GPT is a family of [[instruction-tuned LLM|instruction-tuned LLMs]] developed by [[OpenAI]] that interacts in a conversational way. It is trained to follow an instruction in a prompt and provide a detailed response. The dialogue format makes it possible for GPT systems to answer follow-up questions, admit their mistakes, challenge incorrect premises, and reject inappropriate requests. It is a sibling model to InstructGPT and is trained using [[reinforcement learning from human feedback]]. ChatGPT is fine-tuned from a model in the [[GPT-3.5]] series, which finished training in early 2022. It has been introduced to get users’ feedback and learn about its strengths and weaknesses. GPT typically refers to the "Generative Pre-trained Transformer" models developed by OpenAI. These models can be used through an interface called ChatGPT which is fine-tuned for conversational tasks. While GPT models can be used for a broad range of applications, ChatGPT is optimized to engage in dialogue with users in a coherent and contextually appropriate manner. The ChatGPT interface is available at https://chat.openai.com or as an independent app. < [[Hands-on LLMs]]/[[3 Instruction-Tuned LLM Systems]] > [[GPT-3.5]]