metadata
license: cc-by-4.0
tags:
- video
- urls
- laion
- commoncrawl
configs:
- config_name: default
default: true
data_files:
- split: train
path: default/*.parquet
size_categories:
- 1B<n<10B
LAION-BVD - 1.3B Video URLs
This repository contains 1.3 billion platform-specific video URLs collected from CommonCrawl. No video content is included, only URLs and associated crawl metadata. These URLs form the source corpus for LAION-BVD (LAION - Big Video Dataset).
Loading the data
import datasets
ds = datasets.load_dataset("LAION/BVD-1.3B-URLs", split="train")
Or with pandas:
import pandas as pd
df = pd.read_parquet("data/", engine="pyarrow")