[ML] Performance optimisation for classification and regression model training #2024
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
valeriy42
approved these changes
Sep 14, 2021
|
The failure was timing related |
tveasey
added a commit
that referenced
this pull request
Sep 15, 2021
… training (#2027) Caching example splits proves to be a good runtime optimisation. Testing using our experiment driver on around 250 small data sets the mean runtime dropped from 380s to 320s. However, the runtime improvements were larger on the larger data sets. I'll attach some benchmarks for large datasets as well. I also removed CImmutableRadixSet which was no longer used since looking up splits is not on the hot path. Note that this does have potential to change results slightly. The source of the difference is that we now store the candidate splits in float precision rather than double precision. This ports #2024 to the incremental training feature branch. This collides with a certain amount of code change so I've essentially reapplied the change to this branch.
|
Training on 80% Higgs 1M runtime before 9057s after 7641s. |
tveasey
added a commit
to tveasey/ml-cpp-1
that referenced
this pull request
Oct 11, 2021
… training (elastic#2024) Caching example splits proves to be a good runtime optimisation. Testing using our experiment driver on around 250 small data sets the mean runtime dropped from 380s to 320s. However, the runtime improvements were larger on the larger data sets. I'll attach some benchmarks for large datasets as well. I also removed CImmutableRadixSet which was no longer used since looking up splits is not on the hot path. Note that this does have potential to change results slightly. The source of the difference is that we now store the candidate splits in float precision rather than double precision.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Caching example splits proves to be a good runtime optimisation. Testing using our experiment driver on around 250 small data sets the mean runtime dropped from 380s to 320s. However, the runtime improvements were larger on the larger data sets. I'll attach some benchmarks for large datasets as well. I also removed
CImmutableRadixSetwhich was no longer used since looking up splits is not on the hot path.Note that this does have potential to change results slightly. The source of the difference is that we now store the candidate splits in float precision rather than double precision.