ml-jku / mhn-react Goto Github PK

Hello, your work is of great help to my research. Now I have a new data set and want to use this model, but the effect is not very good. May I ask which parameters play a key role and influence on the model performance? Looking forward to your reply .

Thank you for providing all the code and baselines!
I am trying to reproduce the results but my Segler model for retrosynthesis is about 2% worse than the results in the paper. This is not critical for our work, but I thought I'd give it a try... I used the parameters you mention in the supplementary of the paper but there is still room for interpretation (e.g., number of epochs). Could you maybe provide all the parameters you used? Thank you so much already!

PS In case you planned to update the license, it is still in state "reviewing" :)