Uh oh!
There was an error while loading. Please reload this page.
[SPARK-21027][ML][PYTHON] Added tunable parallelism to one vs. rest in both Scala mllib and Pyspark - #19110
[SPARK-21027][ML][PYTHON] Added tunable parallelism to one vs. rest in both Scala mllib and Pyspark#19110WeichenXu123 wants to merge 25 commits into
Conversation
…st classification. Added a parallelism parameter to the scala implementation of one vs. rest for python persistence but have not yet used it to tune the scala parallelism implementation.
…n of the one vs. rest algorithm.
…ts for testing that parallelism doesn't affect the output.
…executor service with a given level of parallelism in a separat trait that OneVsRest inherits from.
…ng OneVsRest and OneVsRest model JavaMLReadable and JavaMLWritable)
SparkQA
commented
Sep 3, 2017
Test build #81350 has finished for PR 19110 at commit
|
WeichenXu123
commented
Sep 3, 2017
Jenkins, test this please. |
SparkQA
commented
Sep 3, 2017
Test build #81352 has finished for PR 19110 at commit
|
24f4499 to
fc6fd5eCompareSparkQA
commented
Sep 4, 2017
Test build #81381 has finished for PR 19110 at commit
|
LGTM Btw, if someone else merges this, then @ajaysaini725 and @WeichenXu123 should both be authors, with @ajaysaini725 as the primary one. |
BryanCutler
left a comment
There was a problem hiding this comment.
Just a few minor comments, I think left over from the previous PR.
| None, "TypeConverters.toString"), | ||
| ("aggregationDepth", "suggested depth for treeAggregate (>= 2).", "2", | ||
| "TypeConverters.toInt"), | ||
| ("parallelism", "number of threads to use when fitting models in parallel (>= 1).", "1", |
There was a problem hiding this comment.
Maybe this should have a more generic description since it is a shared param?
| /** | ||
| * @group expertSetParam | ||
| * The implementation of parallel one vs. rest runs the classification for | ||
| * each class in a separate threads. |
There was a problem hiding this comment.
from the previous PR, @group expertSetParam should be at the bottom
SparkQA
commented
Sep 6, 2017
Test build #81436 has finished for PR 19110 at commit
|
SparkQA
commented
Sep 6, 2017
Test build #81438 has finished for PR 19110 at commit
|
MLnick
commented
Sep 6, 2017
FYI I went ahead and merged #16774 - the doc on the shared param trait is a little more detailed there which I slightly prefer. @WeichenXu123 you will just need to resolve the small merge conflict that introduces. |
WeichenXu123
commented
Sep 6, 2017
@MLnick Conflict resolved. Thanks! |
SparkQA
commented
Sep 6, 2017
Test build #81456 has finished for PR 19110 at commit
|
BryanCutler
commented
Sep 6, 2017
LGTM! |
WeichenXu123
commented
Sep 9, 2017
Thanks @MLnick@BryanCutler . Would you mind helping review another similar PR #19122 ? We need some other features but blocking on that PR. Thanks! |
| Mixin for param parallelism: number of threads to use when fitting models in parallel. | ||
| """ | ||
| parallelism = Param(Params._dummy(), "parallelism", "the number of threads to use when running parallel algorithms.", typeConverter=TypeConverters.toInt) |
There was a problem hiding this comment.
nit: Is this out of date? It's missing the "(>= 1)" from the code gen file.
jkbradley
commented
Sep 11, 2017
Other than that 1 item, this looks ready |
SparkQA
commented
Sep 12, 2017
Test build #81653 has finished for PR 19110 at commit
|
jkbradley
commented
Sep 12, 2017
LGTM |
| * each class in a separate threads. | ||
| * | ||
| * @group expertSetParam | ||
| */ |
There was a problem hiding this comment.
Thanks! I create a PR to fix this.
## What changes were proposed in this pull request? add missing since tag for `setParallelism` in apache#19110 ## How was this patch tested? N/A Author: WeichenXu <weichen.xu@databricks.com> Closesapache#19214 from WeichenXu123/minor01.
What changes were proposed in this pull request?
Added tunable parallelism to the pyspark implementation of one vs. rest classification. Added a parallelism parameter to the Scala implementation of one vs. rest along with functionality for using the parameter to tune the level of parallelism.
I take this PR #18281 over because the original author is busy but we need merge this PR soon.
After this been merged, we can close#18281 .
How was this patch tested?
Test suite added.