Fig. 4From: Diversity oriented Deep Reinforcement Learning for targeted molecule generationGeneral schema of RNN-based Predictor architecture. SMILES encoding is transformed into an integer vector to be used as inputBack to article page