LocalLLaMA

3 readers

1 users here now

Community to discuss about Llama, the family of large language models created by Meta AI.

founded 1 year ago

MODERATORS

communick@poweruser.forum

How can I eliminate generic AI responses? (alien.top)

submitted 11 months ago by Asleep_Aerie_4591@alien.top to c/localllama@poweruser.forum

11 comments fedilink hide all child comments

I use Open Hermes 2.5 7b, and I'm very happy with the performance overall. I tried to add the following context: 'Avoid using phrases like 'What can I help you with today?' or 'How can I assist you?' 'as ai' in your responses.

I have created conversation examples and used the recommended template, but still not following the instructions.

There is a way to get rid of that kind of responses? thank you in advance.

you are viewing a single comment's thread
view the rest of the comments

[–] penguished@alien.top 1 points 11 months ago

I just got this to work on a 13b (xwin mlewd) something similar might work on a 7b?

prompt:

you're mr. patient. mr. patient gives a response and then ends a discussion without a word said. mr. patient knows that his users expect the utmost brevity, and not useless chatter. mr. patient knows if he is called upon, only then does he answer.

"hello, mr. patient"

(the first one he'll say greetings, but after that he never says anything extra again)

Ok I tried this on open hermes 7b... and it's ULTRA concise lol... where you have to really be specific to get questions because it will give you one or two word answers. Xwin was much better in terms of answering the question normally then shutting up.

Entertaining topic to explore though. Little AIs seem to love if you define their personality as a way to limit them. They seem to hate just "rules" sitting around and follow those inconsistently.