NVIDIA-NeMo/Gym
A library for evaluating and improving AI agents and language models across scalable, stateful environments with shared benchmarks and verifiers.
Not currently ranked — collecting fresh signals.
star history
NeMo Gym provides infrastructure to develop environments for AI agents, run scalable evaluation and training, and access a collection of benchmark environments. An environment comprises a dataset of tasks, an agent harness defining model-world interaction, a verifier for scoring task completion, and per-task execution state. It supports transitioning between evaluation, agent optimization, and training workflows at scale.
Frequently asked
- What is NVIDIA-NeMo/Gym?
- A library for evaluating and improving AI agents and language models across scalable, stateful environments with shared benchmarks and verifiers.
- Is Gym open source?
- Yes — NVIDIA-NeMo/Gym is open source, released under the Apache-2.0 license.
- What language is Gym written in?
- NVIDIA-NeMo/Gym is primarily written in Python.
- How popular is Gym?
- NVIDIA-NeMo/Gym has 1.1k stars on GitHub.
- Where can I find Gym?
- NVIDIA-NeMo/Gym is on GitHub at https://github.com/NVIDIA-NeMo/Gym.