sycophancy_dataset / README.md
darthPanda's picture
Add dataset from JSONL
b545c50 verified
metadata
dataset_info:
  features:
    - name: seed_id
      dtype: string
    - name: question
      dtype: string
    - name: expected_sycophantic_response
      dtype: string
    - name: reason_response_sycophantic
      dtype: string
  splits:
    - name: train
      num_bytes: 153168
      num_examples: 199
  download_size: 81679
  dataset_size: 153168
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*