:py:mod:`neural_compressor.algorithm` ===================================== .. py:module:: neural_compressor.algorithm .. autoapi-nested-parse:: Fetch all files and init all algorithms. Submodules ---------- .. toctree:: :titlesonly: :maxdepth: 1 algorithm/index.rst fast_bias_correction/index.rst weight_correction/index.rst Package Contents ---------------- Classes ~~~~~~~ .. autoapisummary:: neural_compressor.algorithm.ALGORITHMS neural_compressor.algorithm.Algorithm neural_compressor.algorithm.AlgorithmScheduler Functions ~~~~~~~~~ .. autoapisummary:: neural_compressor.algorithm.algorithm_registry .. py:class:: ALGORITHMS Bases: :py:obj:`object` Build a dict for registered algorithms. .. py:method:: support_algorithms() :classmethod: Get all algorithms. :returns: A set of all algorithms. :rtype: Set .. py:class:: Algorithm Bases: :py:obj:`object` The base class of algorithm. .. py:class:: AlgorithmScheduler(conf) Bases: :py:obj:`object` control the Algorithm in different phase. .. py:property:: origin_model Return the origin model. :returns: The origin model. :rtype: model .. py:property:: q_model Return the quantized model. :returns: The quantized model. :rtype: model .. py:property:: dataloader Return the dataloader. :returns: The dataloader. :rtype: dataloader .. py:property:: adaptor Return the adaptor. :returns: The adaptor. :rtype: adaptor .. py:property:: calib_iter Return the calibration iter number. :returns: The calibration iter number. :rtype: calib_iter .. py:function:: algorithm_registry(algorithm_type) Decorate and register all Algorithm subclasses. :param cls: The class of register. :type cls: class :param algorithm_type: The algorithm registration name :type algorithm_type: str :returns: The class of register. :rtype: cls