WooooDyy/AgentGym
A framework and benchmark for training, evolving, and evaluating LLM-based agents across diverse interactive environments using reinforcement learning.
Not currently ranked — collecting fresh signals.
star history
AgentGym provides a unified platform for developing and benchmarking LLM-based agents across multiple environments including WebArena, MiniWob++, and SciWorld. The framework enables reinforcement learning training of agents in interactive multi-turn decision-making settings. It releases a fine-tuned model (AgentEvol-7B), trajectory datasets (AgentTraj-L), and evaluation benchmarks (AgentEval) to assess agent capabilities across diverse tasks.
Frequently asked
- What is WooooDyy/AgentGym?
- A framework and benchmark for training, evolving, and evaluating LLM-based agents across diverse interactive environments using reinforcement learning.
- Is AgentGym open source?
- Yes — WooooDyy/AgentGym is open source, released under the MIT license.
- What language is AgentGym written in?
- WooooDyy/AgentGym is primarily written in Python.
- How popular is AgentGym?
- WooooDyy/AgentGym has 816 stars on GitHub.
- Where can I find AgentGym?
- WooooDyy/AgentGym is on GitHub at https://github.com/WooooDyy/AgentGym.