Fix SFT integration test launch failure#1470
Fix SFT integration test launch failure#1470finbarrtimbers merged 2 commits intomainallenai/open-instruct:mainfrom finbarr/fix-integration-testsallenai/open-instruct:finbarr/fix-integration-testsCopy head branch name to clipboard
Conversation
…val_jobs flag The SFT integration test set --push_to_hub false but left try_launch_beaker_eval_jobs at its default of True, causing a ValueError during the mason.py cache step. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Summary of ChangesHello @finbarrtimbers, 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 addresses a critical failure in the SFT integration test within the CI pipeline. By explicitly disabling the attempt to launch Beaker evaluation jobs when pushing to the hub is also disabled, it resolves a validation error that was preventing the test from completing successfully, thereby restoring the stability and reliability of the continuous integration process for this test. 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
|
There was a problem hiding this comment.
Code Review
This pull request addresses a launch failure in the SFT integration test. The issue stemmed from a validation check in finetune.py that correctly flags an inconsistency: attempting to launch Beaker evaluation jobs (try_launch_beaker_eval_jobs=True by default) while not pushing the model to the Hugging Face Hub (push_to_hub=false). The fix introduces --try_launch_beaker_eval_jobs false to the test script's launch command, which resolves the ValueError. The change is correct and effectively fixes the bug.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The SFT integration test was failing in CI because
--push_to_hub falsewas set without also setting--try_launch_beaker_eval_jobs false. Sincetry_launch_beaker_eval_jobsdefaults toTrue, the__post_init__validation infinetune.pyraised aValueErrorduring mason.py's dataset caching step.Added
--try_launch_beaker_eval_jobs falseto the SFT integration test script.Manually triggered run: https://github.com/allenai/open-instruct/actions/runs/21910196366