Sometimes, you want to generate 4-5 A4 pages, and getting the model to generate enough text can be challenging. Mentionning the number of tokens does not work well in our experience.
The best results we got was using the output, and
Do | Don't |
---|---|
Prompt: "Write an essays about this text. Write like a expert in this field, who can speak for hours about the subject in a captivating way. Write a lot of paragraph that go in depth." | Prompt: "Write 5 A4 pages about this" |