Fig. 5: Molecular performance in RL process.
From: Token-Mol 1.0: tokenized drug design with large language models
a Key metrics such as reward score, Vina score, QED, and SA score during the process of RL. The 2D structure of molecules from the different stages in the RL are also displayed. b The binding modes of selected molecules predicted with QVina2. c Change trends in the predicted affinities of selected molecules for their respective targets using different docking methods. Source data in (a) are provided as a Source Data file.
