Viliam comments on Are LLMs being trained using LessWrong text?