Predicates the best VM for a machine learning model workload.

./prepare.sh

Evaluate on current VM.

./evaluate.sh

We implement and adapt methods from the following publications:

• https://people.eecs.berkeley.edu/~brecht/papers/hogwildTR.pdf
• http://www.stat.ucdavis.edu/~chohsieh/wildSGD.pdf

We strongly rely on previous implementations for easy reproducibility.

• https://github.com/pytorch/examples/tree/master/mnist_hogwild