CITATION.cff

# This CITATION.cff file was generated with cffinit.
# Visit https://bit.ly/cffinit to generate yours today!

cff-version: 1.2.0
title: 'MUCH : A Multilingual Claim Hallucination Benchmark'
message: >-
  If you use this dataset, please cite it using the metadata
  from this file.
type: dataset
authors:
  - given-names: Jérémie
    family-names: Dentan
    orcid: 'https://orcid.org/0009-0001-5561-8030'
  - given-names: Alexi
    family-names: Canesse
    orcid: 'https://orcid.org/0009-0005-1494-4736'
  - given-names: Davide
    family-names: Buscaldi
    orcid: 'https://orcid.org/0000-0003-1112-3789'
  - given-names: Aymen
    family-names: Shabou
    orcid: 'https://orcid.org/0000-0001-8933-7053'
  - given-names: Sonia
    family-names: Vanier
    orcid: 'https://orcid.org/0000-0001-6390-8882'
identifiers:
  - type: url
    value: 'https://arxiv.org/abs/2511.17081'
repository-code: 'https://github.com/orailix/much'
repository-artifact: 'https://huggingface.co/datasets/orailix/MUCH'
abstract: >-
  Claim-level Uncertainty Quantification (UQ) is a promising
  approach to mitigate the lack of reliability in Large
  Language Models (LLMs). We introduce MUCH, the first
  claim-level UQ benchmark designed for fair and
  reproducible evaluation of future methods under realistic
  conditions. It includes 4,873 samples across four European
  languages (English, French, Spanish, and German) and four
  instruction-tuned open-weight LLMs. Unlike prior
  claim-level benchmarks, we release 24 generation logits
  per token, facilitating the development of future
  white-box methods without re-generating data. Moreover, in
  contrast to previous benchmarks that rely on manual or
  LLM-based segmentation, we propose a new deterministic
  algorithm capable of segmenting claims using as little as
  0.2% of the LLM generation time. This makes our
  segmentation approach suitable for real-time monitoring of
  LLM outputs, ensuring that MUCH evaluates UQ methods under
  realistic deployment constraints. Finally, our evaluations
  show that current methods still have substantial room for
  improvement in both performance and efficiency.
license: Apache-2.0