zju-fangyin / kcl

Code and Data for the paper: Molecular Contrastive Learning with Chemical Element Knowledge Graph [AAAI 2022]

About me

• 🎓 I recently earned my Ph.D. degree in Computer Science from Zhejiang University in June 2024.

• 🌐 Do visit my personal homepage for more insights into my work.

• 📩 For further communication, please email me at:

Wishing you a fantastic day! 🥰

kcl's Issues

ImportError

作者您好，我按照您提供的版本信息配置环境，并直接使用了你8.58G的pretrain数据，但是运行bash finetune是出现以下错误importerrror

代码和数据

这是一项有意义的研究，期待代码和数据的更新~

关于使用知识图谱嵌入的问题

作者您好，我又来请教问题了- - 您在readme中Knowledge feature initialization部分提到如果想要自己训练，可以运行code/initial 的load.py，但是在该文件中是通过36行代码：
relation_emb, entity_emb = load('RotatE_128_64.pkl', total_entity2id=data.entity2id, total_relation2id=data.relation2id, e_dim=128, r_dim=64)
实现的，而这里的'RotatE_128_64.pkl'文件本身就包含了您使用化学周期表初始化的嵌入值，如下图

请问您能给出获得'RotatE_128_64.pkl'这个文件的例子吗，entity和relation的值是怎么得到的呢？

关于data的预处理

您好，在数据预处理阶段，运行data/graph_utils.py 应该会得到8.58G的mdb文件，但是我使用您最新的代码获得的mdb文件只有2.41G，数据量远远小于你之前的数据，请问这是什么原因呢？

Code and data missing

Please provide the code and data asap.

The file 'pretrain.py' and 'Set2Set_0910_2302_78000th_epoch.pkl' are missing.

The KEY python script pretrain.py is missing.
I found a file named pretrain.py in code/data/pretrain.py. But obviously, it is not the one for pre-training the model.

In addition, could you provide the pre-trained model weightSet2Set_0910_2302_78000th_epoch.pkl so that I can start with fine-tuning?

KCL/code/script/finetune.sh

Line 8 in 004f568

--readout_path ./dump/Pretrain/gnn-kmpnn-model/Set2Set_0910_2302_78000th_epoch.pkl \

请问可以给出一个适用于回归任务的示例吗？

您好，请问可以给出一个适用于回归任务的示例吗？
类似于分类任务所给出的这种示例

Ref: #5

Status as of today:
Releases
No releases published

code and data

please provide code and data

Please release the code (and data)

Thanks for making the code (and data) available. Github now only provides a zip file to download if the code is released. Please release the code. Thanks.

预训练模型

里面的Set2Set和weightedsum是什么模型

#2 Code and data

Please release asap. It has been four months since your paper was published. Time is of essence. Thanks!