metadata
dataset_info:
features:
- name: snippet
dtype: string
- name: scenario
dtype: string
- name: question
dtype: string
- name: history
dtype: string
- name: label
dtype:
class_label:
names:
'0': 'Yes'
'1': 'No'
'2': Irrelevant
'3': Clarification needed
- name: follow_up_question
dtype: string
- name: tree_id
dtype: string
- name: utterance_id
dtype: string
- name: source_url
dtype: string
splits:
- name: train
num_bytes: 13275703
num_examples: 19788
- name: validation
num_bytes: 1405201
num_examples: 2102
- name: test
num_bytes: 1418434
num_examples: 2270
download_size: 1981472
dataset_size: 16099338
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*
source_datasets:
- nikhilweee/sharc_modified
sharc
ShARC (modified by Verma et al.): answer a rule question given a user scenario and follow-up Q&A.
Original data: nikhilweee/sharc_modified. Repackaged as parquet for tasksource by scripts/upload_repackaged.py.