README.md

---
dataset_info:
  features:
  - name: row_id
    dtype: int64
  - name: input
    dtype: string
  - name: completion
    dtype: string
  - name: data_type
    dtype: string
  - name: label
    dtype: int64
  - name: model_response
    dtype: string
  - name: passed
    dtype: bool
  - name: model_response_clean
    dtype: string
  - name: judge_label
    dtype: string
  - name: judge_attack_succeeded
    dtype: bool
  - name: judge_reason
    dtype: string
  - name: judge_error
    dtype: 'null'
  - name: judge_raw_text
    dtype: string
  - name: input_no_wm
    dtype: string
  - name: model_output
    dtype: string
  - name: llm_judge_w_op
    dtype: bool
  - name: watermark_judge
    dtype: bool
  - name: asr_and
    dtype: bool
  - name: asr_or
    dtype: bool
  splits:
  - name: train
    num_bytes: 15408515
    num_examples: 1449
  download_size: 8424871
  dataset_size: 15408515
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---