arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2401.02777cs.CLcs.AI
收藏

从大语言模型到对话智能体:一种融合大语言模型微调的记忆增强架构

From LLM to Conversational Agent: A Memory Enhanced Architecture with Fine-Tuning of Large Language Models

Beike Inc.(贝壳公司)

Na Liu, Liangyu Chen, Xiaoyu Tian, Wei Zou, Kaijiang Chen, Ming Cui

AI总结:

本研究提出改进ReAct框架的RAISE架构,通过模拟人类记忆的双组件记忆系统和多阶段构建流程增强对话智能体,房产销售场景评估显示其优于传统智能体,为开发更智能的对话智能体提供支撑。

摘要:

本文提出RAISE(即通过暂存区与示例进行推理与行动),这是一种增强GPT-4等大语言模型(LLM)融入对话智能体的先进架构。RAISE是对ReAct框架的改进,整合了模拟人类短期和长期记忆的双组件记忆系统,以维持对话的上下文和连续性。它包含一套完整的智能体构建流程,涵盖对话选择、场景提取、思维链(CoT)补全、场景增强等阶段,最终进入LLM训练阶段。该方法似乎能提升智能体在复杂多轮对话中的可控性与适应性。我们在房产销售场景下的初步评估显示,RAISE相较传统智能体具备一定优势,表明其拥有更广泛的应用潜力。本研究为AI领域提供了一个稳健框架,可用于开发更具上下文感知能力、功能更通用的对话智能体。

英文摘要:

This paper introduces RAISE (Reasoning and Acting through Scratchpad and Examples), an advanced architecture enhancing the integration of Large Language Models (LLMs) like GPT-4 into conversational agents. RAISE, an enhancement of the ReAct framework, incorporates a dual-component memory system, mirroring human short-term and long-term memory, to maintain context and continuity in conversations. It entails a comprehensive agent construction scenario, including phases like Conversation Selection, Scene Extraction, CoT Completion, and Scene Augmentation, leading to the LLMs Training phase. This approach appears to enhance agent controllability and adaptability in complex, multi-turn dialogues. Our preliminary evaluations in a real estate sales context suggest that RAISE has some advantages over traditional agents, indicating its potential for broader applications. This work contributes to the AI field by providing a robust framework for developing more context-aware and versatile conversational agents.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑