[diegoakel]

What does this PR do?

Removes the calculation of load balancing loss for Qwen MoE models when outside training mode.

Fixes #42100

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline,
  Pull Request section?
• Was this discussed/approved via a Github issue or the forum? Please add a link
  to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.
• Did you write any new necessary tests?

Who can review?

@Rocketknight1

[github-actions]

[For maintainers] Suggested jobs to run (before merge)

run-slow: qwen3_moe, qwen3_omni_moe, qwen3_vl_moe

[Rocketknight1]

Posted in #42100 as well, but do you know how output_router_logits is getting set? This should only happen when that's set to True.

[ArthurZucker]

Thanks @diegoakel! On main the load-balancing loss is only computed when output_router_logits is enabled (see modular_qwen3_moe.py), and @Rocketknight1's question in #42100 about how it got turned on at inference was never resolved, so I'm closing this. Happy to reopen if you have a reproducer on current main.