Inventor(s)

NAFollow

Abstract

Current LLMs struggle with real-time knowledge, tracking elapsed time, chronological context, and generating time-sensitive responses, leading to user burden and limited applicability in time-sensitive domains. This disclosures describes techniques for integrating real-time temporal context into large language model (LLM) processing of user queries, addressing their inherent lack of temporal awareness. The techniques described herein automatically obtain current time from a reliable time source as well as relevant temporal context such as conversation history between an LLM-powered chatbot and a user that provides a query. The time information together with the user query is provided to the LLM which can thereby generate a response in a time-aware manner. The techniques can result in more accurate, relevant, and natural responses from LLMs.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.

Share

COinS