ML-PEG (Machine Learning Performance and Extrapolation Guide) Github Repository

Repository to locally host a ML-PEG (Machine Learning Performance Guide) application as an interactive dashboard. ML-PEG is a comprehensive benchmarking framework and interactive performance guide for evaluating Machine Learning Interatomic Potentials (MLIPs) across diverse systems and properties beyond only energies and forces. The interactive performance guide, allowing users to explore and compare MLIP performance and deep dive into errors, connecting performance (or the lack of) to the underlying chemistry and physics. Please note that this code is currently available as an alpha release which is still under development.