---
dataset_info:
  features:
  - name: row_id
    dtype: int64
  - name: sample_idx
    dtype: int64
  - name: problem
    dtype: string
  - name: auto_zeroed
    dtype: bool
  - name: score
    dtype: int64
  - name: max_score
    dtype: int64
  - name: error_step
    dtype: string
  - name: error_category
    dtype: string
  - name: correction
    dtype: string
  - name: raw_judge_output
    dtype: string
  - name: judge_model
    dtype: string
  - name: sampling_params
    struct:
    - name: n
      dtype: int64
    - name: temperature
      dtype: float64
    - name: top_p
      dtype: float64
    - name: max_tokens
      dtype: int64
    - name: seed
      dtype: int64
  splits:
  - name: train
    num_bytes: 11452458
    num_examples: 512
  download_size: 5054424
  dataset_size: 11452458
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---