Add flash attn backend, duplicates #314

[yubofredwang]

Motivation

This feature is authored by @timmy-feng. The original PR is #314. Please see details in that PR.

Modifications

Supports flash attention as backend

Related Issues

Accuracy Test

Benchmark & Profiling

Checklist

• Format your code according to the Code Formatting with Pre-Commit.
• Add unit tests as outlined in the Running Unit Tests.
• Update documentation / docstrings / example tutorials as needed, according to Writing Documentation.
• Provide throughput / latency benchmark results and accuracy evaluation results as needed, according to Benchmark and Profiling and Accuracy Results.
• For reviewers: If you haven't made any contributions to this PR and are only assisting with merging the main branch, please remove yourself as a co-author when merging the PR.
• Please feel free to join our Slack channel at https://sgl-fru7574.slack.com/archives/C09784E3EN6 to discuss your PR.

[gemini-code-assist]

Summary of Changes

Hello @yubofredwang, 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 Flash Attention as an alternative backend for the attention mechanism, aiming to enhance computational efficiency. It refactors dependency management to use pyproject.toml and introduces a new class, LlamaFlashAttention, to encapsulate the Flash Attention logic. The changes are thoroughly tested with a new unit test file that compares the behavior of the new Flash Attention implementation against the existing standard attention.

Highlights

• Flash Attention Backend Integration: Introduced Flash Attention as a new backend option for improved performance in attention mechanisms.
• Dependency Management Update: Migrated project dependencies from requirements.txt to pyproject.toml, adding flash-attn as a new dependency.
• New Flash Attention Class: Implemented LlamaFlashAttention to leverage flash_attn_func for efficient attention computation.
• Comprehensive Testing: Added a dedicated test suite (test_flash_attention.py) to validate the correctness and similarity of outputs and gradients between the standard and Flash Attention implementations.

🧠 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 introduces support for a Flash Attention backend, which is a great performance enhancement. The project's packaging has also been modernized by migrating from requirements.txt to pyproject.toml. However, the review identified a critical issue in the LlamaFlashAttention implementation: it does not correctly handle padded input sequences, which can lead to incorrect model outputs. The accompanying tests are insufficient as they are structured in a way that hides this bug. I've provided detailed comments and suggestions to address these points.