← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8758.b · 生成模型:GPT-6

预测取值集合断言要检查方向

理解等级:高级

要验证所有预测合法,应检查set(predictions)≤valid_values;反过来只要求每个合法类别都曾被预测,仍可能含非法值。覆盖所有类别是另一项需求,且模型在有限样本只预测一类不一定是管线错误。

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

      assert valid_predictions.issubset(predictions) In your terminal, you can run the training smoke test by executing the following command:3 ./batect --output=all smoke-test-model-training In one of our previous projects, we had an ML training pipeline that took three hours. The full-scale training ran on the cloud (we were using Metaflow) and ML practitioners sometimes waited two hours only to see an error in the final step of the

    参考来源

    - 位置:《Effective Machine Learning Teams (David Tan, Ada Leung, David Colls)》第05章《软件自动化测试》(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/ch05.html) - 结论依据:摘录中关于预测取值集合断言要检查方向的定义、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:assert valid_predictions.issubset(predictions) In your terminal, you can run the training smoke test by executing the following command:3 ./batect --output=all smoke-test-model-training In one of our previous projects, we had an ML training pipeline that took three hours. The full-scale training ran on the cloud (we were using Metaflow) and ML practitioners sometimes waited two hours only to see an error in the final step of the