Simplify TransitionCriteria classes by replacing MaxGenerationParallelism with MinTrials #4819
+250
−636
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
added 9 commits
January 23, 2026 13:12
…racked changes) Differential Revision: D90802656
Differential Revision: D86066476
Differential Revision: D91011637
Differential Revision: D91011862
Differential Revision: D91011958
Summary: wip Differential Revision: D91064377
Differential Revision: D91141100
Summary: This cleans up a todo from long ago, this is unique enough as it uses all components of a GeneratorSpec Differential Revision: D91229383
Summary: After discussing with Jason Chow, AEPysch no longer interfaces with Ax, instead it directly leverages BoTorch. There are no plans atm to integrate directly with Ax, but if that is taken on in the future, it would be better to re-onboard them using GenNode based GSs given the opportunity for a clean slate. This allows us to remove special casing for AEPysch from GS codebase Differential Revision: D91156800
meta-cla
bot
added
the
CLA Signed
…lism with MinTrials (facebook#4819) Summary: Pull Request resolved: facebook#4819 Pros of this change: * Reduce number of transition criterion classes --> may be easier for folks to wrap their head around the options with fewer options * In the future we want to move this type of budget to concurrency or something like that, so we can go ahead and improve the stack clarity now Cons of this change: * maxgenerationparallelism was a pretty explict name that defined what that criterion was for, without it you have to reason about the inputs of mintrials to understand how to configure parallelism * some maxgen special casing need to be checked using tc.block_gen_if_met property -- this is mainly for test cases (okay imo) and gs init where we construct the transition_to property for steps Note: Given we only have MinTrials as subclass of TrialBasedCriterion, we could actually compress this even further to only have TrialBasedCriterion if we would like. Totally open to that, but kept it as is as it could make developing new trialbased criterion easier to have the base class. Differential Revision: D91358715
mgarrard
force-pushed
the
export-D91358715
branch
from
418a881 to
33276b0
Compare
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:
Pros of this change:
Cons of this change:
Note:
Given we only have MinTrials as subclass of TrialBasedCriterion, we could actually compress this even further to only have TrialBasedCriterion if we would like. Totally open to that, but kept it as is as it could make developing new trialbased criterion easier to have the base class.
Differential Revision: D91358715