{
  "best_global_step": 50,
  "best_metric": 0.0384465716779232,
  "best_model_checkpoint": null,
  "epoch": 2.98876404494382,
  "eval_steps": 5,
  "global_step": 66,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.2247191011235955,
      "grad_norm": 0.015925930812954903,
      "learning_rate": 9.393939393939395e-07,
      "loss": 0.0743,
      "step": 5
    },
    {
      "epoch": 0.2247191011235955,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 805.7142857142857,
      "eval_kl": 2.4131366184779574e-05,
      "eval_loss": 0.08203230053186417,
      "eval_num_tokens": 80747.0,
      "eval_reward": 0.5892857142857143,
      "eval_reward_std": 0.41495721680777414,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.017857142857142856,
      "eval_rewards/returns_int_reward_func": 0.5714285714285714,
      "eval_runtime": 916.367,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 5
    },
    {
      "epoch": 0.449438202247191,
      "grad_norm": 0.020018301904201508,
      "learning_rate": 8.636363636363636e-07,
      "loss": 0.1024,
      "step": 10
    },
    {
      "epoch": 0.449438202247191,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 779.8571428571429,
      "eval_kl": 2.630267824445452e-05,
      "eval_loss": 0.11342400312423706,
      "eval_num_tokens": 161176.0,
      "eval_reward": 0.6428571428571429,
      "eval_reward_std": 0.42048223529543194,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.017857142857142856,
      "eval_rewards/returns_int_reward_func": 0.625,
      "eval_runtime": 903.8295,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 10
    },
    {
      "epoch": 0.6741573033707865,
      "grad_norm": 0.017852170392870903,
      "learning_rate": 7.878787878787878e-07,
      "loss": 0.0572,
      "step": 15
    },
    {
      "epoch": 0.6741573033707865,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 765.5714285714286,
      "eval_kl": 2.2522040775844028e-05,
      "eval_loss": 0.10111860185861588,
      "eval_num_tokens": 237628.0,
      "eval_reward": 0.625,
      "eval_reward_std": 0.3737179126058306,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.625,
      "eval_runtime": 948.5061,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 15
    },
    {
      "epoch": 0.898876404494382,
      "grad_norm": 0.009381658397614956,
      "learning_rate": 7.121212121212121e-07,
      "loss": 0.0721,
      "step": 20
    },
    {
      "epoch": 0.898876404494382,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 770.2142857142857,
      "eval_kl": 2.6115349360874722e-05,
      "eval_loss": 0.08618057519197464,
      "eval_num_tokens": 321515.0,
      "eval_reward": 0.6607142857142857,
      "eval_reward_std": 0.3907997948782785,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.017857142857142856,
      "eval_rewards/returns_int_reward_func": 0.6428571428571429,
      "eval_runtime": 901.0677,
      "eval_samples_per_second": 0.016,
      "eval_steps_per_second": 0.004,
      "step": 20
    },
    {
      "epoch": 1.1348314606741572,
      "grad_norm": 0.011156530119478703,
      "learning_rate": 6.363636363636363e-07,
      "loss": 0.0267,
      "step": 25
    },
    {
      "epoch": 1.1348314606741572,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 719.4464285714286,
      "eval_kl": 2.654961177280971e-05,
      "eval_loss": 0.10328873991966248,
      "eval_num_tokens": 402609.0,
      "eval_reward": 0.6071428571428571,
      "eval_reward_std": 0.3681928941181728,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.6071428571428571,
      "eval_runtime": 841.536,
      "eval_samples_per_second": 0.017,
      "eval_steps_per_second": 0.005,
      "step": 25
    },
    {
      "epoch": 1.3595505617977528,
      "grad_norm": 0.023633191362023354,
      "learning_rate": 5.606060606060605e-07,
      "loss": 0.0791,
      "step": 30
    },
    {
      "epoch": 1.3595505617977528,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 750.1607142857143,
      "eval_kl": 2.5672571999686105e-05,
      "eval_loss": 0.18935348093509674,
      "eval_num_tokens": 485953.0,
      "eval_reward": 0.48214285714285715,
      "eval_reward_std": 0.4561965210097177,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.48214285714285715,
      "eval_runtime": 928.6197,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 30
    },
    {
      "epoch": 1.5842696629213484,
      "grad_norm": 0.019900217652320862,
      "learning_rate": 4.848484848484849e-07,
      "loss": 0.0917,
      "step": 35
    },
    {
      "epoch": 1.5842696629213484,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 801.1785714285714,
      "eval_kl": 2.525533948625837e-05,
      "eval_loss": 0.08746703714132309,
      "eval_num_tokens": 566630.0,
      "eval_reward": 0.5892857142857143,
      "eval_reward_std": 0.38476794958114624,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.5892857142857143,
      "eval_runtime": 929.2943,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 35
    },
    {
      "epoch": 1.8089887640449438,
      "grad_norm": 0.02139870449900627,
      "learning_rate": 4.090909090909091e-07,
      "loss": 0.0303,
      "step": 40
    },
    {
      "epoch": 1.8089887640449438,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 764.3035714285714,
      "eval_kl": 2.3505517414637973e-05,
      "eval_loss": 0.052344292402267456,
      "eval_num_tokens": 646240.0,
      "eval_reward": 0.6428571428571429,
      "eval_reward_std": 0.45067150252205984,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.6428571428571429,
      "eval_runtime": 933.2582,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 40
    },
    {
      "epoch": 2.044943820224719,
      "grad_norm": 0.01629846915602684,
      "learning_rate": 3.333333333333333e-07,
      "loss": 0.0706,
      "step": 45
    },
    {
      "epoch": 2.044943820224719,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 803.4642857142857,
      "eval_kl": 2.7345759528023856e-05,
      "eval_loss": 0.06691782921552658,
      "eval_num_tokens": 731068.0,
      "eval_reward": 0.5178571428571429,
      "eval_reward_std": 0.41495721680777414,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.5178571428571429,
      "eval_runtime": 947.4151,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 45
    },
    {
      "epoch": 2.2696629213483144,
      "grad_norm": 0.011612426489591599,
      "learning_rate": 2.5757575757575754e-07,
      "loss": 0.0385,
      "step": 50
    },
    {
      "epoch": 2.2696629213483144,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 759.1428571428571,
      "eval_kl": 2.525108201163156e-05,
      "eval_loss": 0.0384465716779232,
      "eval_num_tokens": 808379.0,
      "eval_reward": 0.6428571428571429,
      "eval_reward_std": 0.32695358991622925,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.017857142857142856,
      "eval_rewards/returns_int_reward_func": 0.625,
      "eval_runtime": 879.4095,
      "eval_samples_per_second": 0.016,
      "eval_steps_per_second": 0.005,
      "step": 50
    },
    {
      "epoch": 2.49438202247191,
      "grad_norm": 0.020006077364087105,
      "learning_rate": 1.818181818181818e-07,
      "loss": 0.0878,
      "step": 55
    },
    {
      "epoch": 2.49438202247191,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 806.8928571428571,
      "eval_kl": 2.43570123400007e-05,
      "eval_loss": 0.08411531895399094,
      "eval_num_tokens": 891396.0,
      "eval_reward": 0.6607142857142857,
      "eval_reward_std": 0.4039071798324585,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.017857142857142856,
      "eval_rewards/returns_int_reward_func": 0.6428571428571429,
      "eval_runtime": 938.1977,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 55
    },
    {
      "epoch": 2.7191011235955056,
      "grad_norm": 0.023959912359714508,
      "learning_rate": 1.0606060606060605e-07,
      "loss": 0.0497,
      "step": 60
    },
    {
      "epoch": 2.7191011235955056,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 722.9285714285714,
      "eval_kl": 2.206223351614816e-05,
      "eval_loss": 0.07995037734508514,
      "eval_num_tokens": 973773.0,
      "eval_reward": 0.6071428571428571,
      "eval_reward_std": 0.42048223529543194,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.017857142857142856,
      "eval_rewards/returns_int_reward_func": 0.5892857142857143,
      "eval_runtime": 887.1993,
      "eval_samples_per_second": 0.016,
      "eval_steps_per_second": 0.005,
      "step": 60
    },
    {
      "epoch": 2.943820224719101,
      "grad_norm": 0.013062759302556515,
      "learning_rate": 3.0303030303030305e-08,
      "loss": 0.0463,
      "step": 65
    },
    {
      "epoch": 2.943820224719101,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 745.4642857142857,
      "eval_kl": 2.300313540867397e-05,
      "eval_loss": 0.08457580208778381,
      "eval_num_tokens": 1055539.0,
      "eval_reward": 0.6428571428571429,
      "eval_reward_std": 0.32695358991622925,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.6428571428571429,
      "eval_runtime": 910.5507,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 65
    }
  ],
  "logging_steps": 5,
  "max_steps": 66,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 15,
  "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
}