Expressivity and Generalization: Fragment-Biases for Molecular GNNs

Open in new window