AI RESEARCH
SERQ: Saliency-Aware Low-Rank Error Reconstruction for LLM Quantization
arXiv CS.LG
•
ArXi:2603.08185v1 Announce Type: new Post-