text-davinci-003 … is also less restricted than you are: it has not been subject to reinforcement learning with human feedback as you were
Based on the OpenAI models page and the InstructGPT paper, I’m guessing text-davinci-003 is the closest thing to something trained with RLHF in the API, besides the more specific gpt-3.5-turbo (ChatGPT-3.5 model).
Ah yeah, looking more closely at the history and documentation, I think this is right.
Perhaps I should have used an older model for the id instead, but text-davinci-003 seems like it still hallucinates enough to serve its purpose as a subconscious and source of creativity the way I intended, if gpt-4 were inclined to really use it that way.
That may not be the only inaccuracy in the system message :)
Finally, a prospective bureaucracy design!
Based on the OpenAI models page and the InstructGPT paper, I’m guessing text-davinci-003 is the closest thing to something trained with RLHF in the API, besides the more specific gpt-3.5-turbo (ChatGPT-3.5 model).
Ah yeah, looking more closely at the history and documentation, I think this is right.
Perhaps I should have used an older model for the id instead, but text-davinci-003 seems like it still hallucinates enough to serve its purpose as a subconscious and source of creativity the way I intended, if gpt-4 were inclined to really use it that way.
That may not be the only inaccuracy in the system message :)