It's beginning to look a lot like the last mile!
So work continues on the Bundle. This week I started to package my code and various logic into modules such that an ECL developer could easily use this bundle as part of a machine learning workflow. I.E. so you can train, test, refine, and consume neural networks on data that lives on the HPCC cluster, all in ECL. At least some of the more popular NN architectures. I plan to design the bundle such that it will be easy to add more and more capabilities (think different layers) to the neural network models. Of course, there are examples that demonstrate how one with Python experience can design any model they desire and train, test, and consume within the HPCC cluster. With GPU acceleration of course.
Since the end of my internship is fast approaching, I also started to add to the documentation, started to clean up the code (emphasis on readability), and am preparing the code for final and public submission. The work will be open source on github.com
Comments