QSVD: Efficient Low-rank Approximation for Unified Query-Key-Value Weight Compression in Low-Precision Vision-Language Models

Ähnliche Einträge

• WSVD: Weighted Low-Rank Approximation for Fast and Efficient Execution of Low-Precision Vision-Language Models
  von: Wang, Haiyu, et al.
  Veröffentlicht: (2026)
• LASER: Loss-Aware Singular-value Decomposition and Rank Allocation for Efficient Low-Precision Vision-Language Models
  von: Wang, Haiyu, et al.
  Veröffentlicht: (2026)
• SLiM: One-shot Quantization and Sparsity with Low-rank Approximation for LLM Weight Compression
  von: Mozaffari, Mohammad, et al.
  Veröffentlicht: (2024)
• Loki: Low-rank Keys for Efficient Sparse Attention
  von: Singhania, Prajwal, et al.
  Veröffentlicht: (2024)
• MLorc: Momentum Low-rank Compression for Memory Efficient Large Language Model Adaptation
  von: Shen, Wei, et al.
  Veröffentlicht: (2025)