Data & Analysis reinforcement-learninggrporeward-modelingosworldalfworldlivecodebenchdynamic-tool-selection

Open-AgentRL

RL training framework for reasoning, tool use, and interactive agent environments.

FollowAgents review · FARS-2.0
Not yet reviewed
See the full review method →

What does this agent do, and when should you use it?

Open-AgentRL brings together three agentic-RL research implementations: DemyAgent, RLAnything, and AutoTool. DemyAgent covers cold-start SFT and GRPO-TCR-based agentic RL, with evaluation workflows for math, science, and coding reasoning. RLAnything performs closed-loop optimization of policy, reward model, and environment for OSWorld computer control, AlfWorld text games, and coding tasks. AutoTool supplies a two-phase dynamic tool-selection framework that first stabilizes tool-integrated trajectories and then refines multi-step selection with KL-regularized Plackett–Luce ranking. It is a research training repository that requires users to supply models, data, configuration, and, for some scenarios, external execution infrastructure rather than a ready-to-deploy chat service.

For DemyAgent, the repository reads SFT, RL, and evaluation data in .parquet format, fine-tunes Qwen models through recipe/demystify/qwen3_4b_sft.sh or qwen2_7b_sft.sh, then runs GRPO-TCR with grpo_tcr_qwen2_7b.sh or grpo_tcr_qwen3_4b.sh. Code-execution tasks use a SandboxFusion /run_code endpoint, configured in sandbox_fusion_tool_config.yaml and verl/utils/reward_score/livecodebench/code_math.py. RLAnything launches training from scenario-specific YAML files through osworld_rl.py, alfworld_rl.py, and coding_rl.py, with matching *_eval.py evaluation programs. The AutoTool framework covers Phase I SFT-plus-RL trajectory stabilization and Phase II ranking-based tool-selection refinement; the repository says its full training data and toolset are still forthcoming.

  1. A research team reproducing agentic-reasoning experiments from 3K SFT and 30K RL datasets, then evaluating on AIME, GPQA-Diamond, or LiveCodeBench-v6.
  2. A team with a local or cloud SandboxFusion endpoint training Qwen-family models that execute code during reasoning.
  3. A team that has configured Volcengine cloud VMs, security groups, and an OSWorld image for RL training or evaluation of GUI computer-control policies.
  4. A researcher who has downloaded AlfWorld data and needs to train or evaluate RL policies in a text-based interactive game.
  5. A group investigating dynamic selection over a large, evolving tool library and willing to work before AutoTool’s full data and toolset are released.

What are this agent's strengths and limitations?

Pros
  • Covers agentic reasoning, GUI computer control, text-game environments, coding, and dynamic tool selection in one repository rather than a single benchmark recipe.
  • DemyAgent documents an end-to-end path from cold-start SFT through GRPO-TCR, including named datasets, model checkpoints, scripts, and configuration fields.
  • RLAnything explicitly combines policy, reward-model, and environment adaptation in a closed loop, with separate training and evaluation entry points for OSWorld, AlfWorld, and coding.
  • AutoTool specifies a two-phase approach with KL-regularized Plackett–Luce ranking for multi-step dynamic tool selection.
Limitations
  • It is not a plug-and-play inference product: adopters must provide models, `.parquet` data, absolute paths, and scenario configuration.
  • Code-execution training depends on SandboxFusion or a Volcano Engine Cloud FaaS endpoint, with endpoint changes required in both tool configuration and reward-scoring code.
  • The documented OSWorld experiment depends on Volcengine security groups, VM images, and a parallel cloud VM pool; infrastructure costs are not stated.
  • AutoTool provides only an example format for its 200K tool-use trajectories; the full training data and toolset are described as coming soon.

How do you install or deploy this agent?

For DemyAgent:

git clone https://github.com/Gen-Verse/Open-AgentRL.git
conda create -n OpenAgentRL python=3.11
conda activate OpenAgentRL
cd Open-AgentRL
bash scripts/install_vllm_sglang_mcore.sh
pip install -e .[vllm]

For RLAnything:

conda create --name rlanything python=3.10
source activate rlanything
pip install -r requirements_rlanything.txt

The supplied material does not state an API-key requirement. Code-execution RL additionally needs a SandboxFusion deployment or a Volcano Engine Cloud FaaS /run_code endpoint; the documented OSWorld setup uses Volcengine cloud VMs.

How do you use this agent?

Before DemyAgent cold-start SFT, set absolute TRAIN_DATA, EVAL_DATA, MODEL_PATH, and SAVE_PATH values in recipe/demystify/qwen3_4b_sft.sh or qwen2_7b_sft.sh, then run:

bash recipe/demystify/qwen3_4b_sft.sh

After configuring the SandboxFusion endpoint, RL data, model, benchmark paths, and default_local_dir, run:

bash recipe/demystify/grpo_tcr_qwen3_4b.sh

For RLAnything, the first training invocation depends on the scenario; for example:

python osworld_rl.py config=configs/osworld_rl.yaml

For a single-node run, set num_node to 1 in the corresponding YAML configuration.

How does this agent compare with similar options?

The repository states that its implementation builds on VeRL and ReTool. The supplied material does not provide a feature-by-feature comparison with either project.

FAQ

Can I deploy this directly as an online agent service?
The supplied material documents training and evaluation workflows, not an online service, unified API, or chat-product deployment.
What external access is needed for code-execution training?
You need either a local SandboxFusion deployment or a Volcano Engine Cloud FaaS `/run_code` API endpoint, configured in the specified files.
Can OSWorld run on one node?
Yes. The documentation says to set `num_node` to `1` in `configs/osworld_rl.yaml`; its reported experiments use parallel cloud VMs.
Is the full AutoTool dataset available now?
The repository includes an example format for the 200K trajectory dataset, but says the full training data and toolset are still to be released.

Related agents