The unsorted_segment_sum’s parameter of the message_aggregation in the actor.py file seems incorrect

On lines 120,122 and 123, tf.math.unsorted_segment_sum(messages, self.outcoming_links, num_segments=self.n_links), 'self.outcoming_links' seems to change to 'self.incoming_links'. Using 'self.incoming_links' seems to conform to the aggregation function described in the paper, which adds up the messages {m (h, h_neighborhood)} of adjacent edges