Metraipolis › archive › dc_9f7153cb0502
See, Think, Use: Diagnosing Perception and Reasoning Failures in VLMs — arXiv:2611.01467
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.01467
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T11:06:27.671Z
- SHA-256 of the archived text
fb25a45efd7a4145621535eeb5d4b1519cd09b748265be217e387b4ee9688b6f- Stable URL of this page
https://metraipolis.com/documento/dc_9f7153cb0502
How to cite
See, Think, Use: Diagnosing Perception and Reasoning Failures in VLMs — arXiv:2611.01467 — arXiv (en), https://arxiv.org/abs/2611.01467 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_9f7153cb0502 (Metraipolis archive dc_9f7153cb0502, sha256 fb25a45efd7a4145621535eeb5d4b1519cd09b748265be217e387b4ee9688b6f)
BibTeX
@misc{dc9f7153cb0502,
title = {See, Think, Use: Diagnosing Perception and Reasoning Failures in VLMs — arXiv:2611.01467},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.01467},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_9f7153cb0502; sha256 fb25a45efd7a4145621535eeb5d4b1519cd09b748265be217e387b4ee9688b6f},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.01467. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
We present See, Think, Use, a protocol that traces each answer of a vision-language model to perception, reasoning or the use of perception. The diagnosis shows a developer which stage to improve and directs test-time compute to the stage that fails. Benchmarks score the final answer, and an answer that rests on the image earns the same credit as a guess or a memorized prior. The protocol asks the model two extra questions per item and needs no access to its weights. A localization probe asks for a box around every instance and measures seeing, and a text-replacement probe answers from a description and measures thinking. Use is the rate at which the answer follows the model's own perception across counterfactual image pairs. The rate needs no answer labels, because the question and its prior are the same on both images. On two model families and six datasets the protocol shows that the knob that helps follows the task family. Answers follow the model's perception except on images that contradict a common prior, and asking the model to localize first repairs this failure when its boxes are right. Allocating compute by the measured stages comes close to the best uniform policy at lower cost and beats a random allocation of the same compute wherever a knob raises the short stage. We will release the code, the prompts and the per-question outputs. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI); Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2611.01467 — 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 2026-11-03T10:59:46.014Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).