---
id: "huggingface-fineweb"
title: "The FineWeb Datasets"
creator: "Guilherme Penedo and coauthors, Hugging Face"
creator_url: "https://arxiv.org/abs/2406.17557"
url: "https://arxiv.org/abs/2406.17557"
level: "expert"
level_max: null
category: "model-training"
subcategory: "training-data"
region: "other"
license: "open"
license_scope: "linked-resource"
models: ["any"]
source_date: null
date_added: "2026-10-18"
---

[Model Training](/model-training/index.md)

# The FineWeb Datasets

by [Guilherme Penedo and coauthors, Hugging Face](https://arxiv.org/abs/2406.17557)

[Read the original →](https://arxiv.org/abs/2406.17557)

Study how filtering and deduplication affect pretraining data.

The FineWeb paper reports controlled comparisons of dataset-construction choices and an educational-text subset. It connects data curation to downstream measurements, with released artifacts for readers who want to investigate the method.

Level: [Expert](/levels/index.md#expert)
Models: any
Category: [Model Training](/model-training/index.md)
Tags: 
Source date: date not stated
Added: 2026-10-18
Last verified: 2026-10-18
Status: ok
Prerequisites: Acts on model-training, infrastructure or research tooling; requires training-stack expertise.
Subcategory: [training data](/model-training/training-data/index.md)
Region: other
Original language: en
Access: free
Resource reuse: open: Source states by/4.0. Applies to the linked resource, not every referenced model or dataset. [Facet definitions](/about/index.md#facets)

Summary in our own words. Rights remain with the original creator. [Credit and correction policy](/about/index.md#credit-policy).

[Plain-text Markdown copy](/model-training/training-data/huggingface-fineweb/index.md)