metadata
license: mit
dataset_info:
features:
- name: 1st Entity
dtype: string
- name: 2nd Entity
dtype: string
- name: Class
dtype: string
- name: Prompt
dtype: string
- name: Knowledge
dtype: string
- name: NC-S
dtype: bool
splits:
- name: train
num_bytes: 1245364331
num_examples: 8709810
download_size: 322125304
dataset_size: 1245364331
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
NeuroComparatives is a corpus of 8.71 million comparisons over 1.74 million pairs of entities. See our NAACL 2024 paper for additional details.