o
    ôj—?  ã                   @   sÈ  d dl mZ d dlmZmZmZmZmZ d dlZd dlm	Z	m
Z
 ddlmZmZ ddlmZ ddlmZ d	d
lmZmZmZ d	dlmZ d	dlmZmZmZ g d¢ZG dd„ dƒZG dd„ de	jƒZ G dd„ de	jƒZ!	d2de"de#de$de$def
dd„Z%dee de&dee d e$ded!e!fd"d#„Z'd$ed%œZ(G d&d'„ d'eƒZ)G d(d)„ d)eƒZ*eƒ ed*e)j+fd+�dd,d-œdee) d e$ded!e!fd.d/„ƒƒZ,eƒ ed*e*j+fd+�dd,d-œdee* d e$ded!e!fd0d1„ƒƒZ-dS )3é    )Úpartial)ÚAnyÚCallableÚListÚOptionalÚSequenceN)ÚnnÚTensoré   )ÚConv2dNormActivationÚSqueezeExcitation)ÚImageClassification)Ú_log_api_usage_onceé   )Úregister_modelÚWeightsÚWeightsEnum)Ú_IMAGENET_CATEGORIES)Ú_make_divisibleÚ_ovewrite_named_paramÚhandle_legacy_interface)ÚMobileNetV3ÚMobileNet_V3_Large_WeightsÚMobileNet_V3_Small_WeightsÚmobilenet_v3_largeÚmobilenet_v3_smallc                   @   sP   e Zd Zdedededededededed	efd
d„Zeded	efdd„ƒZ	dS )ÚInvertedResidualConfigÚinput_channelsÚkernelÚexpanded_channelsÚout_channelsÚuse_seÚ
activationÚstrideÚdilationÚ
width_multc
           
      C   sP   |   ||	¡| _|| _|   ||	¡| _|   ||	¡| _|| _|dk| _|| _|| _d S )NÚHS)	Úadjust_channelsr   r   r   r    r!   Úuse_hsr#   r$   )
Úselfr   r   r   r    r!   r"   r#   r$   r%   © r*   úo/var/www/html/academy/tools/voices/clone/sv-venv/lib/python3.10/site-packages/torchvision/models/mobilenetv3.pyÚ__init__   s   

zInvertedResidualConfig.__init__Úchannelsc                 C   s   t | | dƒS )Né   )r   )r-   r%   r*   r*   r+   r'   /   s   z&InvertedResidualConfig.adjust_channelsN)
Ú__name__Ú
__module__Ú__qualname__ÚintÚboolÚstrÚfloatr,   Ústaticmethodr'   r*   r*   r*   r+   r      s,    þýüûúùø	÷

ör   c                	       s^   e Zd Zeeejd�fdededej	f dedej	f f‡ fdd„Z
ded	efd
d„Z‡  ZS )ÚInvertedResidual)Úscale_activationÚcnfÚ
norm_layer.Úse_layerc                    s*  t ƒ  ¡  d|j  krdkstdƒ‚ tdƒ‚|jdko"|j|jk| _g }|jr,tj	ntj
}|j|jkrC| t|j|jd||d�¡ |jdkrJdn|j}| t|j|j|j||j|j||d�¡ |jrut|jd dƒ}| ||j|ƒ¡ | t|j|jd|d d�¡ tj|Ž | _|j| _|jdk| _d S )Nr   r
   zillegal stride value©Úkernel_sizer:   Úactivation_layer)r=   r#   r$   Úgroupsr:   r>   é   r.   )Úsuperr,   r#   Ú
ValueErrorr   r    Úuse_res_connectr(   r   Ú	HardswishÚReLUr   Úappendr   r$   r   r!   r   Ú
SequentialÚblockÚ_is_cn)r)   r9   r:   r;   Úlayersr>   r#   Úsqueeze_channels©Ú	__class__r*   r+   r,   6   sT   
ÿûÿøÿÿÿzInvertedResidual.__init__ÚinputÚreturnc                 C   s   |   |¡}| jr||7 }|S ©N)rH   rC   )r)   rN   Úresultr*   r*   r+   Úforwardn   s   
zInvertedResidual.forward)r/   r0   r1   r   ÚSElayerr   ÚHardsigmoidr   r   ÚModuler,   r	   rR   Ú__classcell__r*   r*   rL   r+   r7   4   s    üþýü8r7   c                       sŒ   e Zd Z				ddee dededeedej	f  d	eedej	f  d
e
deddf‡ fdd„Zdedefdd„Zdedefdd„Z‡  ZS )r   éè  Nçš™™™™™É?Úinverted_residual_settingÚlast_channelÚnum_classesrH   .r:   ÚdropoutÚkwargsrO   c              
      sÈ  t ƒ  ¡  t| ƒ |stdƒ‚t|tƒrtdd„ |D ƒƒs!tdƒ‚|du r't}|du r3t	t
jddd�}g }|d	 j}	| td
|	d
d|t
jd�¡ |D ]
}
| ||
|ƒ¡ qJ|d j}d| }| t||d|t
jd�¡ t
j|Ž | _t
 d¡| _t
 t
 ||¡t
jdd�t
j|dd�t
 ||¡¡| _|  ¡ D ]L}t|t
jƒr³t
jj|jdd� |jdur²t
j |j¡ q•t|t
jt
jfƒrËt
j  |j¡ t
j |j¡ q•t|t
jƒrát
j !|jd	d¡ t
j |j¡ q•dS )a.  
        MobileNet V3 main class

        Args:
            inverted_residual_setting (List[InvertedResidualConfig]): Network structure
            last_channel (int): The number of channels on the penultimate layer
            num_classes (int): Number of classes
            block (Optional[Callable[..., nn.Module]]): Module specifying inverted residual building block for mobilenet
            norm_layer (Optional[Callable[..., nn.Module]]): Module specifying the normalization layer to use
            dropout (float): The droupout probability
        z1The inverted_residual_setting should not be emptyc                 S   s   g | ]}t |tƒ‘qS r*   )Ú
isinstancer   )Ú.0Úsr*   r*   r+   Ú
<listcomp>’   s    z(MobileNetV3.__init__.<locals>.<listcomp>zDThe inverted_residual_setting should be List[InvertedResidualConfig]Ngü©ñÒMbP?g{®Gáz„?)ÚepsÚmomentumr   é   r
   )r=   r#   r:   r>   éÿÿÿÿé   r   r<   T)Úinplace)Úprg   Úfan_out)Úmode)"rA   r,   r   rB   r^   r   ÚallÚ	TypeErrorr7   r   r   ÚBatchNorm2dr   rF   r   rD   r    rG   ÚfeaturesÚAdaptiveAvgPool2dÚavgpoolÚLinearÚDropoutÚ
classifierÚmodulesÚConv2dÚinitÚkaiming_normal_ÚweightÚbiasÚzeros_Ú	GroupNormÚones_Únormal_)r)   rY   rZ   r[   rH   r:   r\   r]   rJ   Úfirstconv_output_channelsr9   Úlastconv_input_channelsÚlastconv_output_channelsÚmrL   r*   r+   r,   v   sx   
ÿþ
úÿ
ûÿ



ü
€€özMobileNetV3.__init__Úxc                 C   s.   |   |¡}|  |¡}t |d¡}|  |¡}|S )Nr   )rn   rp   ÚtorchÚflattenrs   ©r)   r‚   r*   r*   r+   Ú_forward_implÑ   s
   


zMobileNetV3._forward_implc                 C   s
   |   |¡S rP   )r†   r…   r*   r*   r+   rR   Û   s   
zMobileNetV3.forward)rW   NNrX   )r/   r0   r1   r   r   r2   r   r   r   rU   r5   r   r,   r	   r†   rR   rV   r*   r*   rL   r+   r   u   s0    ùþýüûúùø	÷[
r   ç      ð?FÚarchr%   Úreduced_tailÚdilatedr]   c                 K   sÂ  |rdnd}|r
dnd}t t|d�}t tj|d�}| dkrÍ|ddddddddƒ|ddd	d
ddddƒ|d
ddd
ddddƒ|d
dddddddƒ|ddddddddƒ|ddddddddƒ|ddddddddƒ|ddddddddƒ|ddddddddƒ|ddddddddƒ|ddddddddƒ|ddddddddƒ|dddd| ddd|ƒ|d| dd| d| ddd|ƒ|d| dd| d| ddd|ƒg}	|d| ƒ}
|	|
fS | dk�rZ|ddddddddƒ|dddd
ddddƒ|d
ddd
ddddƒ|d
dddddddƒ|ddddddddƒ|ddddddddƒ|ddddddddƒ|ddddddddƒ|ddd d| ddd|ƒ|d| dd!| d| ddd|ƒ|d| dd!| d| ddd|ƒg}	|d"| ƒ}
|	|
fS td#| › �ƒ‚)$Nr
   r   )r%   r   é   rd   FÚREé@   é   éH   é   é(   Téx   éð   éP   r&   éÈ   é¸   ià  ép   i   é    iÀ  i   r   éX   é`   é0   é�   i   i@  i   zUnsupported model type )r   r   r'   rB   )rˆ   r%   r‰   rŠ   r]   Zreduce_dividerr$   Ú
bneck_confr'   rY   rZ   r*   r*   r+   Ú_mobilenet_v3_confß   sN     ñ
î  õþrž   rY   rZ   ÚweightsÚprogressrO   c                 K   sP   |d urt |dt|jd ƒƒ t| |fi |¤Ž}|d ur&| |j|dd�¡ |S )Nr[   Ú
categoriesT)r    Ú
check_hash)r   ÚlenÚmetar   Úload_state_dictÚget_state_dict)rY   rZ   rŸ   r    r]   Úmodelr*   r*   r+   Ú_mobilenet_v3  s   r¨   )r   r   )Úmin_sizer¡   c                   @   s~   e Zd Zedeedd�i e¥dddddd	œid
dddœ¥d�Zedeeddd�i e¥dddddd	œid
dddœ¥d�ZeZ	dS )r   zChttps://download.pytorch.org/models/mobilenet_v3_large-8738ca79.pthéà   ©Ú	crop_sizeiªS ú^https://github.com/pytorch/vision/tree/main/references/classification#mobilenetv3-large--smallúImageNet-1Kg¦›Ä °‚R@gö(\�ÂÕV@©zacc@1zacc@5g-²�ï§ÆË?gw¾Ÿ/5@zJThese weights were trained from scratch by using a simple training recipe.©Ú
num_paramsÚrecipeÚ_metricsÚ_opsÚ
_file_sizeÚ_docs©ÚurlÚ
transformsr¤   zChttps://download.pytorch.org/models/mobilenet_v3_large-5c1a4163.pthéè   )r¬   Úresize_sizezHhttps://github.com/pytorch/vision/issues/3995#new-recipe-with-reg-tuningg¨ÆK7‰ÑR@gNbX9$W@g¬Zd5@a/  
                These weights improve marginally upon the results of the original paper by using a modified version of
                TorchVision's `new training recipe
                <https://pytorch.org/blog/how-to-train-state-of-the-art-models-using-torchvision-latest-primitives/>`_.
            N)
r/   r0   r1   r   r   r   Ú_COMMON_METAÚIMAGENET1K_V1ÚIMAGENET1K_V2ÚDEFAULTr*   r*   r*   r+   r   (  sL    
ÿþÿôýÿþÿôýr   c                   @   sF   e Zd Zedeedd�i e¥dddddd	œid
dddœ¥d�ZeZdS )r   zChttps://download.pytorch.org/models/mobilenet_v3_small-047dcff4.pthrª   r«   iÍ& r­   r®   g˜nƒÀêP@g}?5^ºÙU@r¯   gÉv¾Ÿ/­?gœÄ °r¨#@z}
                These weights improve upon the results of the original paper by using a simple training recipe.
            r°   r·   N)	r/   r0   r1   r   r   r   r¼   r½   r¿   r*   r*   r*   r+   r   T  s(    
ÿþÿôýr   Ú
pretrained)rŸ   T)rŸ   r    c                 K   ó2   t  | ¡} tdi |¤Ž\}}t||| |fi |¤ŽS )a³  
    Constructs a large MobileNetV3 architecture from
    `Searching for MobileNetV3 <https://arxiv.org/abs/1905.02244>`__.

    Args:
        weights (:class:`~torchvision.models.MobileNet_V3_Large_Weights`, optional): The
            pretrained weights to use. See
            :class:`~torchvision.models.MobileNet_V3_Large_Weights` below for
            more details, and possible values. By default, no pre-trained
            weights are used.
        progress (bool, optional): If True, displays a progress bar of the
            download to stderr. Default is True.
        **kwargs: parameters passed to the ``torchvision.models.mobilenet.MobileNetV3``
            base class. Please refer to the `source code
            <https://github.com/pytorch/vision/blob/main/torchvision/models/mobilenetv3.py>`_
            for more details about this class.

    .. autoclass:: torchvision.models.MobileNet_V3_Large_Weights
        :members:
    r   N)r   )r   Úverifyrž   r¨   ©rŸ   r    r]   rY   rZ   r*   r*   r+   r   l  ó   
r   c                 K   rÁ   )a³  
    Constructs a small MobileNetV3 architecture from
    `Searching for MobileNetV3 <https://arxiv.org/abs/1905.02244>`__.

    Args:
        weights (:class:`~torchvision.models.MobileNet_V3_Small_Weights`, optional): The
            pretrained weights to use. See
            :class:`~torchvision.models.MobileNet_V3_Small_Weights` below for
            more details, and possible values. By default, no pre-trained
            weights are used.
        progress (bool, optional): If True, displays a progress bar of the
            download to stderr. Default is True.
        **kwargs: parameters passed to the ``torchvision.models.mobilenet.MobileNetV3``
            base class. Please refer to the `source code
            <https://github.com/pytorch/vision/blob/main/torchvision/models/mobilenetv3.py>`_
            for more details about this class.

    .. autoclass:: torchvision.models.MobileNet_V3_Small_Weights
        :members:
    r   N)r   )r   rÂ   rž   r¨   rÃ   r*   r*   r+   r   ‹  rÄ   r   )r‡   FF).Ú	functoolsr   Útypingr   r   r   r   r   rƒ   r   r	   Zops.miscr   r   rS   Ztransforms._presetsr   Úutilsr   Ú_apir   r   r   Ú_metar   Ú_utilsr   r   r   Ú__all__r   rU   r7   r   r4   r5   r3   rž   r2   r¨   r¼   r   r   r½   r   r   r*   r*   r*   r+   Ú<module>   sˆ    	Akÿÿÿÿÿ
ÿ1ÿþýüû
úþ,ÿÿÿÿþÿÿÿÿþ