metadata
dataset_info:
features:
- name: id
dtype: string
- name: prompt
dtype: string
- name: latents
dtype:
array4_d:
shape:
- 2
- 4
- 96
- 96
dtype: float16
- name: prompt_embed
dtype:
array3_d:
shape:
- 1
- 77
- 1024
dtype: float16
splits:
- name: train
num_bytes: 24758291628
num_examples: 80215
download_size: 23968503840
dataset_size: 24758291628
configs:
- config_name: default
data_files:
- split: train
path: data/train-*