Jérémy Scheurer comments on Imitation Learning from Language Feedback