Metraipolis › archive › dc_246795ffb0bf

Accelerating Unified Multimodal Models with Core-Expansion Routing and Unified Computation Scheduling — arXiv:2608.29291

Source: arXiv (en)
Original: https://arxiv.org/abs/2608.29291
License (as archived): CC0-1.0
Archived: 2026-10-17T07:44:13.103Z
SHA-256 of the archived text: 9c247c2c2f998c6733e2a9483357b5e516f1ed995cacde4ac9085c8868542172
Stable URL of this page: https://metraipolis.com/documento/dc_246795ffb0bf

How to cite

Accelerating Unified Multimodal Models with Core-Expansion Routing and Unified Computation Scheduling — arXiv:2608.29291 — arXiv (en), https://arxiv.org/abs/2608.29291 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_246795ffb0bf (Metraipolis archive dc_246795ffb0bf, sha256 9c247c2c2f998c6733e2a9483357b5e516f1ed995cacde4ac9085c8868542172)

BibTeX

@misc{dc246795ffb0bf,
  title = {Accelerating Unified Multimodal Models with Core-Expansion Routing and Unified Computation Scheduling — arXiv:2608.29291},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.29291},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_246795ffb0bf; sha256 9c247c2c2f998c6733e2a9483357b5e516f1ed995cacde4ac9085c8868542172},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.29291. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Unified multimodal models jointly support understanding and generation, but incur substantial redundant computation across tokens, layers, and generation timesteps. Through token-importance probing, we identify an asymmetric core-expansion structure: understanding exhibits a stable importance component, while generation largely shares this component but requires progress-dependent corrections. We therefore propose CE-Router, which uses a task-shared core scorer and progress-conditioned generation expansions, optimized through generation decomposition and cross-task core alignment. At inference, CE-Router compacts token computation and supplies a learned routing signal to Unified Computation Scheduling, which coordinates layer skipping, FFN pruning, diffusion-head cache reuse, and denoising-step early exit. Experiments on two representative UMM architectures demonstrate consistent quality--efficiency improvements across both tasks, retaining 98.03\% of dense understanding performance with a 1.93$\times$ end-to-end inference speedup.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.29291 — 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 2026-10-17T07:28:26.536Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).