Releases: materialsvirtuallab/megnet
Releases · materialsvirtuallab/megnet
v1.2.2
v1.2.2
- Revising plotting of multifidelity work
v1.2.1 release
- Fix molecular graph to enable flexible ring size.
- Fix typo
v1.1.9
v1.1.9
- Add multi-fidelity model examples
- Add sample weights for model training
- Add default optimizer gradient norm clip
v1.1.8
v1.1.8
- Bug fix of megnet descriptors
v1.1.7
v1.1.7
- Update the model training mechanism and move Gaussian expansion to tensorflow, training speed up 100%
v1.1.6
v1.1.6
- minor fix to include more linear readout option
- add data type control
- refactor local_env
v1.1.5
v1.1.5
- Code refactor and reformat
- Fix tensorflow and numpy type compatibility issues
v1.1.4
v1.1.4
- Update to tensorflow.keras API instead of using keras
v1.1.3
v1.1.3
- Download mvl_models from figshare if not present