do the same thing you did with 27b because it does a lot better.
#3
by drmcbride - opened
do the same thing you did with 27b because it does a lot better.
I think bigger models are harder to steer with small finetunes, still this one is pretty usable for storytelling/roleplay as far as I can see. I don't know if heretic has an option to activate all experts on every token like llmcompressor, that would be one avenue to try if it does, or if it can be added.