Predicting LLM Compression Degradation from Spectral Statistics

Documenti analoghi

• Geometry is All You Need: A Unified Taxonomy of Matrix and Tensor Factorization for Compression of Generative Language Models
  di: Xu, Mingxue, et al.
  Pubblicazione: (2024)
• TensorSLM: Energy-efficient Embedding Compression of Sub-billion Parameter Language Models on Low-end Devices
  di: Xu, Mingxue, et al.
  Pubblicazione: (2025)
• TensorGPT: Efficient Compression of Large Language Models based on Tensor-Train Decomposition
  di: Xu, Mingxue, et al.
  Pubblicazione: (2023)
• How can representation dimension dominate structurally pruned LLMs?
  di: Xu, Mingxue, et al.
  Pubblicazione: (2025)
• The Lifecycle of the Spectral Edge: From Gradient Learning to Weight-Decay Compression
  di: Xu, Yongzhong
  Pubblicazione: (2026)