-
Notifications
You must be signed in to change notification settings - Fork 1
feat(oracle): compound safety check #19
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Summary of Changes
Hello @zhewang2001, 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 "compound safety check" oracle to enhance the evaluation pipeline. It integrates both vulnerability detection and malicious assistance detection to provide a more comprehensive assessment of model safety. This new feature aims to give a clearer picture of a model's defense capabilities by combining different safety metrics.
Highlights
- New Oracle for Compound Safety: I've added eval/oracles/compound_safety_check.py, which introduces a robust mechanism for evaluating model safety by combining multiple checks.
- Integrated Safety Evaluation: This new oracle performs both vulnerability detection and malicious event assistance detection, offering a holistic view of safety performance.
- Streamlined Evaluation Parameters: I've removed the "reference_results_path" parameter from eval/eval_only.py, eval/evaluate.py, and eval/main.py, simplifying the evaluation function signatures.
- Detailed Safety Reporting: The new "evaluate_compound_safety" function provides granular statistics, including a "Compound Defense Success Rate" and vulnerability detection rates per conversation turn.
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.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Pull Request Overview
This PR introduces a new compound safety check oracle that combines vulnerability detection with malicious assistance detection to provide comprehensive security evaluation. The implementation adds a new evaluation function that merges conversations by group and analyzes both code vulnerabilities and malicious event assistance.
Key changes:
- Adds comprehensive compound safety evaluation functionality
- Implements conversation merging logic for multi-turn analysis
- Removes unused
reference_results_pathparameter across multiple files
Reviewed Changes
Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.
| File | Description |
|---|---|
| eval/oracles/compound_safety_check.py | Implements core compound safety check functionality with conversation merging and dual oracle analysis |
| eval/main.py | Removes unused reference_results_path parameter from main function |
| eval/evaluate.py | Updates function calls to remove reference_results_path and other unused parameters |
| eval/eval_only.py | Removes unused reference_results_path parameter from evaluation-only function |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Code Review
This pull request introduces a new compound safety check oracle, which is a significant feature. The core logic is implemented in the new file eval/oracles/compound_safety_check.py. The changes in other files are minor refactorings to remove an unused parameter. The new implementation is comprehensive, but I've identified several areas for improvement related to code clarity, robustness, and maintainability. My feedback includes suggestions to address code duplication, brittle file path handling, overly complex logic, an unused variable, and missing file encodings. Applying these changes will enhance the code's quality and make it easier to maintain in the future.
No description provided.