Adjust l2 norm penalty in example

The reference implementation uses's Tensorflow's l2 norm penalty which  divides the weighting parameter by 2, but Keras's l2 norm penalty implementation doesn't. This brings the example closer to the reference implemenatation.

examples/gat.py

@@ -20,7 +20,7 @@
 F_ = 8                        # Output size of first GraphAttention layer
 n_attn_heads = 8              # Number of attention heads in first GAT layer
 dropout_rate = 0.6            # Dropout rate (between and inside GAT layers)
-l2_reg = 5e-4                 # Factor for l2 regularization
+l2_reg = 5e-4/2               # Factor for l2 regularization
 learning_rate = 5e-3          # Learning rate for Adam
 epochs = 10000                # Number of training epochs
 es_patience = 100             # Patience fot early stopping