Hogwild ! and other Classic Online Convex Optimization algorithms implementation for OCO course.

Raw data (mnist_train.csv and mnist_test.csv) can be downloaded from https://pjreddie.com/projects/mnist-in-csv/, and should be placed in the data/raw/ folder. Run python src/data_utils.py to load, normalize and save the data as numpy arrays to the disk, so they can be loaded faster afterwards. You should now be able to train the algorithms implemented in algos.py by running python src/algos.py.