.detach() on bias init

[tcmyxc]

make init bias better,
I think detach is more safer than data

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Improvement in model bias initialization for object detection.

📊 Key Changes

• Changed how bias tensors are manipulated in the YOLO model; now they are detached before updates.
• Ensured the updated biases are re-wrapped as a parameter, allowing gradient updates.

🎯 Purpose & Impact

• Purpose: To modify the bias initialization process in a way that is safer for gradient calculations, likely aimed at improving training stability.
• Impact: These changes should result in more stable and reliable model training performance, but they may require users to update their code if relying on internals of the YOLO model. 🚀

[github-actions]

👋 Hello @tcmyxc, thank you for submitting a YOLOv5 🚀 PR! To allow your work to be integrated as seamlessly as possible, we advise you to:

• ✅ Verify your PR is up-to-date with upstream/master. If your PR is behind upstream/master an automatic GitHub Actions merge may be attempted by writing /rebase in a new comment, or by running the following code, replacing 'feature' with the name of your local branch:

git remote add upstream https://github.com/ultralytics/yolov5.git
git fetch upstream
# git checkout feature  # <--- replace 'feature' with local branch name
git merge upstream/master
git push -u origin -f

• ✅ Verify all Continuous Integration (CI) checks are passing.
• ✅ Reduce changes to the absolute minimum required for your bug fix or feature addition. "It is not daily increase but daily decrease, hack away the unessential. The closer to the source, the less wastage there is." -Bruce Lee

[glenn-jocher]

@tcmyxc yes you are right, this method is better! PR is merged. Thank you for your contributions to YOLOv5 🚀 and Vision AI ⭐