知识卡片
少样本示范不更新模型权重
few-shot prompting在当前上下文提供示例,借示例说明任务或输出形式。它不等同于梯度微调;效果还受示例代表性、顺序和相似性影响,换请求后若不再提供示例也不能假设效果被永久学入参数。
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
- · 大语言模型入门
原书出处 · 1 处
- · 查看原书依据
how to solve it. This is because the task is simple and clearly defined. In many cases, the tasks might be not so easy to describe in natural language. We can then add some examples in our prompt consisting of either outputs or input-output pairs. While this is called few-shot learning colloquially, the language model is not updated in any way through this prompting technique.
参考来源
- 位置:《Designing Large Language Model Applications》第01章《大语言模型入门》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch01.html)
- 结论依据:摘录中关于少样本示范不更新模型权重的定义、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:how to solve it. This is because the task is simple and clearly defined. In many cases, the tasks might be not so easy to describe in natural language. We can then add some examples in our prompt consisting of either outputs or input-output pairs. While this is called few-shot learning colloquially, the language model is not updated in any way through this prompting technique.