This repo is to facilitate experiments of deep learning.

Most implementations here are based on some common dependencies so that they can be modified easily.

It is composed of:

• datasets: fast dataset loading and preprocessing
• logging: better logging tools
• losses: library of different losses
• pretrained_models: quick training of different models on different datasets. (or just load from some place)
• visualization: visualization tools to better understand the network