i>ddlZddlZddlmcmZddlmZmZddlmZ ddede de de d e d e d efd Z ddede de de d e d e d efd Z ejjdGddej ZejjdGddeZy)N)nnTensor)_log_api_usage_onceinputp block_sizeinplaceepstrainingreturnctjjs-tjjst t |dks|dkDrt d|d|jdk7rt d|jd|r|dk(r|S|j\}}}} t|| |}|dzd k(rt d |d ||z| z|dz||z d z| |z d zzzz } tj||||z d z| |z d zf|j|j } | j| tj| |dzgdzd } tj | d||f|dz} d | z } | j#|| j%zz } |r"|j'| j'| |S|| z| z}|S)a Implements DropBlock2d from `"DropBlock: A regularization method for convolutional networks" `. Args: input (Tensor[N, C, H, W]): The input tensor or 4-dimensions with the first one being its batch i.e. a batch with ``N`` rows. p (float): Probability of an element to be dropped. block_size (int): Size of the block to drop. inplace (bool): If set to ``True``, will do this operation in-place. Default: ``False``. eps (float): A value added to the denominator for numerical stability. Default: 1e-6. training (bool): apply dropblock if is ``True``. Default: ``True``. Returns: Tensor[N, C, H, W]: The randomly zeroed tensor after dropblock. ?4drop probability has to be between 0 and 1, but got .z#input should be 4 dimensional. Got dimensions.rrblock size should be odd. Got  which is even.dtypedevicevalue)rrstride kernel_sizepadding)torchjit is_scripting is_tracingr drop_block2d ValueErrorndimsizeminemptyrr bernoulli_Fpad max_pool2dnumelsummul_) rrr r r r NCHWgammanoisenormalize_scales T/var/www/html/engine/venv/lib/python3.12/site-packages/torchvision/ops/drop_block.pyr%r% s& 99 ! ! #EII,@,@,BL)3w!c'OPQsRSTUU zzQ>uzzl,WXX qCx JAq!QZA&JA~9*_UVVUQYJMq:~/Aa*nWXFX.YZ [E KKAq:~11z>A3EFekkbgbnbn oE U EE%*/*Q.a 8E LLvJ ;S]gkl]l mE IEkkmsUYY['89O 5/ L / Lctjjs-tjjst t |dks|dkDrt d|d|jdk7rt d|jd|r|dk(r|S|j\}}}} } t||| | }|dzd k(rt d |d ||z| z| z|d z||z d z| |z d zz| |z d zzzz } tj||||z d z| |z d z| |z d zf|j|j} | j| tj| |dzgdzd } tj | d|||f|dz} d | z } | j#|| j%zz } |r"|j'| j'| |S|| z| z}|S)a  Implements DropBlock3d from `"DropBlock: A regularization method for convolutional networks" `. Args: input (Tensor[N, C, D, H, W]): The input tensor or 5-dimensions with the first one being its batch i.e. a batch with ``N`` rows. p (float): Probability of an element to be dropped. block_size (int): Size of the block to drop. inplace (bool): If set to ``True``, will do this operation in-place. Default: ``False``. eps (float): A value added to the denominator for numerical stability. Default: 1e-6. training (bool): apply dropblock if is ``True``. Default: ``True``. Returns: Tensor[N, C, D, H, W]: The randomly zeroed tensor after dropblock. rrrrz#input should be 5 dimensional. Got rrrrrrrr)rrrr)r!r"r#r$r drop_block3dr&r'r(r)r*rrr+r,r- max_pool3dr/r0r1)rrr r r r r2r3Dr4r5r6r7r8s r9r?r?:s & 99 ! ! #EII,@,@,BL)3w!c'OPQsRSTUU zzQ>uzzl,WXX qCx JJLMAq!QZAq)JA~9*_UVVUQY] A 1z>A3E!j.[\J\2]abeoaorsas2tu vE KK Aq:~!1z>A#5q:~7IJRWR]R]fkfrfr E U EE%*/*Q.a 8E LL ij*j-Q[eij[j E IEkkmsUYY['89O 5/ L / Lr:r%c VeZdZdZd dededededdf fd Zd edefd Z de fd Z xZ S) DropBlock2dz# See :func:`drop_block2d`. rr r r r NcZt|||_||_||_||_yN)super__init__rr r r selfrr r r __class__s r9rGzDropBlock2d.__init__ws* $ r:rct||j|j|j|j|j Sz Args: input (Tensor): Input feature map on which some areas will be randomly dropped. Returns: Tensor: The tensor after DropBlock layer. )r%rr r r r rIrs r9forwardzDropBlock2d.forward0E4664??DLL$((TXTaTabbr:c|jjd|jd|jd|jd}|S)Nz(p=z , block_size=z , inplace=))rJ__name__rr r )rIss r9__repr__zDropBlock2d.__repr__sC~~&&'s466(-?PPZ[_[g[gZhhi jr:Fư>) rR __module__ __qualname____doc__floatintboolrGrrNstrrT __classcell__rJs@r9rCrCrsS%S4e`dcVcc#r:rCr?c JeZdZdZd dededededdf fd Zd edefd Z xZ S) DropBlock3dz# See :func:`drop_block3d`. rr r r r Nc*t|||||yrE)rFrGrHs r9rGzDropBlock3d.__init__s J5r:rct||j|j|j|j|j SrL)r?rr r r r rMs r9rNzDropBlock3d.forwardrOr:rU) rRrWrXrYrZr[r\rGrrNr^r_s@r9rarasG6%6S646e6`d6cVccr:ra)FrVT)r!torch.fxtorch.nn.functionalr functionalr,rutilsrrZr[r\r%r?fxwrapModulerCrar:r9rls 'ko. ..),.7;.JO.cg. .dko2 22),27;2JO2cg2 2j n"))8 nc+cr: