tf_agents.agents.ppo.ppo_utils.nested_kl_divergence

Given two nested distributions, sum the KL divergences of the leaves.

tf_agents.agents.ppo.ppo_utils.nested_kl_divergence(
    nested_from_distribution: tf_agents.typing.types.NestedDistribution,
    nested_to_distribution: tf_agents.typing.types.NestedDistribution,
    outer_dims: Sequence[int] = ()
) -> tf_agents.typing.types.Tensor