Github user akopich commented on the issue: https://github.com/apache/spark/pull/18924 BTW. Seems like `updateLambda` method relies (in older version as well) on `batchSize` only because this is `an optimization to avoid batch.count`. Shouldn't we rather use `nonEmptyDocsN` instead since we compute it efficietly now? But that is going to change logic...
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org