afe.ir.transform.requantization_fusion
======================================

.. py:module:: afe.ir.transform.requantization_fusion


Attributes
----------

.. autoapisummary::

   afe.ir.transform.requantization_fusion.AttributesMutatorCallable
   afe.ir.transform.requantization_fusion.RequantizationCheckCallable


Classes
-------

.. autoapisummary::

   afe.ir.transform.requantization_fusion.FuseRequantizations


Functions
---------

.. autoapisummary::

   afe.ir.transform.requantization_fusion.convert_to_arith_folded_requantization
   afe.ir.transform.requantization_fusion.can_convert_to_arith_folded_requantization


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

.. py:function:: convert_to_arith_folded_requantization(requant: ml_kernels.requantization.BaseRequantization[numpy.ndarray]) -> Optional[ml_kernels.requantization.ArithFoldedRequantization[numpy.ndarray]]

   Convert the given requantization to an ArithFoldedRequantization, if possible.
   Only convert if the ArithFoldedRequantization is exactly equivalent, including
   rounding and saturation behavior.

   :param requant: Requantization to convert

   :returns: An ArithFoldedRequantization that is equivalent to the input.  None if there
             is no equivalent ArithFoldedRequantization.


.. py:function:: can_convert_to_arith_folded_requantization(requant: ml_kernels.requantization.BaseRequantization[numpy.ndarray]) -> bool

   Decide whether the given requantization can be converted to an ArithFoldedRequantization.


.. py:data:: AttributesMutatorCallable

.. py:data:: RequantizationCheckCallable

.. py:class:: FuseRequantizations



   A compiler pass that does fusion on RequantizeOp nodes, such as the ones
   that are inserted during quantization.  RequantizeOp nodes are fused into
   the preceding node where possible.