{
  "best_global_step": 6,
  "best_metric": 0.012996690347790718,
  "best_model_checkpoint": "/mnt/nvme5n1p1/trained_grpo_distill_14b_rl_70_s3/v3-20250330-200345/checkpoint-6",
  "epoch": 1.8421052631578947,
  "eval_steps": 6,
  "global_step": 8,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": false,
  "log_history": [
    {
      "clip_ratio": 0.0,
      "completion_length": 10352.974609375,
      "epoch": 0.21052631578947367,
      "grad_norm": 0.13259537518024445,
      "kl": 0.0,
      "learning_rate": 1.6666666666666667e-05,
      "loss": -0.11016345024108887,
      "memory(GiB)": 186.69,
      "response_clip_ratio": 0.11328125,
      "reward": -0.002658387296833098,
      "reward_std": 0.06134121119976044,
      "rewards/CosineReward": -0.0026579967816360295,
      "rewards/RepetitionPenalty": -3.8975886695880035e-07,
      "step": 1,
      "train_speed(iter/s)": 0.000241
    },
    {
      "clip_ratio": 0.0,
      "epoch": 0.42105263157894735,
      "grad_norm": 0.1320001482963562,
      "kl": 0.0,
      "learning_rate": 3.3333333333333335e-05,
      "loss": -0.11016345024108887,
      "memory(GiB)": 186.69,
      "step": 2,
      "train_speed(iter/s)": 0.000466
    },
    {
      "clip_ratio": 1.3441811461234465e-05,
      "completion_length": 10439.369140625,
      "epoch": 0.631578947368421,
      "grad_norm": 0.08990391343832016,
      "kl": 9.50181856751442e-07,
      "learning_rate": 5e-05,
      "loss": -0.06604708731174469,
      "memory(GiB)": 186.69,
      "response_clip_ratio": 0.13671875,
      "reward": 0.0006296975770965219,
      "reward_std": 0.07172460854053497,
      "rewards/CosineReward": 0.0006298604130279273,
      "rewards/RepetitionPenalty": -1.6200439745261974e-07,
      "step": 3,
      "train_speed(iter/s)": 0.00035
    },
    {
      "clip_ratio": 1.70210253145342e-05,
      "epoch": 0.8421052631578947,
      "grad_norm": 0.0967094898223877,
      "kl": 1.1101365089416504e-05,
      "learning_rate": 6.666666666666667e-05,
      "loss": -0.06727766245603561,
      "memory(GiB)": 186.69,
      "step": 4,
      "train_speed(iter/s)": 0.000458
    },
    {
      "clip_ratio": 1.675608473306056e-05,
      "completion_length": 10092.408203125,
      "epoch": 1.2105263157894737,
      "grad_norm": 0.142837256193161,
      "kl": 0.00017762184143066406,
      "learning_rate": 8.333333333333334e-05,
      "loss": -0.09315311908721924,
      "memory(GiB)": 186.69,
      "response_clip_ratio": 0.119140625,
      "reward": -0.005135859013535082,
      "reward_std": 0.07994875870645046,
      "rewards/CosineReward": -0.005134060338605195,
      "rewards/RepetitionPenalty": -1.7973881654143042e-06,
      "step": 5,
      "train_speed(iter/s)": 0.000387
    },
    {
      "epoch": 1.4210526315789473,
      "grad_norm": 0.18263348937034607,
      "learning_rate": 0.0001,
      "loss": -0.1041698157787323,
      "memory(GiB)": 186.69,
      "step": 6,
      "train_speed(iter/s)": 0.000459
    },
    {
      "epoch": 1.4210526315789473,
      "eval_clip_ratio": 4.069424539920874e-05,
      "eval_completion_length": 12289.0,
      "eval_kl": 0.04833984375,
      "eval_loss": -0.5377416610717773,
      "eval_response_clip_ratio": 1.0,
      "eval_reward": 0.012996690347790718,
      "eval_reward_std": 0.08769983053207397,
      "eval_rewards/CosineReward": 0.012996694073081017,
      "eval_rewards/RepetitionPenalty": 0.0,
      "eval_runtime": 1030.1122,
      "eval_samples_per_second": 0.001,
      "eval_steps_per_second": 0.001,
      "step": 6
    },
    {
      "clip_ratio": 0.0005237623976199757,
      "completion_length": 10448.94921875,
      "epoch": 1.631578947368421,
      "grad_norm": 0.1291271299123764,
      "kl": 0.017406463623046875,
      "learning_rate": 9.991540791356342e-05,
      "loss": -0.051375165581703186,
      "memory(GiB)": 186.69,
      "response_clip_ratio": 0.1484375,
      "reward": 0.004909618757665157,
      "reward_std": 0.08167182095348835,
      "rewards/CosineReward": 0.004909833543933928,
      "rewards/RepetitionPenalty": -2.1478646772266075e-07,
      "step": 7,
      "train_speed(iter/s)": 0.000382
    },
    {
      "clip_ratio": 0.1706484742462635,
      "epoch": 1.8421052631578947,
      "grad_norm": 0.26641014218330383,
      "kl": 0.089599609375,
      "learning_rate": 9.966191788709716e-05,
      "loss": -0.05105742812156677,
      "memory(GiB)": 186.69,
      "step": 8,
      "train_speed(iter/s)": 0.000433
    }
  ],
  "logging_steps": 1,
  "max_steps": 60,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 15,
  "save_steps": 2,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}