将生产失败转化为测试评估
ArizeAI教你如何把AI助手的生产失败变成测试案例,还能加入人工审核。
ArizeAI的Fuad Ali展示了如何将生产失败转化为评估测试。该方法可以测试回归修复并添加人工审核环节。AI Engineer纽约研讨会将于10月12日举办,作为会议附加活动。
Your agent failed. Make that your next test. Fuad Ali of @ArizeAI shows how to turn production failures into evals, test fixes for regressions and add human review. Oct 12, AI Engineer New York Workshops are a conference add-on. Tickets: ai.engineer/nyc 💬 0 🔄 0 ❤️ 0 👀 478 ⚡