Comments explaining tests.

tests/bricks/test_bn.py

@@ -13,6 +13,7 @@
 
 
 def random_unif(rng, dim, low=1, high=10):
+    """Generate some floatX uniform random numbers."""
     return (rng.uniform(low, high, size=dim)
             .astype(theano.config.floatX))
 
@@ -61,6 +62,7 @@ def check(input_dim, expected_shape, broadcastable=None, save_memory=True):
 
 
 def apply_setup(input_dim, broadcastable, save_memory):
+    """Common setup code."""
     bn = BatchNormalization(input_dim, broadcastable, save_memory,
                             epsilon=1e-4)
     bn.initialize()
@@ -73,6 +75,7 @@ def apply_setup(input_dim, broadcastable, save_memory):
 
 
 def test_batch_normalization_inference_apply():
+    """Test that BatchNormalization.apply works in inference mode."""
     def check(input_dim, variable_dim, broadcastable=None, save_memory=True):
         bn, x, y = apply_setup(input_dim, broadcastable, save_memory)
         rng = numpy.random.RandomState((2015, 12, 16))
@@ -179,6 +182,7 @@ def normalize(x):
 
 
 def test_batch_normalization_image_size_setter():
+    """Test that setting image_size on a BatchNormalization works."""
     bn = BatchNormalization()
     bn.image_size = (5, 4)
     assert bn.input_dim == (None, 5, 4)
@@ -187,6 +191,7 @@ def test_batch_normalization_image_size_setter():
 
 
 def test_spatial_batch_normalization():
+    """Smoke test for SpatialBatchNormalization."""
     def check(*input_dim):
         sbn = SpatialBatchNormalization(input_dim)
         sbn.initialize()
@@ -204,6 +209,7 @@ def check(*input_dim):
 
 
 def test_raise_exception_spatial():
+    """Test that SpatialBatchNormalization raises an expected exception."""
     # Work around a stupid bug in nose2 that unpacks the tuple into
     # separate arguments.
     yield assert_raises, (ValueError, SpatialBatchNormalization, (5,))
@@ -222,6 +228,7 @@ def do_not_fail(*input_dim):
 
 
 def test_batch_normalization_inside_convolutional_sequence():
+    """Test that BN bricks work in ConvolutionalSequences."""
     conv_seq = ConvolutionalSequence(
         [Convolutional(filter_size=(3, 3), num_filters=4),
          BatchNormalization(broadcastable=(False, True, True)),
@@ -256,6 +263,7 @@ def test_batch_normalization_inside_convolutional_sequence():
 
 
 def test_batch_normalized_mlp_construction():
+    """Test that BatchNormalizedMLP performs construction correctly."""
     mlp = BatchNormalizedMLP([Tanh(), Tanh()], [5, 7, 9])
     assert all(isinstance(a, Sequence) for a in mlp.activations)
     assert all(isinstance(a.children[0], BatchNormalization)
@@ -265,6 +273,7 @@ def test_batch_normalized_mlp_construction():
 
 
 def test_batch_normalized_mlp_allocation():
+    """Test that BatchNormalizedMLP performs allocation correctly."""
     mlp = BatchNormalizedMLP([Tanh(), Tanh()], [5, 7, 9])
     mlp.allocate()
     assert mlp.activations[0].children[0].input_dim == 7
@@ -282,6 +291,7 @@ def test_batch_normalized_mlp_transformed():
 
 
 def test_batch_normalized_mlp_save_memory_propagated():
+    """Test that setting save_memory on a BatchNormalizedMLP works."""
     mlp = BatchNormalizedMLP([Tanh(), Tanh()], [5, 7, 9],
                              save_memory=False)
     assert not any(act.children[0].save_memory for act in mlp.activations)