update decorator to work better with sphinx

tensorpack/dataflow/dataset/ilsvrc.py

@@ -31,4 +31,4 @@ class ILSVRCMeta(object):
 
 if __name__ == '__main__':
     meta = ILSVRCMeta()
-    print meta.get_synset_words_1000()
+    print(meta.get_synset_words_1000())

tensorpack/dataflow/dftools.py

@@ -5,7 +5,7 @@
 import sys, os
 from scipy.misc import imsave
 
-from ..utils.utils import mkdir_p
+from ..utils.fs import mkdir_p
 
 # TODO name_func to write label?
 def dump_dataset_images(ds, dirname, max_count=None, index=0):

tensorpack/models/_common.py

@@ -23,38 +23,35 @@ def layer_register(summary_activation=False):
             Can be overriden when creating the layer.
     """
     def wrapper(func):
-        class WrapedObject(object):
-            def __init__(self, func):
-                self.f = func
+        @wraps(func)
+        def wrapped_func(*args, **kwargs):
+            name = args[0]
+            assert isinstance(name, six.string_types), \
+                    'name must be the first argument. Args: {}'.format(str(args))
+            args = args[1:]
 
-            @wraps(func)
-            def __call__(self, *args, **kwargs):
-                name = args[0]
-                assert isinstance(name, six.string_types), \
-                        'name must be either the first argument. Args: {}'.format(str(args))
-                args = args[1:]
+            do_summary = kwargs.pop(
+                'summary_activation', summary_activation)
+            inputs = args[0]
+            with tf.variable_scope(name) as scope:
+                outputs = func(*args, **kwargs)
+                if scope.name not in _layer_logged:
+                    # log shape info and add activation
+                    logger.info("{} input: {}".format(
+                        scope.name, get_shape_str(inputs)))
+                    logger.info("{} output: {}".format(
+                        scope.name, get_shape_str(outputs)))
 
-                do_summary = kwargs.pop(
-                    'summary_activation', summary_activation)
-                inputs = args[0]
-                with tf.variable_scope(name) as scope:
-                    outputs = self.f(*args, **kwargs)
-                    if scope.name not in _layer_logged:
-                        # log shape info and add activation
-                        logger.info("{} input: {}".format(
-                            scope.name, get_shape_str(inputs)))
-                        logger.info("{} output: {}".format(
-                            scope.name, get_shape_str(outputs)))
-
-                        if do_summary:
-                            if isinstance(outputs, list):
-                                for x in outputs:
-                                    add_activation_summary(x, scope.name)
-                            else:
-                                add_activation_summary(outputs, scope.name)
-                        _layer_logged.add(scope.name)
-                    return outputs
-        return WrapedObject(func)
+                    if do_summary:
+                        if isinstance(outputs, list):
+                            for x in outputs:
+                                add_activation_summary(x, scope.name)
+                        else:
+                            add_activation_summary(outputs, scope.name)
+                    _layer_logged.add(scope.name)
+                return outputs
+        wrapped_func.f = func   # attribute to access the underlining function object
+        return wrapped_func
     return wrapper
 
 def shape2d(a):

tensorpack/models/fc.py

@@ -15,6 +15,9 @@ __all__ = ['FullyConnected']
 def FullyConnected(x, out_dim,
                    W_init=None, b_init=None,
                    nl=tf.nn.relu, use_bias=True):
+    """
+    Fully-Connected layer
+    """
     x = batch_flatten(x)
     in_dim = x.get_shape().as_list()[1]