LaViDa-R1: Advancing Reasoning for Unified Multimodal Diffusion Language Models
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917275538292736 |
|---|---|
| author | Li, Shufan Zhu, Yuchen Gu, Jiuxiang Liu, Kangning Lin, Zhe Chen, Yongxin Tao, Molei Grover, Aditya Kuen, Jason |
| author_facet | Li, Shufan Zhu, Yuchen Gu, Jiuxiang Liu, Kangning Lin, Zhe Chen, Yongxin Tao, Molei Grover, Aditya Kuen, Jason |
| contents | Diffusion language models (dLLMs) recently emerged as a promising alternative to auto-regressive LLMs. The latest works further extended it to multimodal understanding and generation tasks. In this work, we propose LaViDa-R1, a multimodal, general-purpose reasoning dLLM. Unlike existing works that build reasoning dLLMs through task-specific reinforcement learning, LaViDa-R1 incorporates diverse multimodal understanding and generation tasks in a unified manner. In particular, LaViDa-R1 is built with a novel unified post-training framework that seamlessly integrates supervised finetuning (SFT) and multi-task reinforcement learning (RL). It employs several novel training techniques, including answer-forcing, tree search, and complementary likelihood estimation, to enhance effectiveness and scalability. Extensive experiments demonstrate LaViDa-R1's strong performance on a wide range of multimodal tasks, including visual math reasoning, reason-intensive grounding, and image editing. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_14147 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | LaViDa-R1: Advancing Reasoning for Unified Multimodal Diffusion Language Models Li, Shufan Zhu, Yuchen Gu, Jiuxiang Liu, Kangning Lin, Zhe Chen, Yongxin Tao, Molei Grover, Aditya Kuen, Jason Computer Vision and Pattern Recognition Diffusion language models (dLLMs) recently emerged as a promising alternative to auto-regressive LLMs. The latest works further extended it to multimodal understanding and generation tasks. In this work, we propose LaViDa-R1, a multimodal, general-purpose reasoning dLLM. Unlike existing works that build reasoning dLLMs through task-specific reinforcement learning, LaViDa-R1 incorporates diverse multimodal understanding and generation tasks in a unified manner. In particular, LaViDa-R1 is built with a novel unified post-training framework that seamlessly integrates supervised finetuning (SFT) and multi-task reinforcement learning (RL). It employs several novel training techniques, including answer-forcing, tree search, and complementary likelihood estimation, to enhance effectiveness and scalability. Extensive experiments demonstrate LaViDa-R1's strong performance on a wide range of multimodal tasks, including visual math reasoning, reason-intensive grounding, and image editing. |
| title | LaViDa-R1: Advancing Reasoning for Unified Multimodal Diffusion Language Models |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2602.14147 |