Metraipolis › archive › dc_423d04797cf8
MOSAIC: Addressing Disproportionate Task-Modality Dependencies via Modality-Partitioned Gradient Routing — arXiv:2610.41004
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.41004
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-02T07:34:51.426Z
- SHA-256 of the archived text
1d613bc63bd1b45fae94c8467d467cfcae6ce4d361a2f36eb0c7259f126d61b0- Stable URL of this page
https://metraipolis.com/documento/dc_423d04797cf8
How to cite
MOSAIC: Addressing Disproportionate Task-Modality Dependencies via Modality-Partitioned Gradient Routing — arXiv:2610.41004 — arXiv (en), https://arxiv.org/abs/2610.41004 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_423d04797cf8 (Metraipolis archive dc_423d04797cf8, sha256 1d613bc63bd1b45fae94c8467d467cfcae6ce4d361a2f36eb0c7259f126d61b0)
BibTeX
@misc{dc423d04797cf8,
title = {MOSAIC: Addressing Disproportionate Task-Modality Dependencies via Modality-Partitioned Gradient Routing — arXiv:2610.41004},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.41004},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_423d04797cf8; sha256 1d613bc63bd1b45fae94c8467d467cfcae6ce4d361a2f36eb0c7259f126d61b0},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.41004. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
In multi-task settings, combining multiple modalities allows task representations to draw on signals that a single modality alone would not capture. However, individual tasks may depend disproportionately on particular modalities, a structure that conventional fusion-based architectures do not model explicitly, instead enforcing a unified representation space. Consequently, a task's gradients can affect parameters tied to modalities it barely relies on, leaving features that compromise between competing demands. Although prior work has explored task-weighting and loss-balancing strategies, their effectiveness is limited by the absence of explicit modality separation, preventing models from optimising each modality according to its task-specific importance. To address these limitations, we propose MOSAIC (MOdality-Separated Architecture with Inter-modal Cross-attention), which preserves unimodal representations while enabling controlled interaction between them, and uses modality-specific losses to route gradients to their corresponding partitions. On the Corpus of Everyday Japanese Conversation (CEJC), MOSAIC improves average macro-F1 by 7.4 points across four tasks with a 72% reduction in backbone parameters relative to a fusion-based baseline. Our results demonstrate that modality partitioning yields consistent gains across tasks while requiring far fewer backbone parameters than a fusion-based design. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG) ; Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2610.41004 — 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 2026-11-02T07:27:12.656Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).