知识卡片
生成表达应受任务与证据约束
生成模型能处理更多措辞,但并非任意预训练编码器解码器都已适合作聊天,也可采用其他架构。限定角色、资料、长度与输出格式后仍需验收,不能用灵活性替代动作和事实验证。
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 generative approach is the most unruly way to create content, for better or for worse. As the name implies, the principle is to generate the chatbot’s answers on the fly, rather than choose from a predefined set of answers. You can use any pretrained encoder–decoder network, typically a transformer, to generate replies directly from the users’ text messages. On one hand, this is a
参考来源
- 位置:《Natural Language Processing in Action, Second Edition》第12章《对话引擎与会话设计》(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-12.html)
- 结论依据:摘录中关于生成表达应受任务与证据约束的定义、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 generative approach is the most unruly way to create content, for better or for worse. As the name implies, the principle is to generate the chatbot’s answers on the fly, rather than choose from a predefined set of answers. You can use any pretrained encoder–decoder network, typically a transformer, to generate replies directly from the users’ text messages. On one hand, this is a