shW5SSKJr SSKJr SSKrSSKJs Jr SSKJr "SS\5r "SS\R5r S \RS \RS \ 4S jr"S S\R5r"SS\R5r"SS\R5r"SS\R5r"SS\R5r"SS\R5r"SS\R5r"SS\R5r"SS\R5r"SS \R5r"S!S"\R5r"S#S$\R5r"S%S&\R5rg)')Any)BaseSparsifierN)nncv^\rSrSrS\\\4SS4U4SjjrS\RS\S\\\4SS4Sjr S r U=r $) ImplementedSparsifier kwargsreturnNc >[TU]US9 g)N)defaults)super__init__)selfr __class__s z/Users/tiagomarins/Projetos/claudeai/copy_bank/venv/lib/python3.13/site-packages/torch/testing/_internal/common_pruning.pyrImplementedSparsifier.__init__ s &)module tensor_namec SURRSRS'URSnUR SS5S-US'g)Nrzlinear1.weight step_count)parametrizationsweightmaskstateget)rrrr linear_states r update_mask!ImplementedSparsifier.update_mask sN45&&q)..q1zz"23 %1%5%5lA%F%J \"r) __name__ __module__ __qualname____firstlineno__dictstrrrrModuler__static_attributes__ __classcell__rs@rrr sX*c3h*D*K"))K#KcSVhK\`KKrrcH\rSrSrSr\S\RSS4Sj5rSr g)MockSparseLinearz This class is a MockSparseLinear class to check convert functionality. It is the same as a normal Linear layer, except with a different type, as well as an additional from_dense method. modr c@U"URUR5nU$)z ) in_features out_features)clsr/linears r from_denseMockSparseLinear.from_denses"S__%%' rr!N) r"r#r$r%__doc__ classmethodrLinearr5r)r!rrr-r-s. RYY+=rr- subset_tensorsuperset_tensorr cSnUHInU[U5:a7[R"X1U5(dUS- nOM8U[U5:aM7 g g)zO Checks to see if all rows in subset tensor are present in the superset tensor rrFT)lentorchequal)r:r;irows rrows_are_subsetrB"sY A#o&&;;sA$677Q #o&&  rcj^\rSrSrSrSU4SjjrS\RS\R4SjrSr U=r $) SimpleLinear2zModel with only Linear layers without biases, some wrapped in a Sequential, some following the Sequential. Used to test basic pruned Linear-Linear fusion.r c @>[TU]5 [R"[R"SSSS9[R"SSSS9[R"SSSS95Ul[R"SSSS9Ul[R"SSSS9Ulg)NFbias )r rr Sequentialr9seqlinear1linear2rrs rrSimpleLinear.__init__6sx == IIa ' IIa ' IIa '  yyAE2 yyBU3 rxclURU5nURU5nURU5nU$N)rOrPrQrrTs rforwardSimpleLinear.forward@. HHQK LLO LLOr)rPrQrOr N r"r#r$r%r7rr>TensorrXr)r*r+s@rrDrD2s.V4%,,rrDcj^\rSrSrSrSU4SjjrS\RS\R4SjrSr U=r $) LinearBiasGzModel with only Linear layers, alternating layers with biases, wrapped in a Sequential. Used to test pruned Linear-Bias-Linear fusion.r c (>[TU]5 [R"[R"SSSS9[R"SSSS9[R"SSSS9[R"SSSS9[R"SSSS95Ulg) NrGrHTrIrKFrM)r rrrNr9rOrRs rrLinearBias.__init__Ksn == IIa & IIa ' IIa & IIa & IIa% (  rrTc(URU5nU$rVrOrWs rrXLinearBias.forwardUs HHQKrrer[r\r+s@rr_r_Gs.O %,,rr_cj^\rSrSrSrSU4SjjrS\RS\R4SjrSr U=r $) LinearActivationZzModel with only Linear layers, some with bias, some in a Sequential and some following. Activation functions modules in between each Linear in the Sequential, and each outside layer. Used to test pruned Linear(Bias)-Activation-Linear fusion.r c >[TU]5 [R"[R"SSSS9[R "5[R"SSSS9[R "5[R"SSSS95Ul[R"SSSS9Ul[R "5Ul [R"SS SS9Ul [R "5Ul g) NrGrHTrIrKFrLrbrM) r rrrNr9ReLUTanhrOrPact1rQact2rRs rrLinearActivation.__init___s == IIa & GGI IIa ' GGI IIa &  yyAD1 GGI yyBU3 GGI rrTcURU5nURU5nURU5nURU5nUR U5nU$rV)rOrPrmrQrnrWs rrXLinearActivation.forwardmH HHQK LLO IIaL LLO IIaLr)rmrnrPrQrOr[r\r+s@rrhrhZs.B %,,rrhcj^\rSrSrSrSU4SjjrS\RS\R4SjrSr U=r $) LinearActivationFunctionalva Model with only Linear layers, some with bias, some in a Sequential and some following. Activation functions modules in between each Linear in the Sequential, and functional activationals are called in between each outside layer. Used to test pruned Linear(Bias)-Activation-Linear fusion.r c >[TU]5 [R"[R"SSSS9[R "5[R"SSSS9[R "5[R"SSSS95Ul[R"SSSS9Ul[R"SS SS9Ul[R"S S SS9Ul [R "5Ul g) NrGrHTrIrKFrLrbrM) r rrrNr9rkrOrPrQlinear3rmrRs rr#LinearActivationFunctional.__init__|s == IIa & GGI IIa ' GGI IIa &  yyAD1 yyAE2 yyBU3 GGI rrTcURU5nURU5n[R"U5nUR U5n[R"U5nUR U5n[R"U5nU$rV)rOrPFrelurQrxrWs rrX"LinearActivationFunctional.forwardsb HHQK LLO FF1I LLO FF1I LLO FF1Ir)rmrPrQrxrOr[r\r+s@rrtrtvs.B %,,rrtcj^\rSrSrSrSU4SjjrS\RS\R4SjrSr U=r $) SimpleConv2dzModel with only Conv2d layers, all without bias, some in a Sequential and some following. Used to test pruned Conv2d-Conv2d fusion.r c &>[TU]5 [R"[R"SSSSSS9[R"SSSSSS95Ul[R"SSSSSS9Ul[R"SSSSSS9Ulg) Nr rbFrI@04r rrrNConv2drOconv2d1conv2d2rRs rrSimpleConv2d.__init__sx == IIaQ . IIb"a / yyRAE: yyRAE: rrTclURU5nURU5nURU5nU$rVrOrrrWs rrXSimpleConv2d.forwardrZrrrrOr[r\r+s@rrrs-1;%,,rrcj^\rSrSrSrSU4SjjrS\RS\R4SjrSr U=r $) Conv2dBiaszModel with only Conv2d layers, some with bias, some in a Sequential and some outside. Used to test pruned Conv2d-Bias-Conv2d fusion.r c T>[TU]5 [R"[R"SSSSSS9[R"SSSSSS9[R"SSSSSS95Ul[R"SSSSSS9Ul[R"SS SSSS9Ulg NrrrbTrIrFrrrrRs rrConv2dBias.__init__s == IIaQ - IIb"a . IIb"a /  yyRAD9 yyRAE: rrTclURU5nURU5nURU5nU$rVrrWs rrXConv2dBias.forwardrZrrr[r\r+s@rrrs-6;%,,rrcj^\rSrSrSrSU4SjjrS\RS\R4SjrSr U=r $) Conv2dActivationa Model with only Conv2d layers, some with bias, some in a Sequential and some following. Activation function modules in between each Sequential layer, functional activations called in-between each outside layer. Used to test pruned Conv2d-Bias-Activation-Conv2d fusion.r c>[TU]5 [R"[R"SSSSSS9[R "5[R"SSSSSS9[R "5[R"SSSSSS9[R "55Ul[R"SSSSSS9Ul[R"SS SSSS9Ul gr) r rrrNrrkrlrOrrrRs rrConv2dActivation.__init__s == IIaQ - GGI IIb"a . GGI IIb"a / GGI  yyRAE: yyRAD9 rrTcURU5nURU5n[R"U5nUR U5n[R "U5nU$rV)rOrr{r|rhardtanhrWs rrXConv2dActivation.forwardsH HHQK LLO FF1I LLO JJqMrrr[r\r+s@rrrs.A :%,,rrcj^\rSrSrSrSU4SjjrS\RS\R4SjrSr U=r $) Conv2dPadBiasaEModel with only Conv2d layers, all with bias and some with padding > 0, some in a Sequential and some following. Activation function modules in between each layer. Used to test that bias is propagated correctly in the special case of pruned Conv2d-Bias-(Activation)Conv2d fusion, when the second Conv2d layer has padding > 0.r c>[TU]5 [R"[R"SSSSSSS9[R "5[R"SSSSSS9[R "5[R"SSSSSSS9[R "5[R"SSSSSSS9[R "5[R"SSSSSS9[R "55 Ul[R"SS SSSSS9Ul[R "5Ul [R"S S SSSSS9Ul [R "5Ul g) NrrrbT)paddingrJFrIrrr) r rrrNrrkrlrOrrmrrnrRs rrConv2dPadBias.__init__s == IIaQ14 8 GGI IIb"a / GGI IIb"aAD 9 GGI IIb"aAD 9 GGI IIb"a . GGI  yyRAqtD GGI yyRAqtD GGI rrTcURU5nURU5nURU5nURU5nUR U5nU$rV)rOrrmrrnrWs rrXConv2dPadBias.forwardrrr)rmrnrrrOr[r\r+s@rrrs.c &%,,rrcj^\rSrSrSrSU4SjjrS\RS\R4SjrSr U=r $) Conv2dPoolzModel with only Conv2d layers, all with bias, some in a Sequential and some following. Activation function modules in between each layer, Pool2d modules in between each layer. Used to test pruned Conv2d-Pool2d-Conv2d fusion.r c n>[TU]5 [R"[R"SSSSSS9[R "SSSS9[R "5[R"SSSSSS9[R"5[R"SSSS95Ul [R"SS SSSS9Ul [R "SSSS9Ul [R "5Ul [R"S S SSSS9Ul [R"S S SSSS9Ulg) NrrrbT kernel_sizerrJrstriderrrr)r rrrNr MaxPool2drkrl AvgPool2drOrmaxpoolaf1rconv2d3rRs rrConv2dPool.__init__s == IIaAD A LLQq! < GGI IIb"!QT B GGI LLQq! <  yyRQM ||!QG 779yyRQM yyRQM rrTc,URU5nURU5nURU5nURU5nUR U5n[ R "USSSS9n[ R"U5nURU5nU$Nrrr) rOrrrrr{ avg_pool2dr|rrWs rrXConv2dPool.forwardsv HHQK LLO LLO HHQK LLO LL!Q ? FF1I LLOr)rrrrrrOr[r\r+s@rrrs.8N  %,,  rrcj^\rSrSrSrSU4SjjrS\RS\R4SjrSr U=r $) Conv2dPoolFlattenFunctionaliaModel with Conv2d layers, all with bias, some in a Sequential and some following, and then a Pool2d and a functional Flatten followed by a Linear layer. Activation functions and Pool2ds in between each layer also. Used to test pruned Conv2d-Pool2d-Flatten-Linear fusion.r c j>[TU]5 [R"[R"SSSSSS9[R "SSSS9[R "5[R"SSSSSS9[R"5[R"SSSS95Ul [R"SSSSSS9Ul [R "5Ul [R"SS SSSS9Ul [R"S 5Ul[R"S S SS 9Ulg) NrrbTrrrrHrG )rr rI)r rrrNrrrkrlrrOrrrAdaptiveAvgPool2davg_poolr9fcrRs rr$Conv2dPoolFlattenFunctional.__init__#s == IIa14 @ LLQq! < GGI IIa14 @ GGI LLQq! <  yyA1adK 779yyBAqtL ,,V4 ))B.rrTc.URU5nURU5n[R"USSSS9nUR U5nUR U5nUR U5n[R"US5nURU5nU$r) rOrr{ max_pool2drrrr>flattenrrWs rrX#Conv2dPoolFlattenFunctional.forward3sz HHQK LLO LL!Q ? HHQK LLO MM!  MM!Q  GGAJr)rrrrrrOr[r\r+s@rrrs.@ /  %,,  rrcj^\rSrSrSrSU4SjjrS\RS\R4SjrSr U=r $) Conv2dPoolFlatteni?a Model with Conv2d layers, all with bias, some in a Sequential and some following, and then a Pool2d and a Flatten module followed by a Linear layer. Activation functions and Pool2ds in between each layer also. Used to test pruned Conv2d-Pool2d-Flatten-Linear fusion.r c >[TU]5 [R"[R"SSSSSS9[R "SSSS9[R "5[R"SSSSSS9[R"5[R"SSSS95Ul [R"SSSSSS9Ul [R "5Ul [R"SS SSSS9Ul [R"S 5Ul[R"5Ul[R""S S SS 9Ulg)NrrbTrrrrHrGr)rr,rrI)r rrrNrrrkrlrrOrrrrrFlattenrr9rrRs rrConv2dPoolFlatten.__init__Es == IIa14 @ LLQq! < GGI IIa14 @ GGI LLQq! <  yyA1adK 779yyBAqtL ,,V4 zz| ))B.rrTc"URU5nURU5n[R"USSSS9nUR U5nUR U5nUR U5nURU5nURU5nU$r) rOrr{rrrrrrrWs rrXConv2dPoolFlatten.forwardVsw HHQK LLO LL!Q ? HHQK LLO MM!  LLO GGAJr)rrrrrrrOr[r\r+s@rrr?s.@ /"  %,,  rrc ^\rSrSrSrS\S\S\S\SS4 U4S jjrS \RS\ \R\R44S jr S r U=r $) LSTMLinearModelibzCContainer module with an encoder, a recurrent module, and a linear. input_dim hidden_dim output_dim num_layersr Nc>[TU]5 [R"XU5Ul[R "X#5UlgrV)r rrLSTMlstmr9r4rrrrrrs rrLSTMLinearModel.__init__es2 GGI:> ii 7 rinputcPURU5up#URU5nXB4$rV)rr4)rroutput_hiddendecodeds rrXLSTMLinearModel.forwardls())E*++f%r)r4r r"r#r$r%r7intrr>r]tuplerXr)r*r+s@rrrbsaM88*-8;>8LO8 8U\\eELL%,,4N.Orrc ^\rSrSrSrS\S\S\S\SS4 U4S jjrS \RS\ \R\R44S jr S r U=r $) LSTMLayerNormLinearModelirz9Container module with an LSTM, a LayerNorm, and a linear.rrrrr Nc>[TU]5 [R"XU5Ul[R "U5Ul[R"X#5UlgrV) r rrrr LayerNormnormr9r4rs rr!LSTMLayerNormLinearModel.__init__usB GGI:> LL, ii 7 rrTcrURU5upURU5nURU5nX4$rV)rrr4)rrTrs rrX LSTMLayerNormLinearModel.forward}s299Q< IIaL KKNxr)r4rrrr+s@rrrrsaC88*-8;>8LO8 8% ell0J*Krr)typingrtorch.ao.pruningrr>torch.nn.functionalr functionalr{rr9r-r]boolrBr(rDr_rhrtrrrrrrrrrr!rrrs.+ KNK ryy  5<< %,, SW  299*&ryy8>299(*ryy8BIIDB"))D  F bii  ryyr