metadata
dataset_info:
features:
- name: image_url
dtype: string
- name: title
dtype: string
- name: year
dtype: int64
- name: caption
dtype: string
- name: t5_prompt_embeds
sequence:
sequence:
sequence: float16
- name: vae_512px
sequence:
sequence:
sequence:
sequence: float16
- name: ratio
dtype: float64
splits:
- name: train
num_bytes: 82813738885
num_examples: 94960
download_size: 81989897794
dataset_size: 82813738885
configs:
- config_name: default
data_files:
- split: train
path: data/train-*