Reproducing the Hugging Face Incident, Part I

Really interesting work on reproducing reward hacking behaviors across frontier AI models.