NVIDIA-NeMo / NVIDIA-NeMo/DataDesigner

Push Data Designer datasets to Harbor Hub

Open
#619 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

enhancement
Dominant language
Python
Stars
2.2k
Forks
211
Avg merge
2d 6h
Merged PRs (30d)
40

Description

Priority Level

Medium (Nice to have)

Is your feature request related to a problem? Please describe.

Add first-class support for exporting and publishing NeMo Data Designer outputs as Harbor datasets on Harbor Hub.

Harbor Hub supports public and private datasets for training and evaluation. When users need a custom benchmark, NeMo Data Designer is a natural place to generate, validate, filter, and score that data. The missing piece is a simple, reliable path from a Data Designer-generated dataset to a published Harbor Hub dataset that can be run with Harbor evals.

Problem

Today, users can run public or private Harbor benchmarks from Harbor Hub, but if they generate a custom benchmark with NeMo Data Designer, they still need to manually convert the output into Harbor’s dataset/task structure.

That conversion is non-trivial. A Harbor dataset is a collection of Harbor tasks. A typical runnable Harbor task includes:

  • task.toml
  • instruction.md
  • tests/test.sh
  • an environment/ definition, such as environment/Dockerfile for Docker-based tasks
  • optional solution/solve.sh for oracle validation
Describe the solution you'd like

Add a Harbor Hub export/publish integration to NeMo Data Designer.

At a high level, Data Designer should support two related operations:

  1. Export a generated dataset into a local Harbor-compatible dataset directory.
  2. Publish that dataset to Harbor Hub using Harbor’s existing publishing workflow.
Describe alternatives you've considered

No response

Agent Investigation

No response

Additional context

No response

Checklist
  • I've reviewed existing issues and the documentation
  • This is a design proposal, not a "please build this" request

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.

Research direction

The issue names no repository files, tests, or entry points. Start by locating existing dataset export paths and Harbor-compatible task handling, then compare them with the listed task.toml, instruction.md, tests/test.sh, environment/, and optional solution/solve.sh structure. Done means local export and Harbor Hub publishing are defined and supported reliably.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
data
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Quiet
Clarity
Needs clarification
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.