torchrl.trainers.algorithms.configs.objectives.FQLLossConfig#
- class torchrl.trainers.algorithms.configs.objectives.FQLLossConfig(_partial_: bool = False, flow_policy: Any = None, actor_network: Any = None, qvalue_network: Any = None, num_qvalue_nets: int = 2, alpha: float = 10.0, q_aggregation: str = 'mean', normalize_q_loss: bool = False, reduction: str = 'mean', gamma: float = 0.99, _target_: str = 'torchrl.trainers.algorithms.configs.objectives.make_fql_loss', _convert_: str = 'object')[source]#
Hydra configuration for
FQLLoss.