metadata
dataset_info:
features:
- name: image
dtype: image
- name: instruction
dtype: string
- name: bbox
sequence: float64
- name: resolution
sequence: int64
- name: source
dtype: string
- name: platform
dtype: string
- name: name
dtype: string
- name: description
dtype: string
- name: type
dtype: string
- name: OCR
dtype: string
- name: language
dtype: string
- name: purpose
dtype: string
- name: expectation
dtype: string
splits:
- name: train
num_bytes: 25626533286.514072
num_examples: 58651
download_size: 2124520843
dataset_size: 25626533286.514072
configs:
- config_name: default
data_files:
- split: train
path: data/train-*