Metraipolis › archive › dc_81857bb3d3ea
Compositional Failure in Audio-Visual LLMs: Late-Layer Prior Dominance Under Cross-modal Conflict — arXiv:2608.27785
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.27785
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T10:47:55.975Z
- SHA-256 of the archived text
d85cbb068205d9921e17e746b7d4d03679e6b20b921eff9d664b648a4bdbb2a3- Stable URL of this page
https://metraipolis.com/documento/dc_81857bb3d3ea
How to cite
Compositional Failure in Audio-Visual LLMs: Late-Layer Prior Dominance Under Cross-modal Conflict — arXiv:2608.27785 — arXiv (en), https://arxiv.org/abs/2608.27785 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_81857bb3d3ea (Metraipolis archive dc_81857bb3d3ea, sha256 d85cbb068205d9921e17e746b7d4d03679e6b20b921eff9d664b648a4bdbb2a3)
BibTeX
@misc{dc81857bb3d3ea,
title = {Compositional Failure in Audio-Visual LLMs: Late-Layer Prior Dominance Under Cross-modal Conflict — arXiv:2608.27785},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.27785},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_81857bb3d3ea; sha256 d85cbb068205d9921e17e746b7d4d03679e6b20b921eff9d664b648a4bdbb2a3},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.27785. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
We study audio-visual conflict as a compositional generalization test for AV-LLMs: the model must combine synchronized but semantically incompatible audio and video evidence and decide whether the pair matches. On VideoLLaMA 2-7B-AV, three alignment configurations remain nearchance on the scored exact-string Yes/No subset of AVHBench, even though their output priors shift substantially. Similarly, off-the-shelf InternVideo2 experienced a 32.3% accuracy decrease specifically under cross-modal conflict, accompanied by a 17.3% instruction-following failure. We call this failure mode prior dominance: late-layer commitment to an internally preferred answer pattern that is weakly grounded in the conflicting inputs. To explain this behavior, we conduct a mechanistic interpretability analysis and find that commitment remains concentrated at 25.5 $\pm$ 1 layers. We show that stronger temporal alignment changes answer bias, but do not improve compositional conflict resolution. Code and data to reproduce our mechanistic audit and behavioral evaluations are available at this https URL . Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-31 in cs.AI). Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2608.27785 — 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 2026-10-17T10:33:42.470Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-31?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).