xt-training

Utilities for training deep learning models in pytorch and scikit-learn, developed by Xtract AI.

The xt-training package has been built by extending the core functionality of pytorch. Core pytorch components (such as datasets/dataloaders, modules, loss functions, optimizers, and schedulers) are defined by the user in the same way, with xt-training providing a layer on top to provide more repeatable, efficient, and better organized code, logging, and experiment management.

Full documentation

Features

Installation

From PyPI

The easiest way to install xt-training is using pip. Assuming you have python (>= 3.5) and pip install on your system, simply run:

pip install xt-training

To upgrade an existing installation, run:

pip install --upgrade xt-training

From source

To install from source, first clone the xt-training repo:

git clone https://github.com/XtractTech/xt-training.git

Then use pip to install the module:

pip install ./xt-training

Next, see Getting Started to learn how to make the most of xt-training.

Contribute

See the Github repo pull requests for ongoing work.

If you wish to make a contribution to the code (or the docs), PRs are more than welcome.

Support

See the Github repo issues page for help.

License

TODO: add a license.