NVIDIA-NeMo / NVIDIA-NeMo/Gym

Implement a sandboxed NOOA runner for CLU

Open
#3,456 0 comments 0 reactions 1 assignee View on GitHub

@init-nikhil is already working on this.

Since Sep 17, 2026.

area:agent area:sandbox dependency-CLU feature
Dominant language
Python
Stars
1.2k
Forks
349
Avg merge
1d 23h
Merged PRs (30d)
324

Description

Parent

Sub-issue of #3453.

Goal

Provide a sandboxed NOOA execution path that can be launched and observed through the CLU evaluation workflow.

Scope

  • Package and configure the NOOA runner for sandboxed execution.
  • Connect the runner to the configured Gym model and resources servers.
  • Preserve per-rollout isolation, logs, trajectories, and failure records.
  • Document the configuration required to launch and reproduce a run.

Acceptance criteria

  • CLU can launch a sandboxed NOOA rollout.
  • Model and resource calls succeed from the sandbox.
  • Concurrent rollouts do not share mutable agent state.
  • Failures and timeouts produce actionable artifacts.
  • A documented smoke evaluation completes end to end.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.