Skip to content

Conversation

@ananthsub
Copy link
Contributor

@ananthsub ananthsub commented Nov 12, 2025

What does this PR do ?

The current recipe examples are not so useful to users seeing the repo for the first time. these were added initially to demonstrate various ways of launching scripts with nemo-run. as a result, they primarily emphasize the YAML config overrides without providing a basic quickstart/intro.

This PR updates the existing llama3 recipe examples in a sequence of scripts to demonstrate how to:

  1. use the recipe configuration in plain python
  2. extend to using yaml config overrides for users who desire this interface
  3. launch through nemo-run for single node (torchrun) and multi-node (slurm)
  4. offer both pretrain + finetune scripts
  5. offer a template sbatch script for users who prefer not to use the nemo-run launcher
  6. a README to walk users through these scripts

these examples are still intended for demonstration and onboarding purposes. we shoudl not have the expectation that every single model supported in the repo will have corresponding examples like this. instead, we will offer more generic training scripts under scripts/ to use out of the box with other recipes (#1390)

Changelog

  • Add specific line by line info of high level changes in this PR.

GitHub Actions CI

See the CI sectionin the Contributing doc for how to trigger the CI. A Nvidia developer will need to approve and trigger the CI for external contributors.

Before your PR is "Ready for review"

Pre checks:

  • Make sure you read and followed Contributor guidelines
  • Did you write any new necessary tests?
  • Did you add or update any necessary documentation?
  • Does the PR affect components that are optional to install? (Ex: Numba, Pynini, Apex etc)
    • Reviewer: Does the PR have correct import guards for all optional libraries?

If you haven't finished some of the above items you can still open "Draft" PR.

Additional Information

  • Related to # (issue)

@copy-pr-bot
Copy link

copy-pr-bot bot commented Nov 12, 2025

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

@ananthsub ananthsub marked this pull request as ready for review November 12, 2025 16:50
@ananthsub
Copy link
Contributor Author

/ok to test 5107a71

@ananthsub ananthsub merged commit 9b6220c into NVIDIA-NeMo:main Nov 26, 2025
48 of 49 checks passed
@ananthsub ananthsub deleted the redo-llama-recipe branch November 26, 2025 08:03
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants