lit_sb3.agents.on_policy

source module lit_sb3.agents.on_policy

OnPolicyAgent — base for advantage-based agents (PPO, A2C, TRPO, …).

On-policy agents receive mini-batches from a pre-collected rollout buffer. One call to :meth:compute_loss processes a single mini-batch.

configure_optimizers is made concrete here: every on-policy agent uses exactly [self.policy.optimizer]. PPO and A2C are identical in this regard.

Classes

• OnPolicyAgent — Abstract base for on-policy agents.

source class OnPolicyAgent(policy: BasePolicy)

Bases : BaseAgent

Abstract base for on-policy agents.

Provides a concrete :meth:configure_optimizers (single policy optimizer). Subclasses must implement :meth:compute_loss.

Methods

• configure_optimizers — Return the single policy optimizer.

• compute_loss — Compute on-policy loss for one rollout mini-batch.

source method OnPolicyAgent.configure_optimizers() → OptimizerLRScheduler

Return the single policy optimizer.

source method OnPolicyAgent.compute_loss(batch: Any) → dict[str, torch.Tensor]

Compute on-policy loss for one rollout mini-batch.