Skip to main content

Typed Thin PyTorch Wrapper

Project description - Typed Thin (Py)Torch Wrapper

T3W is a lightweight framework for training PyTorch models written by Yuyao Huang during his PhD at Tongji University.

  • T3W is "typed". It leverages stronger and static typing compared to normal python code for clearer architecture and less bugs. The programming model is object-oriented. Users (you) are required to implement interfaces as subclasses and inject them as dependencies.
  • T3W is "thin". With the philosophy "less is more" in mind, it leverages a minimal codebase in a self-contained single python script that basically only depends on PyTorch to run. The plugin system under interface ISideEffect makes T3W not only thin, but also highly extensible.
  • T3W stands with "PyTorch".

See the concise example

If you feel like using, you can either

  • install it with pip install t3w, pip install t3w[common], or pip install t3w[all], where the [common] tag will install dependencies for commonly used side effects (like tqdm etc), and [all] tag will install dependencies for all supported side effects. Note that the mnist example requires installing t3w[common].
  • just download and integrate the source file into your own project if you feel like freezing the version and/or ready to make some of your favorite hacks at low-level.

API documentation is currently available at We are going to add more detailed user guide in the future.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

t3w-0.1.2.tar.gz (20.6 kB view hashes)

Uploaded Source

Built Distribution

t3w-0.1.2-py3-none-any.whl (20.9 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page