Metraipolis › archive › dc_b227a59eff2e
Beyond the Sequentiality Bottleneck: A Hybrid Diffusion-Autoregressive Framework for Efficient Multimodal Reasoning — arXiv:2610.28584
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.28584
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-21T07:07:10.155Z
- SHA-256 of the archived text
74afd6c15385dd2edb93ee621c50de4ea20f55c36fd60eb7e59ea58a8edef968- Stable URL of this page
https://metraipolis.com/documento/dc_b227a59eff2e
How to cite
Beyond the Sequentiality Bottleneck: A Hybrid Diffusion-Autoregressive Framework for Efficient Multimodal Reasoning — arXiv:2610.28584 — arXiv (en), https://arxiv.org/abs/2610.28584 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_b227a59eff2e (Metraipolis archive dc_b227a59eff2e, sha256 74afd6c15385dd2edb93ee621c50de4ea20f55c36fd60eb7e59ea58a8edef968)
BibTeX
@misc{dcb227a59eff2e,
title = {Beyond the Sequentiality Bottleneck: A Hybrid Diffusion-Autoregressive Framework for Efficient Multimodal Reasoning — arXiv:2610.28584},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.28584},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_b227a59eff2e; sha256 74afd6c15385dd2edb93ee621c50de4ea20f55c36fd60eb7e59ea58a8edef968},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.28584. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Autoregressive models (ARMs) face a "Sequentiality Bottleneck" in reasoning, while diffusion language models (DLMs) offer parallelism but lack semantic coherence. To addresses both limitations, we propose Collaborative Reasoning Asymmetric Learning (CoRAL), a hybrid framework via an Asymmetric Planner-Executor (APE) architecture: a DLM Planner generates K diverse reasoning trajectories in parallel, while a large ARM Executor selects the optimal plan and synthesizes the final answer. To train the Planner effectively, we introduce Reward-Guided Denoising for CoT (RDC), which integrates outcome-based rewards into the diffusion process to steer it toward goal-directed reasoning. Experiments on vision-language benchmarks show that CoRAL significantly outperforms prior diffusion-based methods and achieves competitive performance against strong ARMs, achieving both superior efficiency and quality. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2610.28584 — 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 2026-10-21T07:05:43.492Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).