Tensorflow vs PyTorch

Tensorflow, PyTorch are currently the most popular deep learning packages.

Caffe2 is intended to be a framework for production edge deployment whereas TensorFlow is more suited towards server production and research. Essentially your target uses are very different. TensorFlow is intended for researchers and servers while Caffe2 is aimed towards mobile phones and other relatively computationally constrained platforms. These development goals are reflected in the designs of each framework. Caffe2 is now part of PyTorch.

• TensorFlow:
  • by Google, for both research and deployment.
  • Github: https://github.com/tensorflow/tensorflow
• pytorch
  • by Facebook, more research-friendly.
  • Github: https://github.com/pytorch/pytorch
• Caffe2
  • by Facebook, more developer-friendly. Used on their mobile apps
  • Caffe2 is now merged with PyTorch(code now lives in PyTorch's repo, maybe a full merge later on?)
  • both PyTorch and Caffe2 support ONNX(https://onnx.ai/) for interchangeability