[Data] optimize dataset.unique() #49296
Merged
Conversation
This file contains 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
wingkitlee0
force-pushed
the
optimize-dataset-unique
branch
2 times, most recently
from
f21dbeb to
a2270a7
Compare
wingkitlee0
force-pushed
the
optimize-dataset-unique
branch
4 times, most recently
from
91c0e5a to
3802155
Compare
wingkitlee0
force-pushed
the
optimize-dataset-unique
branch
from
2882ed5 to
5c5cc7f
Compare
|
@raulchen this should be ready for review. thanks |
raulchen
approved these changes
Dec 28, 2024
wingkitlee0
force-pushed
the
optimize-dataset-unique
branch
2 times, most recently
from
706c2ae to
80f6cb2
Compare
- small clean up to private functions in ds.aggregate Signed-off-by: Kit Lee <[email protected]>
wingkitlee0
force-pushed
the
optimize-dataset-unique
branch
from
80f6cb2 to
4001a64
Compare
raulchen
reviewed
Dec 31, 2024
Signed-off-by: Hao Chen <[email protected]>
|
@raulchen Happy new year! it looks like something failed in the premerge (after the docstring commit), but I don't have access to see what's going on. |
|
Looks unrelated. Retrying failed jobs. |
roshankathawate
pushed a commit
to roshankathawate/ray
that referenced
this pull request
Jan 7, 2025
## Why are these changes needed? The current implementation uses `groupby(column).count()` that causes a full sort. The new implementation uses `AggregateFn` which uses `groupby(None)` and `set()` to aggregate unique values. The time complexity should be O(N / parallelism) according to `ds.aggregate()`. It's about 10x faster in my local test. Some part of `test_unique` is removed because it was designed for the original implementation. ## Related issue number Closes ray-project#49298 --------- Signed-off-by: Kit Lee <[email protected]> Signed-off-by: Hao Chen <[email protected]> Co-authored-by: Hao Chen <[email protected]>
roshankathawate
pushed a commit
to roshankathawate/ray
that referenced
this pull request
Jan 9, 2025
## Why are these changes needed? The current implementation uses `groupby(column).count()` that causes a full sort. The new implementation uses `AggregateFn` which uses `groupby(None)` and `set()` to aggregate unique values. The time complexity should be O(N / parallelism) according to `ds.aggregate()`. It's about 10x faster in my local test. Some part of `test_unique` is removed because it was designed for the original implementation. ## Related issue number Closes ray-project#49298 --------- Signed-off-by: Kit Lee <[email protected]> Signed-off-by: Hao Chen <[email protected]> Co-authored-by: Hao Chen <[email protected]> Signed-off-by: Roshan Kathawate <[email protected]>
anyadontfly
pushed a commit
to anyadontfly/ray
that referenced
this pull request
Feb 13, 2025
## Why are these changes needed? The current implementation uses `groupby(column).count()` that causes a full sort. The new implementation uses `AggregateFn` which uses `groupby(None)` and `set()` to aggregate unique values. The time complexity should be O(N / parallelism) according to `ds.aggregate()`. It's about 10x faster in my local test. Some part of `test_unique` is removed because it was designed for the original implementation. ## Related issue number Closes ray-project#49298 --------- Signed-off-by: Kit Lee <[email protected]> Signed-off-by: Hao Chen <[email protected]> Co-authored-by: Hao Chen <[email protected]> Signed-off-by: Puyuan Yao <[email protected]>
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.
Why are these changes needed?
The current implementation uses
groupby(column).count()that causes a full sort. The new implementation usesAggregateFnwhich usesgroupby(None)andset()to aggregate unique values.The time complexity should be O(N / parallelism) according to
ds.aggregate().It's about 10x faster in my local test.
Some part of
test_uniqueis removed because it was designed for the original implementation.Related issue number
Closes #49298
Checks
git commit -s) in this PR.scripts/format.shto lint the changes in this PR.method in Tune, I've added it in
doc/source/tune/api/under thecorresponding
.rstfile.