add npu qwen3-next example and warning of ep size

[addsubmuldiv]

PR type

• Bug Fix
• New Feature
• Document Updates
• More Models or Datasets Support

PR information

1. Add NPU-based training example for Qwen3_Next
   • add a NPU megatron training example for Qwen3_Next
2. Add warning information about Expert Parallel (EP) size limitation
   • Warning: The number of experts in Qwen3_Next is larger than 128, which is the maximum limit supported by fused ops in MindSpeed. If the number of experts allocated to each rank exceeds 128, the fused ops module will report error information.

Experiment results

Paste your experiment result here(if needed).

[gemini-code-assist]

Summary of Changes

Hello @addsubmuldiv, 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 introduces support for the Qwen3-Next model on NPU by providing a new training example script. It also enhances the Megatron argument parsing to include a crucial warning for Mixture of Experts (MoE) configurations on NPU, guiding users to correctly set the expert_model_parallel_size to prevent potential issues and ensure efficient training when dealing with a large number of experts.

Highlights

• New NPU Example for Qwen3-Next: A new example script has been added to demonstrate training the Qwen3-Next model on NPU using Megatron SFT with LoRA.
• NPU Expert Parallelism Warning: Implemented a warning mechanism in the Megatron arguments to alert users when the number of experts exceeds the recommended limit per Expert Parallelism (EP) group on NPU, suggesting an appropriate expert_model_parallel_size.

🧠 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 adds an example script for training Qwen3-Next on Ascend NPUs and includes a warning for expert parallelism size configuration. The changes are generally good. I have one suggestion to improve the clarity of the warning message in swift/megatron/arguments/megatron_args.py.

[gemini-code-assist]

The formula (num_experts / MAX_NPU_EXPERTS_PER_EP) in the warning message can be misleading because it uses float division, which might not match the integer ceiling division used to calculate required_ep. For example, if num_experts is 129 and MAX_NPU_EXPERTS_PER_EP is 128, required_ep would be 2, but 129 / 128 is 1.0078....

To avoid confusion, I suggest simplifying the message to directly state the required value.

                                   f'Please set expert_model_parallel_size (EP) to at least {required_ep}.')