原书出处
Designing Large Language Model Applications
Organize your prompt to place the cacheable components at the beginning of the prompt, i.e., the prompt prefix. Ultimately, caching can be very valuable in reducing inference time, especially in settings where instructions are repeated for a large number of calls, or the context window contains data like API documentation or RAG output that needs to persist across multiple calls. Next, we’ll explore the early
原书定位详情
来源:a19c7933-4eda-4176-aebd-30fe7e1ad292
正文位置:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch09.html
EPUB SHA-256:1b4d2347fec2f949fcddfbaf4fad597a1ae20887a096cfae64c59a619b4781bf