原书出处
Reliable Machine Learning
10 Continuous ML
A stable baseline is a model C that is not influenced by either A or B, and is allowed to serve a certain amount of traffic so that we can use those results as a comparison. The basic idea is to then look at (A-C) against (B-C) rather than A against B directly, with the idea that this will allow us to see any changes more clearly.
原书定位详情
来源:a19c7933-4eda-4176-aebd-30fe7e1ad292
正文位置:extracted/Reliable Machine Learning/_epub-src/OEBPS/ch10.xhtml
EPUB SHA-256:1201425671b4d9d06d56459bced6f75ec74a7aac380ea24f5b573c80fee7be65