Replies: 1 comment
|
Oh actually, what you saw is expected. you only sent one summary request, and definitely the cache hit was zero. There is no cache before your first question. If you ask a follow-up question, the cache hit will go up for sure. Feel free to try it. interesting. well it is by design: avoid carrying full text (cache) every time when users asked a follow up question that doesn't require full text. My initial motivation is to save user's tokens, and users can choose to send full text or do partial retrieval. But I found the tokens used by cache hit is cheaper? thanks for letting me know it. I will try to find a good way to take advantage of it. I may update my retrieval strategy in the next release. |
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
您好,我在部署完deepseek的API后,用了一次“总结全文”,发现命中缓存很低,请问是什么原因造成的呢?有没有什么好的解决办法?

All reactions