知识卡片
模型调试需区分目标和实现错误
模型效果不理想可来自数据、目标、优化、实现或容量等问题。超参数搜索是经验比较的一部分,不能承担修复全部问题;先保存数据划分、配置与实际结果,才能解释变化来自哪里。
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
- · 查看原书依据
A model starts the training process with random parameter values and adjusts them throughout, whereas hyperparameters are outside of the model and are set by the analyst before the model training. The values of hyperparameters are expected to improve the model accuracy but it is not always the case. They may have adverse effects on the model. The hyperparameter tuning is empirical in nature and
参考来源
- 位置:《Machine Learning Fundamentals: Concepts, Models, and Applications》第03章《机器学习流程与数据管理》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Machine Learning Fundamentals: Concepts, Models, and Applications/_epub-src/OEBPS/xhtml/14_chapter03.xhtml)
- 结论依据:摘录中关于模型调试需区分目标和实现错误的定义、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:A model starts the training process with random parameter values and adjusts them throughout, whereas hyperparameters are outside of the model and are set by the analyst before the model training. The values of hyperparameters are expected to improve the model accuracy but it is not always the case. They may have adverse effects on the model. The hyperparameter tuning is empirical in nature and