Xuhan Liu

@cthoyt Because the ZINC database is so big, I cannot provide all of the files here. If you want to run it, you have to download all of the ZINC...

I am afraid that it is not possible to provide the model publicly because of it big size of model file. If you want it you can send email to...

I guess it is caused by the classification model in your environment. self.clf should by a scikit-learn based machine learning model, not int type. Before you run the "agent.py", you...

I am very happy that you are interested in my project. This paper will be accepted by Journal of Cheminformatics soon. Frankly speaking, it is not easy to provide the...

By the way, the version 3 will be released soon. It is more fantastic model that the previous version. I hope you can look forward to it a little bit.

I am afraid that you have to downgrade the version of RDKit. Because in the previous version, I always use Chem but not Allchem module to invoke MolFromSmiles().

PyCharm did no remind me of this error. I have changed in the "environ.py". Thank you very much!

I checked this part of code in my own machine (PyTorch = 1.0 or 1.3 stable), but I did not find such a bug. Could you tell me the version...

Thanks for reminding me this bug. It is caused by the duplication of 'EOS' in vocabulary defined in the file of 'voc.txt'. I have modified the 'util.py' and removed the...

Currently, the "multi_task" method is only compatible with DNN (i.e. MT_DNN) methods. For the MT_DNN algorithm, it can be implemented with DNN model in the "multi_task" method. For the other...