LLaDA2.0-Uni: Unifying Multimodal Understanding and Generation with Diffusion Large Language Model
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866917429800599552 |
|---|---|
| author | AI, Inclusion Bie, Tiwei Chen, Haoxing Chen, Tieyuan Cheng, Zhenglin Cui, Long Gan, Kai Huang, Zhicheng Lan, Zhenzhong Li, Haoquan Li, Jianguo Lin, Tao Qin, Qi Wang, Hongjun Wang, Xiaomei Wu, Haoyuan Xin, Yi Zhao, Junbo |
| author_facet | AI, Inclusion Bie, Tiwei Chen, Haoxing Chen, Tieyuan Cheng, Zhenglin Cui, Long Gan, Kai Huang, Zhicheng Lan, Zhenzhong Li, Haoquan Li, Jianguo Lin, Tao Qin, Qi Wang, Hongjun Wang, Xiaomei Wu, Haoyuan Xin, Yi Zhao, Junbo |
| contents | We present LLaDA2.0-Uni, a unified discrete diffusion large language model (dLLM) that supports multimodal understanding and generation within a natively integrated framework. Its architecture combines a fully semantic discrete tokenizer, a MoE-based dLLM backbone, and a diffusion decoder. By discretizing continuous visual inputs via SigLIP-VQ, the model enables block-level masked diffusion for both text and vision inputs within the backbone, while the decoder reconstructs visual tokens into high-fidelity images. Inference efficiency is enhanced beyond parallel decoding through prefix-aware optimizations in the backbone and few-step distillation in the decoder. Supported by carefully curated large-scale data and a tailored multi-stage training pipeline, LLaDA2.0-Uni matches specialized VLMs in multimodal understanding while delivering strong performance in image generation and editing. Its native support for interleaved generation and reasoning establishes a promising and scalable paradigm for next-generation unified foundation models. Codes and models are available at https://github.com/inclusionAI/LLaDA2.0-Uni. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_20796 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | LLaDA2.0-Uni: Unifying Multimodal Understanding and Generation with Diffusion Large Language Model AI, Inclusion Bie, Tiwei Chen, Haoxing Chen, Tieyuan Cheng, Zhenglin Cui, Long Gan, Kai Huang, Zhicheng Lan, Zhenzhong Li, Haoquan Li, Jianguo Lin, Tao Qin, Qi Wang, Hongjun Wang, Xiaomei Wu, Haoyuan Xin, Yi Zhao, Junbo Computer Vision and Pattern Recognition We present LLaDA2.0-Uni, a unified discrete diffusion large language model (dLLM) that supports multimodal understanding and generation within a natively integrated framework. Its architecture combines a fully semantic discrete tokenizer, a MoE-based dLLM backbone, and a diffusion decoder. By discretizing continuous visual inputs via SigLIP-VQ, the model enables block-level masked diffusion for both text and vision inputs within the backbone, while the decoder reconstructs visual tokens into high-fidelity images. Inference efficiency is enhanced beyond parallel decoding through prefix-aware optimizations in the backbone and few-step distillation in the decoder. Supported by carefully curated large-scale data and a tailored multi-stage training pipeline, LLaDA2.0-Uni matches specialized VLMs in multimodal understanding while delivering strong performance in image generation and editing. Its native support for interleaved generation and reasoning establishes a promising and scalable paradigm for next-generation unified foundation models. Codes and models are available at https://github.com/inclusionAI/LLaDA2.0-Uni. |
| title | LLaDA2.0-Uni: Unifying Multimodal Understanding and Generation with Diffusion Large Language Model |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2604.20796 |