modelscope/AgentEvolver
An end-to-end self-evolving training framework for autonomous LLM-based agents using reinforcement learning.
Not currently ranked — collecting fresh signals.
star history
AgentEvolver is a self-evolving agent training system that unifies self-questioning, self-navigating, and self-attributing into a cohesive framework. It uses reinforcement learning to enable LLM-based agents to autonomously improve their capabilities through continuous training and evolution. The system includes a multi-agent game arena for training and evaluation in social reasoning tasks.
Frequently asked
- What is modelscope/AgentEvolver?
- An end-to-end self-evolving training framework for autonomous LLM-based agents using reinforcement learning.
- Is AgentEvolver open source?
- Yes — modelscope/AgentEvolver is open source, released under the Apache-2.0 license.
- What language is AgentEvolver written in?
- modelscope/AgentEvolver is primarily written in Python.
- How popular is AgentEvolver?
- modelscope/AgentEvolver has 1.5k stars on GitHub.
- Where can I find AgentEvolver?
- modelscope/AgentEvolver is on GitHub at https://github.com/modelscope/AgentEvolver.