Actor-Critic Methods

Asynchronous Advantage Actor-Critic

Distributed architecture where multiple agents train in parallel on independent environments, periodically sharing their gradients to accelerate learning.

← Wstecz