Fix dtype issue with hidream o1

[comfyanonymous]

No description provided.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 9a7f6b44-4e12-439d-ae87-afe63d582613

📥 Commits

Reviewing files that changed from the base of the PR and between 8e53f00 and 9086fd0.

📒 Files selected for processing (1)

• comfy/text_encoders/qwen35.py

---

📝 Walkthrough

Walkthrough

This PR modifies the patch embedding forward pass in Qwen35VisionPatchEmbed by removing an explicit dtype cast. The input tensor is now passed directly to self.proj without casting to the convolution weight's dtype, then reshaped to the target embedding dimension. The change simplifies the projection path while reducing code by one net line.

🚥 Pre-merge checks | ✅ 3 | ❌ 2

❌ Failed checks (1 warning, 1 inconclusive)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.
Description check	❓ Inconclusive	No description was provided, making it impossible to evaluate relevance to the changeset.	Add a description explaining the dtype issue, why it occurs, and how this fix resolves it for better context.

✅ Passed checks (3 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title directly describes the main change: fixing a dtype issue with Qwen 3.5 model integration.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}