An implementation of the Hogwild! algorithm for asynchronous SGD that interfaces with sci-kit learn.
A requirements.txt file is included and dependencies can be installed via:
pip install -r requirements.txt
from hogwildsgd import HogWildRegressor
hwsgd = HogWildRegressor(n_epochs = 5,
batch_size = 1,
chunk_size = 32,
learning_rate = .001,
verbose=2)
hwsgd.fit(X,y)
For an example, see the test.py file.