LARC advances continuous-space reasoning in multimodal models
Original title: Latent Reasoning in Continuous Space for Unified Multimodal Models
LARC reasons over continuous hidden states to fix token bottlenecks, boosting image generation quality and test-time scaling on GenEval.
Unified Multimodal Models fail to leverage chain-of-thought for image generation due to discrete token vs continuous image space mismatch. LARC uses curriculum supervised fine-tuning followed by information-gain reinforcement learning. Experiments show consistent gains in generation quality, prompt following, and compositional fidelity across text-to-image benchmarks.
- 7% GenEval improvement over BAGEL
- Demonstrates test-time scaling
- Two-stage training: curriculum SFT then RL
- Applies to diverse multimodal tasks
From the original article
1 min read at huggingface.co
Unified Multimodal Models (UMMs) achieve remarkable success on diverse multimodal understanding and generation tasks, yet they still struggle to effectively leverage language-based chain-of-thought reasoning for image generation. As a result, existing reasoning-augmented UMMs often fail to inherit the test-time scaling and fine-grained controllability observed in large language models.
By Byungwoo Jeon, Yoonwoo Jeong, Hyunseok Lee · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.