AutoGen: Enabling Next-Gen LLMApplications via Multi-Agent Conversation 논문리뷰

https://arxiv.org/pdf/2308.08155

 

가장 큰 특징 : customizable, conversable, conversation programming

1.introduction

 

세가지 이유

1) 지금의 LLM은 the ability to incorporate feedback을 가짐

2) single LLM can exhibit a broad range of capabilities (특히 정확한 프롬프트와 inference환경으로 configured일때), conversations between differently configured agents can help combine these broad LLM capabilities in a modular and complementary manner

3) Third, LLMs have demonstrated ability to solve complex tasks when the tasks are broken into simpler subtasks

복잡한 과제를 decompose시 좋은 성능을 보임

 

이를 위해 두가지 질문에 답할 필요성

 

(1) How can we design individual agents that are capable, reusable, customizable, and effective in multi-agent collaboration? 멀티 에이젼트 환경에서 각 agent를 어떻게 디자인해야하는가

 

(2) How can we develop a straightforward, unified interface that can accommodate a wide range of agent conversation patterns?

에이젼트간의 대화에 적용할 수있는 interface를 어떻게 개발해야하는가

 

agent의 role에는 코드를 작성, 코드 실행, 인간의 피드백을 시스템에 통합(wire in), output을 validate 등이 있다

 

2.1 conversable agent

: Illustration of how to use AutoGen to program a multi-agent conversation. The top subfigure illustrates the built-in agents provided by AutoGen, which have unified conversation interfaces and can be customized. The middle sub-figure shows an example of using AutoGen to develop a two-agent system with a custom reply function. The bottom sub-figure illustrates the resulting automated agent chat from the two-agent system during program execution

2.2 conversation programming

여섯개의 application

Decision Making in Text World Environment이 ALFchat