Skip to content

Conversation

@parambole
Copy link
Collaborator

Description

Adding the required base_mlp_dim to avoid the error

ValueError: For a fully MoE model, base_mlp_dim must be equal to base_moe_mlp_dim. Received base_mlp_dim=7168 and base_moe_mlp_dim=1536.

Tests

Locally

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.

Copy link
Collaborator

@gobbleturk gobbleturk left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

just noting we should probably have an AOT test for every model to catch things like this

@copybara-service copybara-service bot merged commit 2c0edb7 into main Nov 18, 2025
46 of 50 checks passed
@copybara-service copybara-service bot deleted the parambole/qwen3-opt branch November 18, 2025 19:36
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants