[ML] Eagerly discard node statistics for leaves which we will never split #1125
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
tveasey
commented
Apr 4, 2020
| counter_t::E_DFTPMEstimatedPeakMemoryUsage) < 6000000); | ||
| BOOST_TEST_REQUIRE(core::CProgramCounters::counter(counter_t::E_DFTPMPeakMemoryUsage) < 1500000); | ||
| counter_t::E_DFTPMEstimatedPeakMemoryUsage) < 4500000); | ||
| BOOST_TEST_REQUIRE(core::CProgramCounters::counter(counter_t::E_DFTPMPeakMemoryUsage) < 1600000); |
There was a problem hiding this comment.
We're not using more memory; however, we now properly account for the memory used by container for the leaf statistics because we resize it before estimating its memory usage.
tveasey
added a commit
to tveasey/ml-cpp-1
that referenced
this pull request
Apr 17, 2020
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Outdated suggestions cannot be applied.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.
This is a simple memory optimisation for regression and classification model training.
Since we limit the maximum tree size, we only need retain statistics for the maximum gain "max tree size" - "current tree size" leaves. Since we add one leaf statistic for each node we add to the tree we only ever need at most "maximum tree size" / 2 node statistics in total.
We use an upper bound for the leaf statistics memory estimates so this also indirectly helps improve our memory estimate accuracy and so helps with #1106. For example, this reduced estimates by around 25% testing.