Difficulty in running cuda based deeplearning (yoloact algo)

Hey I am using jetson nano nx2 -

I installed jetson-nx-jp451-sd-card-image.

I wanted to run cuda based yoloact(deep learning) stuff. sing l4t-pytorch:r32.5.0-pth1.7-py3 docker image

I get torch.cuda.is_available() True outside docker

Inside docker torch.cuda.is_available() — false

I run docker as below docker run -it --rm --net=host --gpus all -e DISPLAY=:0 --runtime nvidia -v /tmp/.X11-unix:/tmp/.X11-unix -e NVIDIA_DRIVER_CAPABILITIES=all -e NVIDIA_VISIBLE_DEVICES=all imagename

dpkg-query --show nvidia-l4t-core — Result dpkg-query --show nvidia-l4t-core

Can you help me how to enable gpu inside ??