Muhan Zhang

Results 65 comments of Muhan Zhang

> 文中没有提到loss function和如何训练,具体是怎么训练的呢? 具体loss function和训练过程请参照图分类部分的[pytorch_DGCNN](https://github.com/muhanzhang/pytorch_DGCNN)代码。

Hi! --only-predict requires you have used --save-model to save a trained model to disk. Your other understanding is correct.

你好。util 文件来自pytorch_DGCNN,需要在使用SEAL前提前下载安装到跟SEAL相同level的文件夹下。你可以执行“bash ./install.sh” 来自动安装pytorch_DGCNN 以及其他所需要的包。谢谢!

Hi, please check out this issue: https://github.com/muhanzhang/SEAL/issues/17

Which networkx version did you use? Sometimes try to replace "g.degree" with "g.degree()" solves this issue.

Hi, please check out this issue: https://github.com/muhanzhang/SEAL/issues/10 And please try to first look for your question in existing issues. Many of them have been answered before. Thanks!

> '\r': command not found Hi,你是不是有在windows和linux之间拷贝过?可以参考这个https://blog.csdn.net/u010416101/article/details/80135293

Hi, the "install.sh" only runs on linux-based systems. If you run it on windows, you need to manually install the packages. However, I would recommend to directly use the latest...

Hi! You need to install pytorch_DGCNN. You can follow the README to type "bash ./install.sh" to install all required libraries.

Hi Xu, feat_dim is the dimension of the one-hot encoding of integer node tags (labels), while attr_dim is dimension of continuous node attribute features which is automatically [inferred ](https://github.com/muhanzhang/pytorch_DGCNN/blob/master/util.py#L139)from input...