统一建模文本理解与图像生成，支持灵活长度输出。
LLaDA-o: An Effective and Length-Adaptive Omni Diffusion Model
- 采用混合扩散框架分离文本与视觉建模，共享注意力结构提升效率。
- 在多模态任务中达顶尖性能，文本到图像生成得分87.04。
- 无需修改架构即可实现长度自适应，适合多场景应用。
我们提出LLaDA-o，一种高效且长度自适应的多模态扩散模型，用于多模态理解和生成。该模型基于混合扩散（MoD）框架，将文本理解的离散掩码扩散与视觉生成的连续扩散解耦，通过共享的简单高效注意力主干连接二者，减少固定条件下的冗余计算。在此基础上，我们进一步引入数据驱动的长度自适应策略，使多模态场景下无需架构调整即可实现灵活长度解码。大量实验表明，LLaDA-o在多模态理解与生成基准上达到当前最优性能，文本到图像生成在DPG-Bench上取得87.04分，验证了统一多模态扩散建模的有效性。代码已开源：https://github.com/ML-GSAI/LLaDA-o。
原文摘要 · Abstract (English)
We present \textbf{LLaDA-o}, an effective and length-adaptive omni diffusion model for multimodal understanding and generation. LLaDA-o is built on a Mixture of Diffusion (MoD) framework that decouples discrete masked diffusion for text understanding and continuous diffusion for visual generation, while coupling them through a shared, simple, and efficient attention backbone that reduces redundant computation for fixed conditions. Building on MoD, we further introduce a data-centric length adaptation strategy that enables flexible-length decoding in multimodal settings without architectural changes. Extensive experiments show that LLaDA-o achieves state-of-the-art performance among omni-diffusion models on multimodal understanding and generation benchmarks, and reaches 87.04 on DPG-Bench for text-to-image generation, supporting the effectiveness of unified omni diffusion modeling. Code is available at https://github.com/ML-GSAI/LLaDA-o.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。