Stop Scoring RAG on Accuracy Alone
中文摘要
评估RAG系统不能仅看准确率，排行榜表现出色并不代表实际体验良好，应重点关注用户反馈。
English Summary
Stop evaluating RAG systems solely on accuracy; prioritize real-world user support tickets over leaderboard rankings to reflect actual performance.
原文节选
Your leaderboard looks fine. Your support tickets do not. Continue reading on Medium »