English

UniCode$^2$: Cascaded Large-scale Codebooks for Unified Multimodal Understanding and Generation

Computer Vision and Pattern Recognition 2025-07-09 v2 Multimedia

Abstract

Unified multimodal large language models (MLLMs) have shown promise in jointly advancing multimodal understanding and generation, with visual codebooks discretizing images into tokens for autoregressive modeling. Existing codebook-based methods either rely on small vocabularies (~16K entries) that lack fine-grained semantics or naively scale up, resulting in low token utilization and unstable training. We propose UniCode2^2, a cascaded codebook framework enabling large-scale, semantically aligned, and stable visual tokenization. By clustering millions of SigLIP sequence embeddings, we build a 500K-entry codebook that preserves vision-language alignment while expanding capacity. Stability is ensured via a cascaded design: a frozen codebook anchors the embedding space, and a trainable codebook refines task-specific semantics. This decoupling promotes high utilization and robust learning. Moreover, the alignment of our visual tokens with textual semantics enables seamless integration with pretrained diffusion decoders, supporting high-quality visual synthesis with minimal adaptation. UniCode^2 delivers strong performance across diverse benchmarks, demonstrating the viability of scaling visual token spaces without sacrificing stability, semantics, or modularity.

Keywords

Cite

@article{arxiv.2506.20214,
  title  = {UniCode$^2$: Cascaded Large-scale Codebooks for Unified Multimodal Understanding and Generation},
  author = {Yanzhe Chen and Huasong Zhong and Yan Li and Zhenheng Yang},
  journal= {arXiv preprint arXiv:2506.20214},
  year   = {2025}
}

Comments

19 pages, 5 figures

R2 v1 2026-07-01T03:32:39.693Z