NVIDIA-NeMo / NVIDIA-NeMo/Automodel

Make StatefulDataloader be DP-aware

Open
#533 7 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

enhancement good first issue
Dominant language
Python
Stars
963
Forks
318
Avg merge
3d 20h
Merged PRs (30d)
143

Description

Is your feature request related to a problem? Please describe.
StatefulDataloader is currently not DP-aware. If you change DP size during a checkpoint then you will get inconsistent results because StatefulDataloader only tracks its own (local to DP group) samples. We need to extend this to enable global dataloader resharding.

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

Start by locating StatefulDataloader and tracing how it records samples during checkpointing within a data-parallel group. Define and validate global resharding behavior for checkpoints restored with a different DP size, including consistent sample results; the issue does not name specific files or tests.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
distributed-systems
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.