metadata
dataset_info:
features:
- name: image
dtype: image
- name: mask
dtype: image
- name: masked_image
dtype: image
- name: mae_embedding
dtype:
array3_d:
shape:
- 1
- 196
- 768
dtype: float32
splits:
- name: train
num_bytes: 89299072580.55
num_examples: 8255
download_size: 89360969643
dataset_size: 89299072580.55
configs:
- config_name: default
data_files:
- split: train
path: data/train-*