Module batch_norm 

Batch Normalization layer with running statistics tracking.

Provides TensorBatchNorm — a full batch normalisation layer that supports both training mode (batch statistics, running stat updates) and inference mode (uses pre-accumulated running statistics). Learnable affine parameters γ (gamma / scale) and β (beta / shift) are optional.

Structs

BatchNormConfig

Configuration for TensorBatchNorm.

BatchNormStats

Aggregate statistics collected across TensorBatchNorm::forward calls.

TensorBatchNorm

Batch normalisation layer.

Enums

NormMode

Execution mode for batch normalisation.