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

Korean

차별화된 언어 모델(DLMs)는 평행성을 제공하지만 정의적 일관성을 부족하게 한다. 두 가지 제한을 해결하기 위해, 우리는 협업적 근거적 비대칭 학습(CoRAL)을 제안하고, APE 아키텍처를 통해 하이브리드 프레임 워크(Sequentiality Bottleneck)에 직면한다: DLM Planner는 K의 다양한 논리 경로를 동시에 생성하고, 큰 ARM Executor는 최적의 계획을 선택하고 최종 해답을 합성한다. 효과적으로 Planner를 훈련시키기 위해, 우리는 CoT(RDC)에 대한 보상 지시된 거부(Reward-Guided Denoising)를 도입하여 결과 기반의 보상을 프로세스에 통합하여 목표 지향적 논리로 이끌

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