Announcing Ultralytics v8.3.20 Release!
Summary
We’re excited to announce the release of Ultralytics version 8.3.20, packed with enhancements for improved efficiency and usability. This update focuses on optimizing training logging, enhancing Docker image compatibility, and providing better documentation for pretrained models.
Key Changes
New Features
- Pretrained Model Documentation: We’ve added examples for using pretrained YOLO models with the Open Images Dataset V7, making it easier to implement advanced AI functionalities.
Improvements
- W&B Integration Fix: Adjusted the Weights & Biases logging to prevent errors when plots are deliberately disabled, optimizing the training process and saving computational resources. PR by @Anzhc
- Docker Update: Updated the base Docker image to a more recent version of PyTorch, offering potential performance boosts and better CUDA support. PR by @glenn-jocher
Purpose & Impact
- Efficiency: The W&B logging fix ensures smoother training operations by avoiding unnecessary plotting.
- Compatibility: The Docker update facilitates more efficient processing with enhanced support for current CUDA features.
- Usability: New code examples for pretrained models simplify the integration of sophisticated AI capabilities.
What’s Changed
- Dockerfile updated to
pytorch/pytorch:2.5.0-cuda12.4-cudnn9-runtimePR by @glenn-jocher - Added usage examples for Open Images Dataset V7 PR by @Y-T-G
- W&B
plots=Falselogging fix PR by @Anzhc
New Contributors
- Welcome @Anzhc for their first contribution! PR
Full Changelog: View here
We encourage you to try out the new release and share your feedback. Your insights are invaluable in helping us improve. Thank you for being part of the Ultralytics community!