Xiaoxing Hu, Wenxi Yao, Chaoyuan Shi, Ziqi Ye, Xiaorong Guo, Peixuan Li, Zhuoran Sun, Ziyang Gong · 2026 · arXiv
DOI: 10.48550/arXiv.2610.28444 · open-access PDF
Os LLMs Omni podem naturalmente localizar evidências audiovisuais e visuais-chave e integrar informações entre modalidades para raciocínio audiovisual. Os benchmarks de raciocínio audiovisual existentes que dependem principalmente de perguntas de múltipla escolha geralmente pontuam modelos baseados exclusivamente na correção da resposta, tornando difícil avaliar se um modelo realmente forma uma cadeia de raciocínio de evidências para a resposta correta. Além disso, atalhos de texto ou unimodal podem permitir que os modelos respondam corretamente sem analisar as evidências audiovisuais necessárias, enfraquecendo ainda mais a confiabilidade dos benchmarks de avaliação.
Machine translation (Meta M2M100); the English original follows.
Omni LLMs can naturally locate key audio and visual evidence and integrate information across modalities for audiovisual reasoning. Existing audio-visual reasoning benchmarks that mainly rely on multiple-choice questions usually score models solely based on answer correctness, making it difficult to assess whether a model truly forms a reasoning chain from evidence to the correct answer. In addition, textual or unimodal shortcuts may allow models to answer correctly without analyzing the necessary audio-visual evidence, further weakening the reliability of evaluation benchmarks. To address the
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.28444, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange