metadata
dataset_info:
features:
- name: context
dtype: string
- name: json_schema
dtype: string
- name: ground_truth
dtype: string
- name: candidate_response
dtype: 'null'
- name: validated_output
dtype: 'null'
- name: source_dataset
dtype: string
- name: source_split
dtype: string
splits:
- name: clean
num_bytes: 243168442
num_examples: 19050
- name: abstention_candidate
num_bytes: 42770539
num_examples: 5006
- name: scrape_failure
num_bytes: 40517131
num_examples: 3996
- name: human_eval
num_bytes: 1099652
num_examples: 100
download_size: 134663444
dataset_size: 327555764
configs:
- config_name: default
data_files:
- split: clean
path: data/clean-*
- split: abstention_candidate
path: data/abstention_candidate-*
- split: scrape_failure
path: data/scrape_failure-*
- split: human_eval
path: data/human_eval-*