diff --git a/jraph/examples/basic.py b/jraph/examples/basic.py
index 0fa4741..66f335c 100644
--- a/jraph/examples/basic.py
+++ b/jraph/examples/basic.py
@@ -57,10 +57,10 @@ def run():
       senders=np.array([0, 1]), receivers=np.array([2, 2]))
   logging.info("Nested graph %r", nested_graph)
 
-  # Creates a GraphsTuple from scratch containing a 2 graphs using an implicit
+  # Creates a GraphsTuple from scratch containing 2 graphs using an implicit
   # batch dimension.
   # The first graph has 3 nodes and 2 edges.
-  # The second graph has 1 nodes and 1 edges.
+  # The second graph has 1 node and 1 edge.
   # Each node has a 4-dimensional feature vector.
   # Each edge has a 5-dimensional feature vector.
   # The graph itself has a 6-dimensional feature vector.
@@ -93,7 +93,7 @@ def run():
   # Creates a padded GraphsTuple from an existing GraphsTuple.
   # The padded GraphsTuple will contain 10 nodes, 5 edges, and 4 graphs.
   # Three graphs are added for the padding.
-  # First an dummy graph which contains the padding nodes and edges and secondly
+  # First a dummy graph which contains the padding nodes and edges and secondly
   # two empty graphs without nodes or edges to pad out the graphs.
   padded_graph = jraph.pad_with_graphs(
       single_graph, n_node=10, n_edge=5, n_graph=4)
@@ -104,7 +104,7 @@ def run():
   single_graph = jraph.unpad_with_graphs(padded_graph)
   logging.info("Unpadded graph %r", single_graph)
 
-  # Creates a GraphsTuple containing a 2 graphs using an explicit batch
+  # Creates a GraphsTuple containing 2 graphs using an explicit batch
   # dimension.
   # An explicit batch dimension requires more memory, but can simplify
   # the definition of functions operating on the graph.
@@ -113,7 +113,7 @@ def run():
   # Using an explicit batch requires padding all feature vectors to
   # the maximum size of nodes and edges.
   # The first graph has 3 nodes and 2 edges.
-  # The second graph has 1 nodes and 1 edges.
+  # The second graph has 1 node and 1 edge.
   # Each node has a 4-dimensional feature vector.
   # Each edge has a 5-dimensional feature vector.
   # The graph itself has a 6-dimensional feature vector.
@@ -125,7 +125,7 @@ def run():
       receivers=np.array([[2, 2], [0, -1]]))
   logging.info("Explicitly batched graph %r", explicitly_batched_graph)
 
-  # Running a graph propagation steps.
+  # Running a graph propagation step.
   # First define the update functions for the edges, nodes and globals.
   # In this example we use the identity everywhere.
   # For Graph neural networks, each update function is typically a neural
@@ -156,6 +156,7 @@ def update_globals_fn(
       aggregated_node_features,
       aggregated_edge_features,
       globals_):
+    """Returns the global features."""
     del aggregated_node_features
     del aggregated_edge_features
     return globals_
@@ -166,8 +167,8 @@ def update_globals_fn(
   aggregate_nodes_for_globals_fn = jraph.segment_sum
   aggregate_edges_for_globals_fn = jraph.segment_sum
 
-  # Optionally define attention logit function and attention reduce function.
-  # This can be used for graph attention.
+  # Optionally define an attention logit function and an attention reduce
+  # function. This can be used for graph attention.
   # The attention function calculates attention weights, and the apply
   # attention function calculates the new edge feature given the weights.
   # We don't use graph attention here, and just pass the defaults.