Eagerly waiting and hoping that they'll do another part soon and discuss R1. Interesting that liang said multimodality is a focus. I hope they can achieve proper cross modal learning transfer that gpt 4o could not.