Qwen Code
Back to Tools
agent-framework ยท tool profile
Verl Agent
verl-agent is an extension of veRL, designed for training LLM/VLM agents via RL. verl-agent is also the official code for paper "Group-in-Group Policy Optimization for LLM Agent Training"
agent-framework Qwen Code
01
At a glance.
A compact read before the deeper capability notes and official setup links.
02
Core features.
Feature cards focus on what the tool helps users do, not generated setup commands.
Group-in-Group Policy Optimization for LLM Agent Training
verl-agent is an extension of veRL, specifically designed for training large language model (LLM) agents via reinforcement learning (RL).
This design makes verl-agent highly scalable for very long-horizon, multi-turn RL training (e.g., tasks in ALFWorld can require up to 50 steps to complete).
verl-agent provides a diverse set of RL algorithms (including our new algorithm GiGPO) and a rich suite of agent environments, enabling the development of reasoning agents in both visual and text-based tasks.
MAS]( which supports stable end-to-end RL post-training of multi-agent LLM systems!
[2025.09] verl-agent-style training pipeline is now supported by OpenManus-RL!
[2025.08] Add Search-R1 experiments and similarity-based GiGPO!
Check out GiGPO's superior performance in Search-R1 experiments here.
04
Agent / Skill / MCP / Workflow fit.
This panel keeps technical format separate from the user-facing AI category.
05
Official setup path.
Generated install snippets are intentionally not mirrored here because they drift. The page links to source-owned setup docs instead.
06
Evidence and adoption notes.
These notes help a user decide whether to investigate the official project further.
Source repository last pushed at 2026-02-27T05:29:25Z.
Generated from source metadata; confirm operational details in the official project before adopting it.
Review the upstream license, maintenance activity, and issue history before using it in production.
Trusted source