AddFeasibility transform
#4547
Open
AddFeasibility transform
#4547
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Summary:
Transform that adds failure-awareness capability to Ax optimization.
This transform enables Ax to learn from deterministic trial failures (ABANDONED trials) and avoid sampling similar parameter configurations that are likely to fail. It achieves this by:
1. Adding a "is_feasible" metric to experiment data based on trial status
- ABANDONED trials get feasibility value of 0.0 (infeasible)
- Other trials get feasibility value of 1.0 (feasible)
2. Adding a feasibility constraint to the optimization config
- The constraint enforces P(is_feasible) >= threshold
- This guides the acquisition function to avoid infeasible regions
NOTE: We should maybe pick a different word than "feasibility" to not be confused with feasibility in the sense of not violating user-specified outcome constraints.
Differential Revision: D85185246
|
@sunnyshen321 has exported this pull request. If you are a Meta employee, you can view the originating Diff in D85185246. |
meta-cla
bot
added
the
CLA Signed
sunnyshen321
pushed a commit
to sunnyshen321/Ax
that referenced
this pull request
Nov 18, 2025
Summary:
Transform that adds failure-awareness capability to Ax optimization.
This transform enables Ax to learn from deterministic trial failures (ABANDONED trials) and avoid sampling similar parameter configurations that are likely to fail. It achieves this by:
1. Adding a "is_feasible" metric to experiment data based on trial status
- ABANDONED trials get feasibility value of 0.0 (infeasible)
- Other trials get feasibility value of 1.0 (feasible)
2. Adding a feasibility constraint to the optimization config
- The constraint enforces P(is_feasible) >= threshold
- This guides the acquisition function to avoid infeasible regions
NOTE: We should maybe pick a different word than "feasibility" to not be confused with feasibility in the sense of not violating user-specified outcome constraints.
Differential Revision: D85185246
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary:
Transform that adds failure-awareness capability to Ax optimization.
This transform enables Ax to learn from deterministic trial failures (ABANDONED trials) and avoid sampling similar parameter configurations that are likely to fail. It achieves this by:
Adding a "is_feasible" metric to experiment data based on trial status
- ABANDONED trials get feasibility value of 0.0 (infeasible)
- Other trials get feasibility value of 1.0 (feasible)
Adding a feasibility constraint to the optimization config
- The constraint enforces P(is_feasible) >= threshold
- This guides the acquisition function to avoid infeasible regions
NOTE: We should maybe pick a different word than "feasibility" to not be confused with feasibility in the sense of not violating user-specified outcome constraints.
Differential Revision: D85185246