3 ms·
Well, that's something that I want to understand better. I know ChatGPT--inspired by InstructGPT--used RLHF to improve their models. Since OpenAI's model is not
by akhayam 3y ago
Well, that's something that I want to understand better. I know ChatGPT--inspired by InstructGPT--used RLHF to improve their models. Since OpenAI's model is not available, how does LangChain make sure it gets similar performance as ChatGPT without using RLHF underneath?
The only other way I can think of is for them to take a core dependency on OpenAI's API which seems like a bad strategy for a promising open-source project like LangChain.
How do you see this evolving?