fisher

Met the same problem with the master branch. I checked the v1 stable version tests, the cpu tests passed but the gpu tests failed in the small_test case. Running GPU...

Check the box position decoding [code](https://github.com/ultralytics/yolov5/blob/c09fb2aa95b6ca86c460aa106e2308805649feb9/models/yolo.py#L62) in the forward method, the detector predicts boxes on neighbouring grid as well

check this the nvidia [deepstream app example](https://github.com/NVIDIA-AI-IOT/deepstream_reference_apps/tree/master/runtime_source_add_delete)

Hi, I met a similar problem while working in the ultralytics/yolov5:v7.0 docker image environment. Everything worked fine while training using 1 GPU; but some problem came up during DDP training....

@glenn-jocher , thanks for reply! I was following this [document](https://docs.ultralytics.com/yolov5/tutorials/clearml_logging_integration/) to test the clearml integration. I had clearml installed and inited successfully. I reproduced this problem with the example coco128...