{
  "best_global_step": 75,
  "best_metric": -0.0067981877364218235,
  "best_model_checkpoint": "outputs/microsoft/Phi-3.5-mini-instruct/countdown/checkpoint-75",
  "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": 1.5520595297857653e-06,
      "learning_rate": 9.466666666666666e-07,
      "loss": 0.0,
      "step": 5
    },
    {
      "epoch": 0.2,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 250.575,
      "eval_kl": 3.934502601623535e-05,
      "eval_loss": 0.015041607432067394,
      "eval_num_tokens": 31688.0,
      "eval_reward": 0.05,
      "eval_reward_std": 0.1,
      "eval_rewards/equation_reward_func": 0.05,
      "eval_rewards/format_reward_func": 0.0,
      "eval_runtime": 283.7484,
      "eval_samples_per_second": 0.07,
      "eval_steps_per_second": 0.018,
      "step": 5
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.023073066025972366,
      "learning_rate": 8.799999999999999e-07,
      "loss": 0.0033,
      "step": 10
    },
    {
      "epoch": 0.4,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 252.7875,
      "eval_kl": 3.967881202697754e-05,
      "eval_loss": -0.00017719810421112925,
      "eval_num_tokens": 62987.0,
      "eval_reward": 0.0375,
      "eval_reward_std": 0.075,
      "eval_rewards/equation_reward_func": 0.0375,
      "eval_rewards/format_reward_func": 0.0,
      "eval_runtime": 282.9691,
      "eval_samples_per_second": 0.071,
      "eval_steps_per_second": 0.018,
      "step": 10
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7865870631794678e-06,
      "learning_rate": 8.133333333333333e-07,
      "loss": 0.0,
      "step": 15
    },
    {
      "epoch": 0.6,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 248.875,
      "eval_kl": 4.363656044006348e-05,
      "eval_loss": 1.7436344705856754e-06,
      "eval_num_tokens": 94591.0,
      "eval_reward": 0.025,
      "eval_reward_std": 0.05,
      "eval_rewards/equation_reward_func": 0.025,
      "eval_rewards/format_reward_func": 0.0,
      "eval_runtime": 282.3086,
      "eval_samples_per_second": 0.071,
      "eval_steps_per_second": 0.018,
      "step": 15
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.0,
      "learning_rate": 7.466666666666667e-07,
      "loss": 0.0138,
      "step": 20
    },
    {
      "epoch": 0.8,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 253.1875,
      "eval_kl": 3.9270520210266115e-05,
      "eval_loss": -0.0007528069545514882,
      "eval_num_tokens": 125922.0,
      "eval_reward": 0.0125,
      "eval_reward_std": 0.025,
      "eval_rewards/equation_reward_func": 0.0125,
      "eval_rewards/format_reward_func": 0.0,
      "eval_runtime": 282.3999,
      "eval_samples_per_second": 0.071,
      "eval_steps_per_second": 0.018,
      "step": 20
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.02649177424609661,
      "learning_rate": 6.800000000000001e-07,
      "loss": 0.0059,
      "step": 25
    },
    {
      "epoch": 1.0,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 252.9875,
      "eval_kl": 3.8030743598937986e-05,
      "eval_loss": 0.006908929906785488,
      "eval_num_tokens": 157014.0,
      "eval_reward": 0.075,
      "eval_reward_std": 0.12886751294136048,
      "eval_rewards/equation_reward_func": 0.075,
      "eval_rewards/format_reward_func": 0.0,
      "eval_runtime": 282.4923,
      "eval_samples_per_second": 0.071,
      "eval_steps_per_second": 0.018,
      "step": 25
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.017524780705571175,
      "learning_rate": 6.133333333333332e-07,
      "loss": 0.0098,
      "step": 30
    },
    {
      "epoch": 1.2,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 251.6625,
      "eval_kl": 3.592073917388916e-05,
      "eval_loss": 0.0063271126709878445,
      "eval_num_tokens": 188533.0,
      "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": 283.0566,
      "eval_samples_per_second": 0.071,
      "eval_steps_per_second": 0.018,
      "step": 30
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.993278601730708e-06,
      "learning_rate": 5.466666666666666e-07,
      "loss": 0.0024,
      "step": 35
    },
    {
      "epoch": 1.4,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 253.1,
      "eval_kl": 4.151761531829834e-05,
      "eval_loss": 0.009441868402063847,
      "eval_num_tokens": 219791.0,
      "eval_reward": 0.05,
      "eval_reward_std": 0.07886751294136048,
      "eval_rewards/equation_reward_func": 0.0375,
      "eval_rewards/format_reward_func": 0.0125,
      "eval_runtime": 282.8829,
      "eval_samples_per_second": 0.071,
      "eval_steps_per_second": 0.018,
      "step": 35
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.774197120445024e-07,
      "learning_rate": 4.8e-07,
      "loss": 0.0106,
      "step": 40
    },
    {
      "epoch": 1.6,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 250.6375,
      "eval_kl": 4.119873046875e-05,
      "eval_loss": 1.6474120911880163e-06,
      "eval_num_tokens": 251168.0,
      "eval_reward": 0.0,
      "eval_reward_std": 0.0,
      "eval_rewards/equation_reward_func": 0.0,
      "eval_rewards/format_reward_func": 0.0,
      "eval_runtime": 282.7407,
      "eval_samples_per_second": 0.071,
      "eval_steps_per_second": 0.018,
      "step": 40
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.944530473949271e-07,
      "learning_rate": 4.1333333333333333e-07,
      "loss": 0.0019,
      "step": 45
    },
    {
      "epoch": 1.8,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 250.8,
      "eval_kl": 3.807544708251953e-05,
      "eval_loss": 0.019552746787667274,
      "eval_num_tokens": 282579.0,
      "eval_reward": 0.05,
      "eval_reward_std": 0.07886751294136048,
      "eval_rewards/equation_reward_func": 0.0375,
      "eval_rewards/format_reward_func": 0.0125,
      "eval_runtime": 282.3595,
      "eval_samples_per_second": 0.071,
      "eval_steps_per_second": 0.018,
      "step": 45
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.0208375733345747,
      "learning_rate": 3.4666666666666665e-07,
      "loss": 0.0081,
      "step": 50
    },
    {
      "epoch": 2.0,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 254.1,
      "eval_kl": 3.8304924964904784e-05,
      "eval_loss": 0.004257645923644304,
      "eval_num_tokens": 313952.0,
      "eval_reward": 0.0125,
      "eval_reward_std": 0.025,
      "eval_rewards/equation_reward_func": 0.0125,
      "eval_rewards/format_reward_func": 0.0,
      "eval_runtime": 282.4466,
      "eval_samples_per_second": 0.071,
      "eval_steps_per_second": 0.018,
      "step": 50
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.027176212519407272,
      "learning_rate": 2.8e-07,
      "loss": 0.0077,
      "step": 55
    },
    {
      "epoch": 2.2,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 251.9625,
      "eval_kl": 4.739761352539063e-05,
      "eval_loss": 0.00685427850112319,
      "eval_num_tokens": 345452.0,
      "eval_reward": 0.025,
      "eval_reward_std": 0.05,
      "eval_rewards/equation_reward_func": 0.0125,
      "eval_rewards/format_reward_func": 0.0125,
      "eval_runtime": 283.0534,
      "eval_samples_per_second": 0.071,
      "eval_steps_per_second": 0.018,
      "step": 55
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.0,
      "learning_rate": 2.1333333333333334e-07,
      "loss": 0.0039,
      "step": 60
    },
    {
      "epoch": 2.4,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 251.025,
      "eval_kl": 3.7276744842529296e-05,
      "eval_loss": 0.0007022842764854431,
      "eval_num_tokens": 376777.0,
      "eval_reward": 0.0125,
      "eval_reward_std": 0.025,
      "eval_rewards/equation_reward_func": 0.0125,
      "eval_rewards/format_reward_func": 0.0,
      "eval_runtime": 283.0247,
      "eval_samples_per_second": 0.071,
      "eval_steps_per_second": 0.018,
      "step": 60
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.01980462484061718,
      "learning_rate": 1.4666666666666666e-07,
      "loss": 0.016,
      "step": 65
    },
    {
      "epoch": 2.6,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 251.0625,
      "eval_kl": 4.115998744964599e-05,
      "eval_loss": 0.010943427681922913,
      "eval_num_tokens": 407811.0,
      "eval_reward": 0.0625,
      "eval_reward_std": 0.125,
      "eval_rewards/equation_reward_func": 0.05,
      "eval_rewards/format_reward_func": 0.0125,
      "eval_runtime": 283.1384,
      "eval_samples_per_second": 0.071,
      "eval_steps_per_second": 0.018,
      "step": 65
    },
    {
      "epoch": 2.8,
      "grad_norm": 8.191254892153665e-07,
      "learning_rate": 8e-08,
      "loss": 0.0001,
      "step": 70
    },
    {
      "epoch": 2.8,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 251.125,
      "eval_kl": 4.7346949577331544e-05,
      "eval_loss": 0.0024404774885624647,
      "eval_num_tokens": 439212.0,
      "eval_reward": 0.0375,
      "eval_reward_std": 0.075,
      "eval_rewards/equation_reward_func": 0.0375,
      "eval_rewards/format_reward_func": 0.0,
      "eval_runtime": 283.0414,
      "eval_samples_per_second": 0.071,
      "eval_steps_per_second": 0.018,
      "step": 70
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.02634088508784771,
      "learning_rate": 1.3333333333333334e-08,
      "loss": 0.0145,
      "step": 75
    },
    {
      "epoch": 3.0,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 252.075,
      "eval_kl": 4.2390823364257815e-05,
      "eval_loss": -0.0067981877364218235,
      "eval_num_tokens": 470623.0,
      "eval_reward": 0.0125,
      "eval_reward_std": 0.025,
      "eval_rewards/equation_reward_func": 0.0125,
      "eval_rewards/format_reward_func": 0.0,
      "eval_runtime": 283.5777,
      "eval_samples_per_second": 0.071,
      "eval_steps_per_second": 0.018,
      "step": 75
    }
  ],
  "logging_steps": 5,
  "max_steps": 75,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 5,
  "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
}