知识卡片
预训练与任务适配职责不同
预训练从广泛数据学习可复用表征与生成规律,微调用后续数据调整任务行为。二者是训练阶段和数据目标的分工;少样本提示则在上下文中示范,不自动修改权重,不能统称成同一种学习过程。
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
pre-training and fine-tuning. An initial large-scale pre-training step initialized the Transformer model with general language capabilities. Subsequently, the pre-trained model could be trained on more concrete tasks, like information extraction or sentiment detection, using a process called fine-tuning. We will cover fine-tuning extensively throughout the book.
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:摘录中关于预训练与任务适配职责不同的定义、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:pre-training and fine-tuning. An initial large-scale pre-training step initialized the Transformer model with general language capabilities. Subsequently, the pre-trained model could be trained on more concrete tasks, like information extraction or sentiment detection, using a process called fine-tuning. We will cover fine-tuning extensively throughout the book.