metadata
dataset_info:
features:
- name: orig_answer
dtype: string
- name: response
dtype: string
- name: response_tokens
sequence: int64
- name: input_tokens
dtype: string
- name: cot_start_position
dtype: int64
- name: original_zero_shot_cot_answer
dtype: string
- name: output_relevances
sequence:
sequence: float64
splits:
- name: train
num_bytes: 139618911
num_examples: 4434
download_size: 104760282
dataset_size: 139618911
configs:
- config_name: default
data_files:
- split: train
path: data/train-*