Madhu Guru 分享了提升 LLM 评估能力的实战经验，强调在完成初步评估系统后，应通过分析生产环境数据来构建失败模式分类法

The best way to get good at evals - Part 3. Let’s talk failure modes taxonomy - that’s the first thing you should build once you have v1 of your evals. Start with your production traces - study the last 500 or 1,000 interactions and look at the failures. Cluster them and name