model suggestion (sory if its wrong place to ask :( )
#1
by
Skorcht
- opened
is it possible for you to fine tune 8b llama 3? so it can be like fim 11b? that would be a very good model. for some reason fim is very good with prompt injection from my testing. for example i put some instructions for the llm above the actual users prompt to ensure it always follows them and it works like a charm. unfortunately no other llm can do this aside from your fim model. thats why i wanted to ask if you could kinda merge it with llama 8b? idk if its that simple and i understand that it is probably a waste of time. Thank you. Just a suggestion...