Newest papers

Beyond the Sequentiality Bottleneck: A Hybrid Diffusion-Autoregressive Framework for Efficient Multimodal Reasoning

Wei Cai, Jian Zhao, Yuchen Yuan, Xuelong Li · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.28584 · cited by 0 works (OpenAlex) · open-access PDF

Arabic

وتشمل النماذج التجميلية (ARMs) "قنبلة التجميل" في التفكير، في حين أن النماذج التجميلية (DLMs) تقدم تناقضات لكنها لا تنتمي إلى التوافق السليمي. لتحديد كل الحدود، ونحن نقدم CoRAL (CoRAL) ، وهي إطار متكامل من خلال إطار APE (Asymmetric Planner-Executor) الأكاديمي: DLM Planner يخلق أطوار التجميل المختلفة K بشكل متوازن، في حين أن ARM Executor الكبير يختار الخطة المثلى وتشمل الإجابة النهائية. لتدريب Planner بشكل فعال، ونحن نقدم Reward-Guided Denoising for CoT (RDC) الذي يضم المكافآت المتعل

Machine translation (Meta M2M100); the English original follows.

Abstract

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.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Artificial intelligence, Planner, Process (computing), Face (sociological concept), Plan (archaeology)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange