知识卡片
yarnit.ai 示例系统与全章结论
作者用虚构站点 yarnit.ai 的纱线点击预测模型走查生产系统:输入含商品文本特征、压成 32×32 的原始图、历史点击与搜索词;标签是「点击/未点击」,设一小时窗口,靠批处理过滤刷点击等欺诈。所谓「持续更新」实则滞后约 12 小时。面对几百 QPS 须在两三成秒返回,故用大量云上副本并监控新鲜度、预测稳定性、特征分布,还预演了无限延迟、全零、全坏等最坏场景与回退手段。全章收束:模型行为由数据而非形式化规格决定,数据一变行为即变。
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
原书出处 · 1 处
- Basic Introduction to Models · 查看原书依据
Here are a few things that we hope you take away: Model behavior is determined by data, rather than by a formal (program) specification. If our data changes, our model behavior will change. Features and training labels are critical inputs to our models.
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:Here are a few things that we hope you take away: Model behavior is determined by data, rather than by a formal (program) specification. If our data changes, our model behavior will change. Features and training labels are critical inputs to our models.