.0001,
      "step": 3300
    },
    {
      "epoch": 188.89,
      "grad_norm": 0.0016547333216294646,
      "learning_rate": 3.7089285825923615e-05,
      "loss": 0.0,
      "step": 3400
    },
    {
      "epoch": 194.44,
      "grad_norm": 0.0006157997995615005,
      "learning_rate": 3.639527095055753e-05,
      "loss": 0.0,
      "step": 3500
    },
    {
      "epoch": 200.0,
      "grad_norm": 0.000979723292402923,
      "learning_rate": 3.568998779509173e-05,
      "loss": 0.0,
      "step": 3600
    },
    {
      "epoch": 205.56,
      "grad_norm": 0.0008198455907404423,
      "learning_rate": 3.497413378288541e-05,
      "loss": 0.0,
      "step": 3700
    },
    {
      "epoch": 211.11,
      "grad_norm": 0.0012410281924530864,
      "learning_rate": 3.424841679035109e-05,
      "loss": 0.0,
      "step": 3800
    },
    {
      "epoch": 216.67,
      "grad_norm": 0.0007560970261693001,
      "learning_rate": 3.351355444696684e-05,
      "loss": 0.0,
      "step": 3900
    },
    {
      "epoch": 222.22,
      "grad_norm": 0.0005174472462385893,
      "learning_rate": 3.277027342564428e-05,
      "loss": 0.0,
      "step": 4000
    },
    {
      "epoch": 227.78,
      "grad_norm": 0.00041007634717971087,
      "learning_rate": 3.201930872415374e-05,
      "loss": 0.0,
      "step": 4100
    },
    {
      "epoch": 233.33,
      "grad_norm": 0.0008007108117453754,
      "learning_rate": 3.126140293831746e-05,
      "loss": 0.0,
      "step": 4200
    },
    {
      "epoch": 238.89,
      "grad_norm": 0.00044247345067560673,
      "learning_rate": 3.0497305527689445e-05,
      "loss": 0.0,
      "step": 4300
    },
    {
      "epoch": 244.44,
      "grad_norm": 0.0001442178909201175,
      "learning_rate": 2.972777207444791e-05,
      "loss": 0.0,
      "step": 4400
    },
    {
      "epoch": 250.0,
      "grad_norm": 0.0005981961148791015,
      "learning_rate": 2.8953563536233525e-05,
      "loss": 0.0,
      "step": 4500
    },
    {
      "epoch": 255.56,
      "grad_norm": 0.00022320376592688262,
      "learning_rate": 2.8175445493671972e-05,
      "loss": 0.0,
      "step": 4600
    },
    {
      "epoch": 261.11,
      "grad_norm": 0.0004137264913879335,
      "learning_rate": 2.7394187393325106e-05,
      "loss": 0.0,
      "step": 4700
    },
    {
      "epoch": 266.67,
      "grad_norm": 0.0003305468999315053,
      "learning_rate": 2.6610561786819204e-05,
      "loss": 0.0,
      "step": 4800
    },
    {
      "epoch": 272.22,
      "grad_norm": 0.00018206202366854995,
      "learning_rate": 2.5825343566902837e-05,
      "loss": 0.0,
      "step": 4900
    },
    {
      "epoch": 277.78,
      "grad_norm": 0.00040663284016773105,
      "learning_rate": 2.5039309201189614e-05,
      "loss": 0.0,
      "step": 5000
    },
    {
      "epoch": 283.33,
      "grad_norm": 0.0004440485790837556,
      "learning_rate": 2.4253235964343676e-05,
      "loss": 0.0,
      "step": 5100
    },
    {
      "epoch": 288.89,
      "grad_norm": 0.0003763148852158338,
      "learning_rate": 2.34679011694671e-05,
      "loss": 0.0,
      "step": 5200
    },
    {
      "epoch": 294.44,
      "grad_norm": 0.0002928711473941803,
      "learning_rate": 2.2684081399449327e-05,
      "loss": 0.0,
      "step": 5300
    },
    {
      "epoch": 300.0,
      "grad_norm": 0.0004030682030133903,
      "learning_rate": 2.1902551739038624e-05,
      "loss": 0.0,
      "step": 5400
    },
    {
      "epoch": 305.56,
      "grad_norm": 0.0004344022599980235,
      "learning_rate": 2.1124085008395054e-05,
      "loss": 0.0,
      "step": 5500
    },
    {
      "epoch": 311.11,
      "grad_norm": 0.00018473710224498063,
      "learning_rate": 2.03494509988827e-05,
      "loss": 0.0,
      "step": 5600
    },
    {
      "epoch": 316.67,
      "grad_norm": 0.00044333594269119203,
      "learning_rate": 1.9579415711857018e-05,
      "loss": 0.0,
      "step": 5700
    },
    {
      "epoch": 322.22,
      "grad_norm": 0.00039796039345674217,
      "learning_rate": 1.881474060119994e-05,
      "loss": 0.0,
      "step": 5800
    },
    {
      "epoch": 327.78,
      "grad_norm": 0.0002483979915268719,
      "learning_rate": 1.8056181820351738e-05,
      "loss": 0.0,
      "step": 5900
    },
    {
      "epoch": 333.33,
      "grad_norm": 0.00021489571372512728,
      "learning_rate": 1.7304489474584307e-05,
      "loss": 0.0,
      "step": 6000
    },
    {
      "epoch": 338.89,
      "grad_norm": 0.00023669018992222846,
      "learning_rate": 1.656040687925519e-05,
      "loss": 0.0,
      "step": 6100
    },
    {
      "epoch": 344.44,
      "grad_norm": 0.00023518071975558996,
      "learning_rate": 1.582466982477587e-05,
      "loss": 0.0,
      "step": 6200
    },
    {
      "epoch": 350.0,
      "grad_norm": 0.00023862645321059972,
      "learning_rate": 1.509800584902108e-05,
      "loss": 0.0,
      "step": 6300
    },
    {
      "epoch": 355.56,
      "grad_norm": 0.00020295938884373754,
      "learning_rate": 1.4381133517898804e-05,
      "loss": 0.0,
      "step": 6400
    },
    {
      "epoch": 361.11,
      "grad_norm": 0.0003517361474223435,
      "learning_rate": 1.3674761714792153e-05,
      "loss": 0.0,
      "step": 6500
    },
    {
      "epoch": 366.67,
      "grad_norm": 0.00022433781123254448,
      "learning_rate": 1.297958893957588e-05,
      "loss": 0.0,
      "step": 6600
    },
    {
      "epoch": 372.22,
      "grad_norm": 0.0002291495620738715,
      "learning_rate": 1.229630261790077e-05,
      "loss": 0.0,
      "step": 6700
    },
    {
      "epoch": 377.78,
      "grad_norm": 0.00021542608737945557,
      "learning_rate": 1.1625578421428714e-05,
      "loss": 0.0,
      "step": 6800
    },
    {
      "epoch": 383.33,
      "grad_norm": 0.00030446049640886486,
      "learning_rate": 1.0968079599690872e-05,
      "loss": 0.0,
      "step": 6900
    },
    {
      "epoch": 388.89,
      "grad_norm": 0.000319374434184283,
      "learning_rate": 1.0324456324229537e-05,
      "loss": 0.0,
      "step": 7000
    },
    {
      "epoch": 394.44,
      "grad_norm": 0.0002203370677307248,
      "learning_rate": 9.695345045672166e-06,
      "loss": 0.0,
      "step": 7100
    },
    {
      "epoch": 400.0,
      "grad_norm": 0.00015890812210272998,
      "learning_rate": 9.081367864373488e-06,
      "loss": 0.0,
      "step": 7200
    },
    {
      "epoch": 405.56,
      "grad_norm": 0.000279374944511801,
      "learning_rate": 8.483131915247968e-06,
      "loss": 0.0,
      "step": 7300
    },
    {
      "epoch": 411.11,
      "grad_norm": 0.00011563805310288444,
      "learning_rate": 7.901228767400859e-06,
      "loss": 0.0,
      "step": 7400
    },
    {
      "epoch": 416.67,
      "grad_norm": 0.00013281428255140781,
      "learning_rate": 7.336233839151693e-06,
      "loss": 0.0,
      "step": 7500
    },
    {
      "epoch": 422.22,
      "grad_norm": 0.0002364695246797055,
      "learning_rate": 6.788705829028483e-06,
      "loss": 0.0,
      "step": 7600
    },
    {
      "epoch": 427.78,
      "grad_norm": 0.00013619402307085693,
      "learning_rate": 6.259186163295438e-06,
      "loss": 0.0,
      "step": 7700
    },
    {
      "epoch": 433.33,
      "grad_norm": 0.0002181720337830484,
      "learning_rate": 5.748198460560475e-06,
      "loss": 0.0,
      "step": 7800
    },
    {
      "epoch": 438.89,
      "grad_norm": 0.0001409970282111317,
      "learning_rate": 5.256248013991857e-06,
      "loss": 0.0,
      "step": 7900
    },
    {
      "epoch": 444.44,
      "grad_norm": 8.899461681721732e-05,
      "learning_rate": 4.78382129165613e-06,
      "loss": 0.0,
      "step": 8000
    },
    {
      "epoch": 450.0,
      "grad_norm": 9.178288746625185e-05,
      "learning_rate": 4.331385455471346e-06,
      "loss": 0.0,
      "step": 8100
    },
    {
      "epoch": 455.56,
      "grad_norm": 0.00016385990602429956,
      "learning_rate": 3.8993878992512415e-06,
      "loss": 0.0,
      "step": 8200
    },
    {
      "epoch": 461.11,
      "grad_norm": 0.00012868938210885972,
      "learning_rate": 3.488255806297311e-06,
      "loss": 0.0,
      "step": 8300
    },
    {
      "epoch": 466.67,
      "grad_norm": 0.00018513025133870542,
      "learning_rate": 3.09839572697605e-06,
      "loss": 0.0,
      "step": 8400
    },
    {
      "epoch": 472.22,
      "grad_norm": 0.00016808818327262998,
      "learning_rate": 2.7301931766992917e-06,
      "loss": 0.0,
      "step": 8500
    },
    {
      "epoch": 477.78,
      "grad_norm": 0.0001630824408493936,
      "learning_rate": 2.384012254705048e-06,
      "loss": 0.0,
      "step": 8600
    },
    {
      "epoch": 483.33,
      "grad_norm": 0.00018219888443127275,
      "learning_rate": 2.0601952840158366e-06,
      "loss": 0.0,
      "step": 8700
    },
    {
      "epoch": 488.89,
      "grad_norm": 0.00011875651398440823,
      "learning_rate": 1.75906247293057e-06,
      "loss": 0.0,
      "step": 8800
    },
    {
      "epoch": 494.44,
      "grad_norm": 0.00015640753554180264,
      "learning_rate": 1.4809115983847266e-06,
      "loss": 0.0,
      "step": 8900
    },
    {
      "epoch": 500.0,
      "grad_norm": 0.00010140336962649599,
      "learning_rate": 1.226017711491867e-06,
      "loss": 0.0,
      "step": 9000
    },
    {
      "epoch": 505.56,
      "grad_norm": 0.00018589019600767642,
      "learning_rate": 9.946328655577624e-07,
      "loss": 0.0,
      "step": 9100
    },
    {
      "epoch": 511.11,
      "grad_norm": 0.00026709603844210505,
      "learning_rate": 7.869858668360042e-07,
      "loss": 0.0,
      "step": 9200
    },
    {
      "epoch": 516.67,
      "grad_norm": 0.00017575189121998847,
      "learning_rate": 6.032820482716001e-07,
      "loss": 0.0,
      "step": 9300
    },
    {
      "epoch": 522.22,
      "grad_norm": 0.0001377611479256302,
      "learning_rate": 4.437030664562969e-07,
      "loss": 0.0,
      "step": 9400
    },
    {
      "epoch": 527.78,
      "grad_norm": 9.088205842999741e-05,
      "learning_rate": 3.084067219964182e-07,
      "loss": 0.0,
      "step": 9500
    },
    {
      "epoch": 533.33,
      "grad_norm": 0.00015840640116948634,
      "learning_rate": 1.975268034707878e-07,
      "loss": 0.0,
      "step": 9600
    },
    {
      "epoch": 538.89,
      "grad_norm": 0.00013717034016735852,
      "learning_rate": 1.1117295513313475e-07,
      "loss": 0.0,
      "step": 9700
    }
  ],
  "logging_steps": 100,
  "max_steps": 10000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 556,
  "save_steps": 500,
  "total_flos": 2.673487599113011e+16,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}