知识卡片
基础设施与流水线:平台、特征生成与「更好即不同」
模型只是更大系统一环,由流水线支撑,也继承传统系统的复杂性。系统多建在 TensorFlow、PyTorch 或云 ML 平台上,虽能快速用上 GPU/TPU,却引入不可控依赖与大量「胶水代码」。作者强调特征生成 bug 是最常见错误源:同一特征在训练与 serving 用不同代码路径计算,预测会悄悄跑偏且缺真值校验。修 bug 也可能闯祸:把误报华氏度的传感器改成摄氏度,模型会把 32 度当成酷寒,故须约定变更告知并监控特征分布。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
- · 基础设施与流水线
原书出处 · 1 处
- Basic Introduction to Models · 查看原书依据
Upgrades and Fixes One particularly subtle area where upstream dependencies can cause issues in our models arises when the upstream system undergoes an upgrade or bug fix. It may seem strange to say that fixing bugs can cause problems. The principle to remember is better is not better, better is different—and different may be bad. This is because any change to the distribution of feature values that our model expects to see as
参考来源
- 位置:《Reliable Machine Learning》第03章《Basic Introduction to Models》(源文件:extracted/Reliable Machine Learning/_epub-src/OEBPS/ch03.xhtml)
- 结论依据:原文明确阐述该主题,并给出定义、实践或反例。
- 原始内容:Upgrades and Fixes One particularly subtle area where upstream dependencies can cause issues in our models arises when the upstream system undergoes an upgrade or bug fix. It may seem strange to say that fixing bugs can cause problems. The principle to remember is better is not better, better is different—and different may be bad. This is because any change to the distribution of feature values that our model expects to see as