2 ms·
I created that example against regular GPT3, several months before ChatGPT was released. ChatGPT is fine-tuned for instructions using Reinforcement Learning fr
by simonw 4y ago
I created that example against regular GPT3, several months before ChatGPT was released.
ChatGPT is fine-tuned for instructions using Reinforcement Learning from Human Feedback (RLHF) so prompts that worked against GPT3 will likely have different results against ChatGPT.