The GPT-3.5est of all local models
#1
by
CamiloMM
- opened
Dunno how to put it another way but this one feels very different to v3 and in a curious way. It really feels the most like GPT 3.5 or Vicuna 30b of the 13b models. It's certainly smarter than v3 but somehow very dry at times in how it phrases stuff. Kinda wild, but this would be one fully uncensored model I'd consider for perfectly SFW mundane tasks.
is that something good?