ModelFineWeb-2
FineWeb2
A large-scale multilingual pretraining corpus released by Hugging Face, covering hundreds of languages and intended for research on training data for large models.
Timeline (last 90 days)
Oct 23
The study used an LLM agent as a corpus debugger to audit the Slovak portion of FineWeb2 (26.5 million documents).
Unverified50%
Oct 23
The largest issue the agent found was machine-generated content farms, accounting for 5.32% of the Slovak corpus (1,408,437 documents).
Unverified50%