Thread count has no effect on gpt4all-j

Open fHachenberg opened this issue 2 years ago • 1 comments

Hi there, thanks for creating this amazing project! I noticed that if I use gpt4all-j in local-ai, I always end up with 4 threads. With gpt2 for instance, it works however. I think the issue is either in this wrapper or the gpt4all-j-cpp-code itself.

May 02 '23 04:05 fHachenberg

Good catch! seems indeed a regression while using the new binding code

May 02 '23 07:05 mudler

confirming. tried to start with 8 threads on startup and it starts only with 4 threads.

May 12 '23 16:05 thebennos

Thanks for the feedback - Looks like a regression, fix is in #249

May 13 '23 07:05 mudler