def model_training(self) -> TorchModelTraining:
if not hasattr(self, "_model_training"):
class_ = load_attr(self.model_task_class, Type[TorchModelTraining])
self._model_training = load_torch_model_training_from_task_id(
class_, self.model_task_id)
return self._model_training
def project_config(self) -> ProjectConfig:
if not hasattr(self, "_project_config"):
self._project_config = deepcopy(
load_attr(self.project, ProjectConfig))
if (self.loss_function == "crm"
and self.project_config.propensity_score_column_name
not in self.project_config.auxiliar_output_columns):
self._project_config.auxiliar_output_columns = [
*self._project_config.auxiliar_output_columns,
Column(self.project_config.propensity_score_column_name,
IOType.NUMBER),
]
return self._project_config
def get_direct_estimator(self, extra_params: dict) -> TorchModelTraining:
assert self.direct_estimator_class is not None
estimator_class = load_attr(self.direct_estimator_class,
Type[TorchModelTraining])
attribute_names = get_attribute_names(estimator_class)
params = {
key: value
for key, value in self.model_training.param_kwargs.items()
if key in attribute_names
}
return estimator_class(**{**params, **extra_params})
def create_agent(self) -> BanditAgent:
bandit_class = load_attr(self.bandit_policy_class, Type[BanditPolicy])
bandit = bandit_class(reward_model=self.get_trained_module(),
**self.bandit_policy_params)
return BanditAgent(bandit)
def module_class(self) -> Type[RecommenderModule]:
if not hasattr(self, "_module_class"):
self._module_class = load_attr(self.recommender_module_class,
Type[RecommenderModule])
return self._module_class