metadata
dataset_info:
features:
- name: instruction
dtype: string
- name: answer
dtype: string
- name: short_answer
dtype: string
- name: long_answer
dtype: string
- name: text
dtype: string
- name: image
dtype: image
- name: quality
dtype: string
- name: source_dataset
dtype: string
- name: original_index
dtype: int64
- name: metadata
struct:
- name: source
dtype: string
- name: config
dtype: string
- name: original_index
dtype: int64
- name: original_split
dtype: string
- name: has_short
dtype: bool
- name: has_long
dtype: bool
- name: answer_type
dtype: string
- name: answer_start_token
dtype: 'null'
- name: answer_end_token
dtype: 'null'
- name: extraction_method
dtype: string
- name: generation_resolution
dtype: string
- name: quality_description
dtype: string
- name: estimated_height
dtype: int64
- name: token_count
dtype: int64
- name: extracted_token_count
dtype: int64
splits:
- name: train
num_bytes: 5060502164
num_examples: 502937
download_size: 4411648649
dataset_size: 5060502164
configs:
- config_name: default
data_files:
- split: train
path: data/train-*