:orphan:

:py:mod:`neural_compressor.tensorflow.quantization.quantize`
============================================================

.. py:module:: neural_compressor.tensorflow.quantization.quantize


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


Functions
~~~~~~~~~

.. autoapisummary::

   neural_compressor.tensorflow.quantization.quantize.quantize_model
   neural_compressor.tensorflow.quantization.quantize.quantize_model_with_single_config



.. py:function:: quantize_model(model: Union[str, tensorflow.keras.Model, neural_compressor.tensorflow.utils.BaseModel], quant_config: Union[neural_compressor.common.base_config.BaseConfig, list], calib_dataloader: Callable = None, calib_iteration: int = 100)

   The main entry to quantize model.

   :param model: a fp32 model to be quantized.
   :param quant_config: single or lists of quantization configuration.
   :param calib_dataloader: a data loader for calibration.
   :param calib_iteration: the iteration of calibration.

   :returns: the quantized model.
   :rtype: q_model


.. py:function:: quantize_model_with_single_config(q_model: neural_compressor.tensorflow.utils.BaseModel, quant_config: neural_compressor.common.base_config.BaseConfig, calib_dataloader: Callable = None, calib_iteration: int = 100)

   Quantize model using single config.

   :param model: a model wrapped by INC TF model class.
   :param quant_config: a quantization configuration.
   :param calib_dataloader: a data loader for calibration.
   :param calib_iteration: the iteration of calibration.

   :returns: the quantized model.
   :rtype: q_model