v0.3.7

[ilyes319]

Main new features

• Finetuning with multiheads : https://mace-docs.readthedocs.io/en/latest/guide/finetuning.html
• Training with multiheads: https://mace-docs.readthedocs.io/en/latest/guide/multihead_training.html

What's Changed

• Add Intel XPU device by @jharrymoore in Add Intel XPU device  #340
• Intel by @ilyes319 in Intel #356
• Fix logic that deals with IsolatedAtom and 1-atom configuration by @bernstei in Fix logic that deals with IsolatedAtom and 1-atom configuration #393
• Fix bug that overwrote REF_* keys when those were the explicitly specified keys for the training reference quantities by @bernstei in Fix bug that overwrote REF_* keys when those were the explicitly specified keys for the training reference quantities #403
• Fix check for patience that fails if swa start was not specified by @bernstei in Fix check for patience that fails if swa start was not specified #405
• fix incorrect conversion of forces from calculator to atoms.arrays for finetuning pbe mp replay head by @bernstei in fix incorrect conversion of forces from calculator to atoms.arrays for finetuning pbe mp replay head #409
• Better fix for PR Fix check for patience that fails if swa start was not specified #405, fix patience check when swa is not active by @bernstei in Better fix for PR #405, fix patience check when swa is not active #411
• fix np.random.choice issue when subselecting from pre-trained configs for multihead fine tune by @bernstei in fix np.random.choice issue when subselecting from pre-trained configs for multihead fine tune #448
• validation loss printed during training is not actually for the specific head by @bernstei in validation loss printed during training is not actually for the specific head #449
• get rid of all stress/n_atoms by @bernstei in get rid of all stress/n_atoms #443
• Stage two weights by @vue1999 in Stage two weights #522
• remove mypy from checks by @ilyes319 in remove mypy from checks #529
• Fix MAE table for universal loss by @ilyes319 in Fix MAE table for universal loss #548
• Create py.typed by @Andrew-S-Rosen in Create py.typed #554
• Improve Log Details by @vue1999 in Improve Log Details #553
• Backward the log changes by @ilyes319 in Backward the log changes #571
• Merge develop, pylint + log changes by @ilyes319 in Merge develop, pylint + log changes #572
• Multihead finetuning and training by @ilyes319 in Multihead finetuning and training #575
• multihead finetuning by @ilyes319 in multihead finetuning #578
• Fix bug about undefined swa by @Hongyu-yu in Fix bug about undefined swa #584
• Fix for Issue 596 by @stenczelt in Fix for Issue 596 #597
• Split download_mace_mp_checkpoint out of mace_mp by @janosh in Split download_mace_mp_checkpoint out of mace_mp #602
• Allow creation of MACECalculator without needing to write a checkpoint to disk. by @CompRhys in Allow creation of MACECalculator without needing to write a checkpoint to disk. #535
• fix swa bug and remove e3nn fixed version by @ilyes319 in fix swa bug and remove e3nn fixed version #589
• Change type hint syntax in MACECalculator to work with python39 by @bernstei in Change type hint syntax in MACECalculator to work with python39 #618
• fix swa bigger than epoch by @ilyes319 in fix swa bigger than epoch #616
• add option to return raw model in mace_mp by @ilyes319 in add option to return raw model in mace_mp #619

New Contributors

• @Hongyu-yu made their first contribution in Fix bug about undefined swa #584
• @CompRhys made their first contribution in Allow creation of MACECalculator without needing to write a checkpoint to disk. #535

Full Changelog: v0.3.6...v0.3.7

---

This discussion was created from the release v0.3.7.

[nihil39]

Thanks for the new version!
Is it normal that in the pip repository (I don't know how to call it) the latest version is still 0.3.6?

https://pypi.org/project/mace-torch/

[ilyes319]

Hey @nihil39, I still need to fix some bugs flagged by @bernstei before merging. Hopefully will happen soon. Please use the main branch and report any bug you find.