Skip to content

Load balancing changes for Deepseek v4#4178

Open
dipakg-lang wants to merge 2 commits into
AI-Hypercomputer:dsv4_model_integratefrom
dipakg-lang:dsv4_model_integrate
Open

Load balancing changes for Deepseek v4#4178
dipakg-lang wants to merge 2 commits into
AI-Hypercomputer:dsv4_model_integratefrom
dipakg-lang:dsv4_model_integrate

Conversation

@dipakg-lang

Copy link
Copy Markdown
Collaborator

Description

Start with a short description of what the PR does and how this is a change from
the past.

The rest of the description includes relevant details and context, examples:

  • why is this change being made,
  • the problem being solved and any relevant context,
  • why this is a good solution,
  • some information about the specific implementation,
  • shortcomings of the solution and possible future improvements.

If the change fixes a bug or a Github issue, please include a link, e.g.,:
FIXES: b/123456
FIXES: #123456

You can also provide a comma-separated list. If you don't want to close a bug but
simply to reference it, use BUGS, e.g.:
BUGS: b/123456

Notice 1: Once all tests pass, the "pull ready" label will automatically be assigned.
This label is used for administrative purposes. Please do not add it manually.

Notice 2: For external contributions, our settings currently require an approval from a MaxText maintainer to trigger CI tests.

Tests

Please describe how you tested this change, and include any instructions and/or
commands to reproduce.

Checklist

Before submitting this PR, please make sure (put X in square brackets):

  • I have performed a self-review of my code. For an optional AI review, add the gemini-review label.
  • I have necessary comments in my code, particularly in hard-to-understand areas.
  • I have run end-to-end tests tests and provided workload links above if applicable.
  • I have made or will make corresponding changes to the doc if needed, including adding new documentation pages to the relevant Table of Contents (toctree directive) as explained in our documentation.

parambole and others added 2 commits June 12, 2026 21:17
This commit introduces full support for DeepSeek V4 by integrating its
compressed attention mechanisms, MoE routing, and architectural layers.

Key changes:
- Add `deepseek4.yml` configuration and `DeepSeek4DecoderLayer` implementation.
- Implement hybrid Hash Routing and Token Routing for MoE layers.
- Add prefix/suffix layer unrolling for non-uniform compression blocks.
- Fix Pydantic validation for base MLP dimensions.
- Bypass MLA instantiation in favor of native CompressedAttention (CSA/HCA).
…alancing for both Deepseek v3 and V4.

Tested by running training loop with new tiny Deeepseek V4 model added as part of the commit,
here are the logs for testing

Without routing logs : https://paste.googleplex.com/6421399878107136

with load balancing logs : https://paste.googleplex.com/6551357300539392
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.

2 participants