PyPI version

Tensorop is a Pytorch wrapper for fast prototyping for research purposes. Main purpose is to bring functionalities that Pytorch or other frameworks lack for some reason and to include best practices being used in research.

Getting Started


Install pytorch and torchvision from

  • Pytorch >= 0.4
  • Torchvision
  • Pandas
  • Numpy


Installation via Pypi

$ pip3 install tensorop

Using with git

$ git clone
$ cd tensorop

To check installation

$ >>> import tensorop; print(tensorop.__version__)

Components (Structure)

  • Vision
  • GANs
  • Models
  • Datasets
  • Layers
  • Loss Functions
  • Numpy utilities
  • tensorop.torch
  • Utilities (I/O)

These are frequently changing once v0.1 is out


There is so much work which needs to be done as of now, PRs are always welcome. Look for issues to get started.

Docs can be found here. These are not updated frequently since the framework is under constant development.