PyTorch is a python based library built to provide flexibility as a deep learning development platform. The workflow of PyTorch is as close as you can get to python’s scientific computing library – numpy.
Much like other frameworks, things like image recognition and timeseries sequences are a frequent use for PyTorch.
Applicable Scenarios and Problems
This demo is pretty basic, and basically the output gives you 5 element tensors. This model is untrained so they don't give you any real feedback that could be useful. For example, however, our sample image is that of a cat and a trained version of this model would give you the same matrix output. It might have 4 of the numbers very close to 0, while a single number would be say .98989 or something around that. This second number would then correspond to our trained models classification of 'Cat' meaning it thinks there is a cat in the picture.
This demonstration does take awhile to compile/build and so therefore it's important to structure your demonstration accordingly. It's good to show that PyTorch does indeed work on the platform with relatively minimal added work and dependencies and modules can be shown.