GRASP: Replace Redundant Layers with Adaptive Singular Parameters for Efficient Model Compression

Documenti analoghi

• Dynamic Attention-Guided Context Decoding for Mitigating Context Faithfulness Hallucinations in Large Language Models
  di: Huang, Yanwen, et al.
  Pubblicazione: (2025)
• Basis Sharing: Cross-Layer Parameter Sharing for Large Language Model Compression
  di: Wang, Jingcun, et al.
  Pubblicazione: (2024)
• Astra: Activation-Space Tail-Eigenvector Low-Rank Adaptation of Large Language Models
  di: Liu, Kainan, et al.
  Pubblicazione: (2026)
• SVFit: Parameter-Efficient Fine-Tuning of Large Pre-Trained Models Using Singular Values
  di: Sun, Chengwei, et al.
  Pubblicazione: (2024)
• xKV: Cross-Layer KV-Cache Compression via Aligned Singular Vector Extraction
  di: Chang, Chi-Chih, et al.
  Pubblicazione: (2025)