Skip to content

Conversation

@google-labs-jules
Copy link
Contributor

This change enhances the MIPROv2 optimizer to tune instructions and the number of examples, not just temperature and demo count. It aligns the optimizer more closely with its intended purpose of performing comprehensive prompt engineering. By expanding the search space, the optimizer can now discover more effective combinations of instructions and examples, leading to improved model performance.

Fixes #425


PR created automatically by Jules for task 14991558155557114068 started by @dosco

The MIPROv2 optimizer was previously only optimizing for model temperature and the number of bootstrapped demos. This change expands the optimizer's search space to include instruction selection and the number of labeled examples, which is closer to the intended implementation described in the DSPy paper.

The key changes are:
- The `compilePython` method now generates a list of instruction candidates.
- The optimization request sent to the Python backend now includes the instruction candidates as a categorical parameter and the number of labeled examples as an integer parameter.
- The trial loop now applies the instruction and labeled example count suggested by the optimizer to a cloned program instance for evaluation.
- The final optimized program is now created with the best-performing instruction and example configuration.
@google-labs-jules
Copy link
Contributor Author

👋 Jules, reporting for duty! I'm here to lend a hand with this pull request.

When you start a review, I'll add a 👀 emoji to each comment to let you know I've read it. I'll focus on feedback directed at me and will do my best to stay out of conversations between you and other bots or reviewers to keep the noise down.

I'll push a commit with your requested changes shortly after. Please note there might be a delay between these steps, but rest assured I'm on the job!

For more direct control, you can switch me to Reactive Mode. When this mode is on, I will only act on comments where you specifically mention me with @jules. You can find this option in the Pull Request section of your global Jules UI settings. You can always switch back!


For security, I will only act on instructions from the user who triggered this task.

New to Jules? Learn more at jules.google/docs.

@dosco dosco marked this pull request as ready for review November 27, 2025 18:23
@dosco dosco merged commit 2f3e6ac into main Nov 27, 2025
5 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

MIPROv2 Optimizer doesn't do anything significant

2 participants