[megatron] support blockwise fp8

[Jintao-Huang]

No description provided.

[gemini-code-assist]

Summary of Changes

Hello @Jintao-Huang, 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 support for blockwise FP8 within the Megatron framework. The primary changes involve adjusting the padding calculation logic to meet the specific requirements of blockwise FP8 and optimizing weight handling by ensuring that gradient tracking is disabled during the loading and exporting of model weights.

Highlights

• Blockwise FP8 Padding: Introduced specific padding requirements for 'blockwise' FP8 recipes, setting the padding factor to 128 in the get_padding_to utility function.
• Gradient Tracking Optimization: Disabled gradient tracking using torch.no_grad() during weight loading and exporting operations in gpt_bridge.py to optimize memory and computation.

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 introduces support for the blockwise FP8 recipe by adjusting sequence padding logic. It also enhances weight loading and exporting efficiency by wrapping these operations in torch.no_grad() contexts, which is a good practice. The changes are logical, but there is a minor opportunity to improve code clarity by replacing a magic number with a named constant.

[gemini-code-assist]

The magic number 128 is introduced here for the blockwise FP8 recipe. To improve readability and maintainability, consider defining this as a named constant at the module level, for example: BLOCKWISE_FP8_PADDING = 128. This would make the code's intent clearer, especially as it relates to hardware or algorithm-specific requirements for the blockwise FP8 format.