silenced SOME (a few) warnings thrown about by pylint. I wont fix any…

… more now.

nn_wtf/input_data.py

@@ -30,35 +30,35 @@
 
 
 def maybe_download(filename, work_directory):
-  """Download the data from Yann's website, unless it's already here."""
-  if not os.path.exists(work_directory):
-    os.mkdir(work_directory)
-  filepath = os.path.join(work_directory, filename)
-  if not os.path.exists(filepath):
-    filepath, _ = urllib.request.urlretrieve(SOURCE_URL + filename, filepath)
-    statinfo = os.stat(filepath)
-    print('Successfully downloaded', filename, statinfo.st_size, 'bytes.')
-  return filepath
+    """Download the data from Yann's website, unless it's already here."""
+    if not os.path.exists(work_directory):
+      os.mkdir(work_directory)
+    filepath = os.path.join(work_directory, filename)
+    if not os.path.exists(filepath):
+        filepath, _ = urllib.request.urlretrieve(SOURCE_URL + filename, filepath)
+        statinfo = os.stat(filepath)
+        print('Successfully downloaded', filename, statinfo.st_size, 'bytes.')
+    return filepath
 
 
 def _read32(bytestream):
-  dt = numpy.dtype(numpy.uint32).newbyteorder('>')
-  return numpy.frombuffer(bytestream.read(4), dtype=dt)[0]
+    dt = numpy.dtype(numpy.uint32).newbyteorder('>')
+    return numpy.frombuffer(bytestream.read(4), dtype=dt)[0]
 
 
 def extract_images(filename):
-  """Extract the images into a 4D uint8 numpy array [index, y, x, depth]."""
-  print('Extracting', filename)
-  with gzip.open(filename) as bytestream:
-    magic = _read32(bytestream)
-    if magic != 2051:
-      raise ValueError(
-          'Invalid magic number %d in MNIST image file: %s' %
-          (magic, filename))
-    num_images = _read32(bytestream)
-    rows = _read32(bytestream)
-    cols = _read32(bytestream)
-    return images_from_bytestream(bytestream, rows, cols, num_images)
+    """Extract the images into a 4D uint8 numpy array [index, y, x, depth]."""
+    print('Extracting', filename)
+    with gzip.open(filename) as bytestream:
+        magic = _read32(bytestream)
+        if magic != 2051:
+            raise ValueError(
+                'Invalid magic number %d in MNIST image file: %s' %
+                (magic, filename))
+        num_images = _read32(bytestream)
+        rows = _read32(bytestream)
+        cols = _read32(bytestream)
+        return images_from_bytestream(bytestream, rows, cols, num_images)
 
 
 def images_from_bytestream(bytestream, rows, cols, num_images):
@@ -92,20 +92,20 @@ def dense_to_one_hot(labels_dense, num_classes=10):
 
 
 def extract_labels(filename, one_hot=False):
-  """Extract the labels into a 1D uint8 numpy array [index]."""
-  print('Extracting', filename)
-  with gzip.open(filename) as bytestream:
-    magic = _read32(bytestream)
-    if magic != 2049:
-      raise ValueError(
-          'Invalid magic number %d in MNIST label file: %s' %
-          (magic, filename))
-    num_items = _read32(bytestream)
-    buf = bytestream.read(num_items)
-    labels = numpy.frombuffer(buf, dtype=numpy.uint8)
-    if one_hot:
-      return dense_to_one_hot(labels)
-    return labels
+    """Extract the labels into a 1D uint8 numpy array [index]."""
+    print('Extracting', filename)
+    with gzip.open(filename) as bytestream:
+        magic = _read32(bytestream)
+        if magic != 2049:
+            raise ValueError(
+                'Invalid magic number %d in MNIST label file: %s' %
+                (magic, filename))
+        num_items = _read32(bytestream)
+        buf = bytestream.read(num_items)
+        labels = numpy.frombuffer(buf, dtype=numpy.uint8)
+        if one_hot:
+            return dense_to_one_hot(labels)
+        return labels
 
 
 TRAIN_IMAGES = 'train-images-idx3-ubyte.gz'

nn_wtf/mnist_graph.py

@@ -1,4 +1,4 @@
-from nn_wtf.neural_network_graph import NeuralNetworkGraph
+from nn_wtf.neural_network_graph import NeuralNetworkGraph, CHANGE_THIS_LEARNING_RATE
 from nn_wtf.neural_network_graph_mixins import SaverMixin, DEFAULT_TRAIN_DIR, SummaryWriterMixin
 
 __author__ = 'Lene Preuss <lp@sinnwerkstatt.com>'
@@ -14,11 +14,19 @@ class MNISTGraph(NeuralNetworkGraph, SaverMixin, SummaryWriterMixin):
     IMAGE_PIXELS = IMAGE_SIZE * IMAGE_SIZE
 
     def __init__(
-        self, verbose=True,
-        learning_rate=0.01, layer_sizes=(128, 32, None),
+        self, input_size=None, layer_sizes=(128, 32, None), output_size=None, learning_rate=CHANGE_THIS_LEARNING_RATE,
+        verbose=True,
         train_dir=DEFAULT_TRAIN_DIR,
-        input_size=None, output_size=None
     ):
+        """The MNISTGraph constructor takes no positional args, in contrast with NeuralNetworkGraph.
+
+        :param input_size: ignored, present for client compatibility
+        :param layer_sizes: tuple of sizes of the neural network hidden layers
+        :param output_size: ignored, present for client compatibility
+        :param learning_rate: learning rate for gradient descent optimizer
+        :param verbose: whether to print some info about the training progress
+        :param train_dir: where to write savepoints and summaries
+        """
         NeuralNetworkGraph.__init__(
             self, self.IMAGE_PIXELS, layer_sizes, self.NUM_CLASSES, learning_rate
         )

nn_wtf/neural_network_graph.py

@@ -14,11 +14,12 @@
 class NeuralNetworkGraph:
 
     def __init__(self, input_size, layer_sizes, output_size, learning_rate=CHANGE_THIS_LEARNING_RATE):
-        """
-        Args:
-          input_size: number of input channels
-          layer_sizes: Sizes of hidden layers in a tuple or list.
-          output_size: Number of output channels.
+        """Initialize a neural network given its geometry.
+
+        :param input_size: number of input channels
+        :param layer_sizes: tuple of sizes of the neural network hidden layers
+        :param output_size: number of output classes
+        :param learning_rate: learning rate for gradient descent optimizer
         """
         self._setup_geometry(input_size, layer_sizes, output_size)
         self.learning_rate = learning_rate
@@ -59,11 +60,8 @@ def fill_feed_dict(self, data_set, batch_size):
               ....
         }
 
-        Args:
-          data_set: The set of images and labels, from input_data.read_data_sets()
-
-        Returns:
-          feed_dict: The feed dictionary mapping from placeholders to values.
+        :param data_set: The set of images and labels, from input_data.read_data_sets()
+        :return The feed dictionary mapping from placeholders to values.
         """
         # Create the feed_dict for the placeholders filled with the next `batch size ` examples.
         input_feed, labels_feed = data_set.next_batch(batch_size)
@@ -94,8 +92,7 @@ def _set_layer_sizes(self, layer_sizes):
     def _build_neural_network(self):
         """Builds a neural network with the given layers and output size.
 
-        Returns:
-          logits: Output tensor with the computed logits.
+        :return Output tensor with the computed logits.
         """
 
         assert self.layers == [], 'build_neural_network() has been called before'

nn_wtf/neural_network_optimizer.py

@@ -8,6 +8,7 @@
 
 
 class NeuralNetworkOptimizer:
+    """Attempts to find the best parameters to a neural network to be trained in the fastest way."""
 
     DEFAULT_LEARNING_RATE = 0.1
     DEFAULT_LAYER_SIZES = (
@@ -75,7 +76,7 @@ def get_network_geometries(self):
                 for l3 in self.layer_sizes[2] if l3 is None or l3 <= l2)
 
     def brute_force_optimize_learning_rate(self):
-        raise NotImplemented()
+        raise NotImplementedError()
 
     def timed_run_training(self, data_sets, geometry, max_steps=10000):
         graph, cpu, wall = timed_run(self.run_training_once, data_sets, geometry, max_steps)
@@ -99,6 +100,6 @@ def run_training_once(self, data_sets, geometry, max_steps):
 
 
 def timed_run(function, *args, **kwargs):
-        start_cpu_time, start_wall_time = time.process_time(), time.time()
-        returned = function(*args, **kwargs)
-        return returned, time.process_time()-start_cpu_time, time.time()-start_wall_time
+    start_cpu_time, start_wall_time = time.process_time(), time.time()
+    returned = function(*args, **kwargs)
+    return returned, time.process_time()-start_cpu_time, time.time()-start_wall_time

nn_wtf/tests/__init__.py

@@ -1 +1,3 @@
 __author__ = 'Lene Preuss <lp@sinnwerkstatt.com>'
+
+"""unit tests for nn_wtf"""

nn_wtf/tests/images_labels_data_set_test.py

@@ -6,6 +6,7 @@
 import unittest
 
 __author__ = 'Lene Preuss <lp@sinnwerkstatt.com>'
+# pylint: disable=missing-docstring
 
 NUM_TRAINING_SAMPLES = 20
 IMAGE_WIDTH = 10

nn_wtf/tests/input_data_test.py

@@ -10,6 +10,7 @@
 import unittest
 
 __author__ = 'Lene Preuss <lene.preuss@gmail.com>'
+# pylint: disable=missing-docstring
 
 
 class InputDataTest(unittest.TestCase):

nn_wtf/tests/neural_network_graph_test.py

@@ -7,6 +7,7 @@
 import unittest
 
 __author__ = 'Lene Preuss <lp@sinnwerkstatt.com>'
+# pylint: disable=missing-docstring
 
 
 class NeuralNetworkGraphTest(unittest.TestCase):

nn_wtf/tests/neural_network_optimizer_test.py

@@ -7,6 +7,7 @@
 import unittest
 
 __author__ = 'Lene Preuss <lene.preuss@gmail.com>'
+# pylint: disable=missing-docstring
 
 
 class NeuralNetworkOptimizerTest(unittest.TestCase):

nn_wtf/tests/predictor_test.py

@@ -4,9 +4,8 @@
 
 import unittest
 
-from nn_wtf.trainer import Trainer
-
 __author__ = 'Lene Preuss <lene.preuss@gmail.com>'
+# pylint: disable=missing-docstring
 
 
 class PredictorTest(unittest.TestCase):