{
  "best_global_step": 55,
  "best_metric": 0.020872214809060097,
  "best_model_checkpoint": "outputs/microsoft/Phi-3.5-mini-instruct/countdown_mcl_1024/checkpoint-15",
  "epoch": 3.0,
  "eval_steps": 5,
  "global_step": 75,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.2,
      "grad_norm": 0.019189711660146713,
      "learning_rate": 9.466666666666666e-07,
      "loss": 0.0336,
      "step": 5
    },
    {
      "epoch": 0.2,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 802.8125,
      "eval_kl": 3.0040740966796875e-05,
      "eval_loss": 0.07697294652462006,
      "eval_num_tokens": 76355.0,
      "eval_reward": 0.0875,
      "eval_reward_std": 0.13273502588272096,
      "eval_rewards/equation_reward_func": 0.075,
      "eval_rewards/format_reward_func": 0.0125,
      "eval_runtime": 1338.4011,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 5
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.01623854786157608,
      "learning_rate": 8.799999999999999e-07,
      "loss": 0.0454,
      "step": 10
    },
    {
      "epoch": 0.4,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 855.5625,
      "eval_kl": 3.3321976661682126e-05,
      "eval_loss": 0.04702009633183479,
      "eval_num_tokens": 152886.0,
      "eval_reward": 0.0625,
      "eval_reward_std": 0.125,
      "eval_rewards/equation_reward_func": 0.0375,
      "eval_rewards/format_reward_func": 0.025,
      "eval_runtime": 1294.6762,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 10
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.0,
      "learning_rate": 8.133333333333333e-07,
      "loss": 0.0494,
      "step": 15
    },
    {
      "epoch": 0.6,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 796.0875,
      "eval_kl": 3.191232681274414e-05,
      "eval_loss": 0.03644490987062454,
      "eval_num_tokens": 230732.0,
      "eval_reward": 0.1125,
      "eval_reward_std": 0.18273502588272095,
      "eval_rewards/equation_reward_func": 0.0875,
      "eval_rewards/format_reward_func": 0.025,
      "eval_runtime": 1335.5012,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 15
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.009749338962137699,
      "learning_rate": 7.466666666666667e-07,
      "loss": 0.0723,
      "step": 20
    },
    {
      "epoch": 0.8,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 810.8,
      "eval_kl": 2.9605627059936523e-05,
      "eval_loss": 0.04239075630903244,
      "eval_num_tokens": 308908.0,
      "eval_reward": 0.0875,
      "eval_reward_std": 0.15386751294136047,
      "eval_rewards/equation_reward_func": 0.05,
      "eval_rewards/format_reward_func": 0.0375,
      "eval_runtime": 1337.0845,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 20
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.0265500470995903,
      "learning_rate": 6.800000000000001e-07,
      "loss": 0.0296,
      "step": 25
    },
    {
      "epoch": 1.0,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 769.025,
      "eval_kl": 2.89231538772583e-05,
      "eval_loss": 0.07508356869220734,
      "eval_num_tokens": 385628.0,
      "eval_reward": 0.1125,
      "eval_reward_std": 0.18273502588272095,
      "eval_rewards/equation_reward_func": 0.1,
      "eval_rewards/format_reward_func": 0.0125,
      "eval_runtime": 1334.3196,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 25
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.0,
      "learning_rate": 6.133333333333332e-07,
      "loss": 0.0476,
      "step": 30
    },
    {
      "epoch": 1.2,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 743.5375,
      "eval_kl": 3.173947334289551e-05,
      "eval_loss": 0.09708776324987411,
      "eval_num_tokens": 463386.0,
      "eval_reward": 0.15,
      "eval_reward_std": 0.25773502588272096,
      "eval_rewards/equation_reward_func": 0.1375,
      "eval_rewards/format_reward_func": 0.0125,
      "eval_runtime": 1333.5969,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 30
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.0,
      "learning_rate": 5.466666666666666e-07,
      "loss": 0.0398,
      "step": 35
    },
    {
      "epoch": 1.4,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 791.5125,
      "eval_kl": 3.103315830230713e-05,
      "eval_loss": 0.04481809586286545,
      "eval_num_tokens": 536555.0,
      "eval_reward": 0.1125,
      "eval_reward_std": 0.20386751294136046,
      "eval_rewards/equation_reward_func": 0.0875,
      "eval_rewards/format_reward_func": 0.025,
      "eval_runtime": 1304.5413,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 35
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.01814551278948784,
      "learning_rate": 4.8e-07,
      "loss": 0.0884,
      "step": 40
    },
    {
      "epoch": 1.6,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 798.8,
      "eval_kl": 3.2714009284973146e-05,
      "eval_loss": 0.06108971685171127,
      "eval_num_tokens": 615157.0,
      "eval_reward": 0.125,
      "eval_reward_std": 0.17886751294136047,
      "eval_rewards/equation_reward_func": 0.1125,
      "eval_rewards/format_reward_func": 0.0125,
      "eval_runtime": 1301.9697,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 40
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.0,
      "learning_rate": 4.1333333333333333e-07,
      "loss": 0.1015,
      "step": 45
    },
    {
      "epoch": 1.8,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 792.85,
      "eval_kl": 2.993345260620117e-05,
      "eval_loss": 0.06866836547851562,
      "eval_num_tokens": 687524.0,
      "eval_reward": 0.1375,
      "eval_reward_std": 0.22673887014389038,
      "eval_rewards/equation_reward_func": 0.075,
      "eval_rewards/format_reward_func": 0.0625,
      "eval_runtime": 1314.7184,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 45
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.01888681948184967,
      "learning_rate": 3.4666666666666665e-07,
      "loss": 0.0441,
      "step": 50
    },
    {
      "epoch": 2.0,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 828.3625,
      "eval_kl": 3.242790699005127e-05,
      "eval_loss": 0.03552088886499405,
      "eval_num_tokens": 764139.0,
      "eval_reward": 0.0375,
      "eval_reward_std": 0.075,
      "eval_rewards/equation_reward_func": 0.0125,
      "eval_rewards/format_reward_func": 0.025,
      "eval_runtime": 1335.2965,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 50
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.012753931805491447,
      "learning_rate": 2.8e-07,
      "loss": 0.0577,
      "step": 55
    },
    {
      "epoch": 2.2,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 838.8,
      "eval_kl": 3.0362606048583983e-05,
      "eval_loss": 0.020872214809060097,
      "eval_num_tokens": 845274.0,
      "eval_reward": 0.05,
      "eval_reward_std": 0.1,
      "eval_rewards/equation_reward_func": 0.0375,
      "eval_rewards/format_reward_func": 0.0125,
      "eval_runtime": 1294.8372,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 55
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.0,
      "learning_rate": 2.1333333333333334e-07,
      "loss": 0.1044,
      "step": 60
    },
    {
      "epoch": 2.4,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 786.7,
      "eval_kl": 3.013014793395996e-05,
      "eval_loss": 0.07739870995283127,
      "eval_num_tokens": 919161.0,
      "eval_reward": 0.1,
      "eval_reward_std": 0.17886751294136047,
      "eval_rewards/equation_reward_func": 0.0875,
      "eval_rewards/format_reward_func": 0.0125,
      "eval_runtime": 1281.0863,
      "eval_samples_per_second": 0.016,
      "eval_steps_per_second": 0.004,
      "step": 60
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.0,
      "learning_rate": 1.4666666666666666e-07,
      "loss": 0.0556,
      "step": 65
    },
    {
      "epoch": 2.6,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 787.15,
      "eval_kl": 3.191828727722168e-05,
      "eval_loss": 0.12674936652183533,
      "eval_num_tokens": 991008.0,
      "eval_reward": 0.125,
      "eval_reward_std": 0.22886751294136048,
      "eval_rewards/equation_reward_func": 0.1,
      "eval_rewards/format_reward_func": 0.025,
      "eval_runtime": 1337.2162,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 65
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.011743688955903053,
      "learning_rate": 8e-08,
      "loss": 0.0575,
      "step": 70
    },
    {
      "epoch": 2.8,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 865.05,
      "eval_kl": 3.235936164855957e-05,
      "eval_loss": 0.022181231528520584,
      "eval_num_tokens": 1072417.0,
      "eval_reward": 0.0375,
      "eval_reward_std": 0.075,
      "eval_rewards/equation_reward_func": 0.025,
      "eval_rewards/format_reward_func": 0.0125,
      "eval_runtime": 1336.7662,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 70
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.012797041796147823,
      "learning_rate": 1.3333333333333334e-08,
      "loss": 0.0873,
      "step": 75
    },
    {
      "epoch": 3.0,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 838.275,
      "eval_kl": 3.303289413452149e-05,
      "eval_loss": 0.026845881715416908,
      "eval_num_tokens": 1151449.0,
      "eval_reward": 0.1125,
      "eval_reward_std": 0.15560638308525085,
      "eval_rewards/equation_reward_func": 0.1,
      "eval_rewards/format_reward_func": 0.0125,
      "eval_runtime": 1328.2121,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 75
    }
  ],
  "logging_steps": 5,
  "max_steps": 75,
  "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
}