Currently we manually set the temperature and keep it the whole chat. Wouldnt it make more sense to let the model decide a temperature itself depending on the topic?
You must log in or register to comment.
“So, in contexts where the top token is 6%, a Min P of 0.1 will only consider tokens that are at least 0.6% probable. But if the top token is 95%, it will only consider tokens at least 9.5% probable.”
Isn’t there a big post on automatic temperature scaling?
Automatic temperature does work. It goes great with min_P. I like it more than mirostat and hope everything implements it.