+
    ùäûjVš  ã                  óÀ  € ^ RI Ht ^ RIt^ RIHt ^ RIt^ RIHt ^ RIH	t	 ^ RI
Ht ^ RIHtHt ^RIHt R&R R	 llt ! R
 R]P$                  4      t ! R R]P$                  4      t ! R R]P$                  4      t ! R R]P$                  4      tR tR t ! R R]P$                  4      t ! R R]P$                  4      t ! R R]P$                  4      t ! R R]P$                  4      t ! R R]P$                  4      t ! R R]P$                  4      t ! R  R!]P$                  4      t ! R" R#]P@                  4      t!]! 4        ! R$ R%]P$                  4      4       t"R# )'é    )ÚannotationsN)ÚLiteral)Ú	rearrange)Ústore_hyperparameters)Ú	to_2tupleÚtrunc_normal_)Úpad_to_multiplec               ó    € V ^8„  d   QhRRRR/# )é   Ú	drop_probÚfloatÚtrainingÚbool© )Úformats   "ÚaC:\Ai_Coding\apps\studio\ComfyUI\.venv\Lib\site-packages\spandrel\architectures\HAT\__arch\HAT.pyÚ__annotate__r      s   € ÷ ñ ˜Eð °4ñ ó    c                óV  € VR8X  g	   V'       g   V # ^V,
          pV P                   ^ ,          3RV P                  ^,
          ,          ,           pV\        P                  ! W@P                  V P
                  R7      ,           pVP                  4        V P                  V4      V,          pV# )z¹Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).

From: https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/layers/drop.py
ç        )ÚdtypeÚdevice©é   )ÚshapeÚndimÚtorchÚrandr   r   Úfloor_Údiv)Úxr   r   Ú	keep_probr   Úrandom_tensorÚoutputs   &&&    r   Ú	drop_pathr%      s…   € ð
 �CÔŸxØˆØ�I•€IØ�W‰W�Q�ZˆM˜DØ	�‰��
õõ €Eð ¤§
¢
¨5¿¹ÈÏÉÔ QÕQ€MØ×ÑÔØ�U‰U�9Ó Õ-€FØ€Mr   c                  ó6   a € ] tR t^"tRtRV 3R lltR tRtV ;t# )ÚDropPathzºDrop paths (Stochastic Depth) per sample  (when applied in main path of residual blocks).

From: https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/layers/drop.py
c                	ó0   <€ \         SV `  4        Wn        R # ©N)ÚsuperÚ__init__r   )Úselfr   Ú	__class__s   &&€r   r+   ÚDropPath.__init__(   s   ø€ Ü‰ÑÔØ"Žr   c                	óB   € \        WP                  V P                  4      # r)   )r%   r   r   ©r,   r!   s   &&r   ÚforwardÚDropPath.forward,   s   € Ü˜ŸN™N¨D¯M©MÓ:Ð:r   )r   r)   ©	Ú__name__Ú
__module__Ú__qualname__Ú__firstlineno__Ú__doc__r+   r1   Ú__static_attributes__Ú__classcell__©r-   s   @r   r'   r'   "   s   ø† ñ÷
#÷;ð ;r   r'   c                  ó<   a € ] tR t^0tRtRR V 3R llltR tRtV ;t# )ÚChannelAttentionz¢Channel attention used in RCAN.
Args:
    num_feat (int): Channel number of intermediate features.
    squeeze_factor (int): Channel squeeze factor. Default: 16.
c               ó   € V ^8„  d   QhRR/# )r   Úsqueeze_factorúint | floatr   )r   s   "r   r   ÚChannelAttention.__annotate__7   s   € ÷ 
ñ 
°ñ 
r   c                	óR  <€ \         SV `  4        \        P                  ! \        P                  ! ^4      \        P
                  ! WV,          ^^ R7      \        P                  ! RR7      \        P
                  ! W,          V^^ R7      \        P                  ! 4       4      V n        R# )r   )ÚpaddingT©ÚinplaceN)	r*   r+   ÚnnÚ
SequentialÚAdaptiveAvgPool2dÚConv2dÚReLUÚSigmoidÚ	attention)r,   Únum_featr?   r-   s   &&&€r   r+   ÚChannelAttention.__init__7   sk   ø€ Ü‰ÑÔÜŸšÜ× Ò  Ó#Ü�IŠI�h¨NÕ :¸AÀqÔIÜ�GŠG˜DÔ!Ü�IŠI�hÕ0°(¸AÀqÔIÜ�JŠJ‹Ló
ˆŽr   c                	ó4   € V P                  V4      pW,          # r)   ©rL   )r,   r!   Úys   && r   r1   ÚChannelAttention.forwardA   s   € Ø�N‰N˜1ÓˆØ�uˆr   rP   )é   r3   r;   s   @r   r=   r=   0   s   ø† ñ÷
ò 
÷ð r   r=   c                  ó8   a € ] tR t^FtRR V 3R llltR tRtV ;t# )ÚCABc               ó    € V ^8„  d   QhRRRR/# ©r   Úcompress_ratior@   r?   r   )r   s   "r   r   ÚCAB.__annotate__G   s    € ÷ 
ñ 
ð $ð
ð $ñ	
r   c                	ó  <€ \         SV `  4        \        P                  ! \        P                  ! WV,          ^^^4      \        P
                  ! 4       \        P                  ! W,          V^^^4      \        W4      4      V n        R# )é   N)r*   r+   rF   rG   rI   ÚGELUr=   Úcab)r,   rM   rX   r?   r-   s   &&&&€r   r+   ÚCAB.__init__G   s_   ø€ ô 	‰ÑÔä—=’=Ü�IŠI�h¨NÕ :¸A¸qÀ!ÓDÜ�GŠG‹IÜ�IŠI�hÕ0°(¸A¸qÀ!ÓDÜ˜XÓ6ó	
ˆŽr   c                	ó$   € V P                  V4      # r)   ©r]   r0   s   &&r   r1   ÚCAB.forwardV   s   € Ø�x‰x˜‹{Ðr   r`   )r[   é   )r4   r5   r6   r7   r+   r1   r9   r:   r;   s   @r   rU   rU   F   s   ø† ÷
ò 
÷ð r   rU   c                  óN   a € ] tR t^ZtRR]P
                  R3V 3R lltR tRtV ;t	# )ÚMlpNr   c                	ó  <€ \         SV `  4        T;'       g    TpT;'       g    Tp\        P                  ! W4      V n        V! 4       V n        \        P                  ! W#4      V n        \        P                  ! V4      V n        R # r)   )	r*   r+   rF   ÚLinearÚfc1ÚactÚfc2ÚDropoutÚdrop)r,   Úin_featuresÚhidden_featuresÚout_featuresÚ	act_layerrk   r-   s   &&&&&&€r   r+   ÚMlp.__init__[   sc   ø€ ô 	‰ÑÔØ#×2Ð2 {ˆØ)×8Ð8¨[ˆÜ—9’9˜[Ó:ˆŒÙ“;ˆŒÜ—9’9˜_Ó;ˆŒÜ—J’J˜tÓ$ˆŽ	r   c                	ó°   € V P                  V4      pV P                  V4      pV P                  V4      pV P                  V4      pV P                  V4      pV# r)   )rg   rh   rk   ri   r0   s   &&r   r1   ÚMlp.forwardk   sH   € Ø�H‰H�Q‹KˆØ�H‰H�Q‹KˆØ�I‰I�a‹LˆØ�H‰H�Q‹KˆØ�I‰I�a‹LˆØˆr   )rh   rk   rg   ri   )
r4   r5   r6   r7   rF   r\   r+   r1   r9   r:   r;   s   @r   rd   rd   Z   s$   ø† ð ØØ—'‘'Ø÷%÷ ð r   rd   c                óÐ   € V P                   w  r#rEV P                  W#V,          WV,          W4      p V P                  ^ ^^^^^4      P                  4       P                  RWV4      pV# )z‚
Args:
    x: (b, h, w, c)
    window_size (int): window size

Returns:
    windows: (num_windows*b, window_size, window_size, c)
éÿÿÿÿ)r   ÚviewÚpermuteÚ
contiguous)r!   Úwindow_sizeÚbÚhÚwÚcÚwindowss   &&     r   Úwindow_partitionr~   t   sh   € ð —‘�J€Aˆ!Ø	�‰ˆq�{Õ" K°kÕ1AÀ;ÓR€Aà	�	‰	�!�Q˜˜1˜a Ó#×.Ñ.Ó0×5Ñ5°b¸+ÐTUÓVð ð €Nr   c                ó   € \        V P                  ^ ,          W#,          V,          V,          ,          4      pV P                  WBV,          W1,          WR4      pVP                  ^ ^^^^^4      P	                  4       P                  WBVR4      pV# )z»
Args:
    windows: (num_windows*b, window_size, window_size, c)
    window_size (int): Window size
    h (int): Height of image
    w (int): Width of image

Returns:
    x: (b, h, w, c)
rt   )Úintr   ru   rv   rw   )r}   rx   rz   r{   ry   r!   s   &&&&  r   Úwindow_reverser�   …   s~   € ô 	ˆG�M‰M˜!Õ ¥¨Õ 3°kÕ AÕBÓC€AØ�‰Ø	�Õ˜QÕ-¨{Èó	€Að 	
�	‰	�!�Q˜˜1˜a Ó#×.Ñ.Ó0×5Ñ5°a¸A¸rÓB€AØ€Hr   c                  ó:   a € ] tR t^˜tRtRV 3R lltRR ltRtV ;t# )ÚWindowAttentiona  Window based multi-head self attention (W-MSA) module with relative position bias.
It supports both of shifted and non-shifted window.

Args:
    dim (int): Number of input channels.
    window_size (tuple[int]): The height and width of the window.
    num_heads (int): Number of attention heads.
    qkv_bias (bool, optional):  If True, add a learnable bias to query, key, value. Default: True
    qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set
    attn_drop (float, optional): Dropout ratio of attention weight. Default: 0.0
    proj_drop (float, optional): Dropout ratio of output. Default: 0.0
c                	óš  <€ \         S	V `  4        Wn        W n        W0n        W,          pT;'       g
    VR,          V n        \        P                  ! \        P                  ! ^V^ ,          ,          ^,
          ^V^,          ,          ^,
          ,          V4      4      V n
        \        P                  ! W^,          VR7      V n        \        P                  ! V4      V n        \        P                  ! W4      V n        \        P                  ! V4      V n        \#        V P                  RR7       \        P$                  ! RR7      V n        R# )ç      à?©Úbiasç{®Gáz”?©Ústd©ÚdimNç      à¿rt   )r*   r+   rŒ   rx   Ú	num_headsÚscalerF   Ú	Parameterr   ÚzerosÚrelative_position_bias_tablerf   Úqkvrj   Ú	attn_dropÚprojÚ	proj_dropr   ÚSoftmaxÚsoftmax)
r,   rŒ   rx   rŽ   Úqkv_biasÚqk_scaler”   r–   Úhead_dimr-   s
   &&&&&&&& €r   r+   ÚWindowAttention.__init__¦   så   ø€ ô 	‰ÑÔØŒØ&ÔØ"ŒØÕ#ˆØ×/Ð/ ¨4¥ˆŒ
ô -/¯LªLÜ�KŠK˜˜[¨�^Õ+¨aÕ/°A¸ÀA½Õ4FÈÕ4JÕKÈYÓWó-
ˆÔ)ô —9’9˜S¨¥'°Ô9ˆŒÜŸš IÓ.ˆŒÜ—I’I˜cÓ'ˆŒ	äŸš IÓ.ˆŒä�d×7Ñ7¸TÕBÜ—z’z bÔ)ˆŽr   c                óº  € VP                   w  rEpV P                  V4      P                  WE^V P                  W`P                  ,          4      P	                  ^^ ^^^4      pV^ ,          V^,          V^,          r©pW€P
                  ,          pW‰P                  RR4      ,          pV P                  VP                  R4      ,          P                  V P                  ^ ,          V P                  ^,          ,          V P                  ^ ,          V P                  ^,          ,          R4      pVP	                  ^^ ^4      P                  4       pW¼P                  ^ 4      ,           pVe‹   VP                   ^ ,          pVP                  WM,          WÐP                  WU4      VP                  ^4      P                  ^ 4      ,           pVP                  RV P                  WU4      pV P                  V4      pMV P                  V4      pV P                  V4      pWº,          P                  ^^4      P                  WEV4      pV P                  V4      pV P                  V4      pV# )z‹
Args:
    x: input features with shape of (num_windows*b, n, c)
    mask: (0/-inf) mask with shape of (num_windows, Wh*Ww, Wh*Ww) or None
éþÿÿÿrt   )r   r“   ÚreshaperŽ   rv   r�   Ú	transposer’   ru   rx   rw   Ú	unsqueezer˜   r”   r•   r–   )r,   r!   ÚrpiÚmaskÚb_Únr|   r“   ÚqÚkÚvÚattnÚrelative_position_biasÚnws   &&&&          r   r1   ÚWindowAttention.forwardÅ   sò  € ð —7‘7‰ˆˆqà�H‰H�Q‹Kß‰W�R˜A˜tŸ~™~¨q·N±NÕ/BÓCß‰W�Q˜˜1˜a Ó#ð 	ð ��FØ��FØ��Fð ˆð —
‘
�NˆØ—;‘;˜r 2Ó&Õ&ˆà!%×!BÑ!BÀ3Ç8Á8ÈBÃ<Õ!P×!UÑ!UØ×Ñ˜QÕ $×"2Ñ"2°1Õ"5Õ5Ø×Ñ˜QÕ $×"2Ñ"2°1Õ"5Õ5Øó"
Ðð
 "8×!?Ñ!?Øˆq�!ó"
ç
‰*‹,ð 	ð ×6Ñ6°qÓ9Õ9ˆàÒØ—‘˜A•ˆBØ—9‘9˜R�X r¯>©>¸1Ó@À4Ç>Á>ØóDç‰i˜‹lõˆDð —9‘9˜R §¡°Ó6ˆDØ—<‘< Ó%‰Dà—<‘< Ó%ˆDà�~‰~˜dÓ#ˆà�X× Ñ   AÓ&×.Ñ.¨r°aÓ8ˆØ�I‰I�a‹LˆØ�N‰N˜1ÓˆØˆr   )
r”   rŒ   rŽ   r•   r–   r“   r’   r�   r˜   rx   )TNr   r   r)   r3   r;   s   @r   rƒ   rƒ   ˜   s   ø† ñ÷*÷>.ò .r   rƒ   c                  ó~   a € ] tR t^ötRt^^ ^^RRRRRRR]P                  ]P                  3R V 3R llltR	 t	R
t
V ;t# )ÚHABaY  Hybrid Attention Block.

Args:
    dim (int): Number of input channels.
    input_resolution (tuple[int]): Input resolution.
    num_heads (int): Number of attention heads.
    window_size (int): Window size.
    shift_size (int): Shift size for SW-MSA.
    mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
    qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True
    qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set.
    drop (float, optional): Dropout rate. Default: 0.0
    attn_drop (float, optional): Attention dropout rate. Default: 0.0
    drop_path (float, optional): Stochastic depth rate. Default: 0.0
    act_layer (nn.Module, optional): Activation layer. Default: nn.GELU
    norm_layer (nn.Module, optional): Normalization layer.  Default: nn.LayerNorm
ç{®Gáz„?ç      @TNr   c               ó    € V ^8„  d   QhRRRR/# rW   r   )r   s   "r   r   ÚHAB.__annotate__	  s    € ÷ :
ñ :
ð $ð:
ð $ñ:
r   c           
     	ó¶  <€ \         SV `  4        Wn        W n        W0n        W@n        WPn        W�n        \        V P                  4      V P
                  8:  d"   ^ V n        \        V P                  4      V n        ^ V P                  u;8:  d   V P
                  8  g    Q R4       h Q R4       hV! V4      V n	        \        V\        V P
                  4      VV
VVVR7      V n        W€n        \        WVR7      V n        VR8”  d   \!        V4      M\"        P$                  ! 4       V n        V! V4      V n        \+        W,          4      p\-        VVVVR7      V n        R# )r   z shift_size must in 0-window_size)rx   rŽ   r™   rš   r”   r–   )rM   rX   r?   r   )rl   rm   ro   rk   N)r*   r+   rŒ   Úinput_resolutionrŽ   rx   Ú
shift_sizeÚ	mlp_ratioÚminÚnorm1rƒ   r   r©   Ú
conv_scalerU   Ú
conv_blockr'   rF   ÚIdentityr%   Únorm2r€   rd   Úmlp)r,   rŒ   r´   rŽ   rx   rµ   rX   r?   r¹   r¶   r™   rš   rk   r”   r%   ro   Ú
norm_layerÚmlp_hidden_dimr-   s   &&&&&&&&&&&&&&&&& €r   r+   ÚHAB.__init__	  s6  ø€ ô& 	‰ÑÔØŒØ 0ÔØ"ŒØ&ÔØ$ŒØ"ŒÜˆt×$Ñ$Ó%¨×)9Ñ)9Ô9àˆDŒOÜ" 4×#8Ñ#8Ó9ˆDÔà�—‘Ö3 4×#3Ñ#3Õ3ð	.à-ó	.Ù3ð	.à-ó	.Ø3ñ   “_ˆŒ
Ü#ØÜ! $×"2Ñ"2Ó3ØØØØØô
ˆŒ	ð %ŒÜØÈô
ˆŒð 1:¸C´œ )Ô,ÄRÇ[Â[Ã]ˆŒÙ “_ˆŒ
Ü˜S�_Ó-ˆÜØØ*ØØô	
ˆŽr   c                	ób  € Vw  rVVP                   w  rxp	Tp
V P                  V4      pVP                  WuWi4      pV P                  VP	                  ^ ^^^4      4      pVP	                  ^ ^^^4      P                  4       P                  WuV,          V	4      pV P                  ^ 8”  d4   \        P                  ! WP                  ) V P                  ) 3RR7      pTpMTpRp\        WÀP                  4      pVP                  RV P                  V P                  ,          V	4      pV P                  WÓVR7      pVP                  RV P                  V P                  V	4      p\        WàP                  WV4      pV P                  ^ 8”  d0   \        P                  ! WÀP                  V P                  3RR7      pMTpVP                  WuV,          V	4      pW P                  V4      ,           W°P                  ,          ,           pWP                  V P                  V P!                  V4      4      4      ,           pV# )r   )ÚshiftsÚdimsN)r¢   r£   )r   r   rt   )r   r¸   ru   rº   rv   rw   rµ   r   Úrollr~   rx   r©   r�   r%   r¹   r½   r¼   )r,   r!   Úx_sizeÚrpi_saÚ	attn_maskrz   r{   ry   Ú_r|   ÚshortcutÚconv_xÚ	shifted_xÚ	x_windowsÚattn_windowsÚattn_xs   &&&&&           r   r1   ÚHAB.forwardE  sã  € Ø‰ˆØ—'‘'‰ˆˆað ˆØ�J‰J�q‹MˆØ�F‰F�1˜Óˆð —‘ §¡¨1¨a°°AÓ!6Ó7ˆØ—‘  1 a¨Ó+×6Ñ6Ó8×=Ñ=¸aÀQÅÈÓJˆð �?‰?˜QÔÜŸ
š
ØŸO™OÐ+¨d¯o©oÐ-=Ð>ÀVôˆIð "‰IàˆIØˆIô %Ø×'Ñ'ó
ˆ	ð —N‘NØ�× Ñ  4×#3Ñ#3Õ3°Qó
ˆ	ð
 —y‘y ¸Y�yÓGˆð $×(Ñ(¨¨T×-=Ñ-=¸t×?OÑ?OÐQRÓSˆÜ" <×1AÑ1AÀ1ÓHˆ	ð �?‰?˜QÔÜ—Z’ZØ§?¡?°D·O±OÐ"DÈ6ô‰Fð ˆFØ—‘˜Q A¥ qÓ)ˆð —~‘~ fÓ-Õ-°¿¹Õ0HÕHˆØ—‘˜tŸx™x¨¯
©
°1«Ó6Ó7Õ7ˆàˆr   )r©   rº   r¹   rŒ   r%   r´   r½   r¶   r¸   r¼   rŽ   rµ   rx   )r4   r5   r6   r7   r8   rF   r\   Ú	LayerNormr+   r1   r9   r:   r;   s   @r   r®   r®   ö   sN   ø† ñð. ØØ&'Ø&(ØØØØØØØØ—'‘'Ø—<‘<÷#:
ò :
÷x3ð 3r   r®   c                  óN   a € ] tR tRtRRR]P
                  3V 3R lltR tRtV ;t	# )ÚOCABi{  TNg       @c
                	óª  <€ \         SV `  4        Wn        W n        W0n        WPn        W,          p
T;'       g
    V
R,          V n        \        W4,          4      V,           V n        V	! V4      V n	        \        P                  ! W^,          VR7      V n        \        P                  ! V P                  V P                  3VV P                  V,
          ^,          R7      V n        \        P                  ! \         P"                  ! W0P                  ,           ^,
          W0P                  ,           ^,
          ,          V4      4      V n        \'        V P$                  RR7       \        P(                  ! R	R7      V n        \        P                  ! W4      V n        V	! V4      V n        \        W,          4      p\1        W\        P2                  R7      V n        R# )
r…   r†   )Úkernel_sizeÚstriderC   rˆ   r‰   r‹   )rl   rm   ro   Nr�   rt   )r*   r+   rŒ   r´   rx   rŽ   r�   r€   Úoverlap_win_sizer¸   rF   rf   r“   ÚUnfoldÚunfoldr�   r   r‘   r’   r   r—   r˜   r•   r¼   rd   r\   r½   )r,   rŒ   r´   rx   Úoverlap_ratiorŽ   r™   rš   r¶   r¾   r›   r¿   r-   s   &&&&&&&&&&  €r   r+   ÚOCAB.__init__~  sX  ø€ ô 	‰ÑÔØŒØ 0ÔØ&ÔØ"ŒØÕ#ˆØ×/Ð/ ¨4¥ˆŒ
Ü # KÕ$?Ó @À;Õ NˆÔá “_ˆŒ
Ü—9’9˜S¨¥'°Ô9ˆŒÜ—i’iØ×.Ñ.°×0EÑ0EÐFØØ×*Ñ*¨[Õ8¸QÕ>ô
ˆŒô -/¯LªLÜ�KŠKØ×4Ñ4Õ4°qÕ8Ø×!6Ñ!6Õ6¸Õ:õ<àóó-
ˆÔ)ô 	�d×7Ñ7¸TÕBÜ—z’z bÔ)ˆŒä—I’I˜cÓ'ˆŒ	á “_ˆŒ
Ü˜S�_Ó-ˆÜØÄrÇwÁwô
ˆŽr   c           	     	ó¬  € Vw  rEVP                   w  rgpTp	V P                  V4      pVP                  WdWX4      pV P                  V4      P	                  WdV^V4      P                  ^^ ^^^4      p
V
^ ,          P                  ^ ^^^4      p\        P                  ! V
^,          V
^,          3^R7      p\        W°P                  4      pVP                  RV P                  V P                  ,          V4      pV P                  V4      p\        VR^VV P                  V P                  R7      P                  4       pV^ ,          V^,          ppVP                   w  pppVP                   w  pppV P                  V P                  ,          pVP	                  VVV P                  V4      P                  ^ ^^^4      pVP	                  VVV P                  V4      P                  ^ ^^^4      pVP	                  VVV P                  V4      P                  ^ ^^^4      pW°P                   ,          pVVP#                  RR4      ,          pV P$                  VP                  R4      ,          P                  V P                  V P                  ,          V P                  V P                  ,          R4      pVP                  ^^ ^4      P                  4       pVVP'                  ^ 4      ,           pV P)                  V4      pVV,          P#                  ^^4      P	                  VVV P                  4      pVP                  RV P                  V P                  V P                  4      p\+        VV P                  WE4      pVP                  WdV,          V P                  4      pV P-                  V4      V	,           pWP/                  V P1                  V4      4      ,           pV# )r[   r‹   z.b (nc ch owh oww) nw -> nc (b nw) (owh oww) ch)ÚncÚchÚowhÚowwrt   rž   )r   r¸   ru   r“   rŸ   rv   r   Úcatr~   rx   rØ   r   rÖ   rw   rŒ   rŽ   r�   r    r’   r¡   r˜   r�   r•   r½   r¼   )r,   r!   rÅ   r¢   rz   r{   ry   rÈ   r|   rÉ   r“   r¦   ÚkvÚ	q_windowsÚ
kv_windowsÚ	k_windowsÚ	v_windowsr¤   Únqr¥   Údr§   r¨   r©   rª   rÍ   s   &&&&                      r   r1   ÚOCAB.forward¯  s[  € Ø‰ˆØ—'‘'‰ˆˆaàˆØ�J‰J�q‹MˆØ�F‰F�1˜Óˆà�h‰h�q‹k×!Ñ! !¨¨1¨aÓ0×8Ñ8¸¸A¸qÀ!ÀQÓGˆØ��F�N‰N˜1˜a  AÓ&ˆÜ�YŠY˜˜A�  A¥Ð'¨QÔ/ˆô %Ø×Ñó
ˆ	ð —N‘NØ�× Ñ  4×#3Ñ#3Õ3°Qó
ˆ	ð —[‘[ “_ˆ
ÜØØ<ØØØ×%Ñ%Ø×%Ñ%ô
÷ ‰*‹,ð 	ð"  *¨!�}¨j¸­m�9ˆ	à—O‘O‰	ˆˆB�Ø—/‘/‰ˆˆ1ˆaØ�H‰H˜Ÿ™Õ&ˆØ×Ñ˜b " d§n¡n°aÓ8×@Ñ@Øˆq�!�Qó
ˆð ×Ñ˜b ! T§^¡^°QÓ7×?Ñ?Øˆq�!�Qó
ˆð ×Ñ˜b ! T§^¡^°QÓ7×?Ñ?Øˆq�!�Qó
ˆð —
‘
�NˆØ�1—;‘;˜r 2Ó&Õ&ˆà!%×!BÑ!BÀ3Ç8Á8ÈBÃ<Õ!P×!UÑ!UØ×Ñ˜t×/Ñ/Õ/Ø×!Ñ! D×$9Ñ$9Õ9Øó"
Ðð
 "8×!?Ñ!?Øˆq�!ó"
ç
‰*‹,ð 	ð Ð,×6Ñ6°qÓ9Õ9ˆà�|‰|˜DÓ!ˆØ˜q�×+Ñ+¨A¨qÓ1×9Ñ9¸"¸bÀ$Ç(Á(ÓKˆð $×(Ñ(Ø�× Ñ  $×"2Ñ"2°D·H±Hó
ˆô ˜<¨×)9Ñ)9¸1Ó@ˆØ�F‰F�1˜!•e˜TŸX™XÓ&ˆà�I‰I�a‹L˜8Õ#ˆà—‘˜Ÿ™ A›Ó'Õ'ˆØˆr   )rŒ   r´   r½   r¸   r¼   rŽ   rÖ   r•   r“   r’   r�   r˜   rØ   rx   )
r4   r5   r6   r7   rF   rÐ   r+   r1   r9   r:   r;   s   @r   rÒ   rÒ   {  s'   ø† ð ØØØ—<‘<÷/
÷bOð Or   rÒ   c            	      ób   a € ] tR tRtRtRRRRRR]P                  RR3	R V 3R	 llltR
 tRt	V ;t
# )ÚAttenBlocksi  aë  A series of attention blocks for one RHAG.

Args:
    dim (int): Number of input channels.
    input_resolution (tuple[int]): Input resolution.
    depth (int): Number of blocks.
    num_heads (int): Number of attention heads.
    window_size (int): Local window size.
    mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
    qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True
    qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set.
    drop (float, optional): Dropout rate. Default: 0.0
    attn_drop (float, optional): Attention dropout rate. Default: 0.0
    drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0
    norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm
    downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None
    use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False.
r°   TNr   Fc               ó    € V ^8„  d   QhRRRR/# rW   r   )r   s   "r   r   ÚAttenBlocks.__annotate__  s$   € ÷ H#ñ H#ð $ðH#ð $ñH#r   c                	óÎ  <€ \         SV `  4        Wn        W n        W0n        VV n        \        P                  ! \        V4       Uu. uFS  p\        TTTTV^,          ^ 8X  d   ^ MV^,          TTTT
TTTT\        V\        4      '       d
   VV,          MTVR7      NKU  	  up4      V n        \        VVVV	VVVV
VR7	      V n        Ve   V! W!VR7      V n        R# RV n        R# u upi )r   )rŒ   r´   rŽ   rx   rµ   rX   r?   r¹   r¶   r™   rš   rk   r”   r%   r¾   )	rŒ   r´   rx   rÙ   rŽ   r™   rš   r¶   r¾   N)rŒ   r¾   )r*   r+   rŒ   r´   ÚdepthÚuse_checkpointrF   Ú
ModuleListÚranger®   Ú
isinstanceÚlistÚblocksrÒ   Úoverlap_attnÚ
downsample)r,   rŒ   r´   rî   rŽ   rx   rX   r?   r¹   rÙ   r¶   r™   rš   rk   r”   r%   r¾   rö   rï   Úir-   s   &&&&&&&&&&&&&&&&&&& €r   r+   ÚAttenBlocks.__init__  s
  ø€ ô* 	‰ÑÔØŒØ 0ÔØŒ
Ø,ˆÔô —m’mô* ˜uœó)ñ( &�Aô' ØØ%5Ø'Ø +Ø%&¨¥U¨a¤Z™q°kÀQÕ6FØ#1Ø#1Ø)Ø'Ø%Ø%ØØ'ä! )¬T×2Ò2ð (¨žlà"Ø)÷#ñ& &ñ)ó
ˆŒô4 !ØØ-Ø#Ø'ØØØØØ!ô

ˆÔð Ò!Ù(Ø °jôˆDŽOð #ˆDŽOùòWs   ÁAC"c                	óÌ   € V P                    F  pV! WVR ,          VR,          4      pK  	  V P                  WVR,          4      pV P                  e   V P                  V4      pV# )rÆ   rÇ   Úrpi_oca)rô   rõ   rö   )r,   r!   rÅ   ÚparamsÚblks   &&&& r   r1   ÚAttenBlocks.forward_  s^   € Ø—;”;ˆCÙ�A˜v hÕ/°¸Õ1DÓEŠAñ ð ×Ñ˜a¨°	Õ):Ó;ˆà�?‰?Ò&Ø—‘ Ó"ˆAØˆr   )rô   rî   rŒ   rö   r´   rõ   rï   ©r4   r5   r6   r7   r8   rF   rÐ   r+   r1   r9   r:   r;   s   @r   rê   rê     s@   ø† ñð< ØØØØØØ—<‘<ØØ÷'H#ò H#÷Tð r   rê   c                  óh   a € ] tR tRtRtRRRRRR]P                  RR^à^R3R	 V 3R
 llltR tRt	V ;t
# )ÚRHAGij  am  Residual Hybrid Attention Group (RHAG).

Args:
    dim (int): Number of input channels.
    input_resolution (tuple[int]): Input resolution.
    depth (int): Number of blocks.
    num_heads (int): Number of attention heads.
    window_size (int): Local window size.
    mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
    qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True
    qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set.
    drop (float, optional): Dropout rate. Default: 0.0
    attn_drop (float, optional): Attention dropout rate. Default: 0.0
    drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0
    norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm
    downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None
    use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False.
    img_size: Input image size.
    patch_size: Patch size.
    resi_connection: The convolutional block before residual connection.
r°   TNr   FÚ1convc               ó    € V ^8„  d   QhRRRR/# rW   r   )r   s   "r   r   ÚRHAG.__annotate__�  s$   € ÷ E
ñ E
ð $ðE
ð $ñE
r   c                	ó¦  <€ \         SV `  4        Wn        W n        \	        R/ R VbRVbRVbRVbRVbRVbRVbRVbRV	bR	V
bR
VbRVbRVbRVbRVbRVbRVbRVb V n        VR8X  d    \        P                  ! W^^^4      V n        M!VR8X  d   \        P                  ! 4       V n        \        VV^ VRR7      V n        \        VV^ VRR7      V n        R# )rŒ   r´   rî   rŽ   rx   rX   r?   r¹   rÙ   r¶   r™   rš   rk   r”   r%   r¾   rö   rï   r  ÚidentityN©Úimg_sizeÚ
patch_sizeÚin_chansÚ	embed_dimr¾   r   )r*   r+   rŒ   r´   rê   Úresidual_grouprF   rI   Úconvr»   Ú
PatchEmbedÚpatch_embedÚPatchUnEmbedÚpatch_unembed)r,   rŒ   r´   rî   rŽ   rx   rX   r?   r¹   rÙ   r¶   r™   rš   rk   r”   r%   r¾   rö   rï   r  r  Úresi_connectionr-   s   &&&&&&&&&&&&&&&&&&&&&&€r   r+   ÚRHAG.__init__�  sG  ø€ ô0 	‰ÑÔàŒØ 0Ôä)ò 
Ùð
á-ð
ñ ð
ñ  ð	
ñ
 $ð
ñ *ð
ñ *ð
ñ "ð
ñ (ð
ñ  ð
ñ ð
ñ ð
ñ ð
ñ  ð
ñ  ð
ñ  "ð!
ñ" "ð#
ñ$ *ð%
ˆÔð* ˜gÔ%ÜŸ	š	 #¨A¨q°!Ó4ˆD�IØ 
Ô*ÜŸš›ˆDŒIä%ØØ!ØØØô
ˆÔô *ØØ!ØØØô
ˆÖr   c                	ó�   € V P                  V P                  V P                  V P                  WV4      V4      4      4      V,           # r)   )r  r  r  r  )r,   r!   rÅ   rû   s   &&&&r   r1   ÚRHAG.forwardÈ  sH   € à×ÑØ—	‘	Ø×&Ñ& t×':Ñ':¸1ÀfÓ'MÈvÓVóóð
 õð	
r   )r  rŒ   r´   r  r  r  rþ   r;   s   @r   r   r   j  sJ   ø† ñðB ØØØØØØ—<‘<ØØØØØ÷-E
ò E
÷N
ð 
r   r   c                  ó6   a € ] tR tRtRtRV 3R lltR tRtV ;t# )r  iÓ  a[  Image to Patch Embedding

Args:
    img_size (int): Image size.  Default: 224.
    patch_size (int): Patch token size. Default: 4.
    in_chans (int): Number of input image channels. Default: 3.
    embed_dim (int): Number of linear projection output channels. Default: 96.
    norm_layer (nn.Module, optional): Normalization layer. Default: None
c                	óT  <€ \         SV `  4        \        V4      p\        V4      pV^ ,          V^ ,          ,          V^,          V^,          ,          .pWn        W n        W`n        V^ ,          V^,          ,          V n        W0n        W@n        Ve   V! V4      V n	        R# RV n	        R# ©r   N)
r*   r+   r   r  r  Úpatches_resolutionÚnum_patchesr	  r
  Únorm©r,   r  r  r	  r
  r¾   r  r-   s   &&&&&& €r   r+   ÚPatchEmbed.__init__Þ  s™   ø€ ô 	‰ÑÔÜ˜XÓ&ˆÜ˜zÓ*ˆ
à�Q�K˜: a�=Õ(Ø�Q�K˜: a�=Õ(ð
Ðð !ŒØ$ŒØ"4ÔØ-¨aÕ0Ð3EÀaÕ3HÕHˆÔà ŒØ"ŒàÒ!Ù" 9Ó-ˆDŽIàˆDŽIr   c                	ó†   € VP                  ^4      P                  ^^4      pV P                  e   V P                  V4      pV# )r   )Úflattenr    r  r0   s   &&r   r1   ÚPatchEmbed.forwardõ  s8   € Ø�I‰I�a‹L×"Ñ" 1 aÓ(ˆØ�9‰9Ò Ø—	‘	˜!“ˆAØˆr   )r
  r  r	  r  r  r  r  ©éà   é   r[   é`   Nr3   r;   s   @r   r  r  Ó  s   ø† ñ÷÷.ð r   r  c                  ó6   a € ] tR tRtRtRV 3R lltR tRtV ;t# )r  iü  a]  Image to Patch Unembedding

Args:
    img_size (int): Image size.  Default: 224.
    patch_size (int): Patch token size. Default: 4.
    in_chans (int): Number of input image channels. Default: 3.
    embed_dim (int): Number of linear projection output channels. Default: 96.
    norm_layer (nn.Module, optional): Normalization layer. Default: None
c                	ó   <€ \         SV `  4        \        V4      p\        V4      pV^ ,          V^ ,          ,          V^,          V^,          ,          .pWn        W n        W`n        V^ ,          V^,          ,          V n        W0n        W@n        R# r  )	r*   r+   r   r  r  r  r  r	  r
  r  s   &&&&&& €r   r+   ÚPatchUnEmbed.__init__  s€   ø€ ô 	‰ÑÔÜ˜XÓ&ˆÜ˜zÓ*ˆ
à�Q�K˜: a�=Õ(Ø�Q�K˜: a�=Õ(ð
Ðð !ŒØ$ŒØ"4ÔØ-¨aÕ0Ð3EÀaÕ3HÕHˆÔà ŒØ"Žr   c                	ó¼   € VP                  ^^4      P                  4       P                  VP                  ^ ,          V P                  V^ ,          V^,          4      pV# r   )r    rw   ru   r   r
  )r,   r!   rÅ   s   &&&r   r1   ÚPatchUnEmbed.forward  sG   € à�K‰K˜˜1Óß‰Z‹\ß‰T�!—'‘'˜!•*˜dŸn™n¨f°Q­i¸À½ÓCð 	
ð
 ˆr   )r
  r  r	  r  r  r  r   r3   r;   s   @r   r  r  ü  s   ø† ñ÷#÷$ð r   r  c                  ó,   a € ] tR tRtRtV 3R ltRtV ;t# )ÚUpsamplei"  z‘Upsample module.

Args:
    scale (int): Scale factor. Supported scales: 2^n and 3.
    num_feat (int): Channel number of intermediate features.
c           
     	ó4  <€ . pW^,
          ,          ^ 8X  d‡   \        \        \        P                  ! V^4      4      4       FX  pVP	                  \
        P                  ! V^V,          ^^^4      4       VP	                  \
        P                  ! ^4      4       KZ  	  MlV^8X  dW   VP	                  \
        P                  ! V^	V,          ^^^4      4       VP	                  \
        P                  ! ^4      4       M\        RV R24      h\        SV `(  ! V!   R# )r   zscale z/ is not supported. Supported scales: 2^n and 3.N)rñ   r€   ÚmathÚlogÚappendrF   rI   ÚPixelShuffleÚ
ValueErrorr*   r+   )r,   r�   rM   ÚmrÈ   r-   s   &&&  €r   r+   ÚUpsample.__init__*  sÔ   ø€ ØˆØ˜Q•YÕ AÔ%Üœ3œtŸxšx¨¨qÓ1Ó2Ö3�Ø—‘œŸš 8¨Q°­\¸1¸aÀÓCÔDØ—‘œŸš¨Ó+Ö,ò 4ð �aŒZØ�H‰H”R—Y’Y˜x¨¨X­°q¸!¸QÓ?Ô@Ø�H‰H”R—_’_ QÓ'Õ(äØ˜˜ÐNÐOóð ô 	‰Ò˜!Ôr   r   )r4   r5   r6   r7   r8   r+   r9   r:   r;   s   @r   r*  r*  "  s   ø† ñ÷ó r   r*  c            4      óà   a € ] tR tRtRt/ tR^@R^R^R^`RR2RR2R	^R
^R^RRRRRRRRRRRRRRRRR]P                  RRRRRRR ^R!R"R#R$R%R&R'^@/R( V 3R) llltR* t	R+ t
R, tR- tR. tR/ tR0 tR1tV ;t# )3ÚHATi:  a«  Hybrid Attention Transformer
    A PyTorch implementation of : `Activating More Pixels in Image Super-Resolution Transformer`.
    Some codes are based on SwinIR.
Args:
    img_size (int | tuple(int)): Input image size. Default 64
    patch_size (int | tuple(int)): Patch size. Default: 1
    in_chans (int): Number of input image channels. Default: 3
    embed_dim (int): Patch embedding dimension. Default: 96
    depths (tuple(int)): Depth of each Swin Transformer layer.
    num_heads (tuple(int)): Number of attention heads in different layers.
    window_size (int): Window size. Default: 7
    mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. Default: 4
    qkv_bias (bool): If True, add a learnable bias to query, key, value. Default: True
    qk_scale (float): Override default qk scale of head_dim ** -0.5 if set. Default: None
    drop_rate (float): Dropout rate. Default: 0
    attn_drop_rate (float): Attention dropout rate. Default: 0
    drop_path_rate (float): Stochastic depth rate. Default: 0.1
    norm_layer (nn.Module): Normalization layer. Default: nn.LayerNorm.
    ape (bool): If True, add absolute position embedding to the patch embedding. Default: False
    patch_norm (bool): If True, add normalization after patch embedding. Default: True
    use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False
    upscale: Upscale factor. 2/3/4/8 for image SR, 1 for denoising and compress artifact reduction
    img_range: Image range. 1. or 255.
    upsampler: The reconstruction reconstruction module. 'pixelshuffle'/'pixelshuffledirect'/'nearest+conv'/None
    resi_connection: The convolutional block before residual connection. '1conv'/'3conv'
r  r  r	  r
  ÚdepthsrŽ   rx   rX   r?   r¹   r¯   rÙ   r…   r¶   r°   r™   Trš   NÚ	drop_rater   Úattn_drop_rateÚdrop_path_rategš™™™™™¹?r¾   ÚapeFÚ
patch_normrï   ÚupscaleÚ	img_rangeç      ð?Ú	upsamplerÚpixelshuffler  r  rM   c               ó$   € V ^8„  d   QhRRRRRR/# )r   rX   r@   r?   r>  zLiteral['pixelshuffle']r   )r   s   "r   r   ÚHAT.__annotate__Y  s/   € ÷ T'ñ T'ð $ðT'ð $ðT'ð4 +ñ5T'r   c               	ó¢  <€ \         S&V `  4        Wpn        V^,          V n        W°n        TpTpVV n        V^8X  d1   R"p\        P                  ! V4      P                  ^^^^4      V n	        M\        P                  ! ^^^^4      V n	        VV n        VV n        V P                  4       pV P                  4       pV P                  RV4       V P                  RV4       \         P"                  ! VV^^^4      V n        \'        V4      V n        W@n        VV n        VV n        W@n        WÀn        \5        TTTTV P.                  '       d   TMRR7      V n        V P6                  P8                  p V P6                  P:                  p!V!V n        \=        TTTTV P.                  '       d   TMRR7      V n        V P,                  '       dI   \         P@                  ! \        P                  ! ^V V4      4      V n!        \E        V PB                  RR7       \         PF                  ! VR7      V n$        \        PJ                  ! ^ V\M        V4      4       U"u. uF  p"V"PO                  4       NK  	  p#p"\         PP                  ! 4       V n)        \U        V P(                  4       F°  p$\W        R#/ RVbR	V!^ ,          V!^,          3bR
VV$,          bRVV$,          bRVbRVbRV	bRV
bRVbRV P2                  bRVbRVbRVbRVbRV#\M        VRV$ 4      \M        VRV$^,            4       bRVbRRbRVbRVbRVbRVb p%V PR                  PY                  V%4       K²  	  V! V P0                  4      V n-        VR8X  d    \         P"                  ! WD^^^4      V n.        M!VR8X  d   \         P^                  ! 4       V n.        V P                  R8X  dz   \         P`                  ! \         P"                  ! VV^^^4      \         Pb                  ! R R!7      4      V n2        \g        VV4      V n4        \         P"                  ! VV^^^4      V n5        V Pm                  V Pn                  4       R# u up"i )$r   Úrelative_position_index_SAÚrelative_position_index_OCANr  rˆ   r‰   )ÚprŒ   r´   rî   rŽ   rx   rX   r?   r¹   rÙ   r¶   r™   rš   rk   r”   r%   r¾   rö   rï   r  r  r  r  r  r?  TrD   )gšwœ¢#¹Ü?gï8EGrùÛ?gB`åÐ"ÛÙ?r   )8r*   r+   rx   rµ   rÙ   r<  r   ÚTensorru   Úmeanr‘   r;  r>  Úcalculate_rpi_saÚcalculate_rpi_ocaÚregister_bufferrF   rI   Ú
conv_firstÚlenÚ
num_layersr
  r9  r:  Únum_featuresr¶   r  r  r  r  r  r  r�   Úabsolute_pos_embedr   rj   Úpos_dropÚlinspaceÚsumÚitemrð   Úlayersrñ   r   r.  r  Úconv_after_bodyr»   rG   Ú	LeakyReLUÚconv_before_upsampler*  ÚupsampleÚ	conv_lastÚapplyÚ_init_weights)'r,   r  r  r	  r
  r5  rŽ   rx   rX   r?   r¹   rÙ   r¶   r™   rš   r6  r7  r8  r¾   r9  r:  rï   r;  r<  r>  r  rM   Ú	num_in_chÚ
num_out_chÚrgb_meanrC  rD  r  r  r!   ÚdprÚi_layerÚlayerr-   s'   &$$$$$$$$$$$$$$$$$$$$$$$$$$           €r   r+   ÚHAT.__init__Y  s1  ø€ ô< 	‰ÑÔà&ÔØ%¨Õ*ˆŒØ*Ôàˆ	Øˆ
à"ˆŒØ�qŒ=Ø/ˆHÜŸš XÓ.×3Ñ3°A°q¸!¸QÓ?ˆD�IäŸš A q¨!¨QÓ/ˆDŒIØˆŒØ"ˆŒð &*×%:Ñ%:Ó%<Ð"Ø&*×&<Ñ&<Ó&>Ð#Ø×ÑÐ9Ð;UÔVØ×ÑÐ:Ð<WÔXô Ÿ)š) I¨y¸!¸QÀÓBˆŒô ˜f›+ˆŒØ"ŒØˆŒØ$ˆŒØ%ÔØ"Œô &ØØ!ØØØ%)§_§_ _‘z¸$ô
ˆÔð ×&Ñ&×2Ñ2ˆØ!×-Ñ-×@Ñ@ÐØ"4ˆÔô *ØØ!ØØØ%)§_§_ _‘z¸$ô
ˆÔð �8�8ˆ8Ü&(§l¢lÜ—’˜A˜{¨IÓ6ó'ˆDÔ#ô ˜$×1Ñ1°tÕ<äŸ
š
 YÔ/ˆŒô $Ÿnšn¨Q°ÄÀFÃÔLó
ÙL˜ˆA�F‰FŽHÑLð 	ð 
ô
 —m’m“oˆŒÜ˜TŸ_™_Ö-ˆGÜò Ùðà"4°QÕ"7Ð9KÈAÕ9NÑ!Oðð ˜W–oðð $ GÖ,ð	ñ
 (ðñ  .ðñ  .ðñ &ðñ ,ðð Ÿ.š.ðñ "ðñ "ðñ ðñ )ðð Ü˜˜x Ð(Ó)¬CØ˜} ¨1¥Ð-ó-ñðñ( &ð)ñ*  ð+ñ,  .ð-ñ. "ð/ñ0 &ð1ñ2 !0ð3ˆEð6 �K‰K×Ñ˜uÖ%ñ9 .ñ: ˜t×0Ñ0Ó1ˆŒ	ð ˜gÔ%Ü#%§9¢9¨YÀ1ÀaÈÓ#KˆDÕ Ø 
Ô*Ü#%§;¢;£=ˆDÔ ð �>‰>˜^Ô+ä(*¯ªÜ—	’	˜) X¨q°!°QÓ7¼¿ºÈdÔ9Só)ˆDÔ%ô % W¨hÓ7ˆDŒMÜŸYšY x°¸QÀÀ1ÓEˆDŒNà�
‰
�4×%Ñ%Ö&ùòi
s   É	Qc                	ó  € \        V\        P                  4      '       dv   \        VP                  R R7       \        V\        P                  4      '       d=   VP
                  e-   \        P                  P                  VP
                  ^ 4       R# R# R# \        V\        P                  4      '       dW   \        P                  P                  VP
                  ^ 4       \        P                  P                  VP                  R4       R# R# )rˆ   r‰   Nr=  )	rò   rF   rf   r   Úweightr‡   ÚinitÚ	constant_rÐ   )r,   r1  s   &&r   r[  ÚHAT._init_weightsï  s¡   € Ü�aœŸ™×#Ò#Ü˜!Ÿ(™(¨Õ-Ü˜!œRŸY™Y×'Ò'¨A¯F©FÒ,>Ü—‘×!Ñ! !§&¡&¨!Ö,ñ -?Ñ'ä˜œ2Ÿ<™<×(Ò(Ü�G‰G×Ñ˜aŸf™f aÔ(Ü�G‰G×Ñ˜aŸh™h¨Ö,ñ )r   c                	óŽ  € \         P                  ! V P                  4      p\         P                  ! V P                  4      p\         P                  ! \         P                  ! W.4      4      p\         P
                  ! V^4      pVR,          VR,          ,
          pVP                  ^^^ 4      P                  4       pVR;;,          V P                  ^,
          ,          uu&   VR;;,          V P                  ^,
          ,          uu&   VR;;,          ^V P                  ,          ^,
          ,          uu&   VP                  R4      pV# )r   ©ºNNNrj  N©rj  Nrj  ©rj  rj  r   ©rj  rj  r   rt   )	r   Úarangerx   ÚstackÚmeshgridr  rv   rw   rR  )r,   Úcoords_hÚcoords_wÚcoordsÚcoords_flattenÚrelative_coordsÚrelative_position_indexs   &      r   rH  ÚHAT.calculate_rpi_saø  sý   € ä—<’< × 0Ñ 0Ó1ˆÜ—<’< × 0Ñ 0Ó1ˆÜ—’œUŸ^š^¨XÐ,@ÓAÓBˆÜŸš v¨qÓ1ˆà˜:Õ&¨¸
Õ)CÕCð 	ð *×1Ñ1Øˆq�!ó
ç
‰*‹,ð 	ð 	˜×  D×$4Ñ$4°qÕ$8Õ8Ó Ø˜×  D×$4Ñ$4°qÕ$8Õ8Ó Ø˜×  A¨×(8Ñ(8Õ$8¸1Õ$<Õ<Ó Ø"1×"5Ñ"5°bÓ"9ÐØ&Ð&r   c                	ó¢  € V P                   pV P                   \        V P                  V P                   ,          4      ,           p\        P                  ! V4      p\        P                  ! V4      p\        P
                  ! \        P                  ! W4.4      4      p\        P                  ! V^4      p\        P                  ! V4      p\        P                  ! V4      p\        P
                  ! \        P                  ! W4.4      4      p\        P                  ! V^4      pVR,          VR,          ,
          p	V	P                  ^^^ 4      P                  4       p	V	R;;,          W,
          ^,           ,          uu&   V	R;;,          W,
          ^,           ,          uu&   V	R;;,          W,           ^,
          ,          uu&   V	P                  R4      p
V
# )r   rk  ri  rl  rm  rt   )rx   r€   rÙ   r   rn  ro  rp  r  rv   rw   rR  )r,   Úwindow_size_oriÚwindow_size_extrq  rr  Ú
coords_oriÚcoords_ori_flattenÚ
coords_extÚcoords_ext_flattenru  rv  s   &          r   rI  ÚHAT.calculate_rpi_oca
  sd  € à×*Ñ*ˆØ×*Ñ*¬S°×1CÑ1CÀd×FVÑFVÕ1VÓ-WÕWˆä—<’< Ó0ˆÜ—<’< Ó0ˆÜ—[’[¤§¢°Ð0DÓ!EÓFˆ
Ü"Ÿ]š]¨:°qÓ9Ðä—<’< Ó0ˆÜ—<’< Ó0ˆÜ—[’[¤§¢°Ð0DÓ!EÓFˆ
Ü"Ÿ]š]¨:°qÓ9Ðð ˜zÕ*Ð-?À
Õ-KÕKð 	ð *×1Ñ1Øˆq�!ó
ç
‰*‹,ð 	ð 	˜× ØÕ-°Õ1õ	
Ó ð 	˜×  OÕ$EÈÕ$IÕIÓ à˜×  OÕ$EÈÕ$IÕIÓ Ø"1×"5Ñ"5°bÓ"9ÐØ&Ð&r   c                	óê  € Vw  r4\         P                  ! ^W4^3VR7      p\        ^ V P                  ) 4      \        V P                  ) V P                  ) 4      \        V P                  ) R4      3p\        ^ V P                  ) 4      \        V P                  ) V P                  ) 4      \        V P                  ) R4      3p^ pV F  pV F  pW…RW4R3&   V^,          pK  	  K  	  \        WPP                  4      p	V	P                  RV P                  V P                  ,          4      p	V	P                  ^4      V	P                  ^4      ,
          p
V
P                  V
^ 8g  R4      P                  V
^ 8H  R4      p
V
# )r   ©r   Nrj  r   rt   g      YÀ)	r   r‘   Úslicerx   rµ   r~   ru   r¡   Úmasked_fill)r,   rÅ   r   rz   r{   Úimg_maskÚh_slicesÚw_slicesÚcntÚmask_windowsrÇ   s   &&&        r   Úcalculate_maskÚHAT.calculate_mask)  si  € à‰ˆÜ—;’;  1¨˜|°5Ô9ˆä�!�d×&Ñ&Ð&Ó'Ü�4×#Ñ#Ð# d§o¡oÐ%5Ó6Ü�4—?‘?Ð" DÓ)ð
ˆô �!�d×&Ñ&Ð&Ó'Ü�4×#Ñ#Ð# d§o¡oÐ%5Ó6Ü�4—?‘?Ð" DÓ)ð
ˆð
 ˆÛˆAÛ�Ø'*˜˜A !˜Ñ$Ø�q•’ó ñ ô
 (Ø×&Ñ&ó
ˆð $×(Ñ(¨¨T×-=Ñ-=À×@PÑ@PÕ-PÓQˆØ ×*Ñ*¨1Ó-°×0FÑ0FÀqÓ0IÕIˆ	Ø×)Ñ)¨)°q©.¸&ÓA×MÑMØ˜‰N˜Có
ˆ	ð Ðr   c                	ó0   € \        WP                  R R7      # )Úreflect)Úmode)r	   rx   r0   s   &&r   Úcheck_image_sizeÚHAT.check_image_sizeH  s   € Ü˜q×"2Ñ"2¸ÔCÐCr   c                	óü  € VP                   ^,          VP                   ^,          3pV P                  W!P                  R7      P                  VP                  4      pRVRV P
                  RV P                  /pV P                  V4      pV P                  '       d   WP                  ,           pV P                  V4      pV P                   F  pV! WV4      pK  	  V P                  V4      pV P                  W4      pV# )r   r�  rÇ   rÆ   rú   )r   r‰  r   Útor   rC  rD  r  r9  rO  rP  rT  r  r  )r,   r!   rÅ   rÇ   rû   ra  s   &&    r   Úforward_featuresÚHAT.forward_featuresK  sÛ   € Ø—'‘'˜!•*˜aŸg™g a�jÐ)ˆð ×'Ñ'¨·g±gÐ'Ó>×AÑAÀ!Ç(Á(ÓKˆ	à˜Ø�d×5Ñ5Ø�t×7Ñ7ð
ˆð ×Ñ˜QÓˆØ�8�8ˆ8Ø×+Ñ+Õ+ˆAØ�M‰M˜!Óˆà—[”[ˆEÙ�a Ó(ŠAñ !ð �I‰I�a‹LˆØ×Ñ˜qÓ)ˆàˆr   c                	óp  € VP                   R ,          w  r#V P                  P                  V4      V n        WP                  ,
          V P                  ,          pV P	                  V4      pV P
                  R8X  dj   V P                  V4      pV P                  V P                  V4      4      V,           pV P                  V4      pV P                  V P                  V4      4      pWP                  ,          V P                  ,           pVRRRW P                  ,          1RW0P                  ,          13,          # ):r   NNr?  rj  N)r   rG  Útype_asr<  rŽ  r>  rK  rU  r’  rW  rY  rX  r;  )r,   r!   ÚHÚWs   &&  r   r1   ÚHAT.forwardd  sï   € Ø�w‰w�r�{‰ˆØ—I‘I×%Ñ% aÓ(ˆŒ	Ø—‘�]˜dŸn™nÕ,ˆØ×!Ñ! !Ó$ˆà�>‰>˜^Ô+à—‘ Ó"ˆAØ×$Ñ$ T×%:Ñ%:¸1Ó%=Ó>ÀÕBˆAØ×)Ñ)¨!Ó,ˆAØ—‘˜tŸ}™}¨QÓ/Ó0ˆAà—‘Õ §¡Õ*ˆà��AÐ)˜Ÿ\™\Õ)Ð)Ð+=¨Q·±Õ-=Ð+=Ð=Õ>Ð>r   )rO  r9  rU  rW  rK  rY  r
  r<  rT  rG  r¶   r  rN  rM  rÙ   r  r:  r  r  rP  rµ   rX  r>  r;  rx   )é   r™  r™  r™  )r4   r5   r6   r7   r8   ÚhyperparametersrF   rÐ   r+   r[  rH  rI  r‰  rŽ  r’  r1   r9   r:   r;   s   @r   r4  r4  :  sg  ø† ñð6 €OðT'ð ðT'ð ð	T'ð
 ðT'ð ðT'ð ðT'ð ðT'ð ðT'ð '(ðT'ð ')ðT'ð ðT'ð ðT'ð ðT'ð ðT'ð  ð!T'ð" ð#T'ð$ ð%T'ð& ð'T'ð( —<‘<ð)T'ð* ð+T'ð, ð-T'ð. ð/T'ð0 ð1T'ð2 ð3T'ð4 .<ð5T'ð6  ð7T'ð8 ÷9T'ò T'òl-ò'ò$'ò>ò>Dò÷2?ð ?r   r4  )r   F)#Ú
__future__r   r,  Útypingr   r   Útorch.nnrF   Úeinopsr   Úspandrel.utilr   Úspandrel.util.timmr   r   Ú__arch_helpers.paddingr	   r%   ÚModuler'   r=   rU   rd   r~   r�   rƒ   r®   rÒ   rê   r   r  r  rG   r*  r4  r   r   r   Ú<module>r£     s%  ðå "ã Ý ã Ý Ý å /ß 7å 5÷ô";ˆr�y‰yô ;ô�r—y‘yô ô,ˆ"�)‰)ô ô(ˆ"�)‰)ô ò4ò"ô&[�b—i‘iô [ô|Bˆ"�)‰)ô BôJCˆ2�9‰9ô CôLf�"—)‘)ô fôRf
ˆ2�9‰9ô f
ôR&�—‘ô &ôR#�2—9‘9ô #ôLˆr�}‰}ô ñ0 Óôx?ˆ"�)‰)ó x?ó òx?r   