Feature request: Implement GroupNormalization-21

[justinchuby]

Describe the issue

https://onnx.ai/onnx/operators/onnx__GroupNormalization.html is not implemented according to the opset support table. Most likely it's just reusing the kernel for microsoft.GroupNorm.

• [Feature Request] [CUDA] GroupNormalization onnx op @ opset 18 #20211
• [Feature Request] Implement GroupNormalization-21 #22697

To reproduce

N/A

Urgency

Support for newer opsets in the exporter and be enabled by this improvement. (pytorch/pytorch#152138)

Platform

Linux

OS Version

N/A

ONNX Runtime Installation

Released Package

ONNX Runtime Version or Commit ID

1.22

ONNX Runtime API

Python

Architecture

X64

Execution Provider

CUDA

Execution Provider Library Version

No response

cc @tianleiwu

[github-actions]

This issue has been automatically marked as stale due to inactivity and will be closed in 30 days if no further activity occurs. If further support is needed, please provide an update and/or more details.

[justinchuby]

GroupNormalization - 21
Version
name: GroupNormalization (GitHub)

domain: main

since_version: 21

function: True

support_level: SupportType.COMMON

shape inference: False

This version of the operator has been available since version 21.

Summary
A GroupNormalization function. Carries out group normalization as described in the paper https://arxiv.org/abs/1803.08494

This operator transforms input according to

y = scale * (x - mean) / sqrt(variance + epsilon) + bias,
where the mean and variance are computed per instance per group of channels, and scale and bias should be specified for each channel. The number of groups num_groups should be divisible by the number of channels so that there are an equal number of channels per group.

The overall computation has two stages: the first stage normalizes the elements to have zero mean and unit variance for each instance in each group, and the second stage scales and shifts the results of the first stage. The floating-point precision used in the first stage is determined by the stash_type attribute. For example, if stash_type is 1, the operator casts all input variables to 32-bit float, performs the computation, and finally casts the normalized results back to the original type of X. The second stage does not depend on stash_type.

When the number of groups is the same as the number of channels, this operator is equivalent to InstanceNormalization. When there is only one group, this operator is equivalent to LayerNormalization.

Attributes
epsilon - FLOAT (default is '1e-05'):

The epsilon value to use to avoid division by zero.

num_groups - INT (required) :

The number of groups of channels. It should be a divisor of the number of channels C.

stash_type - INT (default is '1'):

The floating-point precision used in stage one of the computation.

Inputs
X (heterogeneous) - T:

Input data tensor. Dimensions for image cases are (N x C x H x W), where N is the batch size, C is the number of channels, and H and W are the height and width of the data. Statistics are computed for every group of channels over C, H, and W. For non-image cases, the dimensions are in the form of (N x C x D1 x D2 ... Dn).

scale (heterogeneous) - T:

Scale tensor of shape (C).

bias (heterogeneous) - T:

Bias tensor of shape (C).

Outputs
Y (heterogeneous) - T:

The output tensor of the same shape as X.

Type Constraints
T in ( tensor(bfloat16), tensor(double), tensor(float), tensor(float16) ):

Constrain input and output types to float tensors.

GroupNormalization - 18 vs 21