version 0.8.2

README.rst

@@ -22,7 +22,7 @@
     </div>
 
 
-NeuPy v0.8.1
+NeuPy v0.8.2
 ============
 
 NeuPy is a python library for prototyping and building neural networks. NeuPy uses Tensorflow as a computational backend for deep learning models.

examples/cnn/mnist_cnn.py

@@ -42,7 +42,7 @@ def load_data():
     # It's suitable for classification with 3 and more classes.
     loss='categorical_crossentropy',
 
-    # Mini-batch size. It defined how many samples will be propagated
+    # Mini-batch size. It defines how many samples will be propagated
     # through the network at once. During the training, weights will
     # be updated after every mini-batch propagation.
     # Note: When number of training samples is not divisible by 128

neupy/__init__.py

@@ -2,4 +2,4 @@
 NeuPy is a Python library for Artificial Neural Networks and Deep Learning.
 """
 
-__version__ = '0.8.1'
+__version__ = '0.8.2'

neupy/algorithms/gd/base.py

@@ -178,24 +178,25 @@ def init_functions(self):
 
         tf_utils.initialize_uninitialized_variables()
 
-        self.functions.update(
-            predict=tf_utils.function(
-                inputs=as_tuple(self.network.inputs),
-                outputs=self.network.outputs,
-                name='optimizer/predict'
-            ),
-            one_training_update=tf_utils.function(
-                inputs=as_tuple(self.network.inputs, self.target),
-                outputs=loss,
-                updates=training_updates,
-                name='optimizer/one-update-step'
-            ),
-            score=tf_utils.function(
-                inputs=as_tuple(self.network.inputs, self.target),
-                outputs=val_loss,
-                name='optimizer/score'
+        with tf.name_scope('optimizer'):
+            self.functions.update(
+                predict=tf_utils.function(
+                    inputs=as_tuple(self.network.inputs),
+                    outputs=self.network.outputs,
+                    name='predict'
+                ),
+                one_training_update=tf_utils.function(
+                    inputs=as_tuple(self.network.inputs, self.target),
+                    outputs=loss,
+                    updates=training_updates,
+                    name='one-update-step'
+                ),
+                score=tf_utils.function(
+                    inputs=as_tuple(self.network.inputs, self.target),
+                    outputs=val_loss,
+                    name='score'
+                ),
             )
-        )
 
     def format_input(self, X):
         X = as_tuple(X)

neupy/algorithms/gd/objectives.py

@@ -6,8 +6,11 @@
 from neupy.utils import asfloat, function_name_scope
 
 
-__all__ = ('mse', 'rmse', 'mae', 'msle', 'rmsle', 'binary_crossentropy',
-           'categorical_crossentropy', 'binary_hinge', 'categorical_hinge')
+__all__ = (
+    'mse', 'rmse', 'mae', 'msle', 'rmsle',
+    'binary_crossentropy', 'categorical_crossentropy',
+    'binary_hinge', 'categorical_hinge',
+)
 
 
 smallest_positive_number = 1e-7  # for 32-bit float numbers

neupy/algorithms/gd/rprop.py

@@ -220,7 +220,8 @@ class IRPROPPlus(RPROP):
     >>> x_train = np.array([[1, 2], [3, 4]])
     >>> y_train = np.array([[1], [0]])
     >>>
-    >>> optimizer = algorithms.IRPROPPlus(Input(2) > Sigmoid(3) > Sigmoid(1))
+    >>> network = Input(2) >> Sigmoid(3) >> Sigmoid(1)
+    >>> optimizer = algorithms.IRPROPPlus(network)
     >>> optimizer.train(x_train, y_train)
 
     References

neupy/architectures/resnet.py

@@ -17,7 +17,7 @@ def conv_name(index):
 
     n_output_filters = 4 * n_input_filters
     main_branch = layers.join(
-        # The main purpose of this 1x1 convolutional layer is to
+        # The main purpose of this 1x1 convolution layer is to
         # reduce number of filters. For instance, for the tensor with
         # 256 filters it can be reduced to 64. This trick allows to
         # reduce computation by factor of 4.
@@ -30,7 +30,7 @@ def conv_name(index):
         layers.BatchNorm(name=bn_name('2a')),
         layers.Relu(),
 
-        # This convolutional layer applies 3x3 filter in order to
+        # This convolution layer applies 3x3 filter in order to
         # extract features.
         layers.Convolution(
             (3, 3, n_input_filters),