more papers announcing that language models are doing mediocre things, proving our own fucking https://www.reddit.com/user/casualcreak