Oddin: Latent Visual Reasoning in Unified Multimodal Models via Self-Distillation from Image Generation
Overview & Synopsis
Unified large multimodal models (ULMMs) are able to perform visual reasoning through interleaved text-image generation, but generating explicit intermediate images incurs significant computational overhead. While visual latent reasoning addresses this, most existing methods rely on vision-language models (VLMs) incapable of visual generation. We propose Oddin, a novel framework enabling latent visual reasoning in ULMMs. Oddin replaces intermediate images with compact latent tokens processed entirely by the understanding branch, reserving the generation branch for the final output image. We train the model via offline self-distillation with three objectives, final image generation losses, text distillation losses, and cross-entropy losses for special tokens. Oddin surpasses interleaved generation baselines and closed-source VLMs on average across seven VBVR-Pro visual reasoning tasks. Qualitative analysis reveals that latent tokens attend to relevant regions in an input image, demonstrating that ULMMs can perform complex visual reasoning without generating intermediate images.
arxiv.org)
← / → for pages • + / - for zoom • 0 to fit width
Powered by Mozilla PDF.js v3.11
Document Information
| Document Title | Oddin: Latent Visual Reasoning in Unified Multimodal Models via Self-Distillation from Image Generation |
|---|---|
| Author / Organization | Jaeyong Ko, Hankyeol Kim, Pilsung Kang |
| Subject & Discipline | cs.CV |
| Categories | Science & Mathematics |
| Publication Year | 2026 |
| License / Rights | arXiv Open Access / Creative Commons |
| Resource Verification | Listed PDF (Verification Pending) |
| Original Host Website |
arXiv
(arxiv.org)
|
| File Format | application/pdf (Portable Document Format) |
| Language | EN |
| Catalog ID | DOC-273075 |
| Original Source Link | Visit Source Page on arxiv.org » |