logoalt Hacker News

Your model already knows the answer: how benchmark answers leak into LLMs

7 points • by fran-mora • today at 4:56 PM • 0 comments • view on HN

Comments