AdaTok: Adaptive Token Compression with Object-Aware Representations for Efficient Multimodal LLMs

Documenti analoghi

• Chat-CBM: Towards Interactive Concept Bottleneck Models with Frozen Large Language Models
  di: He, Hangzhou, et al.
  Pubblicazione: (2025)
• Exploiting Inherent Class Label: Towards Robust Scribble Supervised Semantic Segmentation
  di: Zhang, Xinliang, et al.
  Pubblicazione: (2025)
• Bridging Information Asymmetry: A Hierarchical Framework for Deterministic Blind Face Restoration
  di: Yao, Zhengjian, et al.
  Pubblicazione: (2026)
• Inter- and Intra-image Refinement for Few Shot Segmentation
  di: Fu, Ourui, et al.
  Pubblicazione: (2025)
• V2C-CBM: Building Concept Bottlenecks with Vision-to-Concept Tokenizer
  di: He, Hangzhou, et al.
  Pubblicazione: (2025)