FS-GNNTR

FS-GNNTR is a few-shot learning framework that utilizes graph embeddings and transformers to predict molecular properties. It addresses low-data challenges in molecular property prediction and demonstrates strong performance on datasets like Tox21 and SIDER.