All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
Most recent change on the bottom.
-e
option topatch_lammps.sh
- PyTorch 1.11 / fusion strategy support
- Full suite of unit tests
- edge duplication issue
- The mapping from LAMMPS to NequIP types is now explicitly provided
- Don't print full config in LAMMPS log
- Edges outside the cutoff but within the skin are no longer processed by the model
allow_tf32
and_jit_bailout_depth
are now respected- Hack to allow freezing on demand for PyTorch < 1.10