scaling-matrices

#scaling-matrices

SigmaScale: LLM Compression with SVD-based Low-Rank Decomposition and Learned Scaling Matrices

arXiv cs.CL ↗ · 5d ago Cached

Introduces SigmaScale, a method that learns auxiliary scaling matrices for SVD-based LLM compression, showing competitive performance on Llama 3.1 8B and Qwen3-8B benchmarks.

0 favorites 0 likes

scaling-matrices

SigmaScale: LLM Compression with SVD-based Low-Rank Decomposition and Learned Scaling Matrices

Submit Feedback