Agent-R1 uses the same environment setup as verl.
Follow the official verl installation guide, but use a recent source checkout of verl rather than the old release used by earlier Agent-R1 versions. Agent-R1 relies on the newer AgentFlow / async rollout / reward-loop APIs and on verl.trainer.config being available as package data.
If you want a broader overview of the base training workflow, the verl quickstart is also useful.
Once the verl environment is working, Agent-R1 should run in the same environment. In practice, that means you can:
- prepare a Python environment with a compatible recent source installation of
verl - clone this repository
- run Agent-R1 commands directly from the repository root
You do not need to install Agent-R1 as a separate package.
The documentation in this repository intentionally does not duplicate a separate environment guide, so that the infrastructure setup stays aligned with verl.