Dataset · Hugging Face

epfml/FineWeb2-HQ

Name: FineWeb2-HQ
Creator: epfml
License: odc-by
Keywords: task_categories:text-generation, language:ru, language:zh, language:de, language:ja, language:es, language:fr, language:it, language:pt, language:pl, language:nl, language:id, language:tr, language:cs, language:vi

FineWeb2-HQ Dataset summary FineWeb2-HQ is a high-quality, model-filtered pretraining dataset derived as a subset of FineWeb2, spanning 20 languages. It enables around 6x faster pretraining compared to the base dataset. FineWeb2-HQ was created by selecting the top 10% quality documents of FineWeb2 in each language, based on scores assigned by a deep learning classifier trained to identify structured and knowledge-rich samples using XLM-RoBERTa embeddings. Validation… See the full description on the dataset page: https://huggingface.co/datasets/epfml/FineWeb2-HQ.

View on Hugging FaceReport an issue

Author / organization	epfml
License (as declared)	odc-by
Downloads	37K (as reported by Hugging Face)
Likes	85
Created	17 February 2025
Last updated at source	19 February 2025

Embed this entry

<iframe src="https://silytics.si/embed/item/f960588c-46c3-4fd3-b865-a92adef87c23" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Free, no tracking. More about feeds and embeds.

More from epfml

epfml/FineWeb-HQDataset

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 14 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.