Xin Dai

I got NaN loss in the training as well. I used `--precision 16` option, which might lead to overflow in this case, see https://github.com/pytorch/pytorch/issues/40497 A closer inspection suggests that in...

@gahdritz Thanks for the reply! It looks like the overflow is a persistent bug in the mixed precision training and maybe related to deepspeed as well. I'll let you know...

Thanks for your reply! To clarify, I tries to search on PDB for metal ion binding: or both result in 87,669 entries. Do you also perform similar queries and compile...

By original dataset, do you mean all the PDB files? That would be too large I guess, so could you share the script used for search and annotate the PDB...

> I see, thanks for your sample code! I'll try to see if the results match with my aforementioned one.