{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.8695652173913043,
  "eval_steps": 500,
  "global_step": 20,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "completion_length": 4.0,
      "epoch": 0.08695652173913043,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 9.931806517013612e-07,
      "loss": 0.0,
      "reward": 4.398228406906128,
      "reward_std": 0.0,
      "rewards/concensus_correctness_reward_func": 2.341999977827072,
      "rewards/consensus_reward_func": 2.0,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.05622836947441101,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.0,
      "rewards/xmlcount_reward_func": 0.0,
      "step": 2
    },
    {
      "completion_length": 4.0,
      "epoch": 0.17391304347826086,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 9.397368756032444e-07,
      "loss": 0.0,
      "reward": 3.4788137674331665,
      "reward_std": 0.0,
      "rewards/concensus_correctness_reward_func": 1.4399999976158142,
      "rewards/consensus_reward_func": 2.0,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.038813844323158264,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.0,
      "rewards/xmlcount_reward_func": 0.0,
      "step": 4
    },
    {
      "completion_length": 4.0,
      "epoch": 0.2608695652173913,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 8.386407858128706e-07,
      "loss": 0.0,
      "reward": 3.389747124398127,
      "reward_std": 0.0,
      "rewards/concensus_correctness_reward_func": 1.8585000038146973,
      "rewards/consensus_reward_func": 1.5,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.031247180071659386,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.0,
      "rewards/xmlcount_reward_func": 0.0,
      "step": 6
    },
    {
      "completion_length": 4.0,
      "epoch": 0.34782608695652173,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 7.008477123264847e-07,
      "loss": 0.0,
      "reward": 4.452260255813599,
      "reward_std": 0.0,
      "rewards/concensus_correctness_reward_func": 2.8999999463558197,
      "rewards/consensus_reward_func": 1.5,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.052260236465372145,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.0,
      "rewards/xmlcount_reward_func": 0.0,
      "step": 8
    },
    {
      "completion_length": 4.0,
      "epoch": 0.43478260869565216,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 5.412896727361662e-07,
      "loss": 0.0,
      "reward": 2.8429267406463623,
      "reward_std": 0.0,
      "rewards/concensus_correctness_reward_func": 1.324999988079071,
      "rewards/consensus_reward_func": 1.5,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.017926768865436316,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.0,
      "rewards/xmlcount_reward_func": 0.0,
      "step": 10
    },
    {
      "completion_length": 4.0,
      "epoch": 0.5217391304347826,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 3.772572564296004e-07,
      "loss": 0.0,
      "reward": 2.9833587873727083,
      "reward_std": 0.0,
      "rewards/concensus_correctness_reward_func": 1.4460000097751617,
      "rewards/consensus_reward_func": 1.5,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.03735882428009063,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.0,
      "rewards/xmlcount_reward_func": 0.0,
      "step": 12
    },
    {
      "completion_length": 4.0,
      "epoch": 0.6086956521739131,
      "grad_norm": 46.81050491333008,
      "kl": 0.0,
      "learning_rate": 2.2652592093878665e-07,
      "loss": 0.0,
      "reward": 2.0572528541088104,
      "reward_std": 0.3401183784008026,
      "rewards/concensus_correctness_reward_func": 1.0319999754428864,
      "rewards/consensus_reward_func": 1.0,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.025252872379496694,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.0,
      "rewards/xmlcount_reward_func": 0.0,
      "step": 14
    },
    {
      "completion_length": 3.75,
      "epoch": 0.6956521739130435,
      "grad_norm": 178.11549377441406,
      "kl": 0.004039056599140167,
      "learning_rate": 1.0542974530180327e-07,
      "loss": 0.0,
      "reward": 6.128282427787781,
      "reward_std": 2.487646604771726,
      "rewards/concensus_correctness_reward_func": 4.874499976634979,
      "rewards/consensus_reward_func": 1.25,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.003782564483117312,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.0,
      "rewards/xmlcount_reward_func": 0.0,
      "step": 16
    },
    {
      "completion_length": 4.0,
      "epoch": 0.782608695652174,
      "grad_norm": 0.003080617170780897,
      "kl": 0.0005997046828269958,
      "learning_rate": 2.7091379149682682e-08,
      "loss": 0.0,
      "reward": 3.0035960860550404,
      "reward_std": 0.0,
      "rewards/concensus_correctness_reward_func": 1.4249999821186066,
      "rewards/consensus_reward_func": 1.5,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.078596081584692,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.0,
      "rewards/xmlcount_reward_func": 0.0,
      "step": 18
    },
    {
      "completion_length": 4.0,
      "epoch": 0.8695652173913043,
      "grad_norm": 46.77695083618164,
      "kl": 0.0031125321984291077,
      "learning_rate": 0.0,
      "loss": 0.0,
      "reward": 2.7333508729934692,
      "reward_std": 0.3535533845424652,
      "rewards/concensus_correctness_reward_func": 1.4600000083446503,
      "rewards/consensus_reward_func": 1.25,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.02335089026018977,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.0,
      "rewards/xmlcount_reward_func": 0.0,
      "step": 20
    },
    {
      "epoch": 0.8695652173913043,
      "step": 20,
      "total_flos": 0.0,
      "train_loss": 1.798409249431643e-06,
      "train_runtime": 590.8726,
      "train_samples_per_second": 0.135,
      "train_steps_per_second": 0.034
    }
  ],
  "logging_steps": 2,
  "max_steps": 20,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 25,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}