Your brother project: GPTSwarm (https://github.com/metauto-ai/GPTSwarm)

[mczhuge]

Dear Authors,

Thank you for sharing your work. Appreciate your contribution to the field. I’d like to bring your attention to a project that seems closely related to your research, which was released about 8 months prior to yours.

Project: https://gptswarm.org/
Code: https://github.com/metauto-ai/GPTSwarm
Paper: https://arxiv.org/abs/2402.16823 (ICML 2024, Oral Presentation)

Additionally, I would personally suggest considering GPTSwarm's approach of (1) constructing agents as graphs. From there, you could focus on (2) improving the swarm intelligence by improving these graphs, either through reinforcement learning or automated prompt improvement, as outlined in our paper.

In my experience, I believe this might be a visible and promising solution for LLM-based swarm intelligence (or MAS).

Best regards,
Mingchen

---

References:

🐝 GPTSwarm is a graph-based framework for LLM-based agents, providing two high-level features:

• It lets you build LLM-based agents from graphs.
• It enables the customized and automatic self-organization of agent swarms with self-improvement capabilities.

About GPTSwarm

At a granular level, GPTSwarm is a library that includes the following components:

Module	Description
swarm.environment	Domain-specific operations, agents, tools, and tasks
swarm.graph	Graph-related functions for creating and executing agent graphs and swarm composite graphs
swarm.llm	Interface for selecting LLM backends and calculating their operational costs
swarm.memory	Index-based memory
swarm.optimizer	Optimization algorithms designed to enhance agent performance and overall swarm efficiency