OpenAI doc says that system messages are not honored by gpt-3.5-turbo. But the chatbot lesson here showed the entire menu with intricate nuances being fed into the system message, and it worked fine. So, should we use system messages with gpt-3.5-turbo? If yes, are there some kinds of things it works with better than others? Or is it better to just put everything in user messages?
The course notebooks use gpt-3.5-turbo. The one you’re pointing out is gpt-3.5-turbo-0301. Please see the details here.
1 Like
Ah, got it - thanks! I misunderstood - I thought gpt-3.5-turbo was pointing to the latest version, and I thought that was the -0301 version. This explains it well. Thanks very much!