Conversation
…on fields, refactoring `configuration_reference.md` to link to it and updating `index.md` and `custom.css`.
…resolvers and support colon syntax for task function resolution.
…, and expanded sandbox configuration options in documentation and API.
…rguments to resolved task metadata, updating the config parity tool.
…ration, and refine tag filtering logic.
Summary of ChangesHello, 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 refactors the configuration system to improve its adaptability and extensibility, particularly for Yardstick and diverse local use cases. It introduces new configuration options, standardizes naming conventions, and decouples environment-specific settings, allowing for more flexible and maintainable evaluation setups. A major enhancement is the introduction of tag-based filtering, which provides powerful new ways to manage task and sample selection. Highlights
🧠 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. Changelog
Activity
Using Gemini Code AssistThe 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
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 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
|
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request introduces a significant number of updates to the configuration system to support Yardstick and local use cases. Key changes include adding new fields to Job and Task models (like description, imagePrefix, args), introducing tag-based filtering for tasks and samples, and making the sandbox registry configurable. The Task.taskFunc field has been consistently renamed to Task.func across the codebase. Extensive documentation has been added or updated to reflect these changes, including a new detailed YAML configuration reference.
My review focuses on ensuring consistency across the implementation and documentation. I've found a couple of minor inconsistencies in the documentation regarding field names. I also noticed one part of the implementation in the Dart resolver that seems to contradict the implementation plan regarding hardcoded setup commands. Overall, this is a well-structured and comprehensive update.
packages/dataset_config_dart/lib/src/resolvers/eval_set_resolver.dart
Outdated
Show resolved
Hide resolved
…lution and expose new parsing utilities.
…rguments into single map fields, removing individual top-level parameters and `JobTask.systemMessage`.
…lists, rename context files to files, skill paths to skills, and add task parameters.
…les` and `setup` for more flexible resource management.
…led list of sub-fields with a concise summary and external reference.
…ns, breaking changes, and documentation improvements.
|
/gemini review |
…ename `context_files` to `files`, and remove `allowed_variants`.
There was a problem hiding this comment.
Code Review
This pull request significantly refactors the configuration system for evaluation runs, impacting both Dart and Python packages. Key changes include introducing new fields for Job (description, image prefix, tag-based task/sample filtering) and JobTask (per-task arguments). Task models now directly support systemMessage, sandboxParameters, and task-level files and setup declarations. Breaking changes involve renaming Task.taskFunc to Task.func, making the sandbox registry configurable, replacing workspace and tests with files and setup, and consolidating various configuration arguments into Job.sandbox and Job.inspectEvalArguments. Variant fields like context_files and skill_paths have been renamed to files and skills respectively, and variant restriction logic has shifted to include-variants/exclude-variants on job tasks. Documentation has been updated with a new YAML configuration reference and guides. Review comments suggest improving the documentation for mcp_servers by adding descriptions and explanations of 'MCP server identifiers' for better user understanding. Additionally, an improvement opportunity was identified in the create_mcp_servers function to include the server's name in the ValueError message when a command is missing, which would aid in debugging problematic configurations.
packages/dash_evals/src/dash_evals/runner/tasks/task_helpers.py
Outdated
Show resolved
Hide resolved
…tart and tutorial content with new getting started, evaluation, and configuration guides.
…n model, and related documentation.
…N, and CSV formats with new `dataset` key and `json_runner`.
|
/gemini review |
There was a problem hiding this comment.
Code Review
This pull request introduces significant updates to the Yardstick configuration, including new features, breaking changes, and documentation updates. The changes span multiple files, primarily focusing on the CHANGELOG.md, IMPLEMENTATION_PLAN.md, and various Dart and Python files within the dataset_config packages. The updates aim to enhance the flexibility and usability of the configuration system for both local use cases and Yardstick integration. The review comments address issues related to documentation requirements and style guide adherence.
| - Updated `docs/reference/configuration_reference.md` with new examples and directory structure. | ||
| - Updated `docs/guides/config.md`. | ||
|
|
||
| ## 11 March, 2025 |
There was a problem hiding this comment.
The date 11 March, 2025 is mentioned in the changelog. It is extremely important to avoid commenting on dates, times, or versions mentioned in non-code areas such as copyright sections, as per the core instructions.
1 participant
Updating config to serve both our local use cases and Yardstick.