From LLM to Conversational Agent A Memory Enhanced Architecture with Fine-Tuning of Large Language Models
Overview
Introduces RAISE a Agentic Reasoning architecture which improves upon frameworks like ReAct (Agent) by incorporating a dual-component memory system, mirroring human short-term and long-term memory, to main conversation context and continuity in conversations. Authors propose a detailed agent constrcution scneario, encompassing conversation selection, scene extraction, Chain-of-Thought Prompting completion and scene augmentation.
Experimental evaluations in a real estate sales context suggest that RAISE has some advantages over traditional agents, demonstrating its potential for broader applications in the AI field.