---
dataset_info:
  features:
  - name: 'Unnamed: 0'
    dtype: int64
  - name: id
    dtype: string
  - name: adv_source_id
    dtype: string
  - name: source_id
    dtype: string
  - name: model
    dtype: string
  - name: decoding
    dtype: string
  - name: repetition_penalty
    dtype: string
  - name: attack
    dtype: string
  - name: domain
    dtype: string
  - name: title
    dtype: string
  - name: prompt
    dtype: string
  - name: text
    dtype: string
  - name: pred
    dtype: string
  splits:
  - name: train
    num_bytes: 354459207
    num_examples: 144000
  download_size: 154484669
  dataset_size: 354459207
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---