EvoTool-OL is an open-source project researching self-evolving agents, using an innovative online tool learning framework that enables agents to dynamically discover, learn, and utilize external tools during interaction. This project aims to address the challenge of autonomous learning and adaptation for agents in unknown and complex environments. It developed by Ids-lab Key features include:
- Online tool discovery and learning: Agents autonomously explore and learn available tools through interactions with the environment.
- Self-evolving toolkits: Agents dynamically adjust and optimize their own toolsets based on task requirements and environmental changes.
- Reinforcement learning and tool utilization: Agents use reinforcement learning algorithms to learn how to effectively utilize tools to complete complex tasks.