QЦiASSKrSSKJr SSKJr SSKJrJrJrJ r SSK r SSK J r SSK J s Jr SSKJs Jr SSK Jr SSKJr SSKJr S S KJrJrJr S S KJr S S KJrJ r /S Qr!"SS\ RD5r#"SS\ RH5r%"SS\ RL5r'"SS\ RD5r(S\ RDS\S\)SS4Sjr*S\+S\ \+\+\+\+4S\+S\\S\)S\S\(4Sjr,S \S!S"S#.r-"S$S%\5r."S&S'\5r/"S(S)\5r0"S*S+\5r1\"5\ "S,\.Rd4S-9SS.S/.S\\.S\)S\S\(4S0jj55r3\"5\ "S,\/Rd4S-9SS.S/.S\\/S\)S\S\(4S1jj55r4\"5\ "S,\0Rd4S-9SS.S/.S\\0S\)S\S\(4S2jj55r5\"5\ "S,\1Rd4S-9SS.S/.S\\1S\)S\S\(4S3jj55r6g)4N) OrderedDict)partial)AnyListOptionalTuple)Tensor)ImageClassification)_log_api_usage_once)register_modelWeights WeightsEnum)_IMAGENET_CATEGORIES)_ovewrite_named_paramhandle_legacy_interface) DenseNetDenseNet121_WeightsDenseNet161_WeightsDenseNet169_WeightsDenseNet201_Weights densenet121 densenet161 densenet169 densenet201c d^\rSrSrSS\S\S\S\S\SS4 U4S jjjrS \\ S\ 4S jr S \\ S\4S jr \ RRS \\ S\ 4Sj5r\ RR S \\ S\ 4Sj5r\ RR S \ S\ 4Sj5rS \ S\ 4SjrSrU=r$) _DenseLayernum_input_features growth_ratebn_size drop_ratememory_efficientreturnNc >[TU]5 [R"U5Ul[R "SS9Ul[R"XU-SSSS9Ul[R"X2-5Ul [R "SS9Ul [R"X2-USSSSS9Ul [U5Ul XPlg)NTinplacer F kernel_sizestridebiasr*r+paddingr,)super__init__nn BatchNorm2dnorm1ReLUrelu1Conv2dconv1norm2relu2conv2floatr#r$)selfr r!r"r#r$ __class__s Z/var/www/html/ai-image-ml/venv/lib/python3.13/site-packages/torchvision/models/densenet.pyr1_DenseLayer.__init__ s ^^$67 WWT* YY1[3HVW`ahmn ^^G$9: WWT* YYw4kqYZdelqr y) 0inputsc[R"US5nURURUR U555nU$Nr )torchcatr8r6r4)r=rBconcated_featuresbottleneck_outputs r? bn_function_DenseLayer.bn_function/s;!IIfa0 JJtzz$**=N2O'PQ  rAinputc<UHnUR(dM g g)NTF) requires_grad)r=rKtensors r?any_requires_grad_DenseLayer.any_requires_grad5s F###rAcD^U4Sjn[R"U/UQ7SS06$)Nc&>TRU5$N)rI)rBr=s r?closure7_DenseLayer.call_checkpoint_bottleneck..closure=s##F+ +rA use_reentrantF)cp checkpoint)r=rKrTs` r?call_checkpoint_bottleneck&_DenseLayer.call_checkpoint_bottleneck;s# ,}}WBuBEBBrAcgrSr=rKs r?forward_DenseLayer.forwardB rAcgrSr\r]s r?r^r_Fr`rAc[U[5(aU/nOUnUR(aVURU5(a@[R R 5(a [S5eURU5nOURU5nURURURU555nURS:a)[R"X@RUR S9nU$)Nz%Memory Efficient not supported in JITr)ptraining) isinstancer r$rOrEjit is_scripting ExceptionrYrIr;r:r9r#Fdropoutrd)r=rK prev_featuresrH new_featuress r?r^r_Ls eV $ $"GM!M  T%;%;M%J%Jyy%%'' GHH $ ? ? N  $ 0 0 ? zz$**TZZ8I-J"KL >>A 99\^^dmm\LrA)r8r;r#r$r4r9r6r:F)__name__ __module__ __qualname____firstlineno__intr<boolr1rr rIrOrErfunusedrY_overload_methodr^__static_attributes__ __classcell__r>s@r?rrsrw 1"% 147 1BE 1RW 1ko 1  1 1!$v,!6! tF|  YYCV CCC  YY T&\ f    YY V    VrArcd^\rSrSrSrSS\S\S\S\S\S\S S 4U4S jjjrS \ S \ 4S jr Sr U=r $) _DenseBlock`r num_layersr r"r!r#r$r%Nc >[T U]5 [U5H-n[X'U--UUUUS9nUR SUS--U5 M/ g)N)r!r"r#r$z denselayer%dr )r0r1ranger add_module) r=r|r r"r!r#r$ilayerr>s r?r1_DenseBlock.__init__csX z"A"_4'#!1 E OONa!e4e <#rA init_featurescU/nUR5Hup4U"U5nURU5 M [R"US5$rD)itemsappendrErF)r=rfeaturesnamerrls r?r^_DenseBlock.forwardwsC!?::[TU]5 [R"U5Ul[R "SS9Ul[R"XSSSS9Ul[R"SSS9Ul g)NTr'r Fr)r )r*r+) r0r1r2r3normr5relur7conv AvgPool2dpool)r=r rr>s r?r1_Transition.__init__s[ NN#56 GGD) II0ST]^ejk LLQq9 rA)rrrr)rnrorprqrrr1rvrwrxs@r?rrs":3:S:T::rArc^\rSrSrSrSS\S\\\\\4S\S\S\S\S \S S 4U4S jjjr S \ S \ 4Sjr Sr U=r $)raDensenet-BC model class, based on `"Densely Connected Convolutional Networks" `_. Args: growth_rate (int) - how many filters to add each layer (`k` in paper) block_config (list of 4 ints) - how many layers in each pooling block num_init_features (int) - the number of filters to learn in the first convolution layer bn_size (int) - multiplicative factor for number of bottle neck layers (i.e. bn_size * k features in the bottleneck layer) drop_rate (float) - dropout rate after each dense layer num_classes (int) - number of classification classes memory_efficient (bool) - If True, uses checkpointing. Much more memory efficient, but slower. Default: *False*. See `"paper" `_. r! block_confignum_init_featuresr"r# num_classesr$r%Nc>>[TU]5 [U5 [R"[ S[R "SUSSSSS94S[R"U54S[R"S S 94S [R"SSS S 94/55Ul Un[U5Hup[U UUUUUS9n URRSU S --U 5 XU--nU [U5S - :wdMP[XS-S9n URRSU S --U 5 US-nM URRS[R"U55 [R "X5UlUR%5GHn ['U [R 5(a+[R(R+U R,5 MN['U [R5(aV[R(R/U R,S 5 [R(R/U R0S5 M['U [R 5(dM[R(R/U R0S5 GM g)Nconv0r-r Fr.norm0relu0Tr'pool0r )r*r+r/)r|r r"r!r#r$z denseblock%d)r rz transition%dnorm5r)r0r1r r2 Sequentialrr7r3r5 MaxPool2dr enumeraterzrlenrLinear classifiermodulesreinitkaiming_normal_weight constant_r,)r=r!rrr"r#rr$ num_featuresrr|blocktransmr>s r?r1DenseNet.__init__s  D! bii+?@bggd34bllqANO     ) &|4MA%#/'#!1 E MM $ $^q1u%=u E'{*BBLC %))#|ijYjk ((1q5)A5I+q0 5"   "..*FG))L>A!RYY''''1Ar~~..!!!((A.!!!&&!,Aryy))!!!&&!, rAxcURU5n[R"USS9n[R"US5n[R "US5nUR U5nU$)NTr')r r r )rriradaptive_avg_pool2drEflattenr)r=rrouts r?r^DenseNet.forwardsU==#ffXt,##C0mmC#ooc" rA)rr)  @riF)rnrorprq__doc__rrrr<rsr1r r^rvrwrxs@r?rrs "2A!#!&:-:-Cc3./:- :-  :-  :-:-:- :-:-xFrArmodelweightsprogressr%c<[R"S5nURUSS9n[UR 55HHnUR U5nU(dMUR S5UR S5-nXEXG'XE MJ URU5 g)Nz]^(.*denselayer\d+\.(?:norm|relu|conv))\.((?:[12])\.(?:weight|bias|running_mean|running_var))$T)r check_hashr r )recompileget_state_dictlistkeysmatchgroupload_state_dict)rrrpattern state_dictkeyresnew_keys r?_load_state_dictrs jjhG''d'KJJOO%&mmC  3iilSYYq\1G",/J  '  *%rAr!rrkwargsc Ub#[US[URS55 [XU40UD6nUb [ XcUS9 U$)Nr categories)rrr)rrmetarr)r!rrrrrrs r? _densenetrsKfmSl9S5TU [0A LV LEuI LrA)rz*https://github.com/pytorch/vision/pull/116z'These weights are ported from LuaTorch.)min_sizerrecipe_docsc N\rSrSr\"S\"\SS90\ESSSSS .0S S S .ES 9r\r Sr g)ri z@ num_params_metrics_ops _file_sizeurl transformsrr\N rnrorprqrrr _COMMON_META IMAGENET1K_V1DEFAULTrvr\rAr?rr sQ J.#>    !##     M GrArc N\rSrSr\"S\"\SS90\ESSSSS .0S S S .ES 9r\r Sr g)riz    "##  !  M GrArc N\rSrSr\"S\"\SS90\ESSSSS .0S S S .ES 9r\r Sr g)ri3z    "##     M GrArc N\rSrSr\"S\"\SS90\ESSSSS .0S S S .ES 9r\r Sr g)riGz    "##     M GrAr pretrained)rT)rrc J[RU5n[SSSX40UD6$)a?Densenet-121 model from `Densely Connected Convolutional Networks `_. Args: weights (:class:`~torchvision.models.DenseNet121_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.DenseNet121_Weights` below for more details, and possible values. By default, no pre-trained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to the ``torchvision.models.densenet.DenseNet`` base class. Please refer to the `source code `_ for more details about this class. .. autoclass:: torchvision.models.DenseNet121_Weights :members: rrr)rverifyrrrrs r?rr[**"((1G R"g J6 JJrAc J[RU5n[SSSX40UD6$)a?Densenet-161 model from `Densely Connected Convolutional Networks `_. Args: weights (:class:`~torchvision.models.DenseNet161_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.DenseNet161_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.densenet.DenseNet`` base class. Please refer to the `source code `_ for more details about this class. .. autoclass:: torchvision.models.DenseNet161_Weights :members: 0)rr$rr{)rrrrs r?rrurrAc J[RU5n[SSSX40UD6$)a?Densenet-169 model from `Densely Connected Convolutional Networks `_. Args: weights (:class:`~torchvision.models.DenseNet169_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.DenseNet169_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.densenet.DenseNet`` base class. Please refer to the `source code `_ for more details about this class. .. autoclass:: torchvision.models.DenseNet169_Weights :members: r)rrrrr)rrrrs r?rrrrAc J[RU5n[SSSX40UD6$)a?Densenet-201 model from `Densely Connected Convolutional Networks `_. Args: weights (:class:`~torchvision.models.DenseNet201_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.DenseNet201_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.densenet.DenseNet`` base class. Please refer to the `source code `_ for more details about this class. .. autoclass:: torchvision.models.DenseNet201_Weights :members: r)rrrrr)rrrrs r?rrrrA)7r collectionsr functoolsrtypingrrrrrEtorch.nnr2torch.nn.functional functionalritorch.utils.checkpointutilsrXrWr transforms._presetsr r _apirrr_metar_utilsrr__all__Moduler ModuleDictrzrrrrsrrrrrrrrrrrrrrr\rAr?r s #-- ##5'66'B >"))>B&"--&>:"--:RryyRj&BII& &t&PT&&S#s*+k "     (&: :  +(+(+(+(,0C0Q0Q!RS<@SWKH%89KDKcfKksKTK0,0C0Q0Q!RS<@SWKH%89KDKcfKksKTK0,0C0Q0Q!RS<@SWKH%89KDKcfKksKTK0,0C0Q0Q!RS<@SWKH%89KDKcfKksKTKrA