personalitylinmult.inference.registry
=====================================

.. py:module:: personalitylinmult.inference.registry

.. autoapi-nested-parse::

   Resolve a model *name* to its checkpoint and load a predictor.

   Deployment surfaces (a Gradio dropdown, a CLI ``--model`` flag) select a model by
   a short name — ``"AVT"`` or ``"A"`` — rather than a checkpoint path. This module
   holds that name-to-checkpoint mapping and hands back a ready
   :class:`~personalitylinmult.inference.predictor.FiPredictor`, caching each so a
   repeated selection does not reload from disk.



Attributes
----------

.. autoapisummary::

   personalitylinmult.inference.registry.logger


Classes
-------

.. autoapisummary::

   personalitylinmult.inference.registry.ModelSpec
   personalitylinmult.inference.registry.ModelRegistry


Module Contents
---------------

.. py:data:: logger

   Module-level logger.

.. py:class:: ModelSpec

   One registered model.

   :param name: Short selector, e.g. ``"AVT"`` or ``"A"``.
   :type name: str
   :param checkpoint: Path to its ``.ckpt``.
   :type checkpoint: Path
   :param description: One-line human summary shown in a UI.
   :type description: str


.. py:class:: ModelRegistry(specs: list[ModelSpec])

   A name-keyed set of models, loaded on demand and cached.

   :param specs: The registered models.
   :type specs: list[ModelSpec]

   :raises ValueError: If two specs share a name.


   .. py:property:: names
      :type: list[str]


      The registered model names, in insertion order.


   .. py:method:: spec(name: str) -> ModelSpec

      Return the spec for a name.

      :param name: A registered model name.
      :type name: str

      :returns: Its spec.
      :rtype: ModelSpec

      :raises KeyError: If the name is not registered.



   .. py:method:: load(name: str) -> personalitylinmult.inference.predictor.FiPredictor

      Load (or return the cached) predictor for a name.

      :param name: A registered model name.
      :type name: str

      :returns: The predictor.
      :rtype: FiPredictor

      :raises KeyError: If the name is not registered.
      :raises FileNotFoundError: If its checkpoint is missing.



   .. py:method:: from_checkpoints(checkpoints: dict[str, pathlib.Path | str]) -> ModelRegistry
      :classmethod:


      Build a registry from a ``{name: checkpoint_path}`` mapping.

      :param checkpoints: Name-to-path mapping.
      :type checkpoints: dict[str, Path | str]

      :returns: The registry (paths are not opened until
                :meth:`load`, so a missing file surfaces at load time, not here).
      :rtype: ModelRegistry