返回全部动态
MergeSE：无需重训练的后训练模型合并工具
原标题：MergeSE: Post-Hoc Model Merging for Software Engineering Tasks Without Retraining
AI 摘要
arXiv 论文介绍了 MergeSE，一个用于软件工程任务的无训练模型合并工具，支持 CLI 和 Web 界面。该工具可合并 HuggingFace 编码器检查点，提供五种合并算法，并内置九种任务类型。实验表明，合并后的模型在未见克隆类型上泛化能力提升 4 倍，合并两个 1.24 亿参数模型在 CPU 上耗时不到 5 秒。
以上摘要由 AI 生成，可能存在误差。事实请以原文为准。
正文节选
Computer Science > Software Engineering Title:MergeSE: Post-Hoc Model Merging for Software Engineering Tasks Without Retraining View PDF HTML (experimental) Abstract:Fine-tuned code models often behave as domain specialists and can degrade sharply under distribution shift: in our clone-detection setting, a model trained on same-language clones drops 71\% F1 on cross-language clones, while multi-task training falls to 0.151 F1 on unseen AI-generated clones. Our companion study shows t
发布时间：2026-08-06 12:00
抓取时间：2026-08-06 21:49
来源机构：arXiv