{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.0012675024260788625,
  "eval_steps": 500,
  "global_step": 8,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0001584378032598578,
      "grad_norm": 7.086519326549023e-05,
      "learning_rate": 2.9998415465061005e-05,
      "loss": 0.0,
      "loss/policy_avg": 9.534414857625961e-08,
      "objective/entropy": 66.7407455444336,
      "objective/kl": 0.0,
      "objective/rlhf_reward": 1.9394512176513672,
      "objective/scores": 1.9393310546875,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "policy/entropy_avg": 0.501089334487915,
      "step": 1,
      "timer/calc_advantages": 1.9189459085464478,
      "timer/calc_loss": 0.7234929800033569,
      "timer/get_reward": 0.5044295787811279,
      "timer/training_step": 5.130897045135498,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0,
      "val/ratio_var": NaN
    },
    {
      "epoch": 0.0003168756065197156,
      "grad_norm": 9.20624828338623,
      "learning_rate": 2.999683093012201e-05,
      "loss": 0.0082,
      "loss/policy_avg": 0.008180337026715279,
      "objective/entropy": 58.23992919921875,
      "objective/kl": 0.04112038388848305,
      "objective/rlhf_reward": 2.586242914199829,
      "objective/scores": 2.59033203125,
      "policy/approxkl_avg": 0.13751985132694244,
      "policy/clipfrac_avg": 0.353515625,
      "policy/entropy_avg": 0.4371030628681183,
      "step": 2,
      "timer/calc_advantages": 1.77765953540802,
      "timer/calc_loss": 0.620478630065918,
      "timer/get_reward": 0.43380415439605713,
      "timer/training_step": 4.58284854888916,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0005027055740356,
      "val/ratio_var": NaN
    },
    {
      "epoch": 0.0004753134097795734,
      "grad_norm": 11.192896842956543,
      "learning_rate": 2.9995246395183014e-05,
      "loss": 0.0083,
      "loss/policy_avg": 0.008318130858242512,
      "objective/entropy": 60.747886657714844,
      "objective/kl": 0.1742033064365387,
      "objective/rlhf_reward": 2.7990851402282715,
      "objective/scores": 2.8162841796875,
      "policy/approxkl_avg": 0.1877279430627823,
      "policy/clipfrac_avg": 0.345703125,
      "policy/entropy_avg": 0.47007349133491516,
      "step": 3,
      "timer/calc_advantages": 2.159975528717041,
      "timer/calc_loss": 0.8658402562141418,
      "timer/get_reward": 0.5989301204681396,
      "timer/training_step": 5.865015029907227,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 0.9998154640197754,
      "val/ratio_var": NaN
    },
    {
      "epoch": 0.0006337512130394313,
      "grad_norm": 10.267037391662598,
      "learning_rate": 2.999366186024402e-05,
      "loss": 0.0119,
      "loss/policy_avg": 0.011887951754033566,
      "objective/entropy": 59.366294860839844,
      "objective/kl": 0.12045621126890182,
      "objective/rlhf_reward": 2.8014278411865234,
      "objective/scores": 2.8134765625,
      "policy/approxkl_avg": 0.2340196967124939,
      "policy/clipfrac_avg": 0.369140625,
      "policy/entropy_avg": 0.4660375118255615,
      "step": 4,
      "timer/calc_advantages": 1.9160572290420532,
      "timer/calc_loss": 0.8047055602073669,
      "timer/get_reward": 0.5554770231246948,
      "timer/training_step": 5.4731550216674805,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.000678539276123,
      "val/ratio_var": NaN
    },
    {
      "epoch": 0.000792189016299289,
      "grad_norm": 5.658022200805135e-05,
      "learning_rate": 2.9992077325305024e-05,
      "loss": 0.0,
      "loss/policy_avg": 9.505311027169228e-08,
      "objective/entropy": 63.056705474853516,
      "objective/kl": 0.0,
      "objective/rlhf_reward": 1.961860179901123,
      "objective/scores": 1.961883544921875,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "policy/entropy_avg": 0.49685803055763245,
      "step": 5,
      "timer/calc_advantages": 1.9551451206207275,
      "timer/calc_loss": 0.7570784091949463,
      "timer/get_reward": 0.5253068208694458,
      "timer/training_step": 5.274328231811523,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0,
      "val/ratio_var": NaN
    },
    {
      "epoch": 0.0009506268195591468,
      "grad_norm": 5.630626201629639,
      "learning_rate": 2.9990492790366028e-05,
      "loss": 0.0036,
      "loss/policy_avg": 0.0035882075317204,
      "objective/entropy": 64.0039291381836,
      "objective/kl": -0.06272067129611969,
      "objective/rlhf_reward": 2.5381531715393066,
      "objective/scores": 2.53204345703125,
      "policy/approxkl_avg": 0.07097644358873367,
      "policy/clipfrac_avg": 0.244140625,
      "policy/entropy_avg": 0.49359244108200073,
      "step": 6,
      "timer/calc_advantages": 1.9771534204483032,
      "timer/calc_loss": 0.7805941700935364,
      "timer/get_reward": 0.5281075835227966,
      "timer/training_step": 5.320864677429199,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.000971794128418,
      "val/ratio_var": NaN
    },
    {
      "epoch": 0.0011090646228190046,
      "grad_norm": 7.572779178619385,
      "learning_rate": 2.9988908255427033e-05,
      "loss": 0.0058,
      "loss/policy_avg": 0.005815813317894936,
      "objective/entropy": 69.00190734863281,
      "objective/kl": 0.07886971533298492,
      "objective/rlhf_reward": 3.3839988708496094,
      "objective/scores": 3.39208984375,
      "policy/approxkl_avg": 0.1876513808965683,
      "policy/clipfrac_avg": 0.34765625,
      "policy/entropy_avg": 0.520235538482666,
      "step": 7,
      "timer/calc_advantages": 2.296631336212158,
      "timer/calc_loss": 0.9676017761230469,
      "timer/get_reward": 0.6649996042251587,
      "timer/training_step": 6.399840831756592,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.000232458114624,
      "val/ratio_var": NaN
    },
    {
      "epoch": 0.0012675024260788625,
      "grad_norm": 11.984278678894043,
      "learning_rate": 2.9987323720488037e-05,
      "loss": 0.0093,
      "loss/policy_avg": 0.009322328492999077,
      "objective/entropy": 66.09063720703125,
      "objective/kl": 0.015054229646921158,
      "objective/rlhf_reward": 2.489471673965454,
      "objective/scores": 2.491485595703125,
      "policy/approxkl_avg": 0.18275578320026398,
      "policy/clipfrac_avg": 0.3046875,
      "policy/entropy_avg": 0.5070106387138367,
      "step": 8,
      "timer/calc_advantages": 2.2399752140045166,
      "timer/calc_loss": 1.0477569103240967,
      "timer/get_reward": 0.7243468761444092,
      "timer/training_step": 6.686029434204102,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0009795427322388,
      "val/ratio_var": NaN
    }
  ],
  "logging_steps": 1,
  "max_steps": 18933,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 8,
  "total_flos": 4025485044940800.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}