TimeNet already helped us out a lot accelerating training the next version(s) of OpenTSLM :)