I tried to use the pip3 install torch torchvision --index-url https://download.pytorch.org/whl/cu126 on Jetson and it seems like it installed correctly with CUDA support
admin2@admin2-desktop:~$ python3 - <<EOF
import torch, torchvision
print("Torch :", torch.__version__)
print("TorchVision:", torchvision.__version__)
print("CUDA avail?:", torch.cuda.is_available())
if torch.cuda.is_available():
print("GPU device :", torch.cuda.get_device_name(0))
EOF
Torch : 2.10.0+cu126
TorchVision: 0.25.0+cu126
CUDA avail?: True
GPU device : Orin
Is this correct? Because I think the old issue was that torch does not have a ARM version prebuilt but now it has.