1

I am writing code where the ChatGPT API in Python responds to some questions. However, I noticed that when I asked these same questions in the ChatGPT website, I got answers which were more accurate on average and these answers also were longer and more detailed. Is there any way to fix this?

This is the current code I'm using.

completion = openai.ChatCompletion.create(
                 model="gpt-3.5-turbo",
                 messages=[{"role": "user",
                            "content": questions[361]}
                          ],
                )
Mark
  • 7,785
  • 2
  • 14
  • 34
  • The website has an implicit system prompt before the user input providing a baseline description of what the AI is supposed to be doing. You should be providing one too. – Charles Duffy Jul 02 '23 at 03:50
  • 2
    (I'm not sure I agree that this is a Python question; you'd get the same behavior using the API from any other language as well). – Charles Duffy Jul 02 '23 at 03:52
  • 1
    I guess 3.5 turbo model is not exactly the model on chat.openai.com. 3.5 turbo is optimized for speed and computation cost. – liginity Jul 02 '23 at 04:02

0 Answers0