-
Notifications
You must be signed in to change notification settings - Fork 177
Description
Thanks for the collection of works about LLM agents, I'd like to kindly request that you consider adding our work AgentSquare to your repository.
AgentSquare is a novel LLM agent search framework that utilizes module evolution and recombination to automatically optimize agent designs across a modular design space consisting of Planning, Reasoning, Tool Use, and Memory modules. We believe that the proposed modular design space and AgentSquare search framework offer a platform for fully exploiting the potential of prior successful designs and consolidating the collective efforts of the research community of LLM agents.
Here's some information about AgentSquare:
AgentSquare: Automatic LLM Agent Search in Modular Design Space
Website (with demos): https://tsinghua-fib-lab.github.io/AgentSquare_website/
Paper: https://arxiv.org/abs/2410.06153
Code: https://github.com/tsinghua-fib-lab/AgentSquare
TLDR: We propose a modular design space of LLM agents and an automatic agent search framework named AgentSquare, which can adaptively search for better-performing agent designs for given tasks.
Key Features:
Unified Framework: We have provided unified exemplar scripts on six popular agent tasks covering embodied, game, web and tool areas. You can easily run different agent designs with only a line of code!
Extensibility: We have provided detailed guidance on how to develop user's own agent designs and tasks based on AgentSquare, demonstrating the high extensibility of AgentSquare framework.
I believe AgentSquare's novel framework and good usability make it a valuable addition to your repository. Thank you for considering this request.