problem
#1
by
ClaudioItaly
- opened
I tried the Moistral-11B-v2-Q4_K_S-imat.gguf model. It has problems with LMstudio. He answers after a long time and goes nuts. Furthermore, he almost always writes in Spanish. I think it's too humid for a model! I don't recommend it
Haha, wet model, if it's like very slow, it's because this model is pretty big, and is not fitting in your GPU.
Not sure about the Spanish though. Use ChatML prompt.
I don't like LMStudio, I'd recommend KoboldCpp + SillyTavern, this is for roleplay after all.
These are rest quants and quality isn't a guarantee on the model part. Author will probably make a new version soon.