{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 6.7272727272727275,
  "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": 463.28125,
      "epoch": 0.7272727272727273,
      "grad_norm": 1.8766214847564697,
      "kl": 0.002575141654233448,
      "learning_rate": 5e-07,
      "loss": 0.0,
      "reward": 1.158743228763342,
      "reward_std": 1.2187582328915596,
      "rewards/concensus_correctness_reward_func": 0.11062500067055225,
      "rewards/consensus_reward_func": 0.3125,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.5765869729220867,
      "rewards/soft_format_reward_func": 0.015625,
      "rewards/strict_format_reward_func": 0.0,
      "rewards/xmlcount_reward_func": 0.14340625703334808,
      "step": 2
    },
    {
      "completion_length": 428.2857142857143,
      "epoch": 1.3636363636363638,
      "grad_norm": 0.9281177520751953,
      "kl": 0.003123825615538018,
      "learning_rate": 4.864543104251586e-07,
      "loss": 0.0,
      "reward": 1.179137361901147,
      "reward_std": 0.9824260558400836,
      "rewards/concensus_correctness_reward_func": 0.05064285759414945,
      "rewards/consensus_reward_func": 0.35714285714285715,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.49353024789265226,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.03571428571428571,
      "rewards/xmlcount_reward_func": 0.24210714868136815,
      "step": 4
    },
    {
      "completion_length": 486.85714285714283,
      "epoch": 2.0,
      "grad_norm": 0.28709423542022705,
      "kl": 0.0025116097115512404,
      "learning_rate": 4.472851273490984e-07,
      "loss": 0.0,
      "reward": 0.8373106122016907,
      "reward_std": 0.921959366117205,
      "rewards/concensus_correctness_reward_func": 0.0,
      "rewards/consensus_reward_func": 0.21428571428571427,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.3933820351958275,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.0,
      "rewards/xmlcount_reward_func": 0.22964285633393697,
      "step": 6
    },
    {
      "completion_length": 616.875,
      "epoch": 2.7272727272727275,
      "grad_norm": 0.6827354431152344,
      "kl": 0.002641352009959519,
      "learning_rate": 3.867370395306068e-07,
      "loss": 0.0,
      "reward": 0.8498772419989109,
      "reward_std": 1.112674504518509,
      "rewards/concensus_correctness_reward_func": 0.06481250002980232,
      "rewards/consensus_reward_func": 0.3125,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.4721897467970848,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.015625,
      "rewards/xmlcount_reward_func": -0.015249999240040779,
      "step": 8
    },
    {
      "completion_length": 412.14285714285717,
      "epoch": 3.3636363636363638,
      "grad_norm": 0.4301053583621979,
      "kl": 0.002733915811404586,
      "learning_rate": 3.1137137178519977e-07,
      "loss": 0.0,
      "reward": 1.1507055972303664,
      "reward_std": 0.9976620588983808,
      "rewards/concensus_correctness_reward_func": 0.0,
      "rewards/consensus_reward_func": 0.5,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.3728484394294875,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.017857142857142856,
      "rewards/xmlcount_reward_func": 0.2600000138793673,
      "step": 10
    },
    {
      "completion_length": 561.7857142857143,
      "epoch": 4.0,
      "grad_norm": 0.3801424503326416,
      "kl": 0.0019332578272691794,
      "learning_rate": 2.2935516363191693e-07,
      "loss": 0.0,
      "reward": 0.7273521189178739,
      "reward_std": 1.0015213574681963,
      "rewards/concensus_correctness_reward_func": 0.0,
      "rewards/consensus_reward_func": 0.21428571428571427,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.5067449510097504,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.0,
      "rewards/xmlcount_reward_func": 0.006321425948824201,
      "step": 12
    },
    {
      "completion_length": 446.09375,
      "epoch": 4.7272727272727275,
      "grad_norm": 0.8057725429534912,
      "kl": 0.003685262781800702,
      "learning_rate": 1.4957614383675767e-07,
      "loss": 0.0,
      "reward": 1.2268882654607296,
      "reward_std": 1.0097248032689095,
      "rewards/concensus_correctness_reward_func": 0.07799999881535769,
      "rewards/consensus_reward_func": 0.375,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.39338826574385166,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.015625,
      "rewards/xmlcount_reward_func": 0.36487501207739115,
      "step": 14
    },
    {
      "completion_length": 472.5,
      "epoch": 5.363636363636363,
      "grad_norm": 0.49996957182884216,
      "kl": 0.0025227711968390004,
      "learning_rate": 8.067960709356478e-08,
      "loss": 0.0,
      "reward": 0.9737584335463387,
      "reward_std": 1.2166244174752916,
      "rewards/concensus_correctness_reward_func": 0.08371428613151823,
      "rewards/consensus_reward_func": 0.2857142857142857,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.4596870222261974,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.0,
      "rewards/xmlcount_reward_func": 0.1446428575686046,
      "step": 16
    },
    {
      "completion_length": 486.42857142857144,
      "epoch": 6.0,
      "grad_norm": 0.5733634829521179,
      "kl": 0.0032351258975852814,
      "learning_rate": 3.013156219837776e-08,
      "loss": 0.0,
      "reward": 1.01033189786332,
      "reward_std": 1.0374784554753984,
      "rewards/concensus_correctness_reward_func": 0.05357142857142857,
      "rewards/consensus_reward_func": 0.21428571428571427,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.3964747488498688,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.017857142857142856,
      "rewards/xmlcount_reward_func": 0.32814285159111023,
      "step": 18
    },
    {
      "completion_length": 499.875,
      "epoch": 6.7272727272727275,
      "grad_norm": 0.5060414671897888,
      "kl": 0.002737829345278442,
      "learning_rate": 3.4096741493194193e-09,
      "loss": 0.0,
      "reward": 0.46602348145097494,
      "reward_std": 0.9370183423161507,
      "rewards/concensus_correctness_reward_func": 0.03368749842047691,
      "rewards/consensus_reward_func": 0.0625,
      "rewards/cumulative_reward_2": 0.0,
      "rewards/final_correctness_reward_func": 0.0,
      "rewards/question_recreation_reward_func": 0.42717973701655865,
      "rewards/soft_format_reward_func": 0.0,
      "rewards/strict_format_reward_func": 0.0,
      "rewards/xmlcount_reward_func": -0.05734375026077032,
      "step": 20
    },
    {
      "epoch": 6.7272727272727275,
      "step": 20,
      "total_flos": 0.0,
      "train_loss": 2.5640933017712086e-06,
      "train_runtime": 1354.6855,
      "train_samples_per_second": 0.236,
      "train_steps_per_second": 0.015
    }
  ],
  "logging_steps": 2,
  "max_steps": 20,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "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": 4,
  "trial_name": null,
  "trial_params": null
}