!EiaSSKJr SSKrSSKJr SSKr/SQr\"SSS9r\"SS S9rSSS jjr SSS jjr SS jr SS jr g)) annotationsN)TypeVar)fuse_conv_bn_evalfuse_conv_bn_weightsfuse_linear_bn_evalfuse_linear_bn_weightsConvTztorch.nn.modules.conv._ConvNd)boundLinearTztorch.nn.Linearc UR(dUR(a [S5e[R"U5nURb UR c [S5e[ URURURUR URURURU5uUlUlU$)aFuse a convolutional module and a BatchNorm module into a single, new convolutional module. Args: conv (torch.nn.modules.conv._ConvNd): A convolutional module. bn (torch.nn.modules.batchnorm._BatchNorm): A BatchNorm module. transpose (bool, optional): If True, transpose the convolutional weight. Defaults to False. Returns: torch.nn.modules.conv._ConvNd: The fused convolutional module. .. note:: Both ``conv`` and ``bn`` must be in eval mode, and ``bn`` must have its running buffers computed. Fusion only for eval!3bn.running_mean and bn.running_var must not be None) trainingAssertionErrorcopydeepcopy running_mean running_varrweightbiaseps)convbn transpose fused_convs R/home/mande/repo/quber/.venv/lib/python3.13/site-packages/torch/nn/utils/fusion.pyrrs$ }} 455t$J ".."8RSS)=      *&Jz cURnUb UROUn Uc[R"U5nUc[R"U5nUc[R"U5n[R"X4-5n U(a"SS/S/[ UR 5S- --n O!SS/S/[ UR 5S- --n XU -RU 5-RUS9n X- U -U-U-RU S9n [RRXR5[RRXR54$)aFuse convolutional module parameters and BatchNorm module parameters into new convolutional module parameters. Args: conv_w (torch.Tensor): Convolutional weight. conv_b (Optional[torch.Tensor]): Convolutional bias. bn_rm (torch.Tensor): BatchNorm running mean. bn_rv (torch.Tensor): BatchNorm running variance. bn_eps (float): BatchNorm epsilon. bn_w (Optional[torch.Tensor]): BatchNorm weight. bn_b (Optional[torch.Tensor]): BatchNorm bias. transpose (bool, optional): If True, transpose the conv weight. Defaults to False. Returns: Tuple[torch.nn.Parameter, torch.nn.Parameter]: Fused convolutional weight and bias. dtype) r#torch zeros_like ones_likersqrtlenshapereshapetonn Parameter requires_grad)conv_wconv_bbn_rmbn_rvbn_epsbn_wbn_brconv_weight_dtypeconv_bias_dtype bn_var_rsqrtr) fused_conv_w fused_conv_bs rrr:sI2 &,&8fll>OO ~!!%( |u% |&;;u~.LB1#V\\!2Q!677Q1#V\\!2Q!677\1::5AAEEFL^|3d:TAEEFL <)=)=> <)=)=> rc >UR(dUR(a [S5e[R"U5nURUR :wa5UR S:wa%[SURSUR 35eUR b URc [S5e[URURUR URURURUR5uUl Ul U$)asFuse a linear module and a BatchNorm module into a single, new linear module. Args: linear (torch.nn.Linear): A Linear module. bn (torch.nn.modules.batchnorm._BatchNorm): A BatchNorm module. Returns: torch.nn.Linear: The fused linear module. .. note:: Both ``linear`` and ``bn`` must be in eval mode, and ``bn`` must have its running buffers computed. r rzaTo fuse, linear.out_features == bn.num_features or bn.num_features == 1, got linear.out_features=z and bn.num_features=r) rrrr out_features num_featuresrrrrrr)linearr fused_linears rrros "++455==(L boo-"//Q2F''-':':&;;PQSQ`Q`Pa c   ".."8RSS-C     .*L* rcURnUb UROUnUc[R"U5nU[R"X4-5-n X R S5R US9-n X- U -U-R US9n [R RXR5[R RXR54$)aFuse linear module parameters and BatchNorm module parameters into new linear module parameters. Args: linear_w (torch.Tensor): Linear weight. linear_b (Optional[torch.Tensor]): Linear bias. bn_rm (torch.Tensor): BatchNorm running mean. bn_rv (torch.Tensor): BatchNorm running variance. bn_eps (float): BatchNorm epsilon. bn_w (torch.Tensor): BatchNorm weight. bn_b (torch.Tensor): BatchNorm bias. Returns: Tuple[torch.nn.Parameter, torch.nn.Parameter]: Fused linear weight and bias. r r") r#r$r%r' unsqueezer+r,r-r.) linear_wlinear_br1r2r3r4r5linear_weight_dtypelinear_bias_dtypebn_scalefused_wfused_bs rrrs.#..*2*>DW##E*ekk%.11H++B/229L2MMG H,t377>O7PG 88  g'='= >@R@R''A r)F)rr r%torch.nn.modules.batchnorm._BatchNormrboolreturnr )r/ torch.Tensorr0torch.Tensor | Noner1rLr2rLr3floatr4rMr5rMrrJrK-tuple[torch.nn.Parameter, torch.nn.Parameter])r>r rrIrKr )rBrLrCrMr1rLr2rLr3rNr4rLr5rLrKrO) __future__rrtypingrr$__all__r r rrrrrrrTs5"    >? )#4 5 # #-## #\2 2 2 2  2  2  2 2232j0 0-0 0f""!" "  "  "  " "3"r