:py:mod:`neural_compressor.utils.pytorch`
=========================================

.. py:module:: neural_compressor.utils.pytorch

.. autoapi-nested-parse::

   Pytorch utilities.



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


Functions
~~~~~~~~~

.. autoapisummary::

   neural_compressor.utils.pytorch.is_int8_model
   neural_compressor.utils.pytorch.load_weight_only
   neural_compressor.utils.pytorch.load
   neural_compressor.utils.pytorch.recover_model_from_json



.. py:function:: is_int8_model(model)

   Check whether the input model is a int8 model.

   :param model: input model
   :type model: torch.nn.Module

   :returns: Return True if the input model is a int8 model.
   :rtype: result(bool)


.. py:function:: load_weight_only(checkpoint_dir, model, layer_wise=False)

   Load model in weight_only mode.

   :param checkpoint_dir: The folder of checkpoint. 'qconfig.json' and
                          'best_model.pt' are needed in This directory.
                          'checkpoint' dir is under workspace folder and
                          workspace folder is define in configure yaml file.
   :type checkpoint_dir: dir/file/dict
   :param model: fp32 model need to do quantization.
   :type model: object

   :returns: quantized model
   :rtype: (object)


.. py:function:: load(checkpoint_dir=None, model=None, layer_wise=False, history_cfg=None, **kwargs)

   Execute the quantize process on the specified model.

   :param checkpoint_dir: The folder of checkpoint. 'best_configure.yaml' and
                          'best_model_weights.pt' are needed in This directory.
                          'checkpoint' dir is under workspace folder and
                          workspace folder is define in configure yaml file.
   :type checkpoint_dir: dir/file/dict
   :param model: fp32 model need to do quantization.
   :type model: object
   :param history_cfg: configurations from history.snapshot file.
   :type history_cfg: object
   :param \*\*kwargs: contains customer config dict and etc.
   :type \*\*kwargs: dict

   :returns: quantized model
   :rtype: (object)


.. py:function:: recover_model_from_json(model, json_file_path, example_inputs)

   Recover ipex model from JSON file.

   :param model: fp32 model need to do quantization.
   :type model: object
   :param json_file_path: configuration JSON file for ipex.
   :type json_file_path: json
   :param example_inputs: example inputs that will be passed to the ipex function.
   :type example_inputs: tuple or torch.Tensor or dict

   :returns: quantized model
   :rtype: (object)