Open nerbivol opened 2 weeks ago
I converted my model to NCNN using the following command:
!yolo export model=jameslahm/yolov10n format=ncnn
The output was:
... ############# pass_level1 ############# pass_level2 ############# pass_level3 ############# pass_level4 ############# pass_level5 ############# pass_ncnn BinaryOp remainder not supported yet BinaryOp floor_divide not supported yet ignore torch.topk torch.topk_57 param dim=-1 ignore torch.topk torch.topk_57 param k=300 ignore torch.topk torch.topk_57 param largest=True ignore torch.topk torch.topk_57 param sorted=True ignore torch.gather torch.gather_39 param dim=1 ignore torch.gather torch.gather_40 param dim=1 ignore torch.topk torch.topk_58 param dim=-1 ignore torch.topk torch.topk_58 param k=300 ignore torch.topk torch.topk_58 param largest=True ignore torch.topk torch.topk_58 param sorted=True ignore torch.gather torch.gather_41 param dim=1 ignore Tensor.to Tensor.to_8 param copy=False ignore Tensor.to Tensor.to_8 param dtype=torch.float NCNN: export success ✅ 10.9s, saved as 'yolov10n_ncnn_model' (9.0 MB) Export complete (15.4s) Results saved to /content Predict: yolo predict task=detect model=yolov10n_ncnn_model imgsz=640 Validate: yolo val task=detect model=yolov10n_ncnn_model imgsz=640 data=None Visualize: https://netron.app 💡 Learn more at https://docs.ultralytics.com/modes/export
However, when I tried to run a prediction with the command:
!yolo predict task=detect model=yolov10n_ncnn_model source=bus.jpg
I received the following error:
Ultralytics YOLOv8.1.34 🚀 Python-3.10.12 torch-2.3.0+cu121 CUDA:0 (Tesla T4, 15102MiB) Loading yolov10n_ncnn_model for NCNN inference... layer torch.topk not exists or registered
Here's an image of the model visualization showing the torch.topk layers:
torch.topk
Are there any possible ways to replace or bypass torch.topk in this model?
Thank you for your interest! Does this issue https://github.com/THU-MIG/yolov10/issues/196 help?
I converted my model to NCNN using the following command:
The output was:
However, when I tried to run a prediction with the command:
I received the following error:
Here's an image of the model visualization showing the
torch.topk
layers:Are there any possible ways to replace or bypass
torch.topk
in this model?