Indexing Pipeline tends to vary greatly in runtime on same machine #6294
demongolem-biz2
started this conversation in
General
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
I am working on a shared GPU machine which generally speaking I am the only one who is running processes on it. Each time I kick it off, the indexing pipeline seems to take a different length of time, perhaps 10s for one run and 80s for another run of the same documents.
I have this code
Are there reasons proper to the Haystack code base which can explain such a great swing?
Beta Was this translation helpful? Give feedback.
All reactions