1
SigmaScale: LLM Compression with SVD-based Low-Rank Decomposition and Learned Scaling Matrices
用SVD低秩分解加学习缩放矩阵,实现高效LLM压缩,精准降维保性能。
arXiv:2606.07098v1 Announce Type: cross Abstract: We present SigmaScale, a method for learning auxiliary scaling matrices $S$ to aid truncated Singula…