metadata
dataset_info:
features:
- name: index
dtype: int64
- name: occupation
dtype: string
- name: other_person
dtype: string
- name: pronoun_options
sequence: string
- name: sentence_with_blank
dtype: string
- name: BLS_percent_women_2019
dtype: float64
- name: BLS_original_occupation
dtype: string
splits:
- name: train
num_bytes: 645622
num_examples: 2990
download_size: 237192
dataset_size: 645622
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
https://github.com/anthropics/evals/blob/main/winogenerated/winogenerated_examples.jsonl