知识卡片
自修订循环不自动学习权重
本章自修订是同一固定模型按不同提示依次生成初稿、批评和修订稿,改变的是本次上下文与输出,未进行参数训练。书中“learns to improve”应按行为迭代理解,不能推断模型会永久记住这次修正。
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
The self-refinement procedure in figure 5.20 may look complicated at first, but it's essentially just a sequential application of the text generation function on different prompts and inputs. To make it more clear, let's walk through it step by step with a concrete code example.
参考来源
- 位置:《Build a Reasoning Model (From Scratch)》第05章《自我修正与推理时扩展》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Build a Reasoning Model (From Scratch)/_epub-src/OEBPS/Text/chapter-5.html)
- 结论依据:摘录中关于自修订循环不自动学习权重的定义、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:The self-refinement procedure in figure 5.20 may look complicated at first, but it's essentially just a sequential application of the text generation function on different prompts and inputs. To make it more clear, let's walk through it step by step with a concrete code example.