metadata
dataset_info:
features:
- name: video
dtype:
video:
decode: false
- name: model_output
dtype: string
- name: input_ids
sequence:
sequence: int32
- name: attention_mask
sequence:
sequence: int8
- name: pixel_values_videos
sequence:
sequence:
sequence:
sequence:
sequence: float32
splits:
- name: train
num_bytes: 2178412096.3333335
num_examples: 50
- name: test
num_bytes: 566446612.6666666
num_examples: 13
download_size: 411680026
dataset_size: 2744858709
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*