Deep Learning Framework

This is a commonly used deep learning framework (or maybe template? ).

My purpose in creating this repo is to enable better and faster initiation of deep learning tasks based on PyTorch.

Getting Started

If you use docker, run:

$ docker pull penrose0v0/dl_normal:latest
$ docker run --gpus all -it --net=host --shm-size=16gb --name -v /{data_dir}:/root/share --name {container_name} penrose0v0/dl_normal:latest

Next, build your python environment:

% python -m venv {venv_path}
% source {venv_path}/bin/activate
% pip install -r requirements.txt

Then, modify the details of the code.

Finally, start training by running python train.py.

Also, some commonly used model architectures (such as U-Net, ResNet) are already prepared in components.py, and is still being updated.