*M3GNet (formation-energy model) *CIF structures (cubic / tetragonal LLZO) *Python scripts using pymatgen + matgl *GPU inference (PyTorch CUDA)
What I learned
- How crystal structures become graphs
- How GNN models make energy predictions
- How small experiments can still reveal stability differences
- How to survive dependency errors and environment chaos
#cif file source: materialsproject.org
