ray-for-the-curious-6.py

import time

@ray.remote
def worker(ps):
    for _ in range(100):  # Arbitrary number (100) of updates
        # First, get the latest parameters. The following
        # method call is non-blocking; it returns a future
        # effectively immediately.
        params_id = ps.get_params.remote()
        # As before, this is a blocking call that waits for
        # the task to finish and then gets the value.
        params = ray.get(params_id)
        # Compute a gradient update. We make a fake update,
        # but in practice this would use an ML library like
        # TensorFlow and would also take in a batch of data.
        # We'll simulate an expensive calculation by adding
        # a sleep call
        grad = np.ones(10)
        time.sleep(0.2)
        # Update all the parameters with the same gradient.
        ps.update_params.remote(grad)