You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
There are some situations where we do not need to load all the fields currently included in the BatchData instance returned by collate_conformers. For example, we do not need to load forces when training only on energies, need neither forces nor energies if we just want to run inference, do not need total charge for some of the models, etc. Should we subclass NNPInput and Metadata for different combinations of datafields?
The text was updated successfully, but these errors were encountered:
There are some situations where we do not need to load all the fields currently included in the
BatchData
instance returned bycollate_conformers
. For example, we do not need to load forces when training only on energies, need neither forces nor energies if we just want to run inference, do not need total charge for some of the models, etc. Should we subclassNNPInput
andMetadata
for different combinations of datafields?The text was updated successfully, but these errors were encountered: