metadata
dataset_info:
features:
- name: type
dtype: string
- name: brief_caption
dtype: string
- name: raw_prompt
dtype: string
- name: sft_prompt
dtype: string
- name: detailed_caption
dtype: string
- name: pth
dtype: string
- name: id
dtype: string
- name: aesthetic
dtype: float64
- name: width
dtype: int64
- name: hash
dtype: string
- name: augmented_prompts
struct:
- name: object_prompts
sequence: string
- name: paraphrases
sequence: string
- name: short_caption
dtype: string
- name: tags
sequence: string
- name: varied_captions
sequence: string
- name: augmented_cots
struct:
- name: object_centric
sequence: string
- name: region_descriptions
sequence: string
- name: step_by_step
dtype: string
- name: tags
sequence: string
- name: image
dtype: string
- name: pil_image
dtype: image
- name: o3_caption
dtype: string
- name: short4o_caption
dtype: string
splits:
- name: train
num_bytes: 504914272
num_examples: 500
download_size: 503234044
dataset_size: 504914272
configs:
- config_name: default
data_files:
- split: train
path: data/train-*