Is it normal that Gemma models do not work with kv-cache?
#19
by
SolidSnacke
- opened
I have run several Gemma models on KoboldCpp and on Oobabooga. None of them worked with kv-cache and gave an error. Is this how it should be?