I am training an Agent using Soft Actor Critic, and this is the performance after 12k steps of training (yes it’s a little violent rn but I will either train for longer, increase frequency or implement imitation learning)