§
    ·¯¸jRS  ã                   óh  — d Z ddlZddlmZ ddlmZmZmZmZm	Z	m
Z
 ddlZddlmZ ddlmc mZ ddlmZ ddlmZmZ ddlmZmZmZmZmZ dd	lmZ dd
lmZm Z  ddl!m"Z"m#Z#m$Z$ dgZ% G d„ dej&        ¦  «        Z' G d„ dej(        ¦  «        Z) G d„ dej*        ¦  «        Z+ G d„ dej&        ¦  «        Z,de-dee.ej/        f         fd„Z0de.de1dee1df         de2de,f
d„Z3d/de.dee.ef         fd„Z4 e# e4d d!d"¬#¦  «         e4d d!d"¬#¦  «         e4¦   «          e4d ¬$¦  «         e4d ¬$¦  «         e4d ¬$¦  «         e4d ¬$¦  «        d%œ¦  «        Z5e"d0de,fd'„¦   «         Z6e"d0de,fd(„¦   «         Z7e"d0de,fd)„¦   «         Z8e"d0de,fd*„¦   «         Z9e"d0de,fd+„¦   «         Z:e"d0de,fd,„¦   «         Z; e$e<d-d.i¦  «         dS )1zËPytorch Densenet implementation w/ tweaks
This file is a copy of https://github.com/pytorch/vision 'densenet.py' (BSD-3-Clause) with
fixed kwargs passthrough and addition of dynamic global avg/max pool.
é    N)ÚOrderedDict)ÚAnyÚDictÚOptionalÚTupleÚTypeÚUnion)ÚList©ÚIMAGENET_DEFAULT_MEANÚIMAGENET_DEFAULT_STD)ÚBatchNormAct2dÚget_norm_act_layerÚ
BlurPool2dÚcreate_classifierÚget_device_dtypeé   )Úbuild_model_with_cfg)ÚMATCH_PREV_GROUPÚ
checkpoint)Úregister_modelÚgenerate_default_cfgsÚregister_model_deprecationsÚDenseNetc                   ó2  ‡ — e Zd ZdZeddddfdedededeej                 d	e	d
e
ddfˆ fd„Zdeej                 dej        fd„Zdeej                 de
fd„Zdeej                 dej        fd„Zdeej        eej                 f         dej        fd„Zˆ xZS )Ú
DenseLayerzbDense layer for DenseNet.

    Implements the bottleneck layer with 1x1 and 3x3 convolutions.
    ç        FNÚnum_input_featuresÚgrowth_rateÚbn_sizeÚ
norm_layerÚ	drop_rateÚgrad_checkpointingÚreturnc	                 óÄ  •— ||dœ}	t          ¦   «                              ¦   «          |                      d ||fi |	¤Ž¦  «        f |                      dt          j        |||z  fddddœ|	¤Ž¦  «        f |                      d |||z  fi |	¤Ž¦  «        f |                      dt          j        ||z  |fd	dddd
œ|	¤Ž¦  «        f t          |¦  «        | _        || _        dS )ad  Initialize DenseLayer.

        Args:
            num_input_features: Number of input features.
            growth_rate: Growth rate (k) of the layer.
            bn_size: Bottleneck size multiplier.
            norm_layer: Normalization layer class.
            drop_rate: Dropout rate.
            grad_checkpointing: Use gradient checkpointing.
        ©ÚdeviceÚdtypeÚnorm1Úconv1r   F©Úkernel_sizeÚstrideÚbiasÚnorm2Úconv2é   ©r,   r-   Úpaddingr.   N)ÚsuperÚ__init__Ú
add_moduleÚnnÚConv2dÚfloatr"   r#   )Úselfr   r   r    r!   r"   r#   r'   r(   ÚddÚ	__class__s             €úd/root/private/project/pallet_detection_v2/.venv/lib/python3.11/site-packages/timm/models/densenet.pyr5   zDenseLayer.__init__   sP  ø€ ð* ¨Ð/Ð/ˆÝ‰Œ×ÒÑÔÐØ�Š˜  Ð,>Ð!EÐ!EÀ"Ð!EÐ!EÑFÔFÐGÐGØ�Š˜¥¤Ø ¨+Ñ 5ð"bØCDÈQÐUZð"bð "bØ^`ð"bð "bñ 	cô 	cð 	dð 	dà�Š˜  ¨G°kÑ,AÐ!HÐ!HÀRÐ!HÐ!HÑIÔIÐJÐJØ�Š˜¥¤Ø�kÑ! ;ð"fØ<=ÀaÐQRÐY^ð"fð "fØbdð"fð "fñ 	gô 	gð 	hð 	hå˜yÑ)Ô)ˆŒØ"4ˆÔÐÐó    Úxsc                 ó€   — t          j        |d¦  «        }|                      |                      |¦  «        ¦  «        }|S )z.Bottleneck function for concatenated features.r   )ÚtorchÚcatr*   r)   )r:   r?   Úconcated_featuresÚbottleneck_outputs       r=   Úbottleneck_fnzDenseLayer.bottleneck_fn=   s9   € å!œI b¨!Ñ,Ô,ÐØ ŸJšJ t§z¢zÐ2CÑ'DÔ'DÑEÔEÐØ Ð r>   Úxc                 ó$   — |D ]}|j         r dS ŒdS )z.Check if any tensor in list requires gradient.TF)Úrequires_grad)r:   rF   Útensors      r=   Úany_requires_gradzDenseLayer.any_requires_gradD   s-   € àð 	ð 	ˆFØÔ#ð Ø�t�tðàˆur>   c                 ó(   ‡ — ˆ fd„}t          |g|¢R Ž S )z5Call bottleneck function with gradient checkpointing.c                  ó.   •— ‰                      | ¦  «        S )N)rE   )r?   r:   s    €r=   Úclosurez6DenseLayer.call_checkpoint_bottleneck.<locals>.closureM   s   ø€ Ø×%Ò% bÑ)Ô)Ð)r>   )r   )r:   rF   rM   s   `  r=   Úcall_checkpoint_bottleneckz%DenseLayer.call_checkpoint_bottleneckK   s4   ø€ ð	*ð 	*ð 	*ð 	*ð 	*õ ˜'Ð& AÐ&Ð&Ð&Ð&r>   c                 óÖ  — t          |t          j        ¦  «        r|g}n|}| j        rX|                      |¦  «        rCt          j                             ¦   «         rt          d¦  «        ‚|                      |¦  «        }n|  	                    |¦  «        }|  
                    |                      |¦  «        ¦  «        }| j        dk    r!t          j        || j        | j        ¬¦  «        }|S )z¦Forward pass.

        Args:
            x: Input features (single tensor or list of tensors).

        Returns:
            New features to be concatenated.
        z%Memory Efficient not supported in JITr   )ÚpÚtraining)Ú
isinstancerA   ÚTensorr#   rJ   ÚjitÚis_scriptingÚ	ExceptionrN   rE   r0   r/   r"   ÚFÚdropoutrQ   )r:   rF   Úprev_featuresrD   Únew_featuress        r=   ÚforwardzDenseLayer.forwardT   sä   € õ �a�œÑ&Ô&ð 	Ø˜CˆMˆMàˆMàÔ"ð 	B t×'=Ò'=¸mÑ'LÔ'Lð 	BÝŒy×%Ò%Ñ'Ô'ð IÝÐ GÑHÔHÐHØ $× ?Ò ?ÀÑ NÔ NÐÐà $× 2Ò 2°=Ñ AÔ AÐà—z’z $§*¢*Ð->Ñ"?Ô"?Ñ@Ô@ˆØŒ>˜AÒÐÝœ9 \°T´^ÈdÌmÐ\Ñ\Ô\ˆLØÐr>   )Ú__name__Ú
__module__Ú__qualname__Ú__doc__r   Úintr   r7   ÚModuler9   Úboolr5   r
   rA   rS   rE   rJ   rN   r	   r[   Ú__classcell__©r<   s   @r=   r   r      s^  ø€ € € € € ðð ð +9Ø!Ø',ØØð5ð 5à #ð5ð ð5ð ð	5ð
 ˜RœYœð5ð ð5ð !%ð5ð 
ð5ð 5ð 5ð 5ð 5ð 5ð@!  U¤\Ô 2ð !°u´|ð !ð !ð !ð !ð 4¨¬Ô#5ð ¸$ð ð ð ð ð'¨D°´Ô,>ð 'À5Ä<ð 'ð 'ð 'ð 'ð˜˜uœ|¨T°%´,Ô-?Ð?Ô@ð ÀUÄ\ð ð ð ð ð ð ð ð r>   r   c                   óŽ   ‡ — e Zd ZdZdZeddddfdededed	ed
eej	                 de
deddfˆ fd„Zdej        dej        fd„Zˆ xZS )Ú
DenseBlockzTDenseNet Block.

    Contains multiple dense layers with concatenated features.
    é   r   FNÚ
num_layersr   r    r   r!   r"   r#   r$   c
           
      óà   •— ||	dœ}
t          ¦   «                              ¦   «          t          |¦  «        D ]7}t          |||z  z   f|||||dœ|
¤Ž}|                      d|dz   z  |¦  «         Œ8dS )a�  Initialize DenseBlock.

        Args:
            num_layers: Number of layers in the block.
            num_input_features: Number of input features.
            bn_size: Bottleneck size multiplier.
            growth_rate: Growth rate (k) for each layer.
            norm_layer: Normalization layer class.
            drop_rate: Dropout rate.
            grad_checkpointing: Use gradient checkpointing.
        r&   )r   r    r!   r"   r#   zdenselayer%dr   N)r4   r5   Úranger   r6   )r:   rh   r   r    r   r!   r"   r#   r'   r(   r;   ÚiÚlayerr<   s                €r=   r5   zDenseBlock.__init__v   s©   ø€ ð. ¨Ð/Ð/ˆÝ‰Œ×ÒÑÔÐÝ�zÑ"Ô"ð 
	=ð 
	=ˆAÝØ" Q¨¡_Ñ4ðà'ØØ%Ø#Ø#5ðð ð ðð ˆEð �OŠO˜N¨a°!©eÑ4°eÑ<Ô<Ð<Ð<ð
	=ð 
	=r>   Úinit_featuresc                 ó¦   — |g}|                       ¦   «         D ]%\  }} ||¦  «        }|                     |¦  «         Œ&t          j        |d¦  «        S )zËForward pass through all layers in the block.

        Args:
            init_features: Initial features from previous layer.

        Returns:
            Concatenated features from all layers.
        r   )ÚitemsÚappendrA   rB   )r:   rm   ÚfeaturesÚnamerl   rZ   s         r=   r[   zDenseBlock.forward›   sZ   € ð "�?ˆØŸ:š:™<œ<ð 	*ð 	*‰KˆD�%Ø ˜5 ™?œ?ˆLØ�OŠO˜LÑ)Ô)Ð)Ð)ÝŒy˜ 1Ñ%Ô%Ð%r>   )r\   r]   r^   r_   Ú_versionr   r`   r   r7   ra   r9   rb   r5   rA   rS   r[   rc   rd   s   @r=   rf   rf   o   sß   ø€ € € € € ðð ð €Hð +9Ø!Ø',ØØð#=ð #=àð#=ð !$ð#=ð ð	#=ð
 ð#=ð ˜RœYœð#=ð ð#=ð !%ð#=ð 
ð#=ð #=ð #=ð #=ð #=ð #=ðJ& U¤\ð &°e´lð &ð &ð &ð &ð &ð &ð &ð &r>   rf   c                   óz   ‡ — e Zd ZdZedddfdededeej                 de	eej                          ddf
ˆ fd„Z
ˆ xZS )	ÚDenseTransitionzfTransition layer between DenseNet blocks.

    Reduces feature dimensions and spatial resolution.
    Nr   Únum_output_featuresr!   Úaa_layerr$   c           
      ó|  •— ||dœ}t          ¦   «                              ¦   «          |                      d ||fi |¤Ž¦  «         |                      dt          j        ||fddddœ|¤Ž¦  «         |�!|                      d ||fd	d
i|¤Ž¦  «         dS |                      dt          j        d
d
¬¦  «        ¦  «         dS )a  Initialize DenseTransition.

        Args:
            num_input_features: Number of input features.
            num_output_features: Number of output features.
            norm_layer: Normalization layer class.
            aa_layer: Anti-aliasing layer class.
        r&   ÚnormÚconvr   Fr+   NÚpoolr-   rg   )r,   r-   )r4   r5   r6   r7   r8   Ú	AvgPool2d)	r:   r   rv   r!   rw   r'   r(   r;   r<   s	           €r=   r5   zDenseTransition.__init__±   sþ   ø€ ð" ¨Ð/Ð/ˆÝ‰Œ×ÒÑÔÐØ�Š˜ 
 
Ð+=Ð DÐ DÀÐ DÐ DÑEÔEÐEØ�Š˜¥¤	ØÐ 3ð!`ØABÈ1ÐSXð!`ð !`Ø\^ð!`ð !`ñ 	aô 	að 	aàÐØ�OŠO˜F H HÐ-@Ð$QÐ$QÈÐ$QÈbÐ$QÐ$QÑRÔRÐRÐRÐRà�OŠO˜F¥B¤L¸QÀqÐ$IÑ$IÔ$IÑJÔJÐJÐJÐJr>   )r\   r]   r^   r_   r   r`   r   r7   ra   r   r5   rc   rd   s   @r=   ru   ru   «   s«   ø€ € € € € ðð ð +9Ø26ØØðKð Kà #ðKð "%ðKð ˜RœYœð	Kð
 ˜t B¤IœÔ/ðKð 
ðKð Kð Kð Kð Kð Kð Kð Kð Kð Kr>   ru   c                   óò  ‡ — e Zd ZdZ	 	 	 	 	 	 	 	 	 	 	 	 	 	 	 	 d+dedeedf         dedededededededeee	j
                          dededededdfˆ fd„Zej        j        d,d edeeef         fd!„¦   «         Zej        j        d-d"eddfd#„¦   «         Zej        j        de	j
        fd$„¦   «         Zd.dededdfd%„Zd&ej        dej        fd'„Zd,d&ej        d(edej        fd)„Zd&ej        dej        fd*„Zˆ xZS )/r   aö  Densenet-BC model class.

    Based on `"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>`_

    Args:
        growth_rate: How many filters to add each layer (`k` in paper).
        block_config: How many layers in each pooling block.
        bn_size: Multiplicative factor for number of bottle neck layers
          (i.e. bn_size * k features in the bottleneck layer).
        drop_rate: Dropout rate before classifier layer.
        proj_drop_rate: Dropout rate after each dense layer.
        num_classes: Number of classification classes.
        memory_efficient: If True, uses checkpointing. Much more memory efficient,
          but slower. Default: *False*. See `"paper" <https://arxiv.org/pdf/1707.06990.pdf>`_.
    é    ©é   é   é   é   éè  r1   Úavgé   Ú ÚreluÚbatchnorm2dNr   FTr   Úblock_config.Únum_classesÚin_chansÚglobal_poolr    Ú	stem_typeÚ	act_layerr!   rw   r"   Úproj_drop_rateÚmemory_efficientÚaa_stem_onlyr$   c                 ó:  •— ||dœ}|| _         || _        t          ¦   «                              ¦   «          t	          |	|¬¦  «        }	d|v }|dz  }|
€t          j        ddd¬¦  «        }n.t          j        t          j        ddd¬¦  «         |
d$|dd	œ|¤ŽgŽ }|r±|x}}d
|v rd|dz  z  }d|v r|nd|dz  z  }t          j        t          dt          j	        ||dfddddœ|¤Žfd |	|fi |¤Žfdt          j	        ||dfddddœ|¤Žfd |	|fi |¤Žfdt          j	        ||dfddddœ|¤Žfd |	|fi |¤Žfd|fg¦  «        ¦  «        | _
        nLt          j        t          dt          j	        ||fdddddœ|¤Žfd |	|fi |¤Žfd|fg¦  «        ¦  «        | _
        t          |dd|rdnd› �¬¦  «        g| _        d}|}t          |¦  «        D ]Ã\  }}t          d$|||||	||dœ|¤Ž}d|dz   › �}| j
                             ||¦  «         |||z  z   }|rdn|
}|t!          |¦  «        dz
  k    rc| xj        t          ||d|z   ¬¦  «        gz  c_        |dz  }t#          d$||dz  |	|dœ|¤Ž}| j
                             d |dz   › �|¦  «         |dz  }ŒÄ| j
                             d! |	|fi |¤Ž¦  «         | xj        t          ||d"¬¦  «        gz  c_        |x| _        | _        t)          | j        | j         fd#|i|¤Ž\  }}|| _        t          j        |¦  «        | _        || _        |                      ¦   «         D ]å} t5          | t
          j	        ¦  «        r%t
          j                             | j        ¦  «         ŒAt5          | t
          j        ¦  «        rKt
          j                             | j        d¦  «         t
          j                             | j         d¦  «         Œ¦t5          | t
          j!        ¦  «        r%t
          j                             | j         d¦  «         ŒædS )%aj  Initialize DenseNet.

        Args:
            growth_rate: How many filters to add each layer (k in paper).
            block_config: How many layers in each pooling block.
            num_classes: Number of classification classes.
            in_chans: Number of input channels.
            global_pool: Global pooling type.
            bn_size: Multiplicative factor for number of bottle neck layers.
            stem_type: Type of stem ('', 'deep', 'deep_tiered').
            act_layer: Activation layer.
            norm_layer: Normalization layer.
            aa_layer: Anti-aliasing layer.
            drop_rate: Dropout rate before classifier layer.
            proj_drop_rate: Dropout rate after each dense layer.
            memory_efficient: If True, uses checkpointing for memory efficiency.
            aa_stem_only: Apply anti-aliasing only to stem.
        r&   )r�   Údeeprg   Nr1   r   )r,   r-   r3   )Úchannelsr-   Útieredr†   Únarrowr€   Úconv0F)r-   r3   r.   Únorm0r*   r)   r0   r/   Úpool0é   r2   zfeatures.normr   )Únum_chsÚ	reductionÚmodule)rh   r   r    r   r!   r"   r#   Ú
denseblockz	features.)r   rv   r!   rw   Ú
transitionÚnorm5zfeatures.norm5Ú	pool_type© )"r‹   rŒ   r4   r5   r   r7   Ú	MaxPool2dÚ
Sequentialr   r8   rq   ÚdictÚfeature_infoÚ	enumeraterf   r6   Úlenru   Únum_featuresÚhead_hidden_sizer   r�   ÚDropoutÚ	head_dropÚ
classifierÚmodulesrR   ÚinitÚkaiming_normal_ÚweightÚBatchNorm2dÚ	constant_r.   ÚLinear)"r:   r   rŠ   r‹   rŒ   r�   r    rŽ   r�   r!   rw   r"   r�   r‘   r’   r'   r(   r;   Ú	deep_stemÚnum_init_featuresÚ	stem_poolÚ
stem_chs_1Ú
stem_chs_2Úcurrent_striderª   rk   rh   ÚblockÚmodule_nameÚtransition_aa_layerÚtransr®   Úmr<   s"                                    €r=   r5   zDenseNet.__init__Þ   s¥  ø€ ðJ ¨Ð/Ð/ˆØ&ˆÔØ ˆŒÝ‰Œ×ÒÑÔÐÝ'¨
¸iÐHÑHÔHˆ
ð ˜iÐ'ˆ	Ø'¨!™OÐØÐÝœ°¸1ÀaÐHÑHÔHˆIˆIåœÝ”¨°1¸aÐ@Ñ@Ô@Ø�ÐDÐ"3¸AÐDÐDÀÐDÐDð(Fð GˆIð ð 	Ø&1Ð1ˆJ˜Ø˜9Ð$Ð$Ø +°Ñ"2Ñ3�
Ø2:¸iÐ2GÐ2GÐ.Ð.ÈQÐR]ÐabÑRbÑMc�
ÝœM­+Ø�"œ) H¨j¸!ÐcÀAÈqÐW\ÐcÐcÐ`bÐcÐcÐdØ˜*˜* ZÐ6Ð6°2Ð6Ð6Ð7Ø�"œ) J°
¸AÐeÀaÐQRÐY^ÐeÐeÐbdÐeÐeÐfØ˜*˜* ZÐ6Ð6°2Ð6Ð6Ð7Ø�"œ) JÐ0AÀ1ÐlÈQÐXYÐ`eÐlÐlÐikÐlÐlÐmØ˜*˜*Ð%6Ð=Ð=¸"Ð=Ð=Ð>Ø˜)Ð$ð7ñ +ô +ñ ô ˆDŒMˆMõ œM­+Ø�"œ) HÐ.?ÐvÈQÐWXÐbcÐjoÐvÐvÐsuÐvÐvÐwØ˜*˜*Ð%6Ð=Ð=¸"Ð=Ð=Ð>Ø˜)Ð$ð7ñ +ô +ñ ô ˆDŒMõ Ð*°aÐ@gÐU^ÐPeÐPQÐPQÐdeÐ@gÐ@gÐhÑhÔhðjˆÔàˆð )ˆÝ& |Ñ4Ô4ð 	1ð 	1‰MˆAˆzÝð 	Ø%Ø#/ØØ'Ø%Ø(Ø#3ð	ð 	ð ð	ð 	ˆEð 1¨¨A©Ð0Ð0ˆKØŒM×$Ò$ [°%Ñ8Ô8Ð8Ø'¨*°{Ñ*BÑBˆLØ*6Ð"D $ $¸HÐØ•C˜Ñ%Ô%¨Ñ)Ò)Ð)ØÐ!Ô!Ý ¸ÐP[Ð^iÑPiÐjÑjÔjð&lñ lÐ!Ô!à !Ñ#�Ý'ð Ø'3Ø(4¸Ñ(9Ø)Ø0ð	ð ð
 ðð �ð ”×(Ò(Ð)=°a¸!±eÐ)=Ð)=¸uÑEÔEÐEØ+¨qÑ0�øð 	Œ× Ò  ¨*¨*°\Ð*HÐ*HÀRÐ*HÐ*HÑIÔIÐIàÐÔ�d¨<À>ÐZjÐkÑkÔkÐlÑlÐÔØ4@Ð@ˆÔ˜DÔ1õ #4ØÔØÔð#
ð #
ð "ð#
ð ð	#
ð #
Ñˆ�Zð 'ˆÔÝœ IÑ.Ô.ˆŒØ$ˆŒð —’‘”ð 	-ð 	-ˆAÝ˜!�RœYÑ'Ô'ð -Ý”×'Ò'¨¬Ñ1Ô1Ð1Ð1Ý˜A�rœ~Ñ.Ô.ð -Ý”×!Ò! !¤(¨AÑ.Ô.Ð.Ý”×!Ò! !¤&¨!Ñ,Ô,Ð,Ð,Ý˜A�rœyÑ)Ô)ð -Ý”×!Ò! !¤&¨!Ñ,Ô,Ð,øð	-ð 	-r>   Úcoarsec                 óB   — t          d|rdn
ddt          fg¬¦  «        }|S )z"Group parameters for optimization.z<^features\.conv[012]|features\.norm[012]|features\.pool[012]z)^features\.(?:denseblock|transition)(\d+))z+^features\.denseblock(\d+)\.denselayer(\d+)Nz^features\.transition(\d+))ÚstemÚblocks)r¦   r   )r:   rÁ   Úmatchers      r=   Úgroup_matcherzDenseNet.group_matcherd  sB   € õ ØPØCIð Ð?Ð?ØFØ.Õ0@ÐAðPð
ñ 
ô 
ˆð ˆr>   Úenablec                 óv   — | j                              ¦   «         D ]}t          |t          ¦  «        r||_        ŒdS )z)Enable or disable gradient checkpointing.N)rq   r¯   rR   r   r#   )r:   rÇ   Úbs      r=   Úset_grad_checkpointingzDenseNet.set_grad_checkpointingp  sF   € ð ”×&Ò&Ñ(Ô(ð 	.ð 	.ˆAÝ˜!�ZÑ(Ô(ð .Ø'-�Ô$øð	.ð 	.r>   c                 ó   — | j         S )zGet the classifier head.)r®   )r:   s    r=   Úget_classifierzDenseNet.get_classifierw  s   € ð ŒÐr>   c                 óü   — t          | ¦  «        }|| _        t          | j        | j        fd|i|¤Ž\  | _        | _        | j                             | j        ¦  «         | j                             | j        ¦  «         dS )zŸReset the classifier head.

        Args:
            num_classes: Number of classes for new classifier.
            global_pool: Global pooling type.
        r¢   N)r   r‹   r   rª   r�   r®   ÚtrainrQ   )r:   r‹   r�   r;   s       r=   Úreset_classifierzDenseNet.reset_classifier|  sŽ   € õ ˜dÑ#Ô#ˆØ&ˆÔÝ,=ØÔ˜tÔ/ð-Nð -NØ;Fð-NØJLð-Nð -NÑ)ˆÔ˜$œ/àÔ×Ò˜tœ}Ñ-Ô-Ð-ØŒ×Ò˜dœmÑ,Ô,Ð,Ð,Ð,r>   rF   c                 ó,   — |                       |¦  «        S )z/Forward pass through feature extraction layers.)rq   ©r:   rF   s     r=   Úforward_featureszDenseNet.forward_featuresŠ  s   € à�}Š}˜QÑÔÐr>   Ú
pre_logitsc                 óˆ   — |                       |¦  «        }|                      |¦  «        }|r|n|                      |¦  «        S )zÊForward pass through classifier head.

        Args:
            x: Feature tensor.
            pre_logits: Return features before final classifier.

        Returns:
            Output tensor.
        )r�   r­   r®   )r:   rF   rÓ   s      r=   Úforward_headzDenseNet.forward_headŽ  sC   € ð ×Ò˜QÑÔˆØ�NŠN˜1ÑÔˆØÐ6ˆqˆq D§O¢O°AÑ$6Ô$6Ð6r>   c                 óZ   — |                       |¦  «        }|                      |¦  «        }|S )zoForward pass.

        Args:
            x: Input tensor.

        Returns:
            Output logits.
        )rÒ   rÕ   rÑ   s     r=   r[   zDenseNet.forwardœ  s/   € ð ×!Ò! !Ñ$Ô$ˆØ×Ò˜aÑ Ô ˆØˆr>   )r~   r   r„   r1   r…   r†   r‡   rˆ   r‰   Nr   r   FTNN©F)T)r…   )r\   r]   r^   r_   r`   r   Ústrr   r   r7   ra   r9   rb   r5   rA   rT   Úignorer   r   rÆ   rÊ   rÌ   rÏ   rS   rÒ   rÕ   r[   rc   rd   s   @r=   r   r   Í   s‘  ø€ € € € € ðð ð$  "Ø,;Ø#ØØ$ØØØ#Ø+Ø26Ø!Ø$&Ø%*Ø!%ØØð#D-ð D-àðD-ð    S œ/ðD-ð ð	D-ð
 ðD-ð ðD-ð ðD-ð ðD-ð ðD-ð ðD-ð ˜t B¤IœÔ/ðD-ð ðD-ð "ðD-ð #ðD-ð ðD-ð$ 
ð%D-ð D-ð D-ð D-ð D-ð D-ðL „YÔð	ð 	 Dð 	°T¸#¸s¸(´^ð 	ð 	ð 	ñ Ôð	ð „YÔð.ð .¨Tð .¸Tð .ð .ð .ñ Ôð.ð „YÔð ¤	ð ð ð ñ Ôðð-ð -¨Cð -¸cð -Èdð -ð -ð -ð -ð  %¤,ð  °5´<ð  ð  ð  ð  ð7ð 7˜eœlð 7¸ð 7ÈÌð 7ð 7ð 7ð 7ð˜œð ¨%¬,ð ð ð ð ð ð ð ð r>   Ú
state_dictr$   c                 ó  — t          j        d¦  «        }t          |                      ¦   «         ¦  «        D ]R}|                     |¦  «        }|r9|                     d¦  «        |                     d¦  «        z   }| |         | |<   | |= ŒS| S )z¾Filter torchvision pretrained state dict for compatibility.

    Args:
        state_dict: State dictionary from torchvision checkpoint.

    Returns:
        Filtered state dictionary.
    z]^(.*denselayer\d+\.(?:norm|relu|conv))\.((?:[12])\.(?:weight|bias|running_mean|running_var))$r   rg   )ÚreÚcompileÚlistÚkeysÚmatchÚgroup)rÚ   ÚpatternÚkeyÚresÚnew_keys        r=   Ú_filter_torchvision_pretrainedræ   ª  s�   € õ ŒjØhñjô j€Gõ �J—O’OÑ%Ô%Ñ&Ô&ð  ð  ˆØ�mŠm˜CÑ Ô ˆØð 	 Ø—i’i ‘l”l S§Y¢Y¨q¡\¤\Ñ1ˆGØ",¨S¤/ˆJ�wÑØ˜3�øØÐr>   Úvariantr   rŠ   .Ú
pretrainedc                 ój   — ||d<   ||d<   t          t          | |ft          d¬¦  «        t          dœ|¤ŽS )a.  Create a DenseNet model.

    Args:
        variant: Model variant name.
        growth_rate: Growth rate parameter.
        block_config: Block configuration.
        pretrained: Load pretrained weights.
        **kwargs: Additional model arguments.

    Returns:
        DenseNet model instance.
    r   rŠ   T)Úflatten_sequential)Úfeature_cfgÚpretrained_filter_fn)r   r   r¦   ræ   )rç   r   rŠ   rè   Úkwargss        r=   Ú_create_densenetrî   ¿  sZ   € ð& (€Fˆ=ÑØ)€Fˆ>ÑÝÝØØðõ ¨DÐ1Ñ1Ô1Ý;ðð ð ðð ð r>   r‡   Úurlc                 ó6   — | dddddt           t          dddd	œ|¥S )
z1Create default configuration for DenseNet models.r„   )r1   éà   rñ   )r›   r›   g      ì?Úbicubiczfeatures.conv0r®   z
apache-2.0)rï   r‹   Ú
input_sizeÚ	pool_sizeÚcrop_pctÚinterpolationÚmeanÚstdÚ
first_convr®   Úlicenser   )rï   rí   s     r=   Ú_cfgrû   Þ  s9   € ð  4°}ÐSYØ¨IÝ%Õ.BØ&°lÈ|ð	ð ð
 ðð r>   ztimm/)r1   é   rü   gffffffî?)Ú	hf_hub_idÚtest_input_sizeÚtest_crop_pct)rý   )zdensenet121.ra_in1kzdensenetblur121d.ra_in1kzdensenet264d.untrainedúdensenet121.tv_in1kzdensenet169.tv_in1kzdensenet201.tv_in1kzdensenet161.tv_in1kFc           	      óZ   — t          dd¬¦  «        }t          dd| it          |fi |¤Ž¤Ž}|S )ztDensenet-121 model from
    `"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>`
    r~   r   ©r   rŠ   Údensenet121rè   )r  ©r¦   rî   ©rè   rí   Ú
model_argsÚmodels       r=   r  r  ø  óD   € õ
  "°?ÐCÑCÔC€JÝÐ`Ð`°zÐ`ÅTÈ*ÐE_ÐE_ÐX^ÐE_ÐE_Ð`Ð`€EØ€Lr>   c           	      óh   — t          dddt          ¬¦  «        }t          dd| it          |fi |¤Ž¤Ž}|S )zŒDensenet-121 w/ blur-pooling & 3-layer 3x3 stem
    `"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>`
    r~   r   r”   )r   rŠ   rŽ   rw   Údensenetblur121drè   )r
  )r¦   r   rî   r  s       r=   r
  r
    sI   € õ
  "°?ÈfÕ_iÐjÑjÔj€JÝÐeÐe¸JÐeÍ$ÈzÐJdÐJdÐ]cÐJdÐJdÐeÐe€EØ€Lr>   c           	      óZ   — t          dd¬¦  «        }t          dd| it          |fi |¤Ž¤Ž}|S )ztDensenet-169 model from
    `"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>`
    r~   )r€   r�   r~   r~   r  Údensenet169rè   )r  r  r  s       r=   r  r    r  r>   c           	      óZ   — t          dd¬¦  «        }t          dd| it          |fi |¤Ž¤Ž}|S )ztDensenet-201 model from
    `"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>`
    r~   )r€   r�   é0   r~   r  Údensenet201rè   )r  r  r  s       r=   r  r    r  r>   c           	      óZ   — t          dd¬¦  «        }t          dd| it          |fi |¤Ž¤Ž}|S )ztDensenet-161 model from
    `"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>`
    r  )r€   r�   é$   r‚   r  Údensenet161rè   )r  r  r  s       r=   r  r     r  r>   c           	      ó\   — t          ddd¬¦  «        }t          dd| it          |fi |¤Ž¤Ž}|S )ztDensenet-264 model from
    `"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>`
    r  )r€   r�   é@   r  r”   )r   rŠ   rŽ   Údensenet264drè   )r  r  r  s       r=   r  r  *  sF   € õ
  "°?ÈfÐUÑUÔU€JÝÐaÐa¸
ÐaÅdÈ:ÐF`ÐF`ÐY_ÐF`ÐF`ÐaÐa€EØ€Lr>   Útv_densenet121r   )r‡   r×   )=r_   rÜ   Úcollectionsr   Útypingr   r   r   r   r   r	   rA   Útorch.nnr7   Útorch.nn.functionalÚ
functionalrW   Útorch.jit.annotationsr
   Ú	timm.datar   r   Útimm.layersr   r   r   r   r   Ú_builderr   Ú_manipulater   r   Ú	_registryr   r   r   Ú__all__ra   r   Ú
ModuleDictrf   r¥   ru   r   r¦   rØ   rS   ræ   r`   rb   rî   rû   Údefault_cfgsr  r
  r  r  r  r  r\   r£   r>   r=   ú<module>r%     s�  ððð ð 
€	€	€	Ø #Ð #Ð #Ð #Ð #Ð #Ø :Ð :Ð :Ð :Ð :Ð :Ð :Ð :Ð :Ð :Ð :Ð :Ð :Ð :Ð :Ð :à €€€Ø Ð Ð Ð Ð Ð Ø Ð Ð Ð Ð Ð Ð Ð Ð Ø &Ð &Ð &Ð &Ð &Ð &à AÐ AÐ AÐ AÐ AÐ AÐ AÐ AØ kÐ kÐ kÐ kÐ kÐ kÐ kÐ kÐ kÐ kÐ kÐ kÐ kÐ kØ *Ð *Ð *Ð *Ð *Ð *Ø 5Ð 5Ð 5Ð 5Ð 5Ð 5Ð 5Ð 5Ø YÐ YÐ YÐ YÐ YÐ YÐ YÐ YÐ YÐ Yàˆ,€ðUð Uð Uð Uð U�”ñ Uô Uð Uðp9&ð 9&ð 9&ð 9&ð 9&�”ñ 9&ô 9&ð 9&ðxKð Kð Kð Kð K�b”mñ Kô Kð KðDZð Zð Zð Zð ZˆrŒyñ Zô Zð Zðz¨tð ¸¸SÀ%Ä,Ð=NÔ8Oð ð ð ð ð*Øðàðð ˜C ˜H”oðð ð	ð ðð ð ð ð>ð ˆcð  T¨#¨s¨(¤^ð ð ð ð ð %Ð$Ø˜4ØØ%°Tð;ñ ;ô ;ð !% ØØ%°Tð!;ñ !;ô !;ð #˜d™fœfØ˜4¨'Ð2Ñ2Ô2Ø˜4¨'Ð2Ñ2Ô2Ø˜4¨'Ð2Ñ2Ô2Ø˜4¨'Ð2Ñ2Ô2ð&ð &ñ ô €ð ðð ¨xð ð ð ñ „ðð ðð °Hð ð ð ñ „ðð ðð ¨xð ð ð ñ „ðð ðð ¨xð ð ð ñ „ðð ðð ¨xð ð ð ñ „ðð ðð °ð ð ð ñ „ðð Ð ˜HØÐ+ð'ñ ô ð ð ð r>   