ml_switcheroo.core.rewriter.attributes
======================================

.. py:module:: ml_switcheroo.core.rewriter.attributes

.. autoapi-nested-parse::

   Attribute Rewriting Logic.

   Handles transformation of attributes (e.g. `torch.float32`),
   and tracks assignment logic to detect stateful variables (e.g. `self.layer = ...`).



Classes
-------

.. autoapisummary::

   ml_switcheroo.core.rewriter.attributes.AttributeMixin


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

.. py:class:: AttributeMixin(semantics: ml_switcheroo.semantics.manager.SemanticsManager, config: ml_switcheroo.config.RuntimeConfig)

   Bases: :py:obj:`ml_switcheroo.core.rewriter.base.BaseRewriter`


   Mixin for transforming attributes and tracking assignments.


   .. py:method:: leave_Assign(original_node: libcst.Assign, updated_node: libcst.Assign) -> libcst.Assign

      Track stateful assignments (e.g. self.layer = Linear(...)) to determine
      if an object variable requires special handling later.



   .. py:method:: leave_Attribute(original: libcst.Attribute, updated: libcst.Attribute) -> libcst.BaseExpression

      Visits attributes (e.g. torch.float32).