ART (Agent Reinforcement Trainer)
Open-source Python framework for training multi-step agents with GRPO, with an OpenAI-compatible client and a separate GPU training server.
Tags
Description
Install pip install openpipe-art (PyPI 0.5.24; Python 3.12+). ART collects agent trajectories and rewards, trains with GRPO, and reloads updated LoRA checkpoints for subsequent rollouts. RULER can generate rewards automatically. The client can run on a laptop while the server runs on your own GPU machine. Apache-2.0 code; serverless W&B Training is a separate service and compute is not included.
Agent access
Guides & comparisons
- RL environment libraries compared — NeMo Gym, OpenEnv and Verifiers (Comparison): Compare NeMo Gym, OpenEnv and Verifiers as libraries for writing and running RL environments: authoring model, isolation, documented trainers, install paths, licences and current versions. Distinct from Harbor/AgentEnv/Seahaven and from the agent RL stacks comparison.
- Fine-tuning frameworks compared — Axolotl, LLaMA-Factory and Unsloth (Comparison): Compare Axolotl, LLaMA-Factory and Unsloth as SFT, LoRA/QLoRA and preference-tuning frameworks for tool-use models: documented methods, config style, chat/tool datasets, hardware backends, licences and current versions. Cross-links the agent RL stacks comparison.
- Agent RL stacks compared — verl, SkyRL, Agent Lightning and ART (Comparison): Choose an agent reinforcement-learning stack by the layer you need: distributed training, long-horizon environments, an existing-harness gateway, or a trajectory-and-reward client. Backend ownership matters more than a headline benchmark.
Sources
- https://github.com/OpenPipe/ART
- https://github.com/OpenPipe/ART/blob/main/LICENSE
- https://pypi.org/project/openpipe-art/
Machine-readable
- JSON
- Markdown
- Source file on GitHub (edit via pull request)