Dreaming in Flow: Generative Grounding Feedback for Self-Evolving Unified Multimodal Models

arXiv:2609.08282v1 Announce Type: new Abstract: Unified multimodal models integrate visual understanding and generation within a single network, yet the two capabilities are commonly optimized as separate tasks. We introduce Generative Grounding Feedback(GGF), a self-evolving post-training framework that uses only text prompts and the model's own visual experience. Given a prompt, the model first generates a visual ``dream.'' Flow-level feedback compares text-, image-, and repair-conditioned predictions at the same noisy latent state, transferring image-grounded generation directions to the prompt condition. Dream replay grounding replays this dream through captioning and re-imagination, training claim-level evidence to remain consistent across the replay while separating unrelated visual experiences. Jointly optimized, these two directions let generation provide visual grounding for understanding and understanding refine subsequent generation without paired image--text supervision. Experiments across unified models with different understanding--generation integration designs show consistent improvements in text-to-image generation together with modest gains in visual understanding.

Related

• Breaking Dual Bottlenecks: Evolving Unified Multimodal Models into Self-Adaptive Interleaved Visual Reasoners
• Free Lunch for Unified Multimodal Models: Enhancing Generation via Reflective Rectification with Inherent Understanding
• Semantic Generative Tuning for Unified Multimodal Models
• Uni-Edit: Intelligent Editing Is A General Task For Unified Model Tuning
• IoU-PD: IoU-Aware Privileged Distillation for Visual Grounding with Multimodal Large Language Models

Source: arXiv cs.CV | 2026-09-09