Fig. 8From: Conditional reduction of the loss value versus reinforcement learning for biassing a de-novo drug design generatorBiassing G using RL for objective 1: a trying different k episodes, same learning rate used for G of 0.0005, and default random seeds. b trying different learning rates, k = 20, and default random seeds. c trying different random seeds, k = 20 and a learning rate of 0.00005Back to article page