📄 中文摘要
与人工智能助手的持续对话中，存在一个问题：语言模型的上下文窗口有限，经过一定数量的交流后，早期的对话内容会逐渐消失。为了应对这一挑战，开始定期请求助手对之前的对话进行摘要，以便在长期项目中保持信息的连贯性。这种方法帮助用户更好地管理与AI的互动，确保重要信息不会被遗忘，从而提高了合作的效率和效果。
叙事记忆还是摘要用于大型语言模型？
出处: ¿Memoria narrativa o resumen para un LLM?
发布: 2026年2月16日
📄 English Summary
¿Memoria narrativa o resumen para un LLM?
Engaging in a continuous dialogue with an AI assistant revealed a significant limitation: language models have a restricted context window, causing earlier parts of the conversation to fade after a certain number of exchanges. To address this issue, the user began to regularly request summaries of previous discussions. This approach allowed for better management of interactions with the AI, ensuring that key information remained accessible and enhancing the overall efficiency and effectiveness of the collaboration on long-term projects.