FineWeb & FineWeb-Edu
Dataset
Web Corpus
Hugging Face's 15-trillion-token state of the art in open pre-training data.
You're looking at a dataset record: a training-data collection — what's in it and which models learned from it.
FineWeb (2024) processed 96 Common Crawl snapshots into 15T tokens with fully published filtering code; FineWeb-Edu used model-based educational scoring to distill the best 1.3T. Together they are the open world's answer to proprietary data pipelines.
How this connects to the rest of the stack
This record's links at a glance — click a group to jump to it.
Related
Sources
Record reviewed as of 2026-01-15 — see Methodology.