SjؠXddlmZddlZddlmZddlmZmZddlm Z m Z m Z ddl m ZddlmZdd lmZgd ZGd d eZGd deZGddeeZGddeZGddeeZGddeZGddeeZGddeZGddeeZGddeZ y))AnyN)Tensor) functionalinit) ParameterUninitializedBufferUninitializedParameter) SyncBatchNorm)LazyModuleMixin)Module) BatchNorm1dLazyBatchNorm1d BatchNorm2dLazyBatchNorm2d BatchNorm3dLazyBatchNorm3dr ceZdZUdZdZgdZeed<eed<edzed<e ed<e ed < dd d dedededzde d e d e d dffdZ ddZ ddZ dZ dZ dfd ZxZS) _NormBasez,Common base of _InstanceNorm and _BatchNorm.)track_running_statsmomentumeps num_featuresaffinerrNrrrTbiasrreturnc ||d} t |||_||_||_||_||_|j r^ttj|fi| |_ |r%ttj|fi| |_ n7|jddn$|jdd|jdd|j r|jdtj|fi| |jdtj|fi| |||jdtj ddtj"i| j%D cic]\} } | dk7s | | c} } |n6|jdd|jdd|jdd|j'ycc} } w) Ndevicedtyperweight running_mean running_varnum_batches_trackedr"r)super__init__rrrrrrtorchemptyr#rregister_parameterregister_bufferzerosonestensorlongitemsreset_parameters) selfrrrrrr!r"rfactory_kwargskv __class__s Y/var/www/html/ia-enhanced/venv/lib/python3.12/site-packages/torch/nn/modules/batchnorm.pyr)z_NormBase.__init__&s%+U; (   #6 ;;#EKK $O$OPDK%ekk,&Q.&QR ''5  # #Hd 3  # #FD 1  # #   L KN K   uzz,I.I     % **)7(<(<(>O1!w,q!tO      6   5  !6 = Ps # G 1G c|jrP|jj|jj d|j jyy)Nr )rr$zero_r%fill_r&r4s r9reset_running_statsz_NormBase.reset_running_stats[sJ  # #    # # %    " "1 %  $ $ * * , $c|j|jrLtj|j|j tj |j yyyN)r>rrones_r#rzeros_r=s r9r3z_NormBase.reset_parameterscsI   " ;; JJt{{ #yy$ DII&% r?ctrA)NotImplementedErrorr4inputs r9_check_input_dimz_NormBase._check_input_dimjs!!r?cZdjdi|jd|jduiS)Nz{{num_features}, eps={eps}, momentum={momentum}, affine={affine}, bias={use_bias}, track_running_stats={track_running_stats}use_bias)format__dict__rr=s r9 extra_reprz_NormBase.extra_reprms: P IIO -- *.))4*?  r?c d|jdd}||dkr|jru|dz} | |vrl|j8|jjt jdk7r |jn$t j dtj || <t |!|||||||y)Nversionrr&metar)r") getrr&r!r*r0r1r(_load_from_state_dict) r4 state_dictprefixlocal_metadatastrict missing_keysunexpected_keys error_msgsrPnum_batches_tracked_keyr8s r9rSz_NormBase._load_from_state_dictus!$$Y5 Ow{0H0H'-/D&D #&j8//;00775<<;OO,,auzz: 23 %        r?h㈵>皙?TTNNrN)__name__ __module__ __qualname____doc___version __constants__int__annotations__floatboolr)r>r3rHrNrS __classcell__r8s@r9rrs6HXM Jdl L!$$(3 3 3 3 $, 3  3 " 3 3  3 j-'"       r?rcfeZdZ ddddedededzdeded ed dffd Zd ed efd ZxZ S) _BatchNormTNrrrrrrrrc@||d} t ||||||fi| d|iyNr r)r(r)) r4rrrrrr!r"rr5r8s r9r)z_BatchNorm.__init__s?%+U;           r?rGc |j||jd}n |j}|jrd|jrX|jL|jj d|jdt |jz }n |j} |jrd}n|jduxr|jdu} tj||jr |jr |jnd|jr |jr |jnd|j|j|||jS)Nr ?T)rHrtrainingrr&add_rhr$r%F batch_normr#rr)r4rGexponential_average_factor bn_trainings r9forwardz_BatchNorm.forwards' e$ == ), &)- & ==T55''3((--a0==(14uT=U=U7V1V.15.  ==K,,4T4;K;Kt;SK || }}(@(@!!$(MMT5M5MD  SW KK II  & HH  r?r\) r`rarbrfrhrir)rryrjrks@r9rmrms!$$(    $,    "     .0 V0 0 r?rmc`eZdZUeed<eed< ddd d fdZd fd Zd dZxZS) _LazyNormBaser#rTrc ||d}t |d||ddfi|ddi||_||_|jr"t di||_|rt di||_|jrttdi||_tdi||_ tj ddtji|jD cic]\} } | dk7s | | c} } |_yycc} } w)Nr rFrr"rKr')r(r)rrr r#rrr$r%r*r0r1r2r&) r4rrrrr!r"rr5r6r7r8s r9r)z_LazyNormBase.__init__s%+U;          #6 ;;0B>BDK2D^D  # # 3 En ED 2D^DD ',||(jj(%3$8$8$:KDAqa7l1a4K (D $ $Ls < C Ccd|js|jdk7rt| yyy)Nr)has_uninitialized_paramsrr(r3)r4r8s r9r3z_LazyNormBase.reset_parameterss0,,.43D3D3I G $ &4J.r?c||jr+|jd|_|jrt |j t s td|j j|jf|jKt |jt s td|jj|jf|jrL|jj|jf|jj|jf|jyy)Nr z-self.weight must be an UninitializedParameterz+self.bias must be an UninitializedParameter)r~shaperr isinstancer#r AssertionError materializerrr$r%r3rFs r9initialize_parametersz#_LazyNormBase.initialize_parameterss  ( ( * % AD {{!$++/EF(G ''):):(<=99(%dii1GH,III))4+<+<*>?''!!--&&(  ,,&&(  ! ! #+ +r?r\r_) r`rarbr rgr)r3rrjrks@r9r{r{sG ""    ** *X' $r?r{ceZdZdZddZy)raApplies Batch Normalization over a 2D or 3D input. Method described in the paper `Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift `__ . .. math:: y = \frac{x - \mathrm{E}[x]}{\sqrt{\mathrm{Var}[x] + \epsilon}} * \gamma + \beta The mean and standard-deviation are calculated per-dimension over the mini-batches and :math:`\gamma` and :math:`\beta` are learnable parameter vectors of size `C` (where `C` is the number of features or channels of the input). By default, the elements of :math:`\gamma` are set to 1 and the elements of :math:`\beta` are set to 0. At train time in the forward pass, the variance is calculated via the biased estimator, equivalent to ``torch.var(input, correction=0)``. However, the value stored in the moving average of the variance is calculated via the unbiased estimator, equivalent to ``torch.var(input, correction=1)``. Also by default, during training this layer keeps running estimates of its computed mean and variance, which are then used for normalization during evaluation. The running estimates are kept with a default :attr:`momentum` of 0.1. If :attr:`track_running_stats` is set to ``False``, this layer then does not keep running estimates, and batch statistics are instead used during evaluation time as well. .. note:: This :attr:`momentum` argument is different from one used in optimizer classes and the conventional notion of momentum. Mathematically, the update rule for running statistics here is :math:`\hat{x}_\text{new} = (1 - \text{momentum}) \times \hat{x} + \text{momentum} \times x_t`, where :math:`\hat{x}` is the estimated statistic and :math:`x_t` is the new observed value. Because the Batch Normalization is done over the `C` dimension, computing statistics on `(N, L)` slices, it's common terminology to call this Temporal Batch Normalization. Args: num_features: number of features or channels :math:`C` of the input eps: a value added to the denominator for numerical stability. Default: 1e-5 momentum: the value used for the running_mean and running_var computation. Can be set to ``None`` for cumulative moving average (i.e. simple average). Default: 0.1 affine: a boolean value that when set to ``True``, this module has learnable affine parameters. Default: ``True`` track_running_stats: a boolean value that when set to ``True``, this module tracks the running mean and variance, and when set to ``False``, this module does not track such statistics, and initializes statistics buffers :attr:`running_mean` and :attr:`running_var` as ``None``. When these buffers are ``None``, this module always uses batch statistics. in both training and eval modes. Default: ``True`` bias: If set to ``False``, the layer will not learn an additive bias (only relevant if :attr:`affine` is ``True``). Default: ``True`` Shape: - Input: :math:`(N, C)` or :math:`(N, C, L)`, where :math:`N` is the batch size, :math:`C` is the number of features or channels, and :math:`L` is the sequence length - Output: :math:`(N, C)` or :math:`(N, C, L)` (same shape as input) Examples:: >>> # With Learnable Parameters >>> m = nn.BatchNorm1d(100) >>> # Without Learnable Parameters >>> m = nn.BatchNorm1d(100, affine=False) >>> input = torch.randn(20, 100) >>> output = m(input) Nc|jdk7r1|jdk7rtd|jdyyNrzexpected 2D or 3D input (got D input)dim ValueErrorrFs r9rHzBatchNorm1d._check_input_dim{@ 99;!  q 0`__ . .. math:: y = \frac{x - \mathrm{E}[x]}{ \sqrt{\mathrm{Var}[x] + \epsilon}} * \gamma + \beta The mean and standard-deviation are calculated per-dimension over the mini-batches and :math:`\gamma` and :math:`\beta` are learnable parameter vectors of size `C` (where `C` is the input size). By default, the elements of :math:`\gamma` are set to 1 and the elements of :math:`\beta` are set to 0. At train time in the forward pass, the standard-deviation is calculated via the biased estimator, equivalent to ``torch.var(input, correction=0)``. However, the value stored in the moving average of the standard-deviation is calculated via the unbiased estimator, equivalent to ``torch.var(input, correction=1)``. Also by default, during training this layer keeps running estimates of its computed mean and variance, which are then used for normalization during evaluation. The running estimates are kept with a default :attr:`momentum` of 0.1. If :attr:`track_running_stats` is set to ``False``, this layer then does not keep running estimates, and batch statistics are instead used during evaluation time as well. .. note:: This :attr:`momentum` argument is different from one used in optimizer classes and the conventional notion of momentum. Mathematically, the update rule for running statistics here is :math:`\hat{x}_\text{new} = (1 - \text{momentum}) \times \hat{x} + \text{momentum} \times x_t`, where :math:`\hat{x}` is the estimated statistic and :math:`x_t` is the new observed value. Because the Batch Normalization is done over the `C` dimension, computing statistics on `(N, H, W)` slices, it's common terminology to call this Spatial Batch Normalization. Args: num_features: :math:`C` from an expected input of size :math:`(N, C, H, W)` eps: a value added to the denominator for numerical stability. Default: 1e-5 momentum: the value used for the running_mean and running_var computation. Can be set to ``None`` for cumulative moving average (i.e. simple average). Default: 0.1 affine: a boolean value that when set to ``True``, this module has learnable affine parameters. Default: ``True`` track_running_stats: a boolean value that when set to ``True``, this module tracks the running mean and variance, and when set to ``False``, this module does not track such statistics, and initializes statistics buffers :attr:`running_mean` and :attr:`running_var` as ``None``. When these buffers are ``None``, this module always uses batch statistics. in both training and eval modes. Default: ``True`` bias: If set to ``False``, the layer will not learn an additive bias (only relevant if :attr:`affine` is ``True``). Default: ``True`` Shape: - Input: :math:`(N, C, H, W)` - Output: :math:`(N, C, H, W)` (same shape as input) Examples:: >>> # With Learnable Parameters >>> m = nn.BatchNorm2d(100) >>> # Without Learnable Parameters >>> m = nn.BatchNorm2d(100, affine=False) >>> input = torch.randn(20, 100, 35, 45) >>> output = m(input) Ncd|jdk7rtd|jdyNzexpected 4D input (got rrrFs r9rHzBatchNorm2d._check_input_dim0 99;! 6uyy{m8LM M r?r_rrKr?r9rrGRNr?rceZdZdZeZddZy)raA :class:`torch.nn.BatchNorm2d` module with lazy initialization. Lazy initialization is done for the ``num_features`` argument of the :class:`BatchNorm2d` that is inferred from the ``input.size(1)``. The attributes that will be lazily initialized are `weight`, `bias`, `running_mean` and `running_var`. Check the :class:`torch.nn.modules.lazy.LazyModuleMixin` for further documentation on lazy modules and their limitations. Args: eps: a value added to the denominator for numerical stability. Default: 1e-5 momentum: the value used for the running_mean and running_var computation. Can be set to ``None`` for cumulative moving average (i.e. simple average). Default: 0.1 affine: a boolean value that when set to ``True``, this module has learnable affine parameters. Default: ``True`` track_running_stats: a boolean value that when set to ``True``, this module tracks the running mean and variance, and when set to ``False``, this module does not track such statistics, and initializes statistics buffers :attr:`running_mean` and :attr:`running_var` as ``None``. When these buffers are ``None``, this module always uses batch statistics. in both training and eval modes. Default: ``True`` bias: If set to ``False``, the layer will not learn an additive bias (only relevant if :attr:`affine` is ``True``). Default: ``True`` Ncd|jdk7rtd|jdyrrrFs r9rHz LazyBatchNorm2d._check_input_dimrr?r_)r`rarbrcrrrHrKr?r9rr8 MNr?rceZdZdZddZy)rahApplies Batch Normalization over a 5D input. 5D is a mini-batch of 3D inputs with additional channel dimension as described in the paper `Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift `__ . .. math:: y = \frac{x - \mathrm{E}[x]}{ \sqrt{\mathrm{Var}[x] + \epsilon}} * \gamma + \beta The mean and standard-deviation are calculated per-dimension over the mini-batches and :math:`\gamma` and :math:`\beta` are learnable parameter vectors of size `C` (where `C` is the input size). By default, the elements of :math:`\gamma` are set to 1 and the elements of :math:`\beta` are set to 0. At train time in the forward pass, the standard-deviation is calculated via the biased estimator, equivalent to ``torch.var(input, correction=0)``. However, the value stored in the moving average of the standard-deviation is calculated via the unbiased estimator, equivalent to ``torch.var(input, correction=1)``. Also by default, during training this layer keeps running estimates of its computed mean and variance, which are then used for normalization during evaluation. The running estimates are kept with a default :attr:`momentum` of 0.1. If :attr:`track_running_stats` is set to ``False``, this layer then does not keep running estimates, and batch statistics are instead used during evaluation time as well. .. note:: This :attr:`momentum` argument is different from one used in optimizer classes and the conventional notion of momentum. Mathematically, the update rule for running statistics here is :math:`\hat{x}_\text{new} = (1 - \text{momentum}) \times \hat{x} + \text{momentum} \times x_t`, where :math:`\hat{x}` is the estimated statistic and :math:`x_t` is the new observed value. Because the Batch Normalization is done over the `C` dimension, computing statistics on `(N, D, H, W)` slices, it's common terminology to call this Volumetric Batch Normalization or Spatio-temporal Batch Normalization. Args: num_features: :math:`C` from an expected input of size :math:`(N, C, D, H, W)` eps: a value added to the denominator for numerical stability. Default: 1e-5 momentum: the value used for the running_mean and running_var computation. Can be set to ``None`` for cumulative moving average (i.e. simple average). Default: 0.1 affine: a boolean value that when set to ``True``, this module has learnable affine parameters. Default: ``True`` track_running_stats: a boolean value that when set to ``True``, this module tracks the running mean and variance, and when set to ``False``, this module does not track such statistics, and initializes statistics buffers :attr:`running_mean` and :attr:`running_var` as ``None``. When these buffers are ``None``, this module always uses batch statistics. in both training and eval modes. Default: ``True`` bias: If set to ``False``, the layer will not learn an additive bias (only relevant if :attr:`affine` is ``True``). Default: ``True`` Shape: - Input: :math:`(N, C, D, H, W)` - Output: :math:`(N, C, D, H, W)` (same shape as input) Examples:: >>> # With Learnable Parameters >>> m = nn.BatchNorm3d(100) >>> # Without Learnable Parameters >>> m = nn.BatchNorm3d(100, affine=False) >>> input = torch.randn(20, 100, 35, 45, 10) >>> output = m(input) Ncd|jdk7rtd|jdyNzexpected 5D input (got rrrFs r9rHzBatchNorm3d._check_input_dimarr?r_rrKr?r9rrrr?rceZdZdZeZddZy)raA :class:`torch.nn.BatchNorm3d` module with lazy initialization. Lazy initialization is done for the ``num_features`` argument of the :class:`BatchNorm3d` that is inferred from the ``input.size(1)``. The attributes that will be lazily initialized are `weight`, `bias`, `running_mean` and `running_var`. Check the :class:`torch.nn.modules.lazy.LazyModuleMixin` for further documentation on lazy modules and their limitations. Args: eps: a value added to the denominator for numerical stability. Default: 1e-5 momentum: the value used for the running_mean and running_var computation. Can be set to ``None`` for cumulative moving average (i.e. simple average). Default: 0.1 affine: a boolean value that when set to ``True``, this module has learnable affine parameters. Default: ``True`` track_running_stats: a boolean value that when set to ``True``, this module tracks the running mean and variance, and when set to ``False``, this module does not track such statistics, and initializes statistics buffers :attr:`running_mean` and :attr:`running_var` as ``None``. When these buffers are ``None``, this module always uses batch statistics. in both training and eval modes. Default: ``True`` bias: If set to ``False``, the layer will not learn an additive bias (only relevant if :attr:`affine` is ``True``). Default: ``True`` Ncd|jdk7rtd|jdyrrrFs r9rHz LazyBatchNorm3d._check_input_dimrr?r_)r`rarbrcrrrHrKr?r9rrfrr?rceZdZdZ ddddedededzded ed edzd ed dffd ZddZ ddZ de d e fdZ e ddZxZS)r aApplies Batch Normalization over a N-Dimensional input. The N-D input is a mini-batch of [N-2]D inputs with additional channel dimension) as described in the paper `Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift `__ . .. math:: y = \frac{x - \mathrm{E}[x]}{ \sqrt{\mathrm{Var}[x] + \epsilon}} * \gamma + \beta The mean and standard-deviation are calculated per-dimension over all mini-batches of the same process groups. :math:`\gamma` and :math:`\beta` are learnable parameter vectors of size `C` (where `C` is the input size). By default, the elements of :math:`\gamma` are sampled from :math:`\mathcal{U}(0, 1)` and the elements of :math:`\beta` are set to 0. The standard-deviation is calculated via the biased estimator, equivalent to `torch.var(input, correction=0)`. Also by default, during training this layer keeps running estimates of its computed mean and variance, which are then used for normalization during evaluation. The running estimates are kept with a default :attr:`momentum` of 0.1. If :attr:`track_running_stats` is set to ``False``, this layer then does not keep running estimates, and batch statistics are instead used during evaluation time as well. .. note:: This :attr:`momentum` argument is different from one used in optimizer classes and the conventional notion of momentum. Mathematically, the update rule for running statistics here is :math:`\hat{x}_\text{new} = (1 - \text{momentum}) \times \hat{x} + \text{momentum} \times x_t`, where :math:`\hat{x}` is the estimated statistic and :math:`x_t` is the new observed value. Because the Batch Normalization is done for each channel in the ``C`` dimension, computing statistics on ``(N, +)`` slices, it's common terminology to call this Volumetric Batch Normalization or Spatio-temporal Batch Normalization. Currently :class:`SyncBatchNorm` only supports :class:`~torch.nn.DistributedDataParallel` (DDP) with single GPU per process. Use :meth:`torch.nn.SyncBatchNorm.convert_sync_batchnorm()` to convert :attr:`BatchNorm*D` layer to :class:`SyncBatchNorm` before wrapping Network with DDP. Args: num_features: :math:`C` from an expected input of size :math:`(N, C, +)` eps: a value added to the denominator for numerical stability. Default: ``1e-5`` momentum: the value used for the running_mean and running_var computation. Can be set to ``None`` for cumulative moving average (i.e. simple average). Default: 0.1 affine: a boolean value that when set to ``True``, this module has learnable affine parameters. Default: ``True`` track_running_stats: a boolean value that when set to ``True``, this module tracks the running mean and variance, and when set to ``False``, this module does not track such statistics, and initializes statistics buffers :attr:`running_mean` and :attr:`running_var` as ``None``. When these buffers are ``None``, this module always uses batch statistics. in both training and eval modes. Default: ``True`` process_group: synchronization of stats happen within each process group individually. Default behavior is synchronization across the whole world bias: If set to ``False``, the layer will not learn an additive bias (only relevant if :attr:`affine` is ``True``). Default: ``True`` Shape: - Input: :math:`(N, C, +)` - Output: :math:`(N, C, +)` (same shape as input) .. note:: Synchronization of batchnorm statistics occurs only while training, i.e. synchronization is disabled when ``model.eval()`` is set or if ``self.training`` is otherwise ``False``. Examples:: >>> # xdoctest: +SKIP >>> # With Learnable Parameters >>> m = nn.SyncBatchNorm(100) >>> # creating process group (optional) >>> # ranks is a list of int identifying rank ids. >>> ranks = list(range(8)) >>> r1, r2 = ranks[:4], ranks[4:] >>> # Note: every rank calls into new_group for every >>> # process group created, even if that rank is not >>> # part of the group. >>> process_groups = [torch.distributed.new_group(pids) for pids in [r1, r2]] >>> process_group = process_groups[0 if dist.get_rank() <= 3 else 1] >>> # Without Learnable Parameters >>> m = nn.BatchNorm3d(100, affine=False, process_group=process_group) >>> input = torch.randn(20, 100, 35, 45, 10) >>> output = m(input) >>> # network is nn.BatchNorm layer >>> sync_bn_network = nn.SyncBatchNorm.convert_sync_batchnorm(network, process_group) >>> # only single gpu per process is currently supported >>> ddp_sync_bn_network = torch.nn.parallel.DistributedDataParallel( >>> sync_bn_network, >>> device_ids=[args.local_rank], >>> output_device=args.local_rank) TNrrrrrr process_grouprrc N||d} t ||||||fi| d| i||_yro)r(r)r) r4rrrrrrr!r"rr5r8s r9r)zSyncBatchNorm.__init__sI%+U;           +r?cd|jdkrtd|jdy)Nrz expected at least 2D input (got rrrFs r9rHzSyncBatchNorm._check_input_dim s/ 99;?? }HUV V r?cB|jddk(r tdy)Nr rz9SyncBatchNorm number of input channels should be non-zero)sizerrFs r9_check_non_zero_input_channelsz,SyncBatchNorm._check_non_zero_input_channelss' ::a=A K  r?rGc |j||j||jd}n |j}|jrt|jrh|j t d|j jd|jd|j jz }n |j} |jrd}n|jduxr|jdu} |jr |jr |jnd}|jr |jr |jnd}|xrL|jxr>tjjxrtjj}|r|jj ddd tj"j%fvr*t'd tj"j%tjj(j*}|j,r |j,}tjj/|}|dkD}|s:t1j2||||j4|j6|||j8S|s t d t;j<||j4|j6|||j8| S) z( Runs the forward pass. Nrqz$num_batches_tracked must not be Noner rrTcudahpuxpuz;SyncBatchNorm expected input tensor to be on GPU or XPU or zbn_training must be True)rHrrrsrr&rrtitemr$r%r* distributed is_availableis_initializedr!type_C_get_privateuse1_backend_namergroupWORLDrget_world_sizerurvr#rrsync_batch_normapply) r4rGrwrxr$r% need_syncr world_sizes r9ryzSyncBatchNorm.forwards e$ ++E2 == ), &)- & ==T55''/$%KLL  $ $ ) )! ,}}$-043K3K3P3P3R-R*-1]]*  ==K,,4T4;K;Kt;SK &*]]d6N6ND  TX %)MMT5M5MD  SW   3  3!!..0 3!!002  ||  668 ) !Qxx==?@B "--3399M!! $ 2 2 **99-HJ"QI<<  *  $%?@@"((  *  r?c *|}t|tjjjj rtjj |j|j|j|j|j||jdu}|jr?tj5|j|_|j|_ ddd|j|_|j |_|j"|_|j$|_t'|dr|j(|_|j+D]'\}}|j-||j/||)~|S#1swYxYw)aaConverts all :attr:`BatchNorm*D` layers in the model to :class:`torch.nn.SyncBatchNorm` layers. Args: module (nn.Module): module containing one or more :attr:`BatchNorm*D` layers process_group (optional): process group to scope synchronization, default is the whole world Returns: The original :attr:`module` with the converted :class:`torch.nn.SyncBatchNorm` layers. If the original :attr:`module` is a :attr:`BatchNorm*D` layer, a new :class:`torch.nn.SyncBatchNorm` layer object will be returned instead. Example:: >>> # Network with nn.BatchNorm layer >>> # xdoctest: +REQUIRES(env:TORCH_DOCTEST_CUDA) >>> module = torch.nn.Sequential( >>> torch.nn.Linear(20, 100), >>> torch.nn.BatchNorm1d(100), >>> ).cuda() >>> # creating process group (optional) >>> # ranks is a list of int identifying rank ids. >>> ranks = list(range(8)) >>> r1, r2 = ranks[:4], ranks[4:] >>> # Note: every rank calls into new_group for every >>> # process group created, even if that rank is not >>> # part of the group. >>> # xdoctest: +SKIP("distributed") >>> process_groups = [torch.distributed.new_group(pids) for pids in [r1, r2]] >>> process_group = process_groups[0 if dist.get_rank() <= 3 else 1] >>> sync_bn_module = torch.nn.SyncBatchNorm.convert_sync_batchnorm(module, process_group) Nrqconfig)rr*nnmodules batchnormrmr rrrrrrno_gradr#r$r%r&rshasattrrnamed_children add_moduleconvert_sync_batchnorm)clsmoduler module_outputnamechilds r9rz$SyncBatchNorm.convert_sync_batchnormysMH fehh..88CC D!HH22##  **[[,3M}}]]_5+1==M()/M&5*0)<)rs *UU8! | | ~H H VL$OYL$^KT*KT\!TmZ!THLN*LN^!NmZ!NHLN*LN^!NmZ!NHmJmr?