Introducing the ChatGPT Memory Project

ChatGPT Memory is a project that addresses the context length limitations of large language models (LLMs) used in AI applications such as OpenAI's ChatGPT and GPT-4. The current default behavior of these LLMs only retrieves the last k messages relevant to the current message from the entire history, which can lead to inaccurate responses when users change topics mid-conversation. To tackle this problem, ChatGPT Memory uses Redis as a vector database to store an embedded conversation history of past user-bot interactions and employs vector search inside the embedding space to "intelligently" look up historical interactions related to the current user message. This approach is more adaptive than the current default behavior because it only retrieves the previous k messages relevant to the current message from the entire history, allowing for better personalized responses.