Contribute Environments#

Help advance RL training for the community by contributing new environments. There are two main ways to get involved:


New Environments

Build novel training environments for coding, reasoning, tool use, games, and more.

Add a New Environment
Benchmark Integrations

Integrate existing benchmarks (SWE-Bench, Tau Bench, and so on) into NeMo Gym.

→ View mini-swe-agent with SWE-Bench as a reference implementation

More detailed guidance coming soon.

https://github.com/NVIDIA-NeMo/Gym/tree/main/responses_api_agents/mini_swe_agent