1
CoCurve: Cross-Module Co-Pruning Curvature for Training-Free Structured LLM Pruning
无需训练即可实现结构化LLM剪枝，跨模块协同曲率方法大幅降低模型冗余。
arXiv:2607.17568v1 Announce Type: cross Abstract: Structured pruning compresses large language models (LLMs) by removing whole computational units, su…