Powered by OpenAIRE graph
Found an issue? Give us feedback
image/svg+xml Jakob Voss, based on art designer at PLoS, modified by Wikipedia users Nina and Beao Closed Access logo, derived from PLoS Open Access logo. This version with transparent background. http://commons.wikimedia.org/wiki/File:Closed_Access_logo_transparent.svg Jakob Voss, based on art designer at PLoS, modified by Wikipedia users Nina and Beao ZENODOarrow_drop_down
image/svg+xml Jakob Voss, based on art designer at PLoS, modified by Wikipedia users Nina and Beao Closed Access logo, derived from PLoS Open Access logo. This version with transparent background. http://commons.wikimedia.org/wiki/File:Closed_Access_logo_transparent.svg Jakob Voss, based on art designer at PLoS, modified by Wikipedia users Nina and Beao
ZENODO
Dataset . 2025
License: CC BY
Data sources: ZENODO
ZENODO
Dataset . 2025
License: CC BY
Data sources: Datacite
ZENODO
Dataset . 2025
License: CC BY
Data sources: Datacite
versions View all 2 versions
addClaim

InfiniQA: A Large-Scale French Question-Answering Dataset with Comprehensive Evaluation

Authors: CHARLET, Théo;

InfiniQA: A Large-Scale French Question-Answering Dataset with Comprehensive Evaluation

Abstract

Nous présentons InfiniQA, un jeu de données de questions-réponses en français à grande échelle, contenant 101 084 paires de questions-réponses validées manuellement. Contrairement aux autres jeux de données français existants qui reposent sur l’extraction ou la traduction, InfiniQA se concentre sur des paires Q/R directes et factuelles, créées nativement en français. Notre cadre d’évaluation complet démontre l’efficacité du calcul des métriques en comparant les réponses de référence aux réponses modifiées selon plusieurs indicateurs, notamment le Exact Match, les scores BLEU et les métriques ROUGE. Le jeu de données couvre des domaines variés comme la culture, l’histoire, la géographie et le sport, avec un vocabulaire de 83 592 mots uniques. InfiniQA établit une nouvelle référence pour l’évaluation des Q/R en français et constitue une ressource précieuse pour l’entraînement et l’évaluation des modèles de langue en français.

We introduce InfiniQA, a large-scale French question-answering dataset containing 101,084 manually validated question-answer pairs. Unlike existing French QA datasets that rely on extraction or translation, InfiniQA focuses on direct, factual Q&A pairs created natively in French. Our comprehensive evaluation framework demonstrates the efficiency of metric computation through ground truth vs modified response comparisons across multiple metrics including Exact Match, BLEU scores, and ROUGE metrics. The dataset covers diverse domains including culture, history, geography, and sports, with a vocabulary of 83,592 unique words. InfiniQA establishes a new benchmark for French QA evaluation and provides a valuable resource for training and evaluating French language models.

Keywords

InfiniQA, Open-source, Dataset

  • BIP!
    Impact byBIP!
    citations
    The total number of citations received by all articles of the researcher of interest.
    0
    popularity
    This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
    Average
    influence
    This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    Average
    impulse
    This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
    Average
Powered by OpenAIRE graph
Found an issue? Give us feedback
citations
The total number of citations received by all articles of the researcher of interest.
BIP!Citations provided by BIP!
popularity
This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
BIP!Popularity provided by BIP!
influence
This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Influence provided by BIP!
impulse
This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
BIP!Impulse provided by BIP!
0
Average
Average
Average