AI People Wiki

Builders & open source

Guilherme Penedo

Machine Learning Research Engineer, Hugging Face

Guilherme Penedo is a Machine Learning Research Engineer at Hugging Face, where Penedo works on datasets for training large language models. Penedo is first author of the FineWeb datasets paper (2024), the FineWeb2 paper (2025) and the earlier RefinedWeb paper for the Falcon models (2023), and has since released datasets including FineWiki and FineTranslations.12345

At a glance

Current role
Machine Learning Research Engineer
Sector
Company
Work listed
3
Sources cited
6
Profile reviewed

Datasets

Penedo was first author of the RefinedWeb paper (2023), which accompanied the Falcon language models, and of the FineWeb paper (2024), describing the FineWeb web-text dataset released by Hugging Face. The FineWeb2 paper (2025) adapts the same processing approach to many languages.534

Penedo's site lists later releases including FineWiki (October 2025), an updated extraction of Wikipedia in more than 300 languages, and FineTranslations (January 2026), a multilingual to English parallel dataset built from FineWeb2.1

Other work

Penedo is a co-author of the SmolLM2 paper (2025) and the first-listed author of the competitive-programming ("Olympic coding") datasets and models released through Hugging Face's Open-R1 project.61

Work and contributions

  • FineWeb Open data

    First author of "The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scale" (2024), the paper describing the FineWeb dataset.31

  • FineWeb2 Open data

    First author of the FineWeb2 paper (2025), which extends the FineWeb pipeline to many languages and was presented as a spotlight at COLM 2025.41

  • RefinedWeb Open data

    First author of "The RefinedWeb Dataset for Falcon LLM" (2023), on building a pretraining dataset from web data alone.51

Timeline

  1. Posted the RefinedWeb dataset paper for Falcon LLM as first author.5
  2. Posted the FineWeb datasets paper as first author.3
  3. Co-authored the SmolLM2 paper.6
  4. Posted the FineWeb2 paper as first author.4
  5. Released FineWiki, an updated extraction of Wikipedia in more than 300 languages.1
  6. Announced FineTranslations, a multilingual to English dataset built from FineWeb2.1

Where they work

Suggest a correction or removal

Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.

References

6 sources · profile reviewed Oct 22, 2026 · data updated Oct 22, 2026

  1. Guilherme Penedo — homepageGuilherme Penedo · Official page · accessed Oct 22, 2026
  2. guipenedo (Guilherme Penedo) on GitHubGitHub · Official page · accessed Oct 22, 2026
  3. The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scale (arXiv 2406.17557)arXiv · Paper · published Jun 25, 2024 · accessed Oct 22, 2026
  4. FineWeb2: One Pipeline to Scale Them All (arXiv 2506.20920)arXiv · Paper · published Jun 26, 2025 · accessed Oct 22, 2026
  5. The RefinedWeb Dataset for Falcon LLM (arXiv 2306.01116)arXiv · Paper · published Jun 1, 2023 · accessed Oct 22, 2026
  6. SmolLM2: When Smol Goes Big — Data-Centric Training of a Small Language Model (arXiv 2502.02737)arXiv · Paper · published Feb 4, 2025 · accessed Oct 22, 2026