← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6759 · 生成模型:GPT-6

少样本提示不自动更新权重

理解等级:中级

在提示中加入示例,可改变同一次上下文条件下的预测行为,通常称上下文学习。它与用这些例子反传更新参数不同,不应说把完整训练集放入提示就是完成了模型训练;示例顺序、长度和任务匹配仍需验证。

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

      Many researchers and their corporate backers are hopeful that increased size will unlock human-like capabilities. And these big tech researchers have been rewarded at each step of the way. One hundred billion parameter models, such as Bloom and InstructGPT, revealed the capacity for LLMs to understand and respond appropriately to complex instructions for creative writing tasks, such as composing a love poem from a Klingon

    参考来源

    - 位置:《Natural Language Processing in Action, Second Edition》第10章《真实世界中的大语言模型》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Natural Language Processing in Action, Second Edition/_epub-src/OEBPS/Text/chapter-10.html) - 结论依据:摘录中关于少样本提示不自动更新权重的定义、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:Many researchers and their corporate backers are hopeful that increased size will unlock human-like capabilities. And these big tech researchers have been rewarded at each step of the way. One hundred billion parameter models, such as Bloom and InstructGPT, revealed the capacity for LLMs to understand and respond appropriately to complex instructions for creative writing tasks, such as composing a love poem from a Klingon