``glass_box_umap``
==================

.. py:module:: glass_box_umap


Subpackages
-----------
.. toctree::
   :titlesonly:
   :maxdepth: 1

   plotting/index.rst


Submodules
----------
.. toctree::
   :titlesonly:
   :maxdepth: 1

   jacobian/index.rst


Overview
--------

.. list-table:: Classes
   :header-rows: 0
   :widths: auto
   :class: summarytable

   * - :py:obj:`GlassBoxUMAP <glass_box_umap.GlassBoxUMAP>`
     - Glass Box UMAP model.
   * - :py:obj:`ParametricUMAP <glass_box_umap.ParametricUMAP>`
     - Parametric UMAP model.




Classes
-------
.. autoclass:: GlassBoxUMAP



   Bases: :py:obj:`glass_box_umap.parametric_umap.ParametricUMAP`



   .. rubric:: Methods:
   .. py:method:: compute_contributions(X: numpy.typing.NDArray[numpy.floating] | torch.Tensor, batch_size: int | None = None, reduction: Literal['l2'] | None = None) -> numpy.typing.NDArray[numpy.float32]

      Compute per-feature contributions to the embedding via Gradient x Input.

      Projects gradients back to raw feature space if PCA preprocessing was used.

      :param X: The input data (same format as passed to fit/transform).
                Shape: (n_samples, n_features).
      :param batch_size: Batch size for Jacobian computation. Defaults to ``self.batch_size``.
      :param reduction: How to reduce contributions across embedding dimensions. If ``"l2"``,
                        takes the L2 norm across components, returning shape
                        (n_samples, n_features). If ``None``, returns the full
                        (n_samples, n_components, n_features) array.

      :returns: Feature contributions array. Shape is (n_samples, n_components, n_features)
                when reduction is ``None``, or (n_samples, n_features) when a reduction
                is applied.


   .. py:method:: compute_jacobian(x: torch.Tensor, batch_size: int = 1024) -> torch.Tensor

      Compute the Jacobian of a model using ``vmap`` + ``jacrev`` with ``functional_call``.

      See :func:`glass_box_umap.jacobian.compute_jacobian` for details.



.. autoclass:: ParametricUMAP





   .. rubric:: Methods:
   .. py:method:: to(device: str | torch.device) -> typing_extensions.Self

      Move the model (if initialized) and update the target device.


   .. py:method:: fit(X: numpy.typing.NDArray[numpy.floating] | torch.Tensor) -> typing_extensions.Self


   .. py:method:: transform(X: numpy.typing.NDArray[numpy.floating] | torch.Tensor, batch_size: int | None = None) -> numpy.typing.NDArray[numpy.floating]


   .. py:method:: fit_transform(X: numpy.typing.NDArray[numpy.floating] | torch.Tensor) -> numpy.typing.NDArray[numpy.floating]


   .. py:method:: save(path: pathlib.Path) -> None


   .. py:method:: load(path: pathlib.Path) -> typing_extensions.Self
      :classmethod: