lit_sb3.agents.off_policy

source module lit_sb3.agents.off_policy

OffPolicyAgent — base for replay-buffer-based agents (SAC, TD3, DQN, …).

Off-policy agents learn from transitions stored in a replay buffer. configure_optimizers is kept abstract because the optimizer set varies significantly across algorithms (SAC has 3, TD3 has 2, DQN has 1).

Classes

• OffPolicyAgent — Abstract base for off-policy agents.

source class OffPolicyAgent(policy: BasePolicy)

Bases : BaseAgent

Abstract base for off-policy agents.

Subclasses must implement both :meth:configure_optimizers and :meth:compute_loss.

Methods

• configure_optimizers — Return the list of optimizers (varies: SAC=3, TD3=2, DQN=1).

• compute_loss — Compute off-policy loss for one replay-buffer batch.

source method OffPolicyAgent.configure_optimizers() → OptimizerLRScheduler

Return the list of optimizers (varies: SAC=3, TD3=2, DQN=1).

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

Compute off-policy loss for one replay-buffer batch.