NVIDIA / NVIDIA/TransformerEngine

[JAX] Integrate grouped tensor support into MaxText

Open
#2,472 2 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

2.17 MoE
Dominant language
Python
Stars
3.5k
Forks
831
Avg merge
3d 11h
Merged PRs (30d)
65

Description

Integrate JAX-based grouped tensor support with MaxText models, ensuring smooth data and workflow compatibility for MoE training and inference scenarios.

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 MaxText models and MoE training and inference, but gives no files, tests, or entry points. Start by locating the MaxText integration points and determining the required grouped-tensor data and workflow compatibility; done means grouped tensors work for both MoE training and inference.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Quiet
Clarity
Needs clarification
Newbie friendliness
30/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.