SDAR-VL: Stable and Efficient Block-wise Diffusion for Vision-Language Understanding

2025-12-16 · Shuang Cheng, Yuhua Jiang, Zineng Zhou, Dawei Liu, Wang Tao, Linfeng Zhang, Biqing Qi, Bowen Zhou arxiv

Block-wise discrete diffusion offers an attractive balance between parallel generation and causal dependency modeling, making it a promising backbone for vision-language modeling. However, its practical adoption has been limited by high training cost, slow convergence, and instability, which have so far kept it behind strong autoregressive (AR) baselines. We present \textbf{SDAR-VL}, the first systematic application of block-wise discrete diffusion to large-scale vision-language understanding (VLU), together with an \emph{integrated framework for efficient and stable training}. This framework unifies three components: (1) \textbf{Asynchronous Block-wise Noise Scheduling} to diversify supervision within each batch; (2) \textbf{Effective Mask Ratio Scaling} for unbiased loss normalization under stochastic masking; and (3) a \textbf{Progressive Beta Noise Curriculum} that increases effective mask coverage while preserving corruption diversity. Experiments on 21 single-image, multi-image, and video benchmarks show that SDAR-VL consistently improves \emph{training efficiency}, \emph{convergence stability}, and \emph{task performance} over conventional block diffusion. On this evaluation suite, SDAR-VL sets a new state of the art among diffusion-based vision-language models and, under matched settings, matches or surpasses strong AR baselines such as LLaVA-OneVision as well as the global diffusion baseline LLaDA-V, establishing block-wise diffusion as a practical backbone for VLU.

📄 PDF Abstract BibTeX arXiv:2512.14068

Code (0)

등록된 구현이 없습니다.

SDAR: A Synergistic Diffusion-AutoRegression Paradigm for Scalable Sequence Generation

2025-10-07 · Shuang Cheng, Yihan Bian, Dawei Liu, Linfeng Zhang 외 arxiv

We propose SDAR, a Synergistic Diffusion-Autoregression paradigm that unifies the training efficiency of autoregressive models with the parallel inference capability of diffusion. Instead of costly end-to-end diffusion t…

S2D2: Fast Decoding for Diffusion LLMs via Training-Free Self-Speculation

2026-03-26 · Ligong Han, Hao Wang, Han Gao, Kai Xu 외 arxiv

Block-diffusion language models offer a promising path toward faster-than-autoregressive generation by combining block-wise autoregressive decoding with within-block parallel denoising. However, in the few-step regime ne…

SLIM-RL: Risk-Budgeted Random-Masking RL for Diffusion LLMs Without Trajectory Slicing

2026-06-30 · Ruikang Zhao, Zhenting Wang, Han Gao, Ligong Han arxiv

Reinforcement learning for diffusion large language models (dLLMs) has largely moved to trajectory-aware methods. The current state of the art, TraceRL, holds that random masking is mismatched with the model's inference …

Reinforcement Learning

Stable-DiffCoder: Pushing the Frontier of Code Diffusion Large Language Model

2026-01-22 · Chenghao Fan, Wen Heng, Bo Li, Sichen Liu 외 arxiv

Diffusion-based language models (DLLMs) offer non-sequential, block-wise generation and richer data reuse compared to autoregressive (AR) models, but existing code DLLMs still lag behind strong AR baselines under compara…

Continual Pretraining Data Augmentation

SimSD: Simple Speculative Decoding in Diffusion Language Models

2026-06-01 · Junxia Cui, Haotian Ye, Runchu Tian, Hongcan Guo 외 arxiv

Diffusion large language models (dLLMs) have recently emerged as a promising alternative to autoregressive (AR) LLMs, offering faster inference through parallel or blockwise decoding. However, their masked language model…