KV-Efficient VLA: A Method to Speed up Vision Language Models with RNN-Gated Chunked KV Cache

Documenti analoghi

• Cross-Self KV Cache Pruning for Efficient Vision-Language Inference
  di: Pei, Xiaohuan, et al.
  Pubblicazione: (2024)
• KVCapsule: Efficient Sequential KV Cache Compression for Vision-Language Models with Asymmetric Redundancy
  di: Huang, Yingbing, et al.
  Pubblicazione: (2026)
• STaR-KV: Spatio-Temporal Adaptive Re-weighting for KV Cache Compression in GUI Vision-Language Models
  di: Han, Yuhang, et al.
  Pubblicazione: (2026)
• AirCache: Activating Inter-modal Relevancy KV Cache Compression for Efficient Large Vision-Language Model Inference
  di: Huang, Kai, et al.
  Pubblicazione: (2025)
• MuKV: Multi-Grained KV Cache Compression for Long Streaming Video Question-Answering
  di: Xiao, Junbin, et al.
  Pubblicazione: (2026)