Metraipolis › archive › dc_bbdd49716e19

Resolution as a First-Class Decision: Task-Conditioned Routing for Efficient Multimodal Large Language Models — arXiv:2609.34942

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.34942
License (as archived)
CC0-1.0
Archived
2026-10-07T23:52:34.275Z
SHA-256 of the archived text
8f56f0ecccf19d7afd95f1cb695e6f2054a30629209808dc2e610865c4996a43
Stable URL of this page
https://metraipolis.com/documento/dc_bbdd49716e19

How to cite

Resolution as a First-Class Decision: Task-Conditioned Routing for Efficient Multimodal Large Language Models — arXiv:2609.34942 — arXiv (en), https://arxiv.org/abs/2609.34942 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bbdd49716e19 (Metraipolis archive dc_bbdd49716e19, sha256 8f56f0ecccf19d7afd95f1cb695e6f2054a30629209808dc2e610865c4996a43)
BibTeX
@misc{dcbbdd49716e19,
  title = {Resolution as a First-Class Decision: Task-Conditioned Routing for Efficient Multimodal Large Language Models — arXiv:2609.34942},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.34942},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bbdd49716e19; sha256 8f56f0ecccf19d7afd95f1cb695e6f2054a30629209808dc2e610865c4996a43},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34942. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

The inference efficiency of Multimodal Large Language Models (MLLMs) is severely constrained by massive visual token sequences induced by high-resolution inputs, with computational cost scaling quadratically. Existing approaches primarily focus on downstream token compression, while overlooking a fundamental upstream inefficiency: input resolution is treated as a static, task-agnostic hyperparameter. We propose Task-Conditioned Resolution Routing (TCRR), which formulates visual compression as a task-conditioned decision and employs a lightweight cross-modal router that conditions backbone visual representations on textual semantics via feature-wise modulation and cross-attention to predict the minimal sufficient compression level per query. To support this, we curate a dataset of 500k samples across 12 task categories, labeled via a teacher-oracle pipeline to approximate Pareto-optimal compression scales. Extensive experiments across diverse architectures show that TCRR achieves a superior efficiency frontier, specifically reducing visual FLOPs by 40.9% and latency by 53.7% on Qwen3-VL-8B while preserving competitive performance. Further analysis of scaling behavior confirms that dynamically routing visual compression enables optimal resource allocation without modifying the MLLM backbone.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-29 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.34942 — 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 2026-10-07T23:29:27.503Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).