知识卡片
基础编码向量不必最适合句检索
BertModel可输出上下文token与池化表示,但并未因此在任意句向量检索上最优。检索相关性需相应池化、训练目标及领域评价;不能由预训练广泛或接口可用就断言vanilla BERT是所有向量库首选。
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
After importing a BERT model, you can display its string representation to get a summary of its structure. This is a good place to start if you are considering designing your own custom bidirectional transformer. But in most cases, you can use BERT directly to create encodings of English text that accurately represent the meaning of most text. A pretrained BERT model is all you
参考来源
- 位置:《Natural Language Processing in Action, Second Edition》第09章《可堆叠的Transformer》(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-9.html)
- 结论依据:摘录中关于基础编码向量不必最适合句检索的定义、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:After importing a BERT model, you can display its string representation to get a summary of its structure. This is a good place to start if you are considering designing your own custom bidirectional transformer. But in most cases, you can use BERT directly to create encodings of English text that accurately represent the meaning of most text. A pretrained BERT model is all you