from v0.7.3 to v0.7.4-pre, HUGE VRAM increase #14900
DefinitlyEvil
announced in
General
Replies: 0 comments
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.
-
My model was able to take 76000 tokens, but now only 68000.
Any similar findings?
Beta Was this translation helpful? Give feedback.
All reactions