Ignore model hyperparameters when saving because they are saved at checkpointing

atomistic-machine-learning / schnetpack

SchNetPack - Deep Neural Networks for Atomistic Systems

Other

774 stars 214 forks source link

Ignore model hyperparameters when saving because they are saved at checkpointing #600

Closed Maltimore closed 8 months ago

Maltimore commented 8 months ago

This avoids a warning during training, and probably saves some disk space. I just ran a training and afterwards an MD with this patch applied and didn't observe problems.