majid nasiri

> `workers_per_gpu` is the `num_workers` for pytorch dataloader, and I think `workers_per_gpu=16` is a little high for your CPU, and I suggest you set the `num_workers` lower Thank you for...

> I see, and suggest you try to set`workers_per_gpu=0` and `persistent_workers=False` to debug. Hey @MeowZheng thanks for your reply and sorry fo late reply. I haven't tested this solution. But...

This happend for my training also :confused: :question: ``` (av) s2@s2:~/avgit/defect_detection/code/ultralytics$ yolo task=detect mode=train model=yolov8x.pt data=23MixedACPOT+DCC2022+SU+AV+T2W-POT-AUG10X.yaml device=\'0,1\' epochs=100 imgsz=1280 batch=24 project="runs/detect" name="POT_23MixedACPOT+DCC2022+SU+AV+T2W-POT-AUG10X_Y8S_i1280x1280_b24" Ultralytics YOLOv8.0.19 🚀 Python-3.8.10 torch-1.13.1+cu116 CUDA:0 (NVIDIA A40,...

> I got the same issues for a week or so, still trying to find the solution Have you fixed the problem, I have same problem :confused:

Further assistance is needed!

Thanks @glenn-jocher I have to try with yolov8 versions. Is there any way to export openvino models with dynamic batch?

Even when using your cms as follows I got error converting onnx format to openvino ``` yolo export model=yolov8n.pt format=onnx imgsz=640 batch=1 mo --input_model yolov8n.onnx --batch 1 # works fine...

Evan using latest version [**Ultralytics YOLOv8.0.53**], This problem still exist ```shell (av) dev@dev:~/avgit/defect_classification/code/ultralytics$ yolo classify train data=/home/dev/avgit/defect_classification/dataset/T2W+PP+DDC2022+SU+AV+VAN#23MXBG-POT-A2XR-CLS model=yolov8x-cls.pt batch=12 project="../../runs/Y8" name="POT_T2W+PP+DDC2022+SU+AV+VAN#23MX-POT-A2XR-CLS_Y8X_i640_b12" imgsz=640 Ultralytics YOLOv8.0.53 🚀 Python-3.8.16 torch-1.13.1+cu116 CUDA:0 (NVIDIA GeForce...

I got same error when evaluating the trained model! ``` python val.py --device 0 --batch 16 --data data/V2_B20.yaml --img 1280 --weights runs/train/PAC_V2_B20_OY9C_i1280_b64/weights/best.pt --conf 0.001 --iou 0.6 val: data=data/V2_B20.yaml, weights=['runs/train/PAC_V2_B20_OY9C_i1280_b64/weights/best.pt'], batch_size=16,...

This happens for me when running script in container, but when running in shell it's working fine