{
  "best_global_step": 15,
  "best_metric": 9.510484915153938e-07,
  "best_model_checkpoint": "outputs/microsoft/Phi-3.5-mini-instruct/aime_mcl_256/checkpoint-15",
  "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": 0.0,
      "learning_rate": 9.393939393939395e-07,
      "loss": 0.0,
      "step": 5
    },
    {
      "epoch": 0.2247191011235955,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 255.30357142857142,
      "eval_kl": 0.0,
      "eval_loss": 0.004398442804813385,
      "eval_num_tokens": 38036.0,
      "eval_reward": 0.017857142857142856,
      "eval_reward_std": 0.03571428571428571,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.017857142857142856,
      "eval_runtime": 214.4611,
      "eval_samples_per_second": 0.065,
      "eval_steps_per_second": 0.019,
      "step": 5
    },
    {
      "epoch": 0.449438202247191,
      "grad_norm": 1.931009819600149e-06,
      "learning_rate": 8.636363636363636e-07,
      "loss": 0.007,
      "step": 10
    },
    {
      "epoch": 0.449438202247191,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 255.82142857142858,
      "eval_kl": 2.9414892196655273e-05,
      "eval_loss": 0.0010968457208946347,
      "eval_num_tokens": 75945.0,
      "eval_reward": 0.03571428571428571,
      "eval_reward_std": 0.07142857142857142,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.017857142857142856,
      "eval_rewards/returns_int_reward_func": 0.017857142857142856,
      "eval_runtime": 214.3026,
      "eval_samples_per_second": 0.065,
      "eval_steps_per_second": 0.019,
      "step": 10
    },
    {
      "epoch": 0.6741573033707865,
      "grad_norm": 0.0,
      "learning_rate": 7.878787878787878e-07,
      "loss": 0.0,
      "step": 15
    },
    {
      "epoch": 0.6741573033707865,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 256.0,
      "eval_kl": 2.4216515677315848e-05,
      "eval_loss": 9.510484915153938e-07,
      "eval_num_tokens": 112173.0,
      "eval_reward": 0.0,
      "eval_reward_std": 0.0,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.0,
      "eval_runtime": 214.2147,
      "eval_samples_per_second": 0.065,
      "eval_steps_per_second": 0.019,
      "step": 15
    },
    {
      "epoch": 0.898876404494382,
      "grad_norm": 6.757839514648367e-07,
      "learning_rate": 7.121212121212121e-07,
      "loss": 0.0023,
      "step": 20
    },
    {
      "epoch": 0.898876404494382,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 256.0,
      "eval_kl": 2.636228288922991e-05,
      "eval_loss": 1.0162302714888938e-06,
      "eval_num_tokens": 154178.0,
      "eval_reward": 0.017857142857142856,
      "eval_reward_std": 0.03571428571428571,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.017857142857142856,
      "eval_runtime": 214.4077,
      "eval_samples_per_second": 0.065,
      "eval_steps_per_second": 0.019,
      "step": 20
    },
    {
      "epoch": 1.1348314606741572,
      "grad_norm": 1.0110848052136134e-06,
      "learning_rate": 6.363636363636363e-07,
      "loss": 0.0,
      "step": 25
    },
    {
      "epoch": 1.1348314606741572,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 255.76785714285714,
      "eval_kl": 2.2883926119123187e-05,
      "eval_loss": 0.0014293499989435077,
      "eval_num_tokens": 193586.0,
      "eval_reward": 0.017857142857142856,
      "eval_reward_std": 0.03571428571428571,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.017857142857142856,
      "eval_runtime": 214.4438,
      "eval_samples_per_second": 0.065,
      "eval_steps_per_second": 0.019,
      "step": 25
    },
    {
      "epoch": 1.3595505617977528,
      "grad_norm": 8.461184393127041e-07,
      "learning_rate": 5.606060606060605e-07,
      "loss": 0.0,
      "step": 30
    },
    {
      "epoch": 1.3595505617977528,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 254.96428571428572,
      "eval_kl": 3.0347279139927457e-05,
      "eval_loss": 0.006671134382486343,
      "eval_num_tokens": 231950.0,
      "eval_reward": 0.017857142857142856,
      "eval_reward_std": 0.03571428571428571,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.017857142857142856,
      "eval_runtime": 214.3165,
      "eval_samples_per_second": 0.065,
      "eval_steps_per_second": 0.019,
      "step": 30
    },
    {
      "epoch": 1.5842696629213484,
      "grad_norm": 7.697835826547816e-07,
      "learning_rate": 4.848484848484849e-07,
      "loss": 0.0,
      "step": 35
    },
    {
      "epoch": 1.5842696629213484,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 256.0,
      "eval_kl": 3.2765524727957586e-05,
      "eval_loss": 1.3170388228900265e-06,
      "eval_num_tokens": 270522.0,
      "eval_reward": 0.0,
      "eval_reward_std": 0.0,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.0,
      "eval_runtime": 214.4369,
      "eval_samples_per_second": 0.065,
      "eval_steps_per_second": 0.019,
      "step": 35
    },
    {
      "epoch": 1.8089887640449438,
      "grad_norm": 0.0,
      "learning_rate": 4.090909090909091e-07,
      "loss": 0.0,
      "step": 40
    },
    {
      "epoch": 1.8089887640449438,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 256.0,
      "eval_kl": 3.102847508021763e-05,
      "eval_loss": 1.2555974535644054e-06,
      "eval_num_tokens": 309066.0,
      "eval_reward": 0.0,
      "eval_reward_std": 0.0,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.0,
      "eval_runtime": 214.4846,
      "eval_samples_per_second": 0.065,
      "eval_steps_per_second": 0.019,
      "step": 40
    },
    {
      "epoch": 2.044943820224719,
      "grad_norm": 8.674070954839408e-07,
      "learning_rate": 3.333333333333333e-07,
      "loss": 0.0,
      "step": 45
    },
    {
      "epoch": 2.044943820224719,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 255.75,
      "eval_kl": 2.7051993778773715e-05,
      "eval_loss": 0.0015409131301566958,
      "eval_num_tokens": 347338.0,
      "eval_reward": 0.017857142857142856,
      "eval_reward_std": 0.03571428571428571,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.017857142857142856,
      "eval_runtime": 214.4059,
      "eval_samples_per_second": 0.065,
      "eval_steps_per_second": 0.019,
      "step": 45
    },
    {
      "epoch": 2.2696629213483144,
      "grad_norm": 3.324863371290121e-07,
      "learning_rate": 2.5757575757575754e-07,
      "loss": 0.0,
      "step": 50
    },
    {
      "epoch": 2.2696629213483144,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 255.91071428571428,
      "eval_kl": 2.6034457342965262e-05,
      "eval_loss": 0.0005461012478917837,
      "eval_num_tokens": 385150.0,
      "eval_reward": 0.017857142857142856,
      "eval_reward_std": 0.03571428571428571,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.017857142857142856,
      "eval_runtime": 214.4834,
      "eval_samples_per_second": 0.065,
      "eval_steps_per_second": 0.019,
      "step": 50
    },
    {
      "epoch": 2.49438202247191,
      "grad_norm": 5.330386443347379e-07,
      "learning_rate": 1.818181818181818e-07,
      "loss": 0.0005,
      "step": 55
    },
    {
      "epoch": 2.49438202247191,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 256.0,
      "eval_kl": 3.3276421683175226e-05,
      "eval_loss": 1.2978928225493291e-06,
      "eval_num_tokens": 422711.0,
      "eval_reward": 0.0,
      "eval_reward_std": 0.0,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.0,
      "eval_runtime": 214.3979,
      "eval_samples_per_second": 0.065,
      "eval_steps_per_second": 0.019,
      "step": 55
    },
    {
      "epoch": 2.7191011235955056,
      "grad_norm": 0.0,
      "learning_rate": 1.0606060606060605e-07,
      "loss": 0.0016,
      "step": 60
    },
    {
      "epoch": 2.7191011235955056,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 256.0,
      "eval_kl": 2.9053006853376117e-05,
      "eval_loss": 1.1453090564828017e-06,
      "eval_num_tokens": 461581.0,
      "eval_reward": 0.0,
      "eval_reward_std": 0.0,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_rewards/returns_int_reward_func": 0.0,
      "eval_runtime": 214.3855,
      "eval_samples_per_second": 0.065,
      "eval_steps_per_second": 0.019,
      "step": 60
    },
    {
      "epoch": 2.943820224719101,
      "grad_norm": 0.0,
      "learning_rate": 3.0303030303030305e-08,
      "loss": 0.0014,
      "step": 65
    },
    {
      "epoch": 2.943820224719101,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 255.67857142857142,
      "eval_kl": 3.472396305629185e-05,
      "eval_loss": 0.001989251933991909,
      "eval_num_tokens": 501354.0,
      "eval_reward": 0.03571428571428571,
      "eval_reward_std": 0.07142857142857142,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.017857142857142856,
      "eval_rewards/returns_int_reward_func": 0.017857142857142856,
      "eval_runtime": 214.4751,
      "eval_samples_per_second": 0.065,
      "eval_steps_per_second": 0.019,
      "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
}