Some failures on the test set

[CQ-zhang-2016]

Wonderful work! I tried to reproduct the evaluation results. But when I tried to predict some samples in the test set, ResGen failed.

Here's what I did: use gen.py, then set the hyperparameters pdb_file and sdf_file to the file path of protein pocket-ligand pair, separately. ResGen returns 'generation failures' on some cases, have you ever encountered such problem and how could I solve it?

[OdinZhang]

Hi, I think you can lower down the probability filter value or use different checkpoints! It is a common issue for AR models!