[TensorRT] ERROR: Network must have at least one output

AI & Data Science Deep Learning (Training & Inference) TensorRT
#23

Another issue I noticed is, I think polygraphy is not supported for Jetson Embedded Environment; I found that it failed to install its dependencies. Although I was able to install it on a host system. I need to deploy this model on one such Jetson device and I think Polygraphy wouldn’t suffice here.

#24

Hi,

Have you tried polygraphy tool mentioned above on your host and facing the same issue with new ONNX model ?

Thank you.

#25

yes that’s exactly what I did and got the same errror.

#26

Could you please share us issue repro new onnx model.

Thank you.

#27

Here it is. Google Drive: Sign-in
Thanks

#28

Could you please provide access.

#29

Sorry for the delay but done so. Please check

#30

Hi,

We could reproduce the same error after trying polygraphy tool as well,
I think the issue here is that the K value is computed at runtime, which is not supported by TRT.

Thank you.

#31

So TensorRT+ONNX isn’t capable of solving this ?

#32

Its currently not supported by TRT.