[v1] add init plugin

[hiyouga]

What does this PR do?

Fixes # (issue)

Before submitting

• Did you read the contributor guideline?
• Did you write any new necessary tests?

[gemini-code-assist]

Summary of Changes

Hello @hiyouga, 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 enhances the flexibility of model initialization and modularizes the sampling architecture within the LlamaFactory project. By introducing a new plugin system for model loading, it allows for more control over how models are placed on devices, which is particularly beneficial for handling large models efficiently. Concurrently, the sampling components have been refactored into a more extensible design, paving the way for easier integration of various sampling backends.

Highlights

• New Model Initialization Plugin System: Introduced a flexible plugin system for model initialization, allowing users to specify how models are loaded onto devices (e.g., on meta device, on rank 0 CPU, or default accelerator). This is managed via a new init_config argument in ModelArguments.
• Refactored Sampling Architecture: The sampling logic has been modularized and refactored. The old chat_sampler.py has been removed and replaced with a more abstract base_sampler.py which defines BaseEngine and BaseSampler classes, allowing for different sampling backends.
• Removed use_fast_processor Argument: The use_fast_processor argument has been removed from ModelArguments and _init_processor in model_loader.py, as Transformers v5 now consistently uses fast tokenizers.
• Enhanced Test Infrastructure: Minor improvements to the test suite, including a refactor of pytest_collection_modifyitems and the addition of a fixture to clean up distributed state after tests, ensuring better test isolation and reliability.

🧠 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 an init_plugin to provide more control over model initialization, which is a valuable addition for handling large models and distributed environments. The implementation of the plugin and its integration into the model loading process is well-structured.

My review has identified a critical issue in the new base_sampler.py where the HuggingFaceEngine class is incomplete, which will cause a runtime error. I also found a high-severity bug in cli_sampler.py related to incorrect handling of asynchronous code. Additionally, I've provided a medium-severity suggestion to improve code style by moving a local import.

Overall, the changes are a good step forward, and after addressing these points, the PR should be in great shape.