{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 280.0,
  "eval_steps": 500,
  "global_step": 43960,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 3.1847133757961785,
      "grad_norm": 3.422844171524048,
      "learning_rate": 1.9872611464968155e-05,
      "loss": 0.9766,
      "step": 500
    },
    {
      "epoch": 6.369426751592357,
      "grad_norm": 4.270854949951172,
      "learning_rate": 1.9745222929936306e-05,
      "loss": 0.1036,
      "step": 1000
    },
    {
      "epoch": 9.554140127388536,
      "grad_norm": 1.767094373703003,
      "learning_rate": 1.961783439490446e-05,
      "loss": 0.0667,
      "step": 1500
    },
    {
      "epoch": 12.738853503184714,
      "grad_norm": 3.3422610759735107,
      "learning_rate": 1.9490445859872614e-05,
      "loss": 0.0507,
      "step": 2000
    },
    {
      "epoch": 15.923566878980893,
      "grad_norm": 2.3477580547332764,
      "learning_rate": 1.9363057324840767e-05,
      "loss": 0.0369,
      "step": 2500
    },
    {
      "epoch": 19.10828025477707,
      "grad_norm": 1.2746729850769043,
      "learning_rate": 1.9235668789808918e-05,
      "loss": 0.0279,
      "step": 3000
    },
    {
      "epoch": 22.29299363057325,
      "grad_norm": 1.4915257692337036,
      "learning_rate": 1.910828025477707e-05,
      "loss": 0.0208,
      "step": 3500
    },
    {
      "epoch": 25.477707006369428,
      "grad_norm": 1.313347339630127,
      "learning_rate": 1.8980891719745225e-05,
      "loss": 0.0162,
      "step": 4000
    },
    {
      "epoch": 28.662420382165607,
      "grad_norm": 1.810874581336975,
      "learning_rate": 1.8853503184713376e-05,
      "loss": 0.0124,
      "step": 4500
    },
    {
      "epoch": 31.847133757961785,
      "grad_norm": 1.6056509017944336,
      "learning_rate": 1.872611464968153e-05,
      "loss": 0.0102,
      "step": 5000
    },
    {
      "epoch": 35.031847133757964,
      "grad_norm": 1.908366084098816,
      "learning_rate": 1.8598726114649684e-05,
      "loss": 0.0082,
      "step": 5500
    },
    {
      "epoch": 38.21656050955414,
      "grad_norm": 0.49258118867874146,
      "learning_rate": 1.8471337579617837e-05,
      "loss": 0.0068,
      "step": 6000
    },
    {
      "epoch": 41.40127388535032,
      "grad_norm": 1.2429087162017822,
      "learning_rate": 1.8343949044585988e-05,
      "loss": 0.0066,
      "step": 6500
    },
    {
      "epoch": 44.5859872611465,
      "grad_norm": 1.1771211624145508,
      "learning_rate": 1.821656050955414e-05,
      "loss": 0.0059,
      "step": 7000
    },
    {
      "epoch": 47.77070063694268,
      "grad_norm": 0.8416564464569092,
      "learning_rate": 1.8089171974522295e-05,
      "loss": 0.0047,
      "step": 7500
    },
    {
      "epoch": 50.955414012738856,
      "grad_norm": 1.357130765914917,
      "learning_rate": 1.796178343949045e-05,
      "loss": 0.0044,
      "step": 8000
    },
    {
      "epoch": 54.140127388535035,
      "grad_norm": 0.038406338542699814,
      "learning_rate": 1.78343949044586e-05,
      "loss": 0.0047,
      "step": 8500
    },
    {
      "epoch": 57.32484076433121,
      "grad_norm": 0.9533712267875671,
      "learning_rate": 1.7707006369426754e-05,
      "loss": 0.0042,
      "step": 9000
    },
    {
      "epoch": 60.50955414012739,
      "grad_norm": 1.7286570072174072,
      "learning_rate": 1.7579617834394907e-05,
      "loss": 0.0038,
      "step": 9500
    },
    {
      "epoch": 63.69426751592356,
      "grad_norm": 0.576912522315979,
      "learning_rate": 1.7452229299363058e-05,
      "loss": 0.0037,
      "step": 10000
    },
    {
      "epoch": 66.87898089171975,
      "grad_norm": 0.5280445218086243,
      "learning_rate": 1.732484076433121e-05,
      "loss": 0.0036,
      "step": 10500
    },
    {
      "epoch": 70.06369426751593,
      "grad_norm": 0.664702296257019,
      "learning_rate": 1.7197452229299365e-05,
      "loss": 0.0029,
      "step": 11000
    },
    {
      "epoch": 73.2484076433121,
      "grad_norm": 0.18195950984954834,
      "learning_rate": 1.707006369426752e-05,
      "loss": 0.0033,
      "step": 11500
    },
    {
      "epoch": 76.43312101910828,
      "grad_norm": 0.419512003660202,
      "learning_rate": 1.694267515923567e-05,
      "loss": 0.0036,
      "step": 12000
    },
    {
      "epoch": 79.61783439490446,
      "grad_norm": 1.1403623819351196,
      "learning_rate": 1.6815286624203824e-05,
      "loss": 0.0033,
      "step": 12500
    },
    {
      "epoch": 82.80254777070064,
      "grad_norm": 1.0261355638504028,
      "learning_rate": 1.6687898089171977e-05,
      "loss": 0.0027,
      "step": 13000
    },
    {
      "epoch": 85.98726114649682,
      "grad_norm": 0.1661386638879776,
      "learning_rate": 1.6560509554140128e-05,
      "loss": 0.0024,
      "step": 13500
    },
    {
      "epoch": 89.171974522293,
      "grad_norm": 0.039521217346191406,
      "learning_rate": 1.643312101910828e-05,
      "loss": 0.0029,
      "step": 14000
    },
    {
      "epoch": 92.35668789808918,
      "grad_norm": 0.047124966979026794,
      "learning_rate": 1.6305732484076436e-05,
      "loss": 0.0028,
      "step": 14500
    },
    {
      "epoch": 95.54140127388536,
      "grad_norm": 0.012884723953902721,
      "learning_rate": 1.617834394904459e-05,
      "loss": 0.0026,
      "step": 15000
    },
    {
      "epoch": 98.72611464968153,
      "grad_norm": 0.41387057304382324,
      "learning_rate": 1.605095541401274e-05,
      "loss": 0.0026,
      "step": 15500
    },
    {
      "epoch": 101.91082802547771,
      "grad_norm": 0.6253780722618103,
      "learning_rate": 1.5923566878980894e-05,
      "loss": 0.0025,
      "step": 16000
    },
    {
      "epoch": 105.09554140127389,
      "grad_norm": 1.3857741355895996,
      "learning_rate": 1.5796178343949047e-05,
      "loss": 0.0026,
      "step": 16500
    },
    {
      "epoch": 108.28025477707007,
      "grad_norm": 0.14438733458518982,
      "learning_rate": 1.56687898089172e-05,
      "loss": 0.0021,
      "step": 17000
    },
    {
      "epoch": 111.46496815286625,
      "grad_norm": 0.0025589189026504755,
      "learning_rate": 1.5541401273885352e-05,
      "loss": 0.0024,
      "step": 17500
    },
    {
      "epoch": 114.64968152866243,
      "grad_norm": 0.008726066909730434,
      "learning_rate": 1.5414012738853506e-05,
      "loss": 0.002,
      "step": 18000
    },
    {
      "epoch": 117.8343949044586,
      "grad_norm": 0.0028913430869579315,
      "learning_rate": 1.528662420382166e-05,
      "loss": 0.002,
      "step": 18500
    },
    {
      "epoch": 121.01910828025478,
      "grad_norm": 0.0131604615598917,
      "learning_rate": 1.5159235668789811e-05,
      "loss": 0.0021,
      "step": 19000
    },
    {
      "epoch": 124.20382165605096,
      "grad_norm": 0.09808602929115295,
      "learning_rate": 1.5031847133757964e-05,
      "loss": 0.0018,
      "step": 19500
    },
    {
      "epoch": 127.38853503184713,
      "grad_norm": 0.02476484701037407,
      "learning_rate": 1.4904458598726114e-05,
      "loss": 0.0015,
      "step": 20000
    },
    {
      "epoch": 130.57324840764332,
      "grad_norm": 0.015077279880642891,
      "learning_rate": 1.4777070063694268e-05,
      "loss": 0.0018,
      "step": 20500
    },
    {
      "epoch": 133.7579617834395,
      "grad_norm": 0.06911738961935043,
      "learning_rate": 1.464968152866242e-05,
      "loss": 0.0022,
      "step": 21000
    },
    {
      "epoch": 136.94267515923568,
      "grad_norm": 0.00874880701303482,
      "learning_rate": 1.4522292993630574e-05,
      "loss": 0.0019,
      "step": 21500
    },
    {
      "epoch": 140.12738853503186,
      "grad_norm": 1.0872219800949097,
      "learning_rate": 1.4394904458598726e-05,
      "loss": 0.0015,
      "step": 22000
    },
    {
      "epoch": 143.31210191082803,
      "grad_norm": 0.0024208752438426018,
      "learning_rate": 1.426751592356688e-05,
      "loss": 0.002,
      "step": 22500
    },
    {
      "epoch": 146.4968152866242,
      "grad_norm": 0.0036262290086597204,
      "learning_rate": 1.4140127388535032e-05,
      "loss": 0.0015,
      "step": 23000
    },
    {
      "epoch": 149.6815286624204,
      "grad_norm": 0.03490140661597252,
      "learning_rate": 1.4012738853503186e-05,
      "loss": 0.0016,
      "step": 23500
    },
    {
      "epoch": 152.86624203821657,
      "grad_norm": 0.7626912593841553,
      "learning_rate": 1.3885350318471338e-05,
      "loss": 0.0015,
      "step": 24000
    },
    {
      "epoch": 156.05095541401275,
      "grad_norm": 0.014237442053854465,
      "learning_rate": 1.375796178343949e-05,
      "loss": 0.0017,
      "step": 24500
    },
    {
      "epoch": 159.23566878980893,
      "grad_norm": 0.005636280868202448,
      "learning_rate": 1.3630573248407644e-05,
      "loss": 0.0016,
      "step": 25000
    },
    {
      "epoch": 162.4203821656051,
      "grad_norm": 0.05069362744688988,
      "learning_rate": 1.3503184713375796e-05,
      "loss": 0.0015,
      "step": 25500
    },
    {
      "epoch": 165.60509554140128,
      "grad_norm": 0.0035111624747514725,
      "learning_rate": 1.337579617834395e-05,
      "loss": 0.0016,
      "step": 26000
    },
    {
      "epoch": 168.78980891719746,
      "grad_norm": 0.002414689166471362,
      "learning_rate": 1.3248407643312102e-05,
      "loss": 0.0014,
      "step": 26500
    },
    {
      "epoch": 171.97452229299364,
      "grad_norm": 0.05457231029868126,
      "learning_rate": 1.3121019108280256e-05,
      "loss": 0.0014,
      "step": 27000
    },
    {
      "epoch": 175.15923566878982,
      "grad_norm": 0.0018200245685875416,
      "learning_rate": 1.2993630573248408e-05,
      "loss": 0.0013,
      "step": 27500
    },
    {
      "epoch": 178.343949044586,
      "grad_norm": 0.37822210788726807,
      "learning_rate": 1.2866242038216562e-05,
      "loss": 0.0014,
      "step": 28000
    },
    {
      "epoch": 181.52866242038218,
      "grad_norm": 0.009616321884095669,
      "learning_rate": 1.2738853503184714e-05,
      "loss": 0.0019,
      "step": 28500
    },
    {
      "epoch": 184.71337579617835,
      "grad_norm": 0.20008017122745514,
      "learning_rate": 1.2611464968152866e-05,
      "loss": 0.0024,
      "step": 29000
    },
    {
      "epoch": 187.89808917197453,
      "grad_norm": 1.0314069986343384,
      "learning_rate": 1.248407643312102e-05,
      "loss": 0.0011,
      "step": 29500
    },
    {
      "epoch": 191.0828025477707,
      "grad_norm": 0.0037857957649976015,
      "learning_rate": 1.2356687898089172e-05,
      "loss": 0.0015,
      "step": 30000
    },
    {
      "epoch": 194.2675159235669,
      "grad_norm": 0.07296785712242126,
      "learning_rate": 1.2229299363057326e-05,
      "loss": 0.0012,
      "step": 30500
    },
    {
      "epoch": 197.45222929936307,
      "grad_norm": 0.25247231125831604,
      "learning_rate": 1.2101910828025478e-05,
      "loss": 0.0015,
      "step": 31000
    },
    {
      "epoch": 200.63694267515925,
      "grad_norm": 0.008463304489850998,
      "learning_rate": 1.1974522292993632e-05,
      "loss": 0.0012,
      "step": 31500
    },
    {
      "epoch": 203.82165605095543,
      "grad_norm": 0.0013085840037092566,
      "learning_rate": 1.1847133757961784e-05,
      "loss": 0.0014,
      "step": 32000
    },
    {
      "epoch": 207.0063694267516,
      "grad_norm": 0.00894926954060793,
      "learning_rate": 1.1719745222929938e-05,
      "loss": 0.0011,
      "step": 32500
    },
    {
      "epoch": 210.19108280254778,
      "grad_norm": 0.016246678307652473,
      "learning_rate": 1.159235668789809e-05,
      "loss": 0.0011,
      "step": 33000
    },
    {
      "epoch": 213.37579617834396,
      "grad_norm": 0.00870199128985405,
      "learning_rate": 1.1464968152866242e-05,
      "loss": 0.0011,
      "step": 33500
    },
    {
      "epoch": 216.56050955414014,
      "grad_norm": 0.3081636130809784,
      "learning_rate": 1.1337579617834396e-05,
      "loss": 0.001,
      "step": 34000
    },
    {
      "epoch": 219.74522292993632,
      "grad_norm": 0.7328362464904785,
      "learning_rate": 1.1210191082802548e-05,
      "loss": 0.0016,
      "step": 34500
    },
    {
      "epoch": 222.9299363057325,
      "grad_norm": 0.0006987511878833175,
      "learning_rate": 1.1082802547770702e-05,
      "loss": 0.0012,
      "step": 35000
    },
    {
      "epoch": 226.11464968152868,
      "grad_norm": 0.016234412789344788,
      "learning_rate": 1.0955414012738854e-05,
      "loss": 0.0011,
      "step": 35500
    },
    {
      "epoch": 229.29936305732485,
      "grad_norm": 0.7393547892570496,
      "learning_rate": 1.0828025477707008e-05,
      "loss": 0.0011,
      "step": 36000
    },
    {
      "epoch": 232.48407643312103,
      "grad_norm": 0.0006552963168360293,
      "learning_rate": 1.070063694267516e-05,
      "loss": 0.0009,
      "step": 36500
    },
    {
      "epoch": 235.6687898089172,
      "grad_norm": 0.003285791492089629,
      "learning_rate": 1.0573248407643314e-05,
      "loss": 0.0011,
      "step": 37000
    },
    {
      "epoch": 238.8535031847134,
      "grad_norm": 0.8827471733093262,
      "learning_rate": 1.0445859872611466e-05,
      "loss": 0.0009,
      "step": 37500
    },
    {
      "epoch": 242.03821656050957,
      "grad_norm": 0.03062291443347931,
      "learning_rate": 1.0318471337579618e-05,
      "loss": 0.001,
      "step": 38000
    },
    {
      "epoch": 245.22292993630575,
      "grad_norm": 1.2000418901443481,
      "learning_rate": 1.0191082802547772e-05,
      "loss": 0.0014,
      "step": 38500
    },
    {
      "epoch": 248.40764331210192,
      "grad_norm": 0.5093538761138916,
      "learning_rate": 1.0063694267515924e-05,
      "loss": 0.0009,
      "step": 39000
    },
    {
      "epoch": 251.59235668789808,
      "grad_norm": 0.005525712389498949,
      "learning_rate": 9.936305732484078e-06,
      "loss": 0.0008,
      "step": 39500
    },
    {
      "epoch": 254.77707006369425,
      "grad_norm": 1.1260795593261719,
      "learning_rate": 9.80891719745223e-06,
      "loss": 0.0011,
      "step": 40000
    },
    {
      "epoch": 257.96178343949043,
      "grad_norm": 0.0011535738594830036,
      "learning_rate": 9.681528662420384e-06,
      "loss": 0.0012,
      "step": 40500
    },
    {
      "epoch": 261.14649681528664,
      "grad_norm": 0.009397489950060844,
      "learning_rate": 9.554140127388536e-06,
      "loss": 0.0012,
      "step": 41000
    },
    {
      "epoch": 264.3312101910828,
      "grad_norm": 0.003204810433089733,
      "learning_rate": 9.426751592356688e-06,
      "loss": 0.0008,
      "step": 41500
    },
    {
      "epoch": 267.515923566879,
      "grad_norm": 0.0012358404928818345,
      "learning_rate": 9.299363057324842e-06,
      "loss": 0.0009,
      "step": 42000
    },
    {
      "epoch": 270.70063694267515,
      "grad_norm": 0.006686298176646233,
      "learning_rate": 9.171974522292994e-06,
      "loss": 0.0011,
      "step": 42500
    },
    {
      "epoch": 273.88535031847135,
      "grad_norm": 0.004616052843630314,
      "learning_rate": 9.044585987261148e-06,
      "loss": 0.001,
      "step": 43000
    },
    {
      "epoch": 277.0700636942675,
      "grad_norm": 0.000694330723490566,
      "learning_rate": 8.9171974522293e-06,
      "loss": 0.0009,
      "step": 43500
    }
  ],
  "logging_steps": 500,
  "max_steps": 78500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 500,
  "save_steps": 500,
  "total_flos": 9164086093283328.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}