A playground for Simulations in Natural Sciences project revolving around Artificial Neural Networks
To make learning faster, I am using my AMD gpu and it's HIP capabilites, link.
For that I will be using their docker setup since fedora is a pain to install ROCm for.
docker pull rocm/pytorch:latest
Warning: This will use 35GB of your storage!!!
I found that using the conda environment included in the folders and then installing pytorch with rocm
support works easiest.
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm5.4.2
- goal is to do real-time object detection
- use the persson tutorial for that