知识卡片
检索器与阅读器解决不同子问题
检索器筛出可能含答案的文本,阅读器再在候选上下文中定位答案跨度。两者可以用不同模型,召回失败会限制阅读器表现;本章RoBERTa-SQuAD2是抽取式模型,不应把它称为生成回答的Transformer。
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
Now, it’s time to set up our indexing models! The QA pipeline includes two main steps: retrieving documents that might be relevant to the query (semantic search) and processing those documents to extract an answer. You will need an EmbeddingRetriever semantic vector index and a generative transformer model.
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:摘录中关于检索器与阅读器解决不同子问题的定义、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:Now, it’s time to set up our indexing models! The QA pipeline includes two main steps: retrieving documents that might be relevant to the query (semantic search) and processing those documents to extract an answer. You will need an EmbeddingRetriever semantic vector index and a generative transformer model.