🚀 Introducing LoongFlow — A Cognitive Evolutionary AI Framework (Open Source)

LoongFlow is an open-source framework designed for cognitive evolutionary agents, combining reasoning and evolutionary search. It aims to make AI systems evolve more intelligently than with random processes. The core of LoongFlow is a "Plan, Execute, Summarize" (PES) loop that uses LLMs. The planning phase uses LLMs to analyze past generations and strategize the next steps. Execution involves generating and testing new solutions based on those plans. Summarization reflects on the outcomes to guide future planning, improving the search process. This framework is suitable for algorithm discovery, ML optimization, and autonomous agent development. Developers can use it for smart search workflows and for developing a hybrid memory system. You can contribute by testing the framework, creating code, or helping with documentation. Feedback and new application ideas are welcomed, encouraging community collaboration. You are encouraged to explore the GitHub repository to contribute and shape the project.