yueliu1999 / dink-net Goto Github PK

Hi, thanks for your awesome work. However, I met some problems when replicating your work. If I did not use the model parameters you provided, the results could not be aligned with the paper. How are these model parameters file trained? I am looking forward to your reply.

非常感谢作者这篇优秀的工作，有个问题想要请教下，在发布的代码中没有看到关于对ogbn-products和ogbn-papers100M数据集的配置(main.py的32-46行)，除此之外，代码中看到有train_batch和train_test两个参数，在跑四个小数据集时默认是false全图训练，如果跑大数据集的话是不是设置为true即可，期待并感谢作者的解答

Hello. In your paper you mentioned that this work "was unified into an end-to-end framework".
However, in your published code: 1) you directly use ogb-supplied features instead of text attributes; 2) your work includes an inevitable pre-training process.
Do you consider this work as "end-to-end" and why? Looking forward to your reply.

Thanks for the inovative and motivative work that the authors provide. However, it seems that this project has no implementation code or the details of implementation. Could the authors provide us with it?

您好，我尝试复现您代码中的预训练过程。按照您论文中appendix C. Design Details & Hyper-parameter Settings的设置来训练模型，但是效果不太好，例如训练cora数据集时，按照pretrain 200 epochs，lr 1e-3; finetune 200 epochs，lr 1e-2的协议，模型输出结果如下所示
预训练
epoch 009 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 019 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 029 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 039 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 049 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 059 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 069 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 079 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 089 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 099 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 109 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 119 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 129 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 139 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 149 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 159 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 169 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 179 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 189 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 199 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84

finetune
epoch 009 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 019 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 029 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 039 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 049 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 059 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 069 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 079 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 089 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 099 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 109 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 119 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 129 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 139 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 149 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 159 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 169 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 179 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 189 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84
epoch 199 ｜ acc:30.24 ｜ nmi:0.00 ｜ ari:0.00 ｜ f1:6.84

预训练和微调基本区别不大，是我哪里还没调试正确吗？谢谢~