+ j^RIt^RIHt^RIHtHtHt^RIt^RIH t ^RIH t ^RIH t ^RI Ht^RIHtHtHt^RIHt^R IHtHt^R IHtHtHtHtHtHt^R IH t H!t!H"t".R"Ot#!RR]4t$!RR]4t%!RR]4t&!RR ]4t'!RR ]4t(]!RR7]!RR3R7RRRRRR/R R!ll44tR#)#N)partial)AnyOptionalUnionTensor) functional)ImageClassification)register_modelWeights WeightsEnum)_IMAGENET_CATEGORIES)_ovewrite_named_paramhandle_legacy_interface) BasicConv2d GoogLeNetGoogLeNet_WeightsGoogLeNetOutputs Inception InceptionAux) _fuse_modules _replace_reluquantize_modelQuantizableGoogLeNetGoogLeNet_QuantizedWeightscfaa]tRt^toV3RlV3RlltV3RlRltRV3RlRlltRtVtV;t #) QuantizableBasicConv2dc*<V^8dQhRS[RS[RR/#argskwargsreturnNr)format __classdict__s"j/home/luca/screen-checking/.venv/lib/python3.14/site-packages/torchvision/models/quantization/googlenet.py __annotate__#QuantizableBasicConv2d.__annotate__"cSTc\<\SV`!V/VB\P!4VnR#N)super__init__nnReLUreluselfr!r" __class__s&*,r'r/QuantizableBasicConv2d.__init__s" $)&)GGI r+c&<V^8dQhRS[RS[/#r xr#r)r%r&s"r'r(r)sFr+clVPV4pVPV4pVPV4pV#r-convbnr2r4r9s&&r'forwardQuantizableBasicConv2d.forwards. IIaL GGAJ IIaLr+c4<V^8dQhRS[S[,RR/#r is_qatr#Nrbool)r%r&s"r'r(r)$s#JJ$J4Jr+c(\V.ROVRR7R#)r<T)inplaceNr;)r)r4rCs&&r' fuse_model!QuantizableBasicConv2d.fuse_model$sd2FDIr+r2r- __name__ __module__ __qualname____firstlineno__r/r?rH__static_attributes____classdictcell__ __classcell__r5r&s@@r'rrs+ JJJr+rcPaa]tRt^(toV3RlV3RlltV3RlRltRtVtV;t#)QuantizableInceptionc*<V^8dQhRS[RS[RR/#rr$)r%r&s"r'r(!QuantizableInception.__annotate__)s"22c2S2T2r+cz<\SV`!VR\/VB\PP 4VnR# conv_blockN)r.r/rr0 quantizedFloatFunctionalcatr3s&*,r'r/QuantizableInception.__init__)s/ $L+ALVL<<//1r+c&<V^8dQhRS[RS[/#r8r)r%r&s"r'r(rW-s(((F(r+c\VPV4pVPPV^4#))_forwardr])r4r9outputss&& r'r?QuantizableInception.forward-s%--"xx||GQ''r+)r] rLrMrNrOr/r?rPrQrRrSs@@r'rUrU(s22(((r+rUcPaa]tRt^2toV3RlV3RlltV3RlRltRtVtV;t#)QuantizableInceptionAuxc*<V^8dQhRS[RS[RR/#rr$)r%r&s"r'r($QuantizableInceptionAux.__annotate__4r*r+ch<\SV`!VR\/VB\P!4VnR#rY)r.r/rr0r1r2r3s&*,r'r/ QuantizableInceptionAux.__init__4s( $L+ALVLGGI r+c&<V^8dQhRS[RS[/#r8r)r%r&s"r'r(ri8sFr+c\P!VR4pVPV4p\P!V^4pVP VP V44pVPV4pVPV4pV#))rnrn) Fadaptive_avg_pool2dr<torchflattenr2fc1dropoutfc2r>s&&r'r?QuantizableInceptionAux.forward8sh ! !!V , IIaL MM!Q  IIdhhqk " LLO HHQKr+rJrerSs@@r'rgrg2sr+rgcfaa]tRt^JtoV3RlV3RlltV3RlRltRV3RlRlltRtVtV;t #) rc*<V^8dQhRS[RS[RR/#rr$)r%r&s"r'r(!QuantizableGoogLeNet.__annotate__Ls";;c;S;T;r+c<\SV`!VR\\\./VB\ P PP4Vn \ P PP4Vn R#)blocksN) r.r/rrUrgrqao quantization QuantStubquant DeQuantStubdequantr3s&*,r'r/QuantizableGoogLeNet.__init__Lsb   13GI`a ek XX**446 xx,,88: r+c&<V^8dQhRS[RS[/#r8)rr)r%r&s"r'r(rySs 5 5 5$4 5r+cVPV4pVPV4pVPV4wrpVPV4pVP;'d VP p\ PP4'd+V'g\P!R4\WV4#VPWV4#)zCScripted QuantizableGoogleNet always returns GoogleNetOutputs Tuple) _transform_inputrrbrtraining aux_logitsrqjit is_scriptingwarningswarnr eager_outputs)r4r9aux1aux2 aux_defineds&& r'r?QuantizableGoogLeNet.forwardSs  ! !! $ JJqM a(  LLOmm77 99 ! ! # # cd#AT2 2%%at4 4r+c4<V^8dQhRS[S[,RR/#rBrD)r%r&s"r'r(ry`s % %$ %4 %r+c VP4F)p\V4\JgKVPV4K+ R#)zFuse conv/bn/relu modules in googlenet model Fuse conv+bn+relu/ conv+relu/conv+bn modules to prepare for quantization. Model is modified in place. Note that this operation does not change numerics and the model after modification is in floating point N)modulestyperrH)r4rCms&& r'rHQuantizableGoogLeNet.fuse_model`s-AAw00 V$ r+)rrr-rKrSs@@r'rrJs(;; 5 5 % % %r+c]tRt^mt]!R]!]^R7RRRRR]RRR R R ]PR R RRRR//RRRRRR/ R7t ] t Rt R#)rzKhttps://download.pytorch.org/models/quantized/googlenet_fbgemm-c81f6644.pth) crop_size num_paramsiemin_size categoriesbackendfbgemmrecipezdhttps://github.com/pytorch/vision/tree/main/references/classification#post-training-quantized-models unquantized_metricsz ImageNet-1Kzacc@1g/tQ@zacc@5g`"YV@_opsg+? _file_sizeg#~j<)@_docsz These weights were produced by doing Post Training Quantization (eager mode) on top of the unquantized weights listed below. )url transformsmetaN)r) rLrMrNrOr rr rr IMAGENET1K_V1IMAGENET1K_FBGEMM_V1DEFAULTrPrr+r'rrms|" Y.#> '  . x | ,:: VV  E &  0#Gr+quantized_googlenet)name pretrainedcrVPRR4'd\P#\P#)quantizeF)getrrrr)r"s&r'rs1zz*e,, ' ; ; 1"00 1r+)weightsrprogressTrFc V^8dQhR\\\\3,,R\R\R\ R\ /#)r rrrr"r#)rrrrrErr)r%s"r'r(r(sT@@ e68IIJ K@@ @  @  @r+c V'd\M\PV4pVPRR4pVeRV9d\ VRR4\ VRR4\ VRR4\ VR\ VP R,44R VP 9d\ VR VP R ,4VPR R 4p\R /VBp\V4V'd \We4VeXVPVPVRR 74V'gRVn RVnRVnV#\ P"!R 4V#)a\GoogLeNet (Inception v1) model architecture from `Going Deeper with Convolutions `__. .. note:: Note that ``quantize = True`` returns a quantized model with 8 bit weights. Quantized models only support inference and run on CPUs. GPU inference is not yet supported. Args: weights (:class:`~torchvision.models.quantization.GoogLeNet_QuantizedWeights` or :class:`~torchvision.models.GoogLeNet_Weights`, optional): The pretrained weights for the model. See :class:`~torchvision.models.quantization.GoogLeNet_QuantizedWeights` 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. quantize (bool, optional): If True, return a quantized version of the model. Default is False. **kwargs: parameters passed to the ``torchvision.models.quantization.QuantizableGoogLeNet`` base class. Please refer to the `source code `_ for more details about this class. .. autoclass:: torchvision.models.quantization.GoogLeNet_QuantizedWeights :members: .. autoclass:: torchvision.models.GoogLeNet_Weights :members: :noindex: rFNtransform_inputT init_weights num_classesrrr)r check_hashz`auxiliary heads in the pretrained googlenet model are NOT pretrained, so make sure to train themr)rrverifyrrlenrpoprrrload_state_dictget_state_dictrrrrr)rrrr"original_aux_logitsrmodels$$$, r' googlenetrs(\.6);LTTU\]G **\59 F * !&*;T BflD9fne<fmSl9S5TU  $ !&)W\\)5L MjjH-G *6 *E%u& g44hSW4XY"$E EJEJ L MMr  Lr+)rrr))r functoolsrtypingrrrrqtorch.nnr0rr rotransforms._presetsr _apir r r _metar_utilsrrrrrrrrrutilsrrr__all__rrUrgrrrr+r'rs'' $677(Cll??  J[ J(9(l0 %9 %F##8*+   @NR@@ @ ,@r+