Incredible model
#2
by
PaleoBob
- opened
This model is outstanding. It is very coherent and sticks to prompts really well.
I'm using the Q4 K M version and it runs quite well on 24GB of VRAM.
The one thing I am noticing is that it is often reluctant to engage in "questionable" actions. It's not a hard censorship, as it's easy enough to convince it.
Might need a line or two in the system prompt to make it a bit more... "eager" by default.
Great model. using the q8 version. No issues with questionable actions other than it might be a little two eager compared to some other models using the same system prompt. Using ChatML preset in LMS.