A reproducibility study of atomistic line graph neural networks for materials property prediction

Use of machine learning has been increasingly popular in materials science as data-driven materials discovery is becoming the new paradigm. Reproducibility of findings is paramount for promoting transparency and accountability...