SyncBatchNorm(num_features: int, eps: float = 1e-05, momentum: float = 0.1, affine: bool = True, track_running_stats: bool = True, process_group: Optional[Any] = None)¶
Applies Batch Normalization over a N-Dimensional input (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 .
The mean and standard-deviation are calculated per-dimension over all mini-batches of the same process groups. and are learnable parameter vectors of size C (where C is the input size). By default, the elements of are sampled from and the elements of are set to 0. The standard-deviation is calculated via the biased estimator, equivalent to torch.var(input, unbiased=False).
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
track_running_statsis set to
False, this layer then does not keep running estimates, and batch statistics are instead used during evaluation time as well.
momentumargument is different from one used in optimizer classes and the conventional notion of momentum. Mathematically, the update rule for running statistics here is , where is the estimated statistic and is the new observed value.
Because the Batch Normalization is done for each channel in the
Cdimension, computing statistics on
(N, +)slices, it’s common terminology to call this Volumetric Batch Normalization or Spatio-temporal Batch Normalization.
DistributedDataParallel(DDP) with single GPU per process. Use
SyncBatchNormbefore wrapping Network with DDP.
num_features – from an expected input of size
eps – a value added to the denominator for numerical stability. Default:
momentum – the value used for the running_mean and running_var computation. Can be set to
Nonefor 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:
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 uses batch statistics instead in both training and eval modes if the running mean and variance are
process_group – synchronization of stats happen within each process group individually. Default behavior is synchronization across the whole world
Output: (same shape as input)
>>> # With Learnable Parameters >>> m = nn.SyncBatchNorm(100) >>> # creating process group (optional) >>> # process_ids is a list of int identifying rank ids. >>> process_group = torch.distributed.new_group(process_ids) >>> # 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)
Helper function to convert all
BatchNorm*Dlayers in the model to
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
>>> # Network with nn.BatchNorm layer >>> module = torch.nn.Sequential( >>> torch.nn.Linear(20, 100), >>> torch.nn.BatchNorm1d(100), >>> ).cuda() >>> # creating process group (optional) >>> # process_ids is a list of int identifying rank ids. >>> process_group = torch.distributed.new_group(process_ids) >>> sync_bn_module = torch.nn.SyncBatchNorm.convert_sync_batchnorm(module, process_group)