Compatibility issues with torch and torchvision

Robotics & Edge Computing Jetson & Embedded Systems Jetson AGX Orin
,
1

Hi, I’m doing some experiments with the transformer model using the jetson agx orin device. My jetpack version is 5.1.1 and cuda version is 11.4, I followed PyTorch for Jetson to install torch-2.0.0+nv23.05-cp38- cp38-linux_aarch64.whl. After that it shows
ERROR: pip’s dependency resolver does not currently take into account all the packages that are installed. This behavior is the source of the following This behavior is the source of the following dependency conflicts.timm 0.9.12 requires torchvision, which is not installed.
I ignored the warning and ran the detection code. I found that cuda_is_available() shows false, but has a backend quantization engine. However, when I was using the torch-1.11.0 version of whl from the previously mentioned link, cuda_is_available() showed true, but did not have the backend quantization engine.
Is there a torchvision for torch-2.0.0 that will allow me to use cuda, or is there another version of torch and torchvision that will allow me to use both cuda and the quantization engine?

2

Hi, I’ve moved this to the Jetson former so the team there can help! Thanks

3

Thanks a lot!

5

Hi,

Please try the packages shared in the below link:

The PyTorch version is v2.2.0.

Thanks.

6

It works! Thanks a lot