metadata
dataset_info:
features:
- name: scenes
sequence:
sequence: float32
- name: language_annotation
dtype: string
splits:
- name: train
num_bytes: 115343530587
num_examples: 39126213
download_size: 17988933809
dataset_size: 115343530587
configs:
- config_name: default
data_files:
- split: train
path: data/train-*