UniWeTok: An Unified Binary Tokenizer with Codebook Size $\mathit{2^{128}}$ for Unified Multimodal Large Language Model

Documenti analoghi

• BitDance: Scaling Autoregressive Generative Models with Binary Tokens
  di: Ai, Yuang, et al.
  Pubblicazione: (2026)
• WeTok: Powerful Discrete Tokenization for High-Fidelity Visual Reconstruction
  di: Zhuang, Shaobin, et al.
  Pubblicazione: (2025)
• BitLM: Unlocking Multi-Token Language Generation with Bitwise Continuous Diffusion
  di: Zhuang, Shaobin, et al.
  Pubblicazione: (2026)
• UniFlow: A Unified Pixel Flow Tokenizer for Visual Understanding and Generation
  di: Yue, Zhengrong, et al.
  Pubblicazione: (2025)
• UniCode$^2$: Cascaded Large-scale Codebooks for Unified Multimodal Understanding and Generation
  di: Chen, Yanzhe, et al.
  Pubblicazione: (2025)