Skip to content

Latest commit

 

History

History
39 lines (28 loc) · 1.75 KB

README.md

File metadata and controls

39 lines (28 loc) · 1.75 KB
Raw

PyJet (WIP)

Build Status Coverage Status license

Custom Pytorch Frontend I gradually build on for my projects. This has pretty much become a multi-year personal project I use to keep a personally customized front-end for deep learning when trying out new projects. Over the years I've added all kinds of cool features like:

  • Input size inferral - no need to define the input size to your layers
  • Interface to train models with multiple or arbitrarily complex optimizer schemes
  • Interface to train models with multiple or arbitrarily complex loss schemes
  • Tracking training stats that can persist state over multiple batches
  • Keras-like training and usage of models
  • Pythonic pipelining and management of data
  • Integration with useful libraries like imgaug and tqdm

Below is the console and graph output from running mnist.py in the examples folder Example Console

Example Console

Installation

To install run the following in your HOME directory.

git clone https://github.com/abhmul/PyJet/
cd PyJet
sudo pip install -e .
cd ..

Update

To update, go to your PyJet installation directory (should be HOME/PyJet if you followed the installation instructions) and run

git pull