Github user wzhfy commented on the issue: https://github.com/apache/spark/pull/19406 Your proposed way is more consistent with the paper and I also think it's more natural to start from 0. The purpose of this patch is to not introduce accuracy regression for other cases, and improve some corner cases. I can't tell which one is better, I think both ways can work.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org