v2024.10.30

• Faster ALIGNN-FF pretrained model with radius cutoff
• MD stability
• Added new colab notebooks
• Updated documentation for installation specially because of DGL GraphBolt issue

v2024.8.30

• matplotlib agg conflict fix

v2024.8.27

• Added MPtraj/mpf models using the 1.5 mil dataset as well as new dft_3d based models

v2024.5.27

• New default ALIGNN-FF model added

v2024.4.20

• Added LMDB feature
• Added distributed data parallel
• New notebooks added

v2024.4.10

• Added InterMat CBM/VBM models

v2024.3.24

• Add environment.yml file for conda based GitHub actions

v2024.3.4

• Removed pytorch-ignite dependency to enable conda-forge build
• Removed accelerate package, instead use torch.DataParallel
• Removed train_folder.py and train_folder_ff.py, instead use train_alignn.py
• Increase coverage by adding a phonon test

v2024.1.14

• Use torch deterministic algorithm.

v2023.08.01

• Added multiple ALIGNN-FF and Catalyst specific pretrained models
• Added options to save best and last models
• Updated restart model path option
• Added line graph on fly option