jieyibi / AMDKD

The implementation code of our paper "Learning Generalizable Models for Vehicle Routing Problems via Knowledge Distillation", accepted at NeurIPS2022.
MIT License
48 stars 10 forks source link

变量疑问 #5

Closed mengchengTang closed 7 months ago

mengchengTang commented 7 months ago

作者您好,最近在复现这篇经典的文章中,有一点疑问,在下面这行代码中: https://github.com/jieyibi/AMDKD/blob/8f51cabf7618b7abe37caf42fd5d2d44074ef384/AMDKD-AM/knowledge_distillation.py#L171 如果是采用AMDKD框架进行训练时,传入的distribution变量是不是应该是采样得到的class_type,而不是预先设定的opts.data_distribution? 期待回复,谢谢!

jieyibi commented 7 months ago

您好,Mengcheng。

是的,生成训练集之前少了一行opts.data_distribution=class_type。可能是我当时删除注释时不小心删掉了,代码已经更新了,谢谢您的提醒。之后我会重新再跑一遍AMDKD-AM,并上传其预训练模型参数。AMDKD-POMO版本不存在这个问题,您可以优先参考AMDKD-POMO,效果也比较好。

谢谢!

mengchengTang commented 7 months ago

好的,谢谢回答!