gSDE uses a fixed noise throughout the trajectory to avoid having erratic behaviours of the actor.
gSDE can be used with SAC, DDPG, PPO and others. We should think in advance how implementing this would modify the workflow, what implementation changes would be needed, what new class should be created etc.
vmoens
commented
2 years ago
gSDE uses a fixed noise throughout the trajectory to avoid having erratic behaviours of the actor. gSDE can be used with SAC, DDPG, PPO and others. We should think in advance how implementing this would modify the workflow, what implementation changes would be needed, what new class should be created etc.