For many use cases like summarization or information extraction, you can get deterministic and mostly non-creative results by adjusting the parameters (temperat

For many use cases like summarization or information extraction, you can get deterministic and mostly non-creative results by adjusting the parameters (temperature, top-p, etc.). This is only possible via the API, though. And it work's most reliably when providing the whole input which should be worked on ("open book" as another commenter called it). I run a task like this for Hacker Jobs [1] and am quite happy with the results so far (there is also an article detailing how it works [2]). If you ask for facts that you hope are somehow remembered by the model itself, it is a different story. [1] https://www.hacker-jobs.com https://www.hacker-jobs.com [2] https://marcotm.com/articles/information-extraction-with-large-language-models-parsing-unstructured-data-with-gpt/ https://marcotm.com/articles/information-extraction-with-lar...