tilelang.quantize.utils
=======================

.. py:module:: tilelang.quantize.utils


Functions
---------

.. autoapisummary::

   tilelang.quantize.utils.gen_quant4
   tilelang.quantize.utils.general_compress
   tilelang.quantize.utils.interleave_weight


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

.. py:function:: gen_quant4(k, n, groupsize=-1)

.. py:function:: general_compress(lowprecision_weight, source_bits=4, storage_dtype=None)

.. py:function:: interleave_weight(qweight, nbits=4, target_dtype='float16')

   Interleave the weight to the target data type.

   :param qweight: _description_
   :type qweight: _type_
   :param nbits: _description_. Defaults to 4.
   :type nbits: int, optional
   :param target_dtype: _description_. Defaults to "float16".
   :type target_dtype: str, optional

   :returns: _description_
   :rtype: _type_

   .. rubric:: Example

   qweight = torch.randint(0, 127, (10, 10), dtype=torch.int8).cuda()
   interleave_weight(qweight, 4, "float16")