Alby M.

I'd be inclined to not make this an extra option and instead just always do this, since the original behavior is an intuitive subset of the behavior this can give...

Hi @tgmaxson , The supported solution to this is to have all of the files concatenated into a single comprehensive dataset (which will get preprocessed once), always specify that dataset,...

This is a good point, thanks--let me think a little more about how this could integrate into other fixes to the dataset architecture.

In the new updated code, there is support for providing multiple validation and test datasets both in the root data module (https://nequip.readthedocs.io/en/latest/api/datamodule.html#nequip.data.datamodule.NequIPDataModule) and, as a convenience feature, in the helper...

You need `StressForceOutput` instead of `ForceOutput` as `pair_allegro` expects the virial to be computed. Please check that you are using the latest `main` branches of `nequip` and `pair_allegro` as well.

Hi @samueldyoung29ctr , Hm, odd. If you start a simulation in LAMMPS from exactly a frame that shows up in your training data, or the one that shows a stress...

> My build of LAMMPS is CPU-only due to resource constraints on the cluster I use. Aha. I don't think we've ever actually tested Kokkos pair_allegro on CPU, nor am...

Hi @OlfwayAdbayIgbay , Can you confirm which Allegro `pair_style` you are using in your LAMMPS script? And which branch of `pair_allegro` you are building, and whether you pulled its latest...

Hi all---this is indeed strange that you are still seeing this issue with what should have been a fix. I'm not sure why... we should have some more patches out...

I'm not sure of this exact error, but in general we do not support Windows systems except inside of Windows Subsystem for Linux (WSL).