Intel® oneAPI Deep Neural Network Developer Guide and Reference
A newer version of this document is available. Customers should click here to go to the newest version.
Visible to Intel only — GUID: GUID-3E0D00A6-2C85-4257-BA62-69559BC0B7B0
Visible to Intel only — GUID: GUID-3E0D00A6-2C85-4257-BA62-69559BC0B7B0
Layer Normalization
General
The layer normalization primitive performs a forward or backward layer normalization operation on a 2-5D data tensor.
Forward
The layer normalization operation performs normalization over the last logical axis of the data tensor and is defined by the following formulas. We show formulas only for 3D data, which are straightforward to generalize to cases of higher dimensions. Variable names follow the standard Naming Conventions.
where
are optional scale and shift for a channel (see dnnl_use_scale, dnnl_use_shift flags),
are mean and variance (see dnnl_use_global_stats flag), and
is a constant to improve numerical stability.
Mean and variance are computed at runtime or provided by a user. When mean and variance are computed at runtime, the following formulas are used:
,
.
The 
and 
tensors are considered learnable.
Difference Between Forward Training and Forward Inference
If mean and variance are computed at runtime (i.e., dnnl_use_global_stats is not set), they become outputs for the propagation kind dnnl_forward_training (because they would be required during the backward propagation). Data layout for mean and variance must be specified during creation of the layer normalization primitive descriptor by passing the memory descriptor for statistics (e.g., by passing stat_desc in dnnl::layer_normalization_forward::primitive_desc()). Mean and variance are not exposed for the propagation kind dnnl_forward_inference.
Backward
The backward propagation computes 
, 
, and 
based on 
, 
, 
, 
, 
, and 
.
The tensors marked with an asterisk are used only when the primitive is configured to use , and (i.e., dnnl_use_scale or dnnl_use_shift are set).
Execution Arguments
Depending on the flags and propagation kind, the layer normalization primitive requires different inputs and outputs. For clarity, a summary is shown below.
Flags |
||||
|---|---|---|---|---|
Inputs : |
Inputs : |
Inputs : |
Same as for dnnl_backward |
|
Inputs : |
Inputs : |
Inputs : |
Same as for dnnl_backward |
|
Inputs : |
Inputs : |
Inputs : |
Not supported |
|
Inputs : |
Inputs : |
Inputs : |
Not supported |
|
Inputs : |
Inputs : |
Inputs : |
Not supported |
When executed, the inputs and outputs should be mapped to an execution argument index as specified by the following table.
Primitive input/output |
Execution argument index |
|---|---|
|
DNNL_ARG_SRC |
|
DNNL_ARG_SCALE |
|
DNNL_ARG_SHIFT |
mean ( |
DNNL_ARG_MEAN |
variance ( |
DNNL_ARG_VARIANCE |
|
DNNL_ARG_DST |
|
DNNL_ARG_DIFF_DST |
|
DNNL_ARG_DIFF_SRC |
|
DNNL_ARG_DIFF_SCALE |
|
DNNL_ARG_DIFF_SHIFT |
|
DNNL_ARG_ATTR_SCALES | DNNL_ARG_SRC |
|
DNNL_ARG_ATTR_SCALES | DNNL_ARG_DST |
)
Implementation Details
General Notes
The different flavors of the primitive are partially controlled by the flags parameter that is passed to the primitive descriptor creation function (e.g., dnnl::layer_normalization_forward::primitive_desc()). Multiple flags can be set using the bitwise OR operator (|).
For forward propagation, the mean and variance might be either computed at runtime (in which case they are outputs of the primitive) or provided by a user (in which case they are inputs). In the latter case, a user must set the dnnl_use_global_stats flag. For the backward propagation, the mean and variance are always input parameters.
Both forward and backward propagation support in-place operations, meaning that
can be used as input and output for forward propagation, and 
can be used as input and output for backward propagation. In case of an in-place operation, the original data will be overwritten. This support is limited to cases when data types of and 
or 
and are identical. Note, however, that backward propagation requires original , hence the corresponding forward propagation should not be performed in-place.
Post-ops and Attributes
Attributes enable you to modify the behavior of the layer normalization primitive. The following attributes are supported by the layer normalization primitive:
Propagation |
Type |
Operation |
Description |
Restrictions |
|---|---|---|---|---|
forward |
attribute |
Scales the corresponding tensor by the given scale factor(s). |
Supported only for int8 layer normalization and one scale per tensor is supported. |
Data Type Support
The operation supports the following combinations of data types:
Propagation |
Source |
Destination |
Scale and Shift |
|---|---|---|---|
forward |
f32, bf16, f16, u8, s8, f64 |
f32, bf16, f16, u8, s8, f64 |
f32, bf16, f16 |
backward |
f32, bf16, f16, f64 |
f32, bf16, f16, f64 |
f32, bf16, f16 |
Mean and Variance data types are always f32 and independent of Source and Destination data types.
Data Representation
Mean and Variance
The mean (
) and variance (
) are separate tensors with number of dimensions equal to (
) and size 
.
The corresponding memory object can have an arbitrary memory format. Unless mean and variance are computed at runtime and not exposed (i.e., propagation kind is dnnl_forward_inference and dnnl_use_global_stats is not set), the user should provide a memory descriptor for statistics when creating the layer normalization primitive descriptor. For best performance, it is advised to use the memory format that follows the data memory format; i.e., if the data format is dnnl_tnc, the best performance can be expected for statistics with the dnnl_tn format and suboptimal for statistics with the dnnl_nt format.
Scale and Shift
If dnnl_use_scale or dnnl_use_shift are used, the scale (
) and shift (
) are separate 1D tensors of shape 
.
Source, Destination, and Their Gradients
The layer normalization primitive works with an arbitrary data tensor; however, it was designed for RNN data tensors (i.e., dnnl_nc, dnnl_tnc, dnnl_ldnc). Unlike CNN data tensors, RNN data tensors have a single feature dimension. Layer normalization performs normalization over the last logical dimension (feature dimension for RNN tensors) across non-feature dimensions.
The layer normalization primitive is optimized for the following memory formats:
Implementation Limitations
Refer to Data Types for limitations related to data types support.
GPU
Only tensors of 6 or fewer dimensions are supported.
Performance Tips
For data tensors
, , , and , use memory formats for which the last logical axis is the last in the physical memory layout.For mean and variance, use the memory format that follows the data memory format; i.e., if the data format is dnnl_tnc, the best performance can be expected for statistics with dnnl_tn and suboptimal for statistics with the dnnl_nt format.
For backward propagation, use the same memory format for
, , and . Different formats are functionally supported but lead to highly suboptimal performance.Use in-place operations whenever possible (see caveats in General Notes).
Example
Layer Normalization Primitive Example
This C++ API example demonstrates how to create and execute a Layer normalization primitive in forward propagation mode.
Key optimizations included in this example:
In-place primitive execution;
Creation of memory objects using the primitive descriptor.