57544b1ff9f97d4da9f64d25c8ea5a3d8d247ffc,rllib/examples/centralized_critic.py,,,#,183

Before Change

            "num_workers": 0,
            "multiagent": {
                "policies": {
                    "pol1": (None, Discrete(6), TwoStepGame.action_space, {}),
                    "pol2": (None, Discrete(6), TwoStepGame.action_space, {}),
                },
                "policy_mapping_fn": lambda x: "pol1" if x == 0 else "pol2",

After Change

    postprocess_fn=centralized_critic_postprocessing,
    loss_fn=loss_with_central_critic,
    before_init=setup_mixins,
    mixins=[
        TorchLR, TorchEntropyCoeffSchedule, TorchKLCoeffMixin,
        CentralizedValueMixin
    ])


def get_policy_class(config):

Instances