YOLOv9-e-convert(gelan) serialize error

[ou525]

Env

• GPU, jetson orin.
• OS, Ubuntu20.04
• Cuda 11.4
• TensorRT 8.6

command: ./yolov9 -s ../yolov9-e1-converted.wts yolov9-e1-converted.engine ge
output:
1 ../yolov9-e1-converted.wts ge yolov9-e1-converted.engine
[10/17/2024-10:44:05] [W] [TRT] The implicit batch dimension mode has been deprecated. Please create the network with NetworkDefinitionCreationFlag::kEXPLICIT_BATCH flag whenever possible.
Loading weights: ../yolov9-e1-converted.wts
Building engine, please wait for a while...
Segmentation fault (core dumped)

But I can generate normally using the model without reparameterization.
./yolov9 -s ../yolov9-e1.wts yolov9-e1.engine e

[wang-xinyu]

@WuxinrongY Please help here.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[WuxinrongY]

can you report your yolov9-e1-converted.wts and the structure of yolov9-e1-converted.pt?

[ou525]

When I use the official original weights for conversion, this is what happens.