mll-lab-nu/RAGEN
A reinforcement learning framework for training LLM-based reasoning agents in interactive, stochastic environments.
Not currently ranked — collecting fresh signals.
star history
RAGEN provides a flexible RL framework focused on training reasoning agents powered by large language models. It includes diagnostic tools to understand agent failure modes and identify issues like reasoning collapse during agent RL training. The framework offers lightweight interventions to stabilize training and improve agent performance in dynamic environments.
Frequently asked
- What is mll-lab-nu/RAGEN?
- A reinforcement learning framework for training LLM-based reasoning agents in interactive, stochastic environments.
- Is RAGEN open source?
- Yes — mll-lab-nu/RAGEN is open source, released under the MIT license.
- What language is RAGEN written in?
- mll-lab-nu/RAGEN is primarily written in Python.
- How popular is RAGEN?
- mll-lab-nu/RAGEN has 2.7k stars on GitHub.
- Where can I find RAGEN?
- mll-lab-nu/RAGEN is on GitHub at https://github.com/mll-lab-nu/RAGEN.