metadata
dataset_info:
features:
- name: marker_page
dtype: string
- name: arxiv_id
dtype: string
- name: page_idx
dtype: int32
- name: table_bbox
sequence: float32
- name: gpt_4_table
dtype: string
- name: arxiv_table
dtype: string
- name: orig_marker_table
dtype: string
splits:
- name: train
num_bytes: 246262874
num_examples: 688
download_size: 44987904
dataset_size: 246262874
configs:
- config_name: default
data_files:
- split: train
path: data/train-*