知识卡片

普通读书笔记卡 · 8761.a · 生成模型:GPT-6

指标测试保留实际结果诊断

理解等级:中级

失败除了给出红灯,还应留下具体指标、样本和混淆情况,帮助区分缺数据、某类错误或阈值不适合。只有assert失败而无上下文,会让团队再次花时间手工重建分析。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      In model tests, it’s not enough for tests to pass or fail. We also want useful visual feedback (e.g., actual metrics, confusion matrices) in our test logs. You can run this test in the repo by running the following command: ./batect model-metrics-test The test passes if the model’s recall score is above the specified threshold. For a refresher on what batect is and how to

    参考来源

    - 位置:《Effective Machine Learning Teams (David Tan, Ada Leung, David Colls)》第06章《模型自动化测试》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Effective Machine Learning Teams (David Tan, Ada Leung, David Colls)/_epub-src/OEBPS/ch06.html) - 结论依据:摘录中关于指标测试保留实际结果诊断的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:In model tests, it’s not enough for tests to pass or fail. We also want useful visual feedback (e.g., actual metrics, confusion matrices) in our test logs. You can run this test in the repo by running the following command: ./batect model-metrics-test The test passes if the model’s recall score is above the specified threshold. For a refresher on what batect is and how to