NVIDIA-NeMo / NVIDIA-NeMo/Gym

feat: Nemo Gym <> VeRL Integration

Open
#165 4 comments 0 reactions 1 assignee View on GitHub

@cmunley1 is already working on this.

Since Apr 20, 2026.

training
Dominant language
Python
Stars
1.2k
Forks
349
Avg merge
1d 21h
Merged PRs (30d)
318

Description

Use cases, pain points, and background
VeRL is a very popular training framework we should integrate with.

Description:
We should develop a generic interface such that VeRL can use any NeMo Gym environments. There should be a tutorial in VeRL docs that demonstrates how to use one or more NeMo Gym environments with VeRL for training. There should also be a corresponding docs page in NeMo Gym docs that links to the tutorial in VeRL.

Design:
What files should be touched? What logic should be written?

Out of scope:
What are some items that this issue could be mistaken to cover that this issue should explicitly NOT cover?

Acceptance Criteria:

  • Implement the Gym + VeRL integration
  • Replicate a Gym baselining run (GRPO on Qwen 3 4B Instruct, 32k seq len, Library judge math verifier, DAPO17k dataset, AIME24 validation) and report the W&B curves

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.