metadata
dataset_info:
features:
- name: answers
struct:
- name: answer_start
sequence: 'null'
- name: text
sequence: string
- name: inputs
dtype: string
- name: targets
dtype: string
- name: true_doc
dtype: string
splits:
- name: train_qa
num_bytes: 980198
num_examples: 6000
- name: train_ic_qa
num_bytes: 4823367
num_examples: 6000
- name: train_recite_qa
num_bytes: 7696718
num_examples: 6000
- name: eval_qa
num_bytes: 1058934
num_examples: 6489
- name: eval_ic_qa
num_bytes: 5212318
num_examples: 6489
- name: eval_recite_qa
num_bytes: 8311634
num_examples: 6489
- name: all_docs
num_bytes: 7497763
num_examples: 10925
- name: all_docs_eval
num_bytes: 13103091
num_examples: 10925
- name: train
num_bytes: 7696718
num_examples: 6000
- name: validation
num_bytes: 8311634
num_examples: 6489
download_size: 40278892
dataset_size: 64692375
configs:
- config_name: default
data_files:
- split: train_qa
path: data/train_qa-*
- split: train_ic_qa
path: data/train_ic_qa-*
- split: train_recite_qa
path: data/train_recite_qa-*
- split: eval_qa
path: data/eval_qa-*
- split: eval_ic_qa
path: data/eval_ic_qa-*
- split: eval_recite_qa
path: data/eval_recite_qa-*
- split: all_docs
path: data/all_docs-*
- split: all_docs_eval
path: data/all_docs_eval-*
- split: train
path: data/train-*
- split: validation
path: data/validation-*