Metraipolis › archive › dc_700daaa6954b

Multimodal Models Can See, But Can They Think? A Study of Turkish Visual Reasoning — arXiv:2611.10831

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.10831
License (as archived)
CC0-1.0
Archived
2026-11-10T18:41:14.257Z
SHA-256 of the archived text
bc3b2cfcaa3b046e65ac3cff9386dacdae33ded3f879504f531a1c0250fa0c52
Stable URL of this page
https://metraipolis.com/documento/dc_700daaa6954b

How to cite

Multimodal Models Can See, But Can They Think? A Study of Turkish Visual Reasoning — arXiv:2611.10831 — arXiv (en), https://arxiv.org/abs/2611.10831 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_700daaa6954b (Metraipolis archive dc_700daaa6954b, sha256 bc3b2cfcaa3b046e65ac3cff9386dacdae33ded3f879504f531a1c0250fa0c52)
BibTeX
@misc{dc700daaa6954b,
  title = {Multimodal Models Can See, But Can They Think? A Study of Turkish Visual Reasoning — arXiv:2611.10831},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.10831},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_700daaa6954b; sha256 bc3b2cfcaa3b046e65ac3cff9386dacdae33ded3f879504f531a1c0250fa0c52},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.10831. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Vision-Language Models have had massive leaps in recent years. Yet, when tested on high-level cognitive tasks like pattern completion, abstract logic, and relational reasoning, they still fall short. This limitation stems from the models' lack of ability to find out underlying rules and maintain constant intermediate steps, rather than merely deciding on the correct option. To research these shortcomings, this study analyzes visual reasoning performance throughout a multilingual evaluation framework encompassing the BİLSEM, MM-IQ, Visual Puzzles, and PuzzleVQA datasets. We evaluate various data mixing strategies and supervised fine-tuning (SFT) methods. We observed that a correct answer does not always mean the reasoning chain is valid. Because of this, we adopt an LLM-as-a-Judge approach to assess the quality of reasoning beyond final answer accuracy. Our experimental results reveal that baseline models often exhibit more stable reasoning compared to fine-tuned counterparts. Most importantly, we observe that SFT can lead to model collapse, manifesting as infinite text loops or empty outputs. The study concludes that current VLMs still struggle with abstract visual reasoning.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2611.10831 — 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 2026-11-10T18:37:37.505Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).