Initialize the FQL model with behavior cloning flow, actor, critic, and target critic. Args: bc_flow (ReFlow): Behavior cloning flow model. actor (OneStepActor): Actor network for action generation. critic (CriticObsAct): Critic network for Q-value estima
(self,
bc_flow: ReFlow,
actor: OneStepActor,
critic: CriticObsAct,
inference_steps: int,
normalize_q_loss: bool,
device)
source not stored for this graph (policy: none)