#Getting consistent results in response length. Rewriting text.
1 messages · Page 1 of 1 (latest)
what are you using?
Tried llama3 and mistral
how many tokens?
Do you mean the prompt with the text?
and after that it start allucinating?
I tried a lot of prompts and it always gives me like 400-600 words back. If you chat with it and ask it to extend the text its almost does it. But the main problem that I want consintent outpust cause I want to automate it with python. So I want to get the results in one prompt.
lets wait for nick he might know better
Well LLMs see tokens not words
I guess you could only try with a better prompt or use another model
Can I set an expedted token count in the response ?
I tried 2 different models with like 10-15 different prompts. No consistency unfortunatelly.
You can set a max output tokens but not minimum output tokens