Code for downstream tasks

[zhfanrui]

Hi there,

Great work!

I’m having trouble reproducing the performance of the downstream tasks you presented. Could you please share the code for those tasks? It would be really helpful.

Thanks!

[toby-clark4]

Which task hasn't been working for you? I've been testing the model and to reproduce some of the results I think a non-default value for alpha in the elastic regression model is needed.

For example for the melting temperature prediction task I used alpha=0.001 (default in scikit-learn is 1) which gives an R² value of 0.75 vs 0.51 with default parameters.