JARVIS-ML is a repository of machine learning (ML) model parameters, descriptors, and ML related input and target data. JARVIS-ML is a part of the NIST-JARVIS project (https://jarvis.nist.gov/). Additional resources available include:
JARVIS-ML introduced Classical Force-field Inspired Descriptors (CFID) as a universal framework to represent a material’s chemistry-structure-charge related data. With the help of CFID and JARVIS-DFT data, several high-accuracy classifications and regression ML models were developed, with applications to fast materials-screening and energy-landscape mapping. Some of the trained property models include formation energies, exfoliation energies, bandgaps, magnetic moments, refractive index, dielectric, thermoelectric, and maximum piezoelectric and infrared modes. Also, several ML interpretability analysis revealed physical-insights beyond intuitive understandings in materials-science. These models, the workflow, dataset etc. are disseminated to enhance the transparency of the work. Recently, JARVIS-ML expanded to include STM-image ML models, as they are reported to directly accelerate experiments. Graph convolution neural network models are being developed for the image and crystal-structures.