Github user jkbradley commented on a diff in the pull request: https://github.com/apache/spark/pull/21183#discussion_r187217859 --- Diff: mllib/src/main/scala/org/apache/spark/ml/clustering/LDA.scala --- @@ -622,11 +623,11 @@ object LocalLDAModel extends MLReadable[LocalLDAModel] { val vectorConverted = MLUtils.convertVectorColumnsToML(data, "docConcentration") val matrixConverted = MLUtils.convertMatrixColumnsToML(vectorConverted, "topicsMatrix") val Row(vocabSize: Int, topicsMatrix: Matrix, docConcentration: Vector, - topicConcentration: Double, gammaShape: Double) = + topicConcentration: Double, gammaShape: Double, seed: Long) = --- End diff -- This will break backwards compatibility of ML persistence (when users try to load LDAModels saved using past versions of Spark). Could you please test this manually by saving a LocalLDAModel using Spark 2.3 and loading it with a build of your PR? You can fix this by checking for the Spark version (in the `metadata`) and loading the seed for Spark >= 2.4.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org