torch-tools 0.0.26

A library of helpers to train, evaluate and visualize deep nets with PyTorch.

Readme

A library of helpers for PyTorch.

Michaël Gharbi <[e-mail]>

Installation

From pip: pip install torch-tools

From source: python setup.py install

Quickstart

To quickly get up and running, use our repo initialization script. In your command line, simple run:

ttools.new

This will prompt you to give a name to your new project, and create the necessary files and folders for you.

Documentation

The documentation webpage can be found here https://torch-tools.readthedocs.io/en/latest/

Demo

For a simple demo, look at the MNIST example in examples/train_mnist.py:

python examples/train_mnist.py data out

Contributors

Dima Smirnov implemented the tensorboard hooks and callbacks.