Buck comments on Redwood Research’s current project

[Buck]

I’d call it our language model adversarial training project, maybe? Your proposal seems fine too