Integrating prior conversation content (plus best general settings)
Hi all,
I have just began testing this model. It's very interesting! I did notice though, that it seems to start each prompt as a new conversation. For instance, I asked a certain prompt about creativity, and it suggested I try some "stream-of-consciousness" writing. I then asked the best way to do that, and it gave a response as if we had never discussed that topic before at all.
Is there any way I can get the model to better integrate the prior conversations into its next response?
Also, what are the best and recommended settings for temperature, min_k, etc.?
Thanks again for publishing such an interesting model!
Edit: Hmm.. actually, I was a little mistaken. It DID reference the prior conversation in the example above. However, I have had instances where it seems to deviate from our conversation, so any advice on maximizing context and prior conversations is appreciated. Thank you!