Metraipolis › archive › dc_98223ee5171e

EDCT-Bench: Uncovering Faithfulness Gaps in VLMs via Explanation-Driven Counterfactual Testing — arXiv:2609.17953

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.17953
License (as archived)
CC0-1.0
Archived
2026-10-15T16:16:16.034Z
SHA-256 of the archived text
7ff0d87cd47071d9f19e99c5eebcc52a5c2f483ad3b89726fc3fcfe7bad92fc7
Stable URL of this page
https://metraipolis.com/documento/dc_98223ee5171e

How to cite

EDCT-Bench: Uncovering Faithfulness Gaps in VLMs via Explanation-Driven Counterfactual Testing — arXiv:2609.17953 — arXiv (en), https://arxiv.org/abs/2609.17953 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_98223ee5171e (Metraipolis archive dc_98223ee5171e, sha256 7ff0d87cd47071d9f19e99c5eebcc52a5c2f483ad3b89726fc3fcfe7bad92fc7)
BibTeX
@misc{dc98223ee5171e,
  title = {EDCT-Bench: Uncovering Faithfulness Gaps in VLMs via Explanation-Driven Counterfactual Testing — arXiv:2609.17953},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.17953},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_98223ee5171e; sha256 7ff0d87cd47071d9f19e99c5eebcc52a5c2f483ad3b89726fc3fcfe7bad92fc7},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.17953. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Vision-Language Models (VLMs) can produce Natural Language Explanations (NLEs) that sound plausible yet remain inconsistent with the visual evidence they cite. We present Explanation-Driven Counterfactual Testing (EDCT), an intervention-based protocol that extracts visual concepts cited in a model's explanation, applies verified minimal edits to them, and tests whether the resulting answer and explanation remain consistent with the edited image. Using this protocol, we create EDCT-Bench, a comprehensive benchmark spanning three complementary domains: knowledge-intensive visual question answering (OK-VQA), safety-critical driving (DriveLM), and 3D spatial reasoning (3DSRBench). Across the evaluated VLMs, EDCT reveals substantial faithfulness gaps, with models frequently producing responses inconsistent with verified visual changes. Finally, our fine-tuning study suggests that EDCT-generated counterfactuals provide high-impact training signals.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-17 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.17953 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-15T15:45:19.737Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-17?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).