I’ll sell it to the highest bidder in the comments.
I can write the stochastic gradient descent part in ten lines of python.
Essentially, there are two components required for most mortals who aim to train neural nets: a unit that efficiently computes derivatives of functions that are compositions of many sub-functions and a unit that runs stochastic gradient descent.
There have been four thousand new frameworks for deep learning thrown on the market the past year, and I bet you were wondering what you needed to jump into this hot marketplace.
