metadata
dataset_info:
features:
- name: scenes
sequence:
sequence: float32
- name: language_annotation
dtype: string
splits:
- name: train
num_bytes: 41820172461
num_examples: 7966221
download_size: 4545543118
dataset_size: 41820172461
configs:
- config_name: default
data_files:
- split: train
path: data/train-*