Glenn Jocher

@git-hamza export requirements are listed in requirements.txt, i.e. https://github.com/ultralytics/yolov5/blob/39d7a93619083cb8e37f5ef7708cf50b34e20ee1/requirements.txt#L30

@mozeqiu could be thermal throttling on your device after extended use, i.e. the thermal management system on embedded devices may reduce performance to reduce overheating. This happens on iPhones for...

@SahinTiryaki 👋 hi, thanks for letting us know about this possible problem with YOLOv5 🚀. We've created a few short guidelines below to help users provide what we need in...

@edaabahar PyTorch Hub and detect.py inference on default images with default model produces identical results: ## How to create a Minimal, Reproducible Example When asking a question, people will be...

@mynameischaos 👋 Hello! Thanks for asking about model anchors. YOLOv5 🚀 uses a new [Ultralytics](https://ultralytics.com/) algorithm called **AutoAnchor** for anchor verification and generation before training starts. Autoanchor will analyse your...

@konioy 👋 Hello, thanks for asking about the differences between [train.py](https://github.com/ultralytics/yolov5/blob/master/train.py), [detect.py](https://github.com/ultralytics/yolov5/blob/master/detect.py) and [val.py](https://github.com/ultralytics/yolov5/blob/master/val.py) in YOLOv5 🚀. There is no bug. These 3 files are designed for different purposes and...

@konioy current master with torch>=1.12.0 is fully reproducible:

@konioy torch>=1.12 should be fully reproducible using single GPU. Multi-GPU is not yet reproducible and we don't have a clear reason why.

@konioy zero val loss typically indicates your validation set has no labels EDIT: if you used --no-val then the above is normal

@konioy use torch>=12.0 for reproducible Single-GPU CUDA trainings runs