{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 34265,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.1459321415541773,
      "grad_norm": 3.8735740724697223,
      "learning_rate": 2.915086081120514e-06,
      "loss": 1.154789306640625,
      "step": 1000
    },
    {
      "epoch": 0.2918642831083546,
      "grad_norm": 3.513073777315236,
      "learning_rate": 5.833090166326233e-06,
      "loss": 0.9386199340820313,
      "step": 2000
    },
    {
      "epoch": 0.43779642466253194,
      "grad_norm": 2.667804789171682,
      "learning_rate": 8.751094251531953e-06,
      "loss": 0.8885218505859375,
      "step": 3000
    },
    {
      "epoch": 0.5837285662167092,
      "grad_norm": 1.8311641416860456,
      "learning_rate": 9.991513345767592e-06,
      "loss": 0.8694163208007812,
      "step": 4000
    },
    {
      "epoch": 0.7296607077708865,
      "grad_norm": 2.250397237666049,
      "learning_rate": 9.936020028278053e-06,
      "loss": 0.85284765625,
      "step": 5000
    },
    {
      "epoch": 0.8755928493250639,
      "grad_norm": 2.6756387164181197,
      "learning_rate": 9.829343371836088e-06,
      "loss": 0.821312744140625,
      "step": 6000
    },
    {
      "epoch": 1.021452024808464,
      "grad_norm": 2.6729990910245593,
      "learning_rate": 9.672589544454328e-06,
      "loss": 0.7774617919921875,
      "step": 7000
    },
    {
      "epoch": 1.1673841663626414,
      "grad_norm": 2.642277321184833,
      "learning_rate": 9.46738398205746e-06,
      "loss": 0.608907958984375,
      "step": 8000
    },
    {
      "epoch": 1.3133163079168186,
      "grad_norm": 2.1544218289818913,
      "learning_rate": 9.215854533761766e-06,
      "loss": 0.6088528442382812,
      "step": 9000
    },
    {
      "epoch": 1.459248449470996,
      "grad_norm": 2.3990917619771497,
      "learning_rate": 8.920609397454381e-06,
      "loss": 0.6051533203125,
      "step": 10000
    },
    {
      "epoch": 1.6051805910251733,
      "grad_norm": 2.5614175915079387,
      "learning_rate": 8.584710074466158e-06,
      "loss": 0.6031138916015625,
      "step": 11000
    },
    {
      "epoch": 1.7511127325793505,
      "grad_norm": 1.5472599757402412,
      "learning_rate": 8.211639623780629e-06,
      "loss": 0.6085886840820313,
      "step": 12000
    },
    {
      "epoch": 1.897044874133528,
      "grad_norm": 2.384688518854998,
      "learning_rate": 7.805266544962458e-06,
      "loss": 0.6080531616210938,
      "step": 13000
    },
    {
      "epoch": 2.042904049616928,
      "grad_norm": 2.0076063347622974,
      "learning_rate": 7.3698046643160645e-06,
      "loss": 0.520530029296875,
      "step": 14000
    },
    {
      "epoch": 2.1888361911711054,
      "grad_norm": 1.529742164922155,
      "learning_rate": 6.909769440229038e-06,
      "loss": 0.33792144775390626,
      "step": 15000
    },
    {
      "epoch": 2.334768332725283,
      "grad_norm": 2.116689548920815,
      "learning_rate": 6.4299311407857035e-06,
      "loss": 0.3347291259765625,
      "step": 16000
    },
    {
      "epoch": 2.48070047427946,
      "grad_norm": 1.8084278631560584,
      "learning_rate": 5.935265379168761e-06,
      "loss": 0.3385657958984375,
      "step": 17000
    },
    {
      "epoch": 2.6266326158336373,
      "grad_norm": 3.29945063597593,
      "learning_rate": 5.430901519764892e-06,
      "loss": 0.33349261474609376,
      "step": 18000
    },
    {
      "epoch": 2.7725647573878147,
      "grad_norm": 1.9290178050582503,
      "learning_rate": 4.9220694899697185e-06,
      "loss": 0.3383995361328125,
      "step": 19000
    },
    {
      "epoch": 2.918496898941992,
      "grad_norm": 1.8157318709036212,
      "learning_rate": 4.414045549219315e-06,
      "loss": 0.3316441650390625,
      "step": 20000
    },
    {
      "epoch": 3.064356074425392,
      "grad_norm": 3.0277622115117175,
      "learning_rate": 3.912097577588397e-06,
      "loss": 0.2421361541748047,
      "step": 21000
    },
    {
      "epoch": 3.2102882159795696,
      "grad_norm": 2.0816758739547914,
      "learning_rate": 3.4214304512770823e-06,
      "loss": 0.1313203887939453,
      "step": 22000
    },
    {
      "epoch": 3.356220357533747,
      "grad_norm": 1.739691217957409,
      "learning_rate": 2.9471320714071095e-06,
      "loss": 0.12995486450195312,
      "step": 23000
    },
    {
      "epoch": 3.502152499087924,
      "grad_norm": 3.1257187709887875,
      "learning_rate": 2.4941206057740675e-06,
      "loss": 0.1304095458984375,
      "step": 24000
    },
    {
      "epoch": 3.6480846406421015,
      "grad_norm": 2.2140995652592785,
      "learning_rate": 2.06709349062457e-06,
      "loss": 0.12645170593261718,
      "step": 25000
    },
    {
      "epoch": 3.7940167821962785,
      "grad_norm": 1.7680132762507759,
      "learning_rate": 1.6704787212769829e-06,
      "loss": 0.12362543487548829,
      "step": 26000
    },
    {
      "epoch": 3.939948923750456,
      "grad_norm": 3.186667247027872,
      "learning_rate": 1.3083889366705216e-06,
      "loss": 0.11934644317626954,
      "step": 27000
    },
    {
      "epoch": 4.085808099233856,
      "grad_norm": 0.9381402976600279,
      "learning_rate": 9.845787739562829e-07,
      "loss": 0.07250856018066407,
      "step": 28000
    },
    {
      "epoch": 4.231740240788033,
      "grad_norm": 1.5081719675375433,
      "learning_rate": 7.024059353355333e-07,
      "loss": 0.03657876205444336,
      "step": 29000
    },
    {
      "epoch": 4.377672382342211,
      "grad_norm": 1.3793894154463466,
      "learning_rate": 4.64796370857008e-07,
      "loss": 0.035053966522216796,
      "step": 30000
    },
    {
      "epoch": 4.523604523896388,
      "grad_norm": 1.9377667427717677,
      "learning_rate": 2.7421393820510846e-07,
      "loss": 0.0349309196472168,
      "step": 31000
    },
    {
      "epoch": 4.669536665450566,
      "grad_norm": 1.109997862535006,
      "learning_rate": 1.326348540874095e-07,
      "loss": 0.03447756195068359,
      "step": 32000
    },
    {
      "epoch": 4.815468807004743,
      "grad_norm": 1.152236730081828,
      "learning_rate": 4.152720214406214e-08,
      "loss": 0.03444968414306641,
      "step": 33000
    },
    {
      "epoch": 4.96140094855892,
      "grad_norm": 1.6457557752833996,
      "learning_rate": 1.8357098688476238e-09,
      "loss": 0.034835365295410155,
      "step": 34000
    },
    {
      "epoch": 5.0,
      "step": 34265,
      "total_flos": 481722301939712.0,
      "train_loss": 0.401988563902201,
      "train_runtime": 92209.7944,
      "train_samples_per_second": 2.973,
      "train_steps_per_second": 0.372
    }
  ],
  "logging_steps": 1000,
  "max_steps": 34265,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 50000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 481722301939712.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}