Github user avulanov commented on the pull request: https://github.com/apache/spark/pull/9614#issuecomment-168854243 @jkbradley If we want to follow the logic "users think they are still using the existing optimizer when they access the optimizer", then there should be two private variables for each optimizer instead of one. The proposed fix creates a new optimizer if user switches from SGD to LBFGS and the mentioned logic breaks. By the way, `FeedForwardTrainer` is `ML` private, so it should not be used by users. In practice, the creation of new optimizer never happens because the public class `MultilayerPerceptronClassifier` uses LBFGS only.
--- 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