Performance & runtime improvements to info-theoretic acquisition functions (1/N) #2748
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
facebook-github-bot
added
the
CLA Signed
Contributor
|
@sdaulton has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
Contributor
|
Thanks! It seems like |
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #2748 +/- ##
=======================================
Coverage 99.99% 99.99%
=======================================
Files 219 219
Lines 20887 20893 +6
=======================================
+ Hits 20886 20892 +6
Misses 1 1 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
Contributor
Author
|
@sdaulton for sure! I currently observe similar things for JES, but I'm not sure whether the found points are actually higher in acquisition function value or not (for either LogEI or JES) |
Contributor
|
That would be interesting to see |
saitcakmak
requested changes
Feb 20, 2025
Contributor
saitcakmak
left a comment
saitcakmak
left a comment
There was a problem hiding this comment.
Hi Carl! This seems like a decent improvement. Just a few comments in-line
hvarfner
force-pushed
the
optimize_posterior_samples
branch
2 times, most recently
from
520aad7 to
43744fa
Compare
hvarfner
force-pushed
the
optimize_posterior_samples
branch
6 times, most recently
from
3ae01cb to
f4f01bf
Compare
saitcakmak
approved these changes
Feb 21, 2025
Contributor
|
@sdaulton has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
hvarfner
force-pushed
the
optimize_posterior_samples
branch
3 times, most recently
from
ef6822d to
6b3002b
Compare
facebook-github-bot
pushed a commit
that referenced
this pull request
Feb 25, 2025
…tions (0/N) - Restructuring of sampling methods (#2753) Summary: Reshuffling of sampling methods that are not directly related to acquisition function optimization (i.e., don't take it as an argument) based on [this discussion](#2748 (comment)). To remove code duplication specifically related to optimization of info-theoretic acquisition functions, this seemed like sensible moves! Pull Request resolved: #2753 Test Plan: Moved unittests and added new one for `boltzmann_sample`, which was used throughout and is once again used in subsequent PRs. ## Related PRs First of a series, like [this one](#2748). Reviewed By: esantorella Differential Revision: D70131981 Pulled By: saitcakmak fbshipit-source-id: 48dd86e7e06006054294d7cd8b9a3d318b0b0ad1
hvarfner
force-pushed
the
optimize_posterior_samples
branch
from
6b3002b to
9d39469
Compare
Contributor
|
@sdaulton has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
1 similar comment
Contributor
|
@sdaulton has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
sdaulton
suggested changes
Feb 27, 2025
hvarfner
force-pushed
the
optimize_posterior_samples
branch
from
a2a2b46 to
d2ee1de
Compare
Contributor
|
@sdaulton has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
hvarfner
added a commit
to hvarfner/botorch
that referenced
this pull request
Nov 19, 2025
…tions (1/N) Summary: A series of improvements directed towards improving the performance of PES & JES, as well as their MultiObj counterparts. ## Motivation As pointed out by SebastianAment in [this paper](https://arxiv.org/pdf/2310.20708), the BoTorch variant of JES, and to an extent PES, is brutally slow an suspiciously ill-performing. To bring them up to their potential, I've added a series of performance improvements: **1. Improvement to get_optimal_samples and optimal_posterior_samples**: As this is an integral part of their efficiency, I've added `suggestions` (similar to `sample_around_best`) to `optimize_posterior_samples`. Marginal runtime improvement in acquisition optimization (sampling time practically unchanged):  Substantial performance improvement: . **2. Added initializer to acquisition funcction optimization**: Similar to KG, ES methods have sensible suggestions for acquisition function optimization in the form of the sampled optima. This drastically reduces the time of acquisition function optimization, which could on occasion take 30+ seconds when `num_restarts` was large `>4`. Benchmarking INC **2b. Multi-objective support for initializer**: By re-naming arguments of the multi-objective variants, we get consistency and support for MO variants. **3. Enabled gradient-based optimization for PES**: The current implementation contains a while-loop which forces the gradients to be recursively computed. This commonly causes NaN gradients, which is why the recommended option is `"with_grad": False` in the tutorial. One `detach()` alleviates this issue, enabling gradient-based optimization. NOTE: this has NOT been ablated, since the non-grad optimization is extremely computationally demanding. X-link: meta-pytorch#2748 Reviewed By: saitcakmak Differential Revision: D69787454 Pulled By: hvarfner
hvarfner
force-pushed
the
optimize_posterior_samples
branch
from
d2ee1de to
7ac20d8
Compare
…tions (1/N) (meta-pytorch#2748) Summary: A series of improvements directed towards improving the performance of PES & JES, as well as their MultiObj counterparts. ## Motivation As pointed out by SebastianAment in [this paper](https://arxiv.org/pdf/2310.20708), the BoTorch variant of JES, and to an extent PES, is brutally slow an suspiciously ill-performing. To bring them up to their potential, I've added a series of performance improvements: **1. Improvement to get_optimal_samples and optimal_posterior_samples**: As this is an integral part of their efficiency, I've added `suggestions` (similar to `sample_around_best`) to `optimize_posterior_samples`. Marginal runtime improvement in acquisition optimization (sampling time practically unchanged):  Substantial performance improvement: . **2. Added initializer to acquisition funcction optimization**: Similar to KG, ES methods have sensible suggestions for acquisition function optimization in the form of the sampled optima. This drastically reduces the time of acquisition function optimization, which could on occasion take 30+ seconds when `num_restarts` was large `>4`. Benchmarking INC **2b. Multi-objective support for initializer**: By re-naming arguments of the multi-objective variants, we get consistency and support for MO variants. **3. Enabled gradient-based optimization for PES**: The current implementation contains a while-loop which forces the gradients to be recursively computed. This commonly causes NaN gradients, which is why the recommended option is `"with_grad": False` in the tutorial. One `detach()` alleviates this issue, enabling gradient-based optimization. NOTE: this has NOT been ablated, since the non-grad optimization is extremely computationally demanding. X-link: meta-pytorch#2748 Test Plan: Unit tests and benchmarking. ## Related PRs First of a couple! Bonus: while benchmarking, I had issues repro'ing the LogEI performance initially. I found that `sample_around_best` made LogEI worse on Mich5. All experiments are otherwise a repro of the settings used in the LogEI paper.  Reviewed By: saitcakmak Differential Revision: D69787454 Pulled By: hvarfner
hvarfner
force-pushed
the
optimize_posterior_samples
branch
from
7ac20d8 to
70d12de
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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.
A series of improvements directed towards improving the performance of PES & JES, as well as their MultiObj counterparts.
Motivation
As pointed out by @SebastianAment in this paper, the BoTorch variant of JES, and to an extent PES, is brutally slow an suspiciously ill-performing. To bring them up to their potential, I've added a series of performance improvements:
1. Improvement to get_optimal_samples and optimal_posterior_samples: As this is an integral part of their efficiency, I've added
.
suggestions(similar tosample_around_best) tooptimize_posterior_samples.Marginal runtime improvement in acquisition optimization (sampling time practically unchanged):
Substantial performance improvement:
2. Added initializer to acquisition funcction optimization: Similar to KG, ES methods have sensible suggestions for acquisition function optimization in the form of the sampled optima. This drastically reduces the time of acquisition function optimization, which could on occasion take 30+ seconds when
num_restartswas large>4.Benchmarking INC
2b. Multi-objective support for initializer: By re-naming arguments of the multi-objective variants, we get consistency and support for MO variants.
3. Enabled gradient-based optimization for PES: The current implementation contains a while-loop which forces the gradients to be recursively computed. This commonly causes NaN gradients, which is why the recommended option is
"with_grad": Falsein the tutorial. Onedetach()alleviates this issue, enabling gradient-based optimization.NOTE: this has NOT been ablated, since the non-grad optimization is extremely computationally demanding.
Test Plan
Unit tests and benchmarking.
Related PRs
First of a couple!
Bonus: while benchmarking, I had issues repro'ing the LogEI performance initially. I found that
sample_around_bestmade LogEI worse on Mich5. All experiments are otherwise a repro of the settings used in the LogEI paper.