Usage of multivariate gaussian model

[zoctipus]

I see that multivariate gaussian model is implemented as explained in doc

Yet in torch.Runner

self._class_mapping = {
            # model
            "gaussianmixin": gaussian_model,
            "deterministicmixin": deterministic_model,
            "shared": shared_model,
            # memory
            "randommemory": RandomMemory,
            # agent
            "ppo": PPO,
            "ippo": IPPO,
            "mappo": MAPPO,
            # trainer
            "sequentialtrainer": SequentialTrainer,
        }

multivariate_gaussian_model is not default provided,
I can import and add to mapping as "multivariategaussianmixin": multivariate_gaussian_model
then I can change the config['model']['policy']['class'] to multivariategaussianmixin
and program can run just fine

I wonder is there anything I need to be mindful for doing this, is there some detail that should be considered ? why isn't provided as default?

similarly, for algorithms like sac, trpo, ... , can I just import the respective algorithm and add in _class_mapping?

Thank you!