Zero Zeng

Have you tried to debug it with GDB? what does the stack trace like?

Sry for the late reply, I don't have much time to debug your code but I think the way you add the input might be incorrect. you can refer to...

@kevinch-nv ^ ^

the reason why it fails here is because it has a zero scale 

I would guess it's a bug in our pytorch-quantization tool. zero or negative scale should never generated. cc @netaz @ttyio

@ttyio ^^

DLA does not support dynamic dimensions. Thus, for wildcard dimensions, the min, max, and opt values of the profile must be equal.. see https://docs.nvidia.com/deeplearning/tensorrt/developer-guide/index.html#dla-supp-lay-rest

5.1.5 is too old. and DLA is only available at Jetson Platform, like Jetson Xavier, NX, Nano, Orin...

@ttyio ^^ Are they open-source?

We don't support float shape tensor now but I know there is some development on it, @ttyio do we have a target release for this feature?