ChatGPT is an AI language model that was trained with a large amount of text from various sources (for example, ChatGPT is a sister model of InstructGPT, which is trained to follow an instruction quickly and provide a detailed answer). We trained this model using reinforcement learning based on human feedback (RLHF), using the same methods as InstructGPT, but with slight differences in the data collection configuration. ChatGPT is optimized based on a model from the GPT-3.5 series. Here are some important high-level concepts to understand.
The first is that the GPT-3, Generative PreTrained Transformer, is a model developed by OpenAI to perform the task of completing the chat. Therefore, if prompted, it will end the message.
