EPLB for fused_deep_moe

[wangyibo1005]

1. add new output : recvCount
2. support dynamic inputs

[gemini-code-assist]

Summary of Changes

Hello @wangyibo1005, 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 upgrades the fused_deep_moe operation by introducing a new recvCount output for better insight into expert distribution and by enabling dynamic input handling for its weight and scale tensors. These changes improve the flexibility and analytical capabilities of the Mixture-of-Experts (MoE) implementation, allowing for more adaptive and detailed processing within the deep learning framework.

Highlights

• New Output: recvCount: The fused_deep_moe operation now includes a new output, recvCount, which provides information about the number of tokens received by each expert. This enhances the observability and analysis capabilities of the MoE layer.
• Dynamic Inputs Support: The fused_deep_moe function has been updated to support dynamic inputs for its weight and scale parameters (gmm1PermutedWeight, gmm1PermutedWeightScale, gmm2Weight, gmm2WeightScale). This is achieved by changing the input types from at::Tensor to at::TensorList, allowing for more flexible model architectures.
• API and Kernel Updates: Significant changes were made across the API and kernel implementations to accommodate the new recvCount output and dynamic input handling. This includes modifications to function signatures, data structures, and the underlying computation logic in the aclnn and AscendC layers.

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 adds support for dynamic inputs and a new recvCount output to the fused_deep_moe operation. The changes are consistently applied across the operator definition, implementation, and kernel files. My review identifies opportunities to improve code maintainability by reducing duplication, and flags a potential risk from disabled checks.