A template in PyTorch for developing a machine learning / deep learning / computer vision project.
The intent of the project is to allow people who is about to develop a ML project to focus on the key idea of their projects, and to save time from debugging all sorts of trivial and unmeaningful bugs. All the concerns are about running code with lowest probability of making mistakes. You don't have to follow my methods, but every programmer indeed should be aware of how mistakes could happen and hence take some precautions accordingly. And my concerns might inspire you a little.
Run
pip install -r requirements.txt
If you the cuda version on your machine is 9.0 or below, which can't install torch 1.0 or torch 1.2, then run:
pip install -r requirements-old.txt