description/abstract: In my project GrainLearning we trained some RNNs and studied their performance. We are planning to write an article about it and I would like to present in the ML SIG our results. I would like some feedback on our methods and what things we still might be missing for our article.