metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
splits:
- name: train
num_bytes: 73233441
num_examples: 2637
download_size: 13593265
dataset_size: 73233441
configs:
- config_name: default
data_files:
- split: train
path: data/arc/train-*