:j;" ddlmZddlmZmZddlZddlmZddlmcm Z ddl m Z ddl m Z ddlmZmZmZdd lmZdd lmZmZgd ZGd d ej0ZGddej0Zdedeedededef dZedddZGddeZGddeZ eedejBfddd deedededefd!Z"eede jBfddd dee dededefd"Z#y)#)partial)AnyOptionalN)ImageClassification)_log_api_usage_once)register_modelWeights WeightsEnum)_IMAGENET_CATEGORIES)_ovewrite_named_paramhandle_legacy_interface) SqueezeNetSqueezeNet1_0_WeightsSqueezeNet1_1_Weights squeezenet1_0 squeezenet1_1c leZdZdededededdf fd Zdej dej fd ZxZS) Fireinplanessqueeze_planesexpand1x1_planesexpand3x3_planesreturnNct|||_tj||d|_tj d|_tj||d|_tj d|_ tj||dd|_ tj d|_ y)Nr  kernel_sizeTinplace)rpadding) super__init__rnnConv2dsqueezeReLUsqueeze_activation expand1x1expand1x1_activation expand3x3expand3x3_activation)selfrrrr __class__s r/home/mcse/projects/srt_converter/srt-converter-venv/lib/python3.12/site-packages/torchvision/models/squeezenet.pyr$z Fire.__init__s   yy>qI "$''$"7>3CQRS$&GGD$9!>3CQR\]^$&GGD$9!xc|j|j|}tj|j |j ||j |j|gdSNr )r)r'torchcatr+r*r-r,r.r2s r0forwardz Fire.forwardsb  # #DLLO 4yy  & &t~~a'8 94;T;TUYUcUcdeUf;g hjk  r1) __name__ __module__ __qualname__intr$r5Tensorr8 __classcell__r/s@r0rrsH::c:S:dg:lp:  %,, r1rc jeZdZd dedededdffd ZdejdejfdZ xZ S) rversion num_classesdropoutrNct|t|||_|dk(rt j t j ddddt jdt jddd tdd d d td d d d td d d d t jddd tdd d d tddddtddddtdd ddt jddd tdd dd |_ n|dk(rt j t j dd ddt jdt jddd td d d d td d d d t jddd td d d d tdd d d t jddd tddddtddddtdd ddtdd dd |_ ntd|dt j d|jd}t j t j||t jdt jd|_|jD]}t!|tj s||ur#t#j$|j&ddnt#j(|j&|j*qt#j,|j*dy)N1_0r!`r)rstrideTr)rrH ceil_mode@ 0ii1_1zUnsupported SqueezeNet version z: 1_0 or 1_1 expectedr r)p)r r gg{Gz?)meanstdr)r#r$rrBr% Sequentialr&r( MaxPool2drfeatures ValueErrorDropoutAdaptiveAvgPool2d classifiermodules isinstanceinitnormal_weightkaiming_uniform_bias constant_)r.rArBrC final_convmr/s r0r$zSqueezeNet.__init__%s D!& e MM !RQq9% 1ERR$S"b"%S"c3' 1ES"c3'S"c3'S"c3'S"c3' 1ES"c3'DM MM !RQq9% 1ERR$S"b"% 1ES"c3'S"c3' 1ES"c3'S"c3'S"c3'S"c3'DM&>wiG\]^ ^YYsD$4$4!D -- JJ !:rwwt/DbFZFZ[aFb  .A!RYY' ?LL>))!((366%NN1661- .r1r2cr|j|}|j|}tj|dSr4)rWr[r5flattenr7s r0r8zSqueezeNet.forward^s/ MM!  OOA }}Q""r1)rEig?) r9r:r;strr<floatr$r5r=r8r>r?s@r0rr$sA7.7.#7.u7._c7.r##%,,#r1rrAweightsprogresskwargsrc |#t|dt|jdt|fi|}|"|j |j |d|S)NrB categoriesT)rk check_hash)rlenmetarload_state_dictget_state_dict)rArjrkrlmodels r0 _squeezenetruds] fmSl9S5TU w )& )E g44hSW4XY Lr1z@https://github.com/pytorch/vision/pull/49#issuecomment-277560717zXThese weights reproduce closely the results of the paper using a simple training recipe.)rnrecipe_docsc TeZdZedeediedddddd id d d  ZeZy)rz>https://download.pytorch.org/models/squeezenet1_0-b66bff10.pth crop_size)r|i  ImageNet-1Kg M@g{GT@zacc@1zacc@5gh|?5?g&1@min_size num_params_metrics_ops _file_sizeurl transformsrqN r9r:r;r rr _COMMON_META IMAGENET1K_V1DEFAULTr1r0rr|T L.#>     !##    M"Gr1rc TeZdZedeediedddddd id d d  ZeZy)rz>https://download.pytorch.org/models/squeezenet1_1-b8a52dc0.pthryrz)ri(r}gX9M@g-'T@r~gtV?g"~@rrNrrr1r0rrrr1r pretrained)rjT)rjrkc Htj|}td||fi|S)aSqueezeNet model architecture from the `SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size `_ paper. Args: weights (:class:`~torchvision.models.SqueezeNet1_0_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.SqueezeNet1_0_Weights` below for more details, and possible values. By default, no pre-trained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to the ``torchvision.models.squeezenet.SqueezeNet`` base class. Please refer to the `source code `_ for more details about this class. .. autoclass:: torchvision.models.SqueezeNet1_0_Weights :members: rE)rverifyrurjrkrls r0rrs(2$**73G ugx :6 ::r1c Htj|}td||fi|S)a/SqueezeNet 1.1 model from the `official SqueezeNet repo `_. SqueezeNet 1.1 has 2.4x less computation and slightly fewer parameters than SqueezeNet 1.0, without sacrificing accuracy. Args: weights (:class:`~torchvision.models.SqueezeNet1_1_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.SqueezeNet1_1_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.squeezenet.SqueezeNet`` base class. Please refer to the `source code `_ for more details about this class. .. autoclass:: torchvision.models.SqueezeNet1_1_Weights :members: rQ)rrrurs r0rrs(6$**73G ugx :6 ::r1)$ functoolsrtypingrrr5torch.nnr% torch.nn.initr^transforms._presetsrutilsr_apir r r _metar _utilsrr__all__Modulerrrhboolrurrrrrrrr1r0rsl 5'66'B m 299 $=#=#@  k "   $'P k K*K*,0E0S0S!TU26;./;BF;Y\;;V;6,0E0S0S!TU26;./;BF;Y\;;V;r1