Settings that worked for me

#1
by luthis - opened

I was getting short single sentences, until I changed to these settings and reloaded the model:
Model Loader: GPTQ-for-LLaMa
wbits 4
groupsize 128
model_type gptq

Sign up or log in to comment