Using the “n” parameter with max_tokens set to 1 will give you a list of probable words to operate on, but I was wondering if there was another way.
Is there a less expensive way to select a specific word from a list of the probable words?
Using the “n” parameter with max_tokens set to 1 will give you a list of probable words to operate on, but I was wondering if there was another way.
Is there a less expensive way to select a specific word from a list of the probable words?
I believe you are looking for logit_bias
Awesome, thanks! I’ll read it.