How to restrict a model to provide short responses | OpenRouter | Page 1

proud musk May 10, 2024, 4:53 PM

#

I tried prompting the model to keep responses within certain word limit, that doesnt work.

When using max tokens I see responses are getting truncated at the end.

I am using a 13b model.

Any suggestions on how to make this work?

velvet forum May 10, 2024, 6:54 PM

#

I'd estimate the amount of word you want to use and the max token count

#

so 100 words is ~400 tokens

proud musk May 11, 2024, 1:10 PM

#

@velvet forum I have now put 25 words limit in prompt for mythomax but still getting longer responses which gets truncated after max token limit of 100.

velvet forum May 11, 2024, 3:18 PM

#

I'd recommend a smarter model IMO. Mythomax is good at roleplay but not necessary with following instruction. Most of these LLM doens't know how to count afaik

proud musk May 12, 2024, 3:32 PM

#

Could you suggest which is the best and most used model for roleplay?

#How to restrict a model to provide short responses