Hi there, I was testing ChatGPT API. The goal is to create an assistant for a fitness app. I used the system content to instruct it to behave as an assistant and to follow precise rules about topics (in my case just diets). When I start chatting, it follows rules… but if I says something like: “…

Similar issue in my post.Ai leaves the role too easily and ignores the “system” content. https://community.openai.com/t/gpt-3-5-turbo-vs-text-davinci-003-chatbot/82806 I love how “text-davinci-003” is working on my ChatBot examples. Gpt-3.5-turbo is a lot chaper, and I’d prefer to use it , but te…

I have read (but haven’t tested) that you can try putting the system message at the end of the array and reduce the temperature. Another variant is repeat the system message periodically and reduce the temperature.

I really hope we can make Gpt-3.5-turbo do these types of roleplay scenarios better. There is so much potential and I’d love to switch over from davinci-003. It’s great to have this system role with a lower weight, but ideally I would want to set an optional weight for each entry to the system role……

Read here for a few more details on the underlying ChatML and what to expect moving forward: “In the near future, we will train our models to be much more steerable via the system message. But to date, we have trained only on a few system messages, so the models pay much more attention to user examp…

Thanks for the information. Don’t get me wrong, GPT-3.5 Turbo is really amazing for general-purpose chat tasks, but for this specific task of impersonating a person, even using the user examples are not yet useful, and the AI leaves the role very easily anyway. If we ask these four questions and th…

FWIW, have been working with the system role testing it today, and here are some mixed results. When I lowered the temp in the last few examples, it was better: You are an expert astrophysicist and a very good Ruby programmer. [Screenshot 2023-03-04 at 9.32.17 PM] You are an expert biologist and…

Here are three more system role examples with the temperature set to 0.1 [Screenshot 2023-03-04 at 10.08.54 PM] [Screenshot 2023-03-04 at 10.08.23 PM] [Screenshot 2023-03-04 at 10.07.54 PM] HTH :slight_smile:

(sorry previous post was deleted) I’m not sure if you found the solution, but I’m addressing the same issue here: [image] Gpt-3.5-turbo vs text-davinci-003 *ChatBot* API Yeah, sorry, I have only used the Playground 1 time about two months ago. I write all my own “playg…

We can confirm that this solution works… My assumption before reading this comment was that ChatGPT’s token limit cuts off the first messages when there is too much data in system message and following user/assistent messages. This somewhat confirms it, although it does not work even when the limit…

System message: how to force ChatGPT API to follow it

API

wfhbrian March 3, 2023, 6:12pm 3

It’s been mentioned that the system role is still being improved. So it is expected to perform better in the future.

In the meantime, you should try placing those instructions in the user role.

Topic		Replies	Views
Inconsistent Outputs Between GPT-3.5 API and ChatGPT Even After Adjusting Temperature Prompting gpt-35-turbo , chatgpt , api	4	2906	December 18, 2023
What is the difference between putting the AI personality in system content and in user content? Prompting gpt-4 , gpt-35-turbo , chatml , chatml-system , chatml-user	8	18184	April 15, 2024
What should be included in the System part of the Prompt? API api	14	43879	March 30, 2024
OpenAI's Chat API is leaking system message to user API	7	1969	December 15, 2023
Is it possible to have a setting to increase the weight of the system prompt? Prompting gpt-4 , chatgpt	12	1887	May 18, 2024

System message: how to force ChatGPT API to follow it

Related topics