Add some training tricks for RLVR #147
Conversation
There was a problem hiding this comment.
Summary of Changes
Hello @hiyuchang, 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 a new AddStrategy called DuplicateInformativeAddStrategy to the trinity framework. This strategy aims to enhance the experience replay mechanism by intelligently selecting and duplicating only those experiences that exhibit significant reward variance within a task, as referenced by the Polaris project. The goal is to ensure that the system primarily processes and learns from more 'informative' data, potentially leading to more effective and efficient learning outcomes.
Highlights
- New Strategy Implementation: I've implemented
DuplicateInformativeAddStrategy, a new experience addition strategy inspired by the Polaris project. This strategy is designed to select and duplicate experiences based on their reward variance. - Experience Filtering and Duplication Logic: The core logic in
DuplicateInformativeAddStrategyfilters incoming experiences. It groups them by task and only considers tasks where the reward variance exceeds a specifiedvariance_threshold. These 'informative' experiences are then duplicated until the total number of experiences reaches the original target size, with duplicated task IDs being offset to maintain uniqueness. - Module Integration: The new
DuplicateInformativeAddStrategyclass has been registered with theADD_STRATEGYregistry under the name 'duplicate_informative' and is now exposed through thetrinity.algorithm.add_strategypackage's__init__.pyfile, making it available for use.
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 in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.
| 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 issue 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 or fill out our survey to provide feedback.
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
-
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. ↩
There was a problem hiding this comment.
Code Review
This pull request introduces a new DuplicateInformativeAddStrategy for adding experiences to the buffer. The new strategy filters experiences based on reward variance and then duplicates the informative ones to reach a target count.
My review has identified a few issues:
- A critical issue with a misleading docstring that doesn't match the implementation.
- Some high-severity issues related to code style (imports inside a method) and performance (inefficient sampling).
- Medium-severity concerns about code duplication.
Please address these points to improve the code's correctness, performance, and maintainability.
|
/unittest-module-algorithm |
1 similar comment
|
/unittest-module-algorithm |
Summary
Tests
Github Test Reporter by CTRF 💚 |
Description
Addstrategy=duplicate_informative(from Polaris)Checklist
Please check the following items before code is ready to be reviewed.