Hey everyone! 👋
Just wanted to share an open-source project I've been working on called LoongFlow.
It’s a framework that helps Agents "self-evolve" using Evolutionary Algorithms. Basically, instead of us manually fixing prompts/code when an Agent fails, LoongFlow lets the Agent mutate and optimize its own logic over time.
We actually released a paper on arXiv about it recently, but I'm more interested in seeing if this is useful for builders here.
Has anyone tried combining EA with Agents before? Would love some feedback on our approach.
Repo: https://github.com/baidu-baige/LoongFlow
Paper: https://arxiv.org/abs/2512.24077
Evolve Agent Development Framework. Contribute to baidu-baige/LoongFlow development by creating an account on GitHub.
The transition from static Large Language Models (LLMs) to self-improving agents is hindered by the lack of structured reasoning in traditional evolutionary approaches. Existing methods often struggle with premature convergence and inefficient exploration in high-dimensional code spaces. To address these challenges, we introduce LoongFlow, a sel...