Language

unilab.algos.torch.hora.sac.HoraSACRuntime¶

class unilab.algos.torch.hora.sac.HoraSACRuntime[source]¶

Bases: OffPolicyRuntime

Resolved HORA-SAC hooks consumed by the generic off-policy script.

Parameters:

learner_cls (type[Any] | None)
algo_type (str | None)
actor_kwargs (dict[str, Any])
supports_symmetry (bool)
actor_cfg (dict[str, Any])

Methods

`__init__`([learner_cls, algo_type, ...])
`build_model_kwargs`(*, obs_dim, critic_obs_dim)	Build shared learner/collector actor kwargs for the HORA-SAC actor.

Attributes

`algo_type`
`supports_symmetry`
`actor_cfg`
`actor_kwargs`

learner_cls¶: alias of HoraSACLearner

algo_type: str | None = 'hora_sac'¶

supports_symmetry: bool = False¶

actor_cfg: dict[str, Any]¶

build_model_kwargs(*, obs_dim, critic_obs_dim)[source]¶

Build shared learner/collector actor kwargs for the HORA-SAC actor.

Parameters:

obs_dim (int)
critic_obs_dim (int)

Return type:

dict[str, Any]

__init__(learner_cls=<class 'unilab.algos.torch.hora.sac_learner.HoraSACLearner'>, algo_type='hora_sac', actor_kwargs=<factory>, supports_symmetry=False, actor_cfg=<factory>)¶

Parameters:

learner_cls (type[Any] | None)
algo_type (str | None)
actor_kwargs (dict[str, Any])
supports_symmetry (bool)
actor_cfg (dict[str, Any])

actor_kwargs: dict[str, Any]¶