Memory in GPT refers to the capability of models like šChatGPT to retain and utilize information across conversations, enhancing personalization and contextual relevance. This system comprises two primary components: saved memories and chat history. Saved memories are explicit detailsāsuch as user preferences, names, or goalsāthat the model remembers for future interactions, either by user request or automatically. Chat history allows the model to reference previous conversations even if information has not been saved as a memory, improving continuity and context. Users maintain control over these features, with options to review, update, or delete memories and to disable memory or chat history reference as desired. Settings enable the management of all saved information, and temporary chats can be initiated for conversations without memory reference.
