Going over Fine Web with a Fine-Tooth Comb: Technical Report of Indexing Fine Web for Problematic Content Search and Retrieval

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Marinas, Inés Altemir, Kucherenko, Anastasiia, Kucharavy, Andrei
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915470466088960
author Marinas, Inés Altemir
Kucherenko, Anastasiia
Kucharavy, Andrei
author_facet Marinas, Inés Altemir
Kucherenko, Anastasiia
Kucharavy, Andrei
contents Large language models (LLMs) rely heavily on web-scale datasets like Common Crawl, which provides over 80\% of training data for some modern models. However, the indiscriminate nature of web crawling raises challenges in data quality, safety, and ethics. Despite the critical importance of training data quality, prior research on harmful content has been limited to small samples due to computational constraints. This project presents a framework for indexing and analyzing LLM training datasets using an ElasticSearch-based pipeline. We apply it to SwissAI's FineWeb-2 corpus (1.5TB, four languages), achieving fast query performance--most searches in milliseconds, all under 2 seconds. Our work demonstrates real-time dataset analysis, offering practical tools for safer, more accountable AI systems.
format Preprint
id arxiv_https___arxiv_org_abs_2508_21788
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Going over Fine Web with a Fine-Tooth Comb: Technical Report of Indexing Fine Web for Problematic Content Search and Retrieval
Marinas, Inés Altemir
Kucherenko, Anastasiia
Kucharavy, Andrei
Computation and Language
Artificial Intelligence
Information Retrieval
Large language models (LLMs) rely heavily on web-scale datasets like Common Crawl, which provides over 80\% of training data for some modern models. However, the indiscriminate nature of web crawling raises challenges in data quality, safety, and ethics. Despite the critical importance of training data quality, prior research on harmful content has been limited to small samples due to computational constraints. This project presents a framework for indexing and analyzing LLM training datasets using an ElasticSearch-based pipeline. We apply it to SwissAI's FineWeb-2 corpus (1.5TB, four languages), achieving fast query performance--most searches in milliseconds, all under 2 seconds. Our work demonstrates real-time dataset analysis, offering practical tools for safer, more accountable AI systems.
title Going over Fine Web with a Fine-Tooth Comb: Technical Report of Indexing Fine Web for Problematic Content Search and Retrieval
topic Computation and Language
Artificial Intelligence
Information Retrieval
url https://arxiv.org/abs/2508.21788