知识卡片

普通读书笔记卡 · 9260 · 生成模型:GPT-6

先缩小模型候选再统一比较

理解等级:中级

先根据标签、输出类型和输入结构筛选可行模型,再考虑数据量、解释需求和计算约束,保留若干候选。没有必要一开始就宣布唯一正确架构,候选应在相同划分与评价目标下比较。

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

      Let us begin by discussing how to choose a model. Selecting the correct model to deploy is a crucial decision in the machine learning pipeline. As we have seen up till now, a large number of machine learning models exist, and each model is suitable for different types of problems and data. Therefore, we must first identify the type of problem that is aiming to

    参考来源

    - 位置:《Machine Learning Fundamentals: Concepts, Models, and Applications》第09章《训练机器学习模型》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Machine Learning Fundamentals: Concepts, Models, and Applications/_epub-src/OEBPS/xhtml/23_chapter09.xhtml) - 结论依据:摘录中关于先缩小模型候选再统一比较的定义、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:Let us begin by discussing how to choose a model. Selecting the correct model to deploy is a crucial decision in the machine learning pipeline. As we have seen up till now, a large number of machine learning models exist, and each model is suitable for different types of problems and data. Therefore, we must first identify the type of problem that is aiming to