Metraipolis › archive › dc_d27615ad12e6

Tracing the Evidence: Faithful Token Attribution Through Vision-Language Reasoning — arXiv:2609.37656

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.37656
License (as archived)
CC0-1.0
Archived
2026-10-07T18:41:26.963Z
SHA-256 of the archived text
4d54119720f8dd9b376c6254b55f036d1cfd6416091eb8d32610b9e5b95a8457
Stable URL of this page
https://metraipolis.com/documento/dc_d27615ad12e6

How to cite

Tracing the Evidence: Faithful Token Attribution Through Vision-Language Reasoning — arXiv:2609.37656 — arXiv (en), https://arxiv.org/abs/2609.37656 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d27615ad12e6 (Metraipolis archive dc_d27615ad12e6, sha256 4d54119720f8dd9b376c6254b55f036d1cfd6416091eb8d32610b9e5b95a8457)
BibTeX
@misc{dcd27615ad12e6,
  title = {Tracing the Evidence: Faithful Token Attribution Through Vision-Language Reasoning — arXiv:2609.37656},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.37656},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d27615ad12e6; sha256 4d54119720f8dd9b376c6254b55f036d1cfd6416091eb8d32610b9e5b95a8457},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.37656. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large vision-language models (LVLMs) exhibit strong reasoning capabilities, yet the visual and textual evidence supporting the generated responses remains difficult to identify. Faithful token attribution explains an LVLM's response by assigning scores that rank image and prompt tokens by how much the model relies on them, such that removing higher-ranked tokens causes the likelihood of the generated response to drop more rapidly. However, existing token-attribution methods have been developed mainly for text-based language models, and our empirical study reveals two challenges when complex multimodal sources are involved. First, the joint image-text attribution can underrepresent visual evidence relative to text, obscuring the image regions supporting the response. Second, visual evidence may influence the generated response through multiple intermediate reasoning paths, while existing methods trace only a limited subset of these paths, causing important visual contributions to be underestimated. Motivated by these insights, we introduce VTrace, a multimodal token-attribution framework that traces input contributions through intermediate reasoning and calibrates attribution scores across modalities. VTrace constructs pairwise attributions that highlight token-specific contributions and aggregates all forward attribution paths in closed form to account for both direct and indirect contributions. Cross-modal calibration then rescales image and text attribution scores using modality contributions estimated from response-likelihood changes, enabling a unified ranking of input tokens. Evaluations against seven baselines across six visual reasoning benchmarks demonstrate the superior attribution faithfulness. Project page: this https URL .

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-30 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.37656 — 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 2026-10-07T18:04:24.897Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-30?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).