Alby M.

OK yeah, back of the envelope that is just a lot of data (assuming all-to-all edges for 50, which is a bit of an exaggeration but probably not much at...

Hi @peastman , That would be great! Obviously caveat (2) still applies, but this would be a good starting point. You'd just replace the `Data(...)` constructor with `AtomicData.from_points`, more or...

Hi @peastman , I'm suggesting that you write a new subclass of `AtomicDataset`. All datasets are expected to be subclasses of `AtomicDataset`, and arbitrary dataset classes with arbitrary `__init__` arguments...

@peastman oh nice! Yeah, the ASE neighborlists are not that fast... you could always try to play with the `dataloader_num_workers` YAML setting. I've never really used it set to anything...

Hi @JonathanSchmidt1 , Thanks for your interest in our code/method for your project! Sounds like an interesting application; please feel free to get in touch by email and let us...

OK, I've merged the latest `develop` -> `horovod`, see https://github.com/mir-group/nequip/pull/211.

If you try this, please run the Horovod unit tests `tests/integration/test_train_horovod.py` and confirm that they (1) are not skipped (i.e. horovod is installed) and (2) pass.

Hi @JonathanSchmidt1 , Surprised that the tests run if the training won't... that sounds like a sign that the tests are broken 😄 Whoops yes I forgot to mention, I...

Aha... here's that "this is very untested" 😁 I think PyTorch / Horovod may be too smart for its own good and reloading transmitted tensors onto different CUDA devices when...

Hi @choant , This is almost correct--- the key is that the parity in the _irrep_ does not describe the sign of the data itself, but rather how that sign...