metadata
dataset_info:
features:
- name: images
list: image
- name: texts
list: string
- name: key
dtype: string
- name: loss
dtype: float32
- name: embedding
sequence: float32
splits:
- name: train
num_bytes: 5297605360
num_examples: 24000
download_size: 4731136341
dataset_size: 5297605360