metadata
dataset_info:
features:
- name: doc_id
dtype: int64
- name: doc
dtype: string
- name: target
dtype: string
- name: arguments
dtype: string
- name: resps
dtype: string
- name: filtered_resps
dtype: string
- name: doc_hash
dtype: string
- name: prompt_hash
dtype: string
- name: target_hash
dtype: string
- name: exact_match
dtype: int64
- name: extracted_answers
dtype: string
- name: source_file
dtype: string
- name: generation
dtype: string
- name: info
dtype: string
- name: evaluation_api_cost
dtype: string
splits:
- name: train
num_bytes: 407734853
num_examples: 3656
download_size: 37840641
dataset_size: 407734853
configs:
- config_name: default
data_files:
- split: train
path: data/train-*