
    |j                    j    d dl mZ d dlZd dlmZ ddlmZ erd dlmZ ddl	m
Z
  G d d	e          ZdS )
    )annotationsN)TYPE_CHECKING   )Quantization)Layer)QuantConfigc                  .     e Zd ZdZd fdZdddZ xZS )QATa  
    Tools used to prepare model for quantization-aware training.
    Args:
        config(QuantConfig): Quantization configuration

    Examples:
        .. code-block:: python

            >>> from paddle.quantization import QAT, QuantConfig
            >>> from paddle.quantization.quanters import FakeQuanterWithAbsMaxObserver
            >>> quanter = FakeQuanterWithAbsMaxObserver(moving_rate=0.9)
            >>> q_config = QuantConfig(activation=quanter, weight=quanter)
            >>> qat = QAT(q_config)
    configr   returnNonec                J    t                                          |           d S )N)super__init__)selfr   	__class__s     g/lsinfo/ai/hellotax_ai/data_center/backend/venv/lib/python3.11/site-packages/paddle/quantization/qat.pyr   zQAT.__init__+   s!             Fmodelr   inplaceboolc                    |j         s
J d            |r|nt          j        |          }| j                            |           |                     || j                   |                     || j                   |S )a  
        Create a model for quantization-aware training.

        The quantization configuration will be propagated in the model.
        And it will insert fake quanters into the model to simulate the quantization.

        Args:
            model(Layer): The model to be quantized.
            inplace(bool): Whether to modify the model in-place.

        Return: The prepared model for quantization-aware training.

        Examples:
            .. code-block:: python

                >>> from paddle.quantization import QAT, QuantConfig
                >>> from paddle.quantization.quanters import FakeQuanterWithAbsMaxObserver
                >>> from paddle.vision.models import LeNet

                >>> quanter = FakeQuanterWithAbsMaxObserver(moving_rate=0.9)
                >>> q_config = QuantConfig(activation=quanter, weight=quanter)
                >>> qat = QAT(q_config)
                >>> model = LeNet()
                >>> quant_model = qat.quantize(model)
                >>> print(quant_model)
                LeNet(
                  (features): Sequential(
                    (0): QuantedConv2D(
                      (weight_quanter): FakeQuanterWithAbsMaxObserverLayer()
                      (activation_quanter): FakeQuanterWithAbsMaxObserverLayer()
                    )
                    (1): ObserveWrapper(
                      (_observer): FakeQuanterWithAbsMaxObserverLayer()
                      (_observed): ReLU()
                    )
                    (2): ObserveWrapper(
                      (_observer): FakeQuanterWithAbsMaxObserverLayer()
                      (_observed): MaxPool2D(kernel_size=2, stride=2, padding=0)
                    )
                    (3): QuantedConv2D(
                      (weight_quanter): FakeQuanterWithAbsMaxObserverLayer()
                      (activation_quanter): FakeQuanterWithAbsMaxObserverLayer()
                    )
                    (4): ObserveWrapper(
                      (_observer): FakeQuanterWithAbsMaxObserverLayer()
                      (_observed): ReLU()
                    )
                    (5): ObserveWrapper(
                      (_observer): FakeQuanterWithAbsMaxObserverLayer()
                      (_observed): MaxPool2D(kernel_size=2, stride=2, padding=0)
                    )
                  )
                  (fc): Sequential(
                    (0): QuantedLinear(
                      (weight_quanter): FakeQuanterWithAbsMaxObserverLayer()
                      (activation_quanter): FakeQuanterWithAbsMaxObserverLayer()
                    )
                    (1): QuantedLinear(
                      (weight_quanter): FakeQuanterWithAbsMaxObserverLayer()
                      (activation_quanter): FakeQuanterWithAbsMaxObserverLayer()
                    )
                    (2): QuantedLinear(
                      (weight_quanter): FakeQuanterWithAbsMaxObserverLayer()
                      (activation_quanter): FakeQuanterWithAbsMaxObserverLayer()
                    )
                  )
                )
        zfQuantization-Aware Training should work on training models. Please set training mode by model.train().)trainingcopydeepcopy_config_specify_convert_to_quant_layers_insert_activation_observers)r   r   r   _models       r   quantizezQAT.quantize.   s    J ~ 	
 	
t	
 	
~ ";t}U';';f%%%%%fdl;;;))&$,???r   )r   r   r   r   )F)r   r   r   r   r   r   )__name__
__module____qualname____doc__r   r!   __classcell__)r   s   @r   r
   r
      si         ! ! ! ! ! !L L L L L L L L Lr   r
   )
__future__r   r   typingr   r!   r   	paddle.nnr   r   r   r
    r   r   <module>r+      s    # " " " " "              " " " " " " $######_ _ _ _ _, _ _ _ _ _r   