An open-source python machine-learning library.

PyTorch is based on the Torch machine-learning library, and was developed and released by Facebook in October 2016. Due to incompatability issues between PyTorch and Caffe2, Microsoft and Facebook worked together to reolve this and officially merged Caffe2 with Pytorch in March 2018. PyTorch makes use of CUDA for GPU accelerated machine learning on NVidia Graphics or Computing Cards. Pytorch can be found at, which includes download/installation instructions, as well as documentation and tutorials.


PyTorch can be installed easily with conda using the command:

conda install pytorch torchvision -c pytorch

or with pip:

pip3 install torch torchvision

PyTorch can be used in AWS, Google Cloud Platform, and Microsoft Azure for machine learning in cloud environments.

Last Updated: April 07, 2019