Fig. 2From: Efficient virtual high-content screening using a distance-aware transformer modelPredicting similarities between two molecules. The L2 norm is used to calculate the distance in latent space based on tokenized SMILES stringsBack to article page