Skip to content
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.

Sign up for GitHub

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

Jump to bottom

Evaluate oracle at target task and fidelity #2727

Closed
wants to merge 2 commits into from
Closed

Evaluate oracle at target task and fidelity #2727

wants to merge 2 commits into from

Commits on Aug 29, 2024

  1. Merge create_single_objective_problem_from_botorch and create_multi_o… 

    …bjective_problem_from_botorch; support constrained MOO (facebook#2722)

Summary:
Pull Request resolved: facebook#2722

Context: These functions have a lot of overlapping functionality. Combining them makes it easier to extend their functionality (for example, by supporting constrained MOO).

This PR:
* Combines `create_single_problem_objective_from_botorch` and `create_multi_objective_problem_from_botorch` into `create_problem_from_botorch`
* Reads `lower_is_better` off the test problem -- BoTorch test problems assume minimization unless `negate` is set to True -- instead of requiring the user to pass it (multi-objective problems used to always have lower_is_better=True, so this is more accurate and prevents mismatch)
* Adds support for constrained MOO problems from Botorch
* Remove stub for constrained MOO problem that only existed to test for an exception that is no longer raised

Differential Revision: D61877865

Reviewed By: Balandat
    @esantorella @facebook-github-bot
    esantorella authored and facebook-github-bot committed Aug 29, 2024
    Configuration menu
    Copy the full SHA
    a10871e View commit details
    Browse the repository at this point in the history

  2. Evaluate oracle at target task and fidelity 

    Summary:
`BenchmarkRunner.evaluate_oracle` now evaluates at the target task and fidelity.

A weakness of this solution is that it requires passing the search space to the runner. Generally, there is too much information that must be on both the problem and the runner; this could be addressed by making the problem an attribute of the runner rather than the other way around.

Reviewed By: sdaulton

Differential Revision: D61884295
    @esantorella @facebook-github-bot
    esantorella authored and facebook-github-bot committed Aug 29, 2024
    Configuration menu
    Copy the full SHA
    f4cb63e View commit details
    Browse the repository at this point in the history