GLM
GLM copied to clipboard
Does this model support temperature and repetition_penalty?
Without temperature, the generation is always fixed. Whitout repetition_penalty, it seems to fall into a loop of repeating last words.
Using model glm-10b-chinese.