From: Diversity oriented Deep Reinforcement Learning for targeted molecule generation
 | DrugEx | REINVENT | ORGANIC | Best configuration |
---|---|---|---|---|
\({\varvec{\tau }}\) | – | – | – | 2 |
\({\varvec{\alpha }}\) | – | – | – | 0.7 |
\({\varvec{\lambda }}\) | – | – | – | 0.9 |
% Desirable | 98.5 | 98.2 | 99.8 | 95.1 |
% Unique | 99.1 | 95.8 | 94.8 | 92.9 |
% Validity | 99.0 | 98.8 | 99.8 | 90.4 |
Diversity | 0.74 | 0.75 | 0.67 | 0.86 |