知识卡片
提示格式也是实验变量
提示模板既影响输出格式,也可能影响模型实际作答。一个样本中“短答案错、长答案对”不足以证明模板效应;应在固定样本、预算和判分协议下做受控比较,并区分格式合规率与答案正确率。
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
- · 评估推理模型
原书出处 · 1 处
- · 查看原书依据
whether a prompt template is well-suited for a given model and task ideally needs to be determined on a larger set of examples before we can draw any conclusions, for instance, the MATH-500 dataset we will evaluate the model on later in this section.
参考来源
- 位置:《Build a Reasoning Model (From Scratch)》第03章《评估推理模型》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Build a Reasoning Model (From Scratch)/_epub-src/OEBPS/Text/chapter-3.html)
- 结论依据:摘录中关于提示格式也是实验变量的定义、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:whether a prompt template is well-suited for a given model and task ideally needs to be determined on a larger set of examples before we can draw any conclusions, for instance, the MATH-500 dataset we will evaluate the model on later in this section.