AI RESEARCH
UniCom: Unified Multimodal Modeling via Compressed Continuous Semantic Representations
arXiv CS.CV
•
ArXi:2603.10702v1 Announce Type: new Current unified multimodal models typically rely on discrete visual tokenizers to bridge the modality gap. However, discretization inevitably discards fine-grained semantic information, leading to suboptimal performance in visual understanding tasks. Conversely, directly modeling continuous semantic representations (e.g., CLIP, SigLIP) poses significant challenges in high-dimensional generative modeling, resulting in slow convergence and