TIGER-AI-Lab/verl-tool
A unified tool-agent training framework that uses reinforcement learning to train LLMs to use diverse tools.
Not currently ranked — collecting fresh signals.
star history
VerlTool extends the verl framework to support training LLM agents that can use external tools. It provides training recipes including Search-R1, NL2SQL, and DAPO for tool-augmented reasoning. The framework supports trajectory-level asynchronous rollout generation and integrates with vllm for efficient inference during training.
Frequently asked
- What is TIGER-AI-Lab/verl-tool?
- A unified tool-agent training framework that uses reinforcement learning to train LLMs to use diverse tools.
- Is verl-tool open source?
- Yes — TIGER-AI-Lab/verl-tool is open source, released under the MIT license.
- What language is verl-tool written in?
- TIGER-AI-Lab/verl-tool is primarily written in Python.
- How popular is verl-tool?
- TIGER-AI-Lab/verl-tool has 1k stars on GitHub.
- Where can I find verl-tool?
- TIGER-AI-Lab/verl-tool is on GitHub at https://github.com/TIGER-AI-Lab/verl-tool.