Github user jkbradley commented on a diff in the pull request:
https://github.com/apache/spark/pull/18924#discussion_r140621444
--- Diff:
mllib/src/main/scala/org/apache/spark/mllib/clustering/LDAOptimizer.scala ---
@@ -462,36 +462,55 @@ final class OnlineLDAOptimizer extends LDAOptimizer {
val expElogbetaBc = batch.sparkContext.broadcast(expElogbeta)
val alpha = this.alpha.asBreeze
val gammaShape = this.gammaShape
+ val optimizeDocConcentration = this.optimizeDocConcentration
+ // We calculate logphat in the same pass as other statistics, but we
only need
+ // it if we are optimizing docConcentration
+ val logphatPartOptionBase = () => if (optimizeDocConcentration)
Some(BDV.zeros[Double](k))
+ else None
- val stats: RDD[(BDM[Double], List[BDV[Double]])] = batch.mapPartitions
{ docs =>
+ val stats: RDD[(BDM[Double], Option[BDV[Double]], Int)] =
batch.mapPartitions { docs =>
--- End diff --
Let's use Long for the doc count since it could overflow for large datasets
and miniBatchFraction
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
