Hi,
I’m trying to use a Jetson Orin with pytorch for YOLO real-time recognition
The code already works on CPU, on a regular notebook, but I’m not able to move it to GPU on the Orin,
I already have Cuda 11.4 installed and Torch 1.12.1, which, from what I found, should support one another, but
torch.cuda.is_available()
returns False
Aditionally my JetPack version is 5.0.2-b231
Hello @jotapedromabreu and welcome to the NVIDIA developer forums!
I highly recommend going through the “Getting Started” instructions on Start Locally | PyTorch and check that you are fulfilling all prerequisites.
In addition to that I will move your post to the Jetson Orin category, they will haave better insight into what requirements are needed to run pytorch with CUDA on that device.
Thanks!
@jotapedromabreu see the PyTorch for Jetson topic for the CUDA-enabled PyTorch wheels for JetPack