[Relax][PyTroch] Add NHWC layout support

[guan404ming]

Why

• The interpolate operation was hardcoded to only support NCHW layout
• Users need flexibility to choose the appropriate layout for their target platform

How

• Added default_image_layout parameter
• Exposed default_image_layout parameter in the public from_fx()

[gemini-code-assist]

Summary of Changes

Hello @guan404ming, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request significantly enhances the PyTorch FX frontend for Relax by introducing flexible support for the NHWC image layout in interpolation operations. Previously, the interpolate operation was rigidly tied to the NCHW layout, which limited its adaptability. By allowing users to specify their preferred image layout, this change improves compatibility with various target platforms and optimizes performance for models that benefit from NHWC data arrangements.

Highlights

• NHWC Layout Support: The TorchFXImporter class now supports specifying the default image layout (NCHW or NHWC) during initialization, addressing the previous hardcoding to NCHW for interpolation operations.
• Dynamic Spatial Dimension Calculation: The _interpolate method within TorchFXImporter has been updated to dynamically determine spatial dimensions based on the configured default_image_layout, ensuring correct behavior for both NCHW and NHWC inputs.
• Public API Exposure: The public from_fx function now exposes a default_image_layout parameter, allowing users to explicitly choose the image layout when importing PyTorch FX graphs, enhancing user flexibility.
• Comprehensive Testing: New unit tests have been added to validate the correct behavior of interpolate with both NCHW (default) and NHWC layouts, covering nearest neighbor and bilinear interpolation modes, and ensuring backward compatibility.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

The pull request successfully adds NHWC layout support for the interpolate operation in the PyTorch FX frontend. The changes correctly introduce a default_image_layout parameter, update the spatial dimension calculation logic, and propagate the chosen layout to the relax.op.image.resize2d operator. Comprehensive test cases have been added to verify both the new functionality and backward compatibility. The implementation is clean and addresses the stated problem effectively.

[gemini-code-assist]

The else branch currently covers "NCHW or other layouts". While NCHW is the default and likely the primary alternative, if only NCHW is explicitly supported besides NHWC, it might be clearer to make this condition elif self.default_image_layout == "NCHW": for explicitness. If other layouts are genuinely intended to fall into this category, a comment explaining this assumption would be beneficial for future maintainability.

Suggested change

	else: # NCHW or other layouts
	spatial_start = 2
	spatial_end = len(shape)
	else: # NCHW
	spatial_start = 2
	spatial_end = len(shape)

[guan404ming]

cc @tlopex @mshr-h

[tlopex]

LGTM. Thank you!

[guan404ming]

Thanks!