[Relax] Add FRelaxInferLayout and TMixedPrecisionPolicy for dynamic_strided_slice

[guan404ming]

Why

The dynamic_strided_slice operator was missing FRelaxInferLayout and TMixedPrecisionPolicy attributes, preventing it from participating in layout transformations and mixed precision optimizations.

How

• Add TMixedPrecisionPolicy attribute with kFollow policy and
• Add InferLayoutDynStridedSlice function that falls back to initial layout (since begin/end/strides are dynamic tensors that cannot be transformed at compile time)

[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 integrates the relax.dynamic_strided_slice operator into the mixed precision transformation system. Previously, this operator was not considered during mixed precision optimizations, which could hinder performance in scenarios utilizing lower precision computations like FP16. By adding the necessary policy and corresponding tests, the operator can now seamlessly participate in mixed precision graphs, contributing to more efficient model execution.

Highlights

• Mixed Precision Support: The relax.dynamic_strided_slice operator now supports mixed precision transformations by registering the TMixedPrecisionPolicy attribute.
• Precision Policy: The TMixedPrecisionPolicy for dynamic_strided_slice is set to MixedPrecisionPolicyKind::kFollow, meaning it will adapt its precision based on its input.
• Test Coverage: New test cases have been added to test_transform_to_mixed_precision.py to validate the correct behavior of dynamic_strided_slice within mixed precision workflows.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

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

This pull request correctly adds the TMixedPrecisionPolicy to the dynamic_strided_slice operator, enabling it for mixed-precision transformations. The added test case is also relevant. However, the expected output in the new test contains a redundant and inefficient sequence of data type casts. I've included a specific comment with a suggestion to simplify the test's expected output, which would reflect a more optimal transformation.

[gemini-code-assist]

The Expected module contains a redundant sequence of casts. The conv2d output lv2 is float32. It is cast to float16 (as lv3), then immediately back to float32 (as lv4), before being used in dynamic_strided_slice. This is inefficient.

Since dynamic_strided_slice is a kFollow operator and its input from the preceding conv2d is already float32, the intermediate casts are unnecessary. The dynamic_strided_slice operator can directly use the lv2 tensor. The expected transformed code should be more optimal.

Suggested change

	lv3: R.Tensor((2, 4, 26, 26), dtype="float16") = R.astype(lv2, dtype="float16")
	lv4: R.Tensor((2, 4, 26, 26), dtype="float32") = R.astype(lv3, dtype="float32")
	gv: R.Tensor(None, dtype="float32", ndim=4) = R.dynamic_strided_slice(lv4, begin, end, strides)
	gv: R.Tensor(None, dtype="float32", ndim=4) = R.dynamic_strided_slice(lv2, begin, end, strides)

[guan404ming]

cc @tlopex @mshr-h

[tlopex]

Why we don't use tvm.ir.assert_structural_equal here?

[guan404ming]

_assert_test is shared helper func in this file. I use it for consistency here. Shall we update to use tvm.ir.assert_structural_equal?

[tlopex]

No worries, I had a look and it's nearly the same

[tlopex]

No worries, I had a look and it's nearly the same

[guan404ming]

No worries, I had a look and it's nearly the same

Thanks!