blip

BLIP is a framework for training and fine-tuning machine learning interatomic potentials, providing reliable uncertainty estimates and minimal computational overhead. It allows users to convert existing machine learning interatomic potentials into Bayesian models, enhancing their robustness and interpretability.