Skip to content
arXiv cs.CV · Papers

Symbiotic-MoE: Unlocking the Synergy between Generation and Understanding

arXiv:2604.07753v2 Announce Type: replace Abstract: Empowering Large Multimodal Models (LMMs) with image generation often leads to catastrophic forgetting in understanding tasks due to severe gradient conflicts. While existing paradigms like Mixture-of-Transformers (MoT) mitigate this conflict through structural isolat