[API] Implement get for multiple objects (#398)

* [API] Implement get for multiple objects

* Small fixes.

examples/alexnet/alexnet.py

@@ -241,7 +241,7 @@ def num_images(batches):
     int: The number of images
   """
   shape_ids = [ra.shape.remote(batch) for batch in batches]
-  return sum([ray.get(shape_id)[0] for shape_id in shape_ids])
+  return sum([shape[0] for shape in ray.get(shape_ids)])
 
 @ray.remote
 def compute_mean_image(batches):
@@ -256,9 +256,8 @@ def compute_mean_image(batches):
   if len(batches) == 0:
     raise Exception("No images were passed into `compute_mean_image`.")
   sum_image_ids = [ra.sum.remote(batch, axis=0) for batch in batches]
-  sum_images = [ray.get(sum_image_id) for sum_image_id in sum_image_ids]
   n_images = num_images.remote(batches)
-  return np.sum(sum_images, axis=0).astype("float64") / ray.get(n_images)
+  return np.sum(ray.get(sum_image_ids), axis=0).astype("float64") / ray.get(n_images)
 
 @ray.remote(num_return_vals=4)
 def shuffle_arrays(first_images, first_labels, second_images, second_labels):

examples/alexnet/driver.py

@@ -93,7 +93,7 @@ if __name__ == "__main__":
     print "Iteration {}: accuracy = {:.3}%".format(iteration, 100 * ray.get(accuracy))
 
     # Fetch the gradients. This blocks until the gradients have been computed.
-    gradient_sets = [ray.get(gradient_id) for gradient_id in gradient_ids]
+    gradient_sets = ray.get(gradient_ids)
     # Average the gradients over all of the tasks.
     mean_gradients = [np.mean([gradient_set[i] for gradient_set in gradient_sets], axis=0) for i in range(len(weights))]
     # Use the gradients to update the network.

examples/lbfgs/README.md

@@ -111,12 +111,12 @@ gradient.
 def full_loss(theta):
   theta_id = ray.put(theta)
   loss_ids = [loss.remote(theta_id, xs_id, ys_id) for (xs_id, ys_id) in batch_ids]
-  return sum([ray.get(loss_id) for loss_id in loss_ids])
+  return sum(ray.get(loss_ids))
 
 def full_grad(theta):
   theta_id = ray.put(theta)
   grad_ids = [grad.remote(theta_id, xs_id, ys_id) for (xs_id, ys_id) in batch_ids]
-  return sum([ray.get(grad_id) for grad_id in grad_ids]).astype("float64") # This conversion is necessary for use with fmin_l_bfgs_b.
+  return sum(ray.get(grad_ids)).astype("float64") # This conversion is necessary for use with fmin_l_bfgs_b.
 ```
 
 Note that we turn `theta` into a remote object with the line `theta_id =

examples/lbfgs/driver.py

@@ -92,13 +92,13 @@ if __name__ == "__main__":
   def full_loss(theta):
     theta_id = ray.put(theta)
     loss_ids = [loss.remote(theta_id, xs_id, ys_id) for (xs_id, ys_id) in batch_ids]
-    return sum([ray.get(loss_id) for loss_id in loss_ids])
+    return sum(ray.get(loss_ids))
 
   # Compute the gradient of the loss on the entire dataset.
   def full_grad(theta):
     theta_id = ray.put(theta)
     grad_ids = [grad.remote(theta_id, xs_id, ys_id) for (xs_id, ys_id) in batch_ids]
-    return sum([ray.get(grad_id) for grad_id in grad_ids]).astype("float64") # This conversion is necessary for use with fmin_l_bfgs_b.
+    return sum(ray.get(grad_ids)).astype("float64") # This conversion is necessary for use with fmin_l_bfgs_b.
 
   # From the perspective of scipy.optimize.fmin_l_bfgs_b, full_loss is simply a
   # function which takes some parameters theta, and computes a loss. Similarly,