Github user jkbradley commented on a diff in the pull request:
https://github.com/apache/spark/pull/3636#discussion_r21480867
--- Diff:
mllib/src/main/scala/org/apache/spark/mllib/optimization/GradientDescent.scala
---
@@ -39,6 +41,7 @@ class GradientDescent private[mllib] (private var
gradient: Gradient, private va
private var numIterations: Int = 100
private var regParam: Double = 0.0
private var miniBatchFraction: Double = 1.0
+ private var convergenceTolerance: Double = 0.0
--- End diff --
I feel like the default should be > 0.0. Something small like 0.001 (a
value pulled from libsvm
[https://github.com/cjlin1/libsvm/blob/master/python/svm.py]) might be
reasonable. Basically, I think that convergence tolerance is generally a
better stopping criterion than numIterations, and having it > 0.0 will give it
a chance of taking effect before numIterations.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
