metadata
dataset_info:
features:
- name: image
dtype: image
- name: latent
dtype:
array3_d:
shape:
- 4
- 32
- 32
dtype: float16
splits:
- name: train
num_bytes: 4793291940
num_examples: 50000
download_size: 4762069423
dataset_size: 4793291940
configs:
- config_name: default
data_files:
- split: train
path: data/train-*