shorten mnist example / update depthwise conv 2d

Author: zsdonghao

example/tutorial_mnist.py

@@ -61,13 +61,13 @@ def main_test_layers(model='relu'):
         network = tl.layers.InputLayer(x, name='input')
         network = tl.layers.DropoutLayer(network, keep=0.8, name='drop1')
         network = tl.layers.DenseLayer(network, n_units=800,
-                                        act = tf.nn.relu, name='relu1')
+                                        act=tf.nn.relu, name='relu1')
         network = tl.layers.DropoutLayer(network, keep=0.5, name='drop2')
         network = tl.layers.DenseLayer(network, n_units=800,
-                                        act = tf.nn.relu, name='relu2')
+                                        act=tf.nn.relu, name='relu2')
         network = tl.layers.DropoutLayer(network, keep=0.5, name='drop3')
         network = tl.layers.DenseLayer(network, n_units=10,
-                                        act = tf.identity,
+                                        act=tf.identity,
                                         name='output')
     elif model == 'dropconnect':
         network = tl.layers.InputLayer(x, name='input')
@@ -79,7 +79,7 @@ def main_test_layers(model='relu'):
                                                 name='dropconnect_relu2')
         network = tl.layers.DropconnectDenseLayer(network, keep = 0.5,
                                                 n_units=10,
-                                                act = tf.identity,
+                                                act=tf.identity,
                                                 name='output')
 
     # To print all attributes of a Layer.
@@ -295,20 +295,18 @@ def main_test_stacked_denoise_AE(model='relu'):
     network = tl.layers.DropoutLayer(network, keep=0.5, name='denoising1')
     # 1st layer
     network = tl.layers.DropoutLayer(network, keep=0.8, name='drop1')
-    network = tl.layers.DenseLayer(network, n_units=800, act = act, name=model+'1')
+    network = tl.layers.DenseLayer(network, n_units=800, act=act, name=model+'1')
     x_recon1 = network.outputs
     recon_layer1 = tl.layers.ReconLayer(network, x_recon=x, n_units=784,
-                                        act = act_recon, name='recon_layer1')
+                                        act=act_recon, name='recon_layer1')
     # 2nd layer
     network = tl.layers.DropoutLayer(network, keep=0.5, name='drop2')
     network = tl.layers.DenseLayer(network, n_units=800, act = act, name=model+'2')
     recon_layer2 = tl.layers.ReconLayer(network, x_recon=x_recon1, n_units=800,
-                                        act = act_recon, name='recon_layer2')
+                                        act=act_recon, name='recon_layer2')
     # 3rd layer
     network = tl.layers.DropoutLayer(network, keep=0.5, name='drop3')
-    network = tl.layers.DenseLayer(network, n_units=10,
-                                            act = tf.identity,
-                                            name='output')
+    network = tl.layers.DenseLayer(network, 10, act=tf.identity, name='output')
 
     # Define fine-tune process
     y = network.outputs
@@ -485,23 +483,20 @@ def main_test_cnn_layer():
     #                     pool = tf.nn.max_pool,
     #                     name ='pool2',)   # output: (?, 7, 7, 64)
     ## Simplified conv API for beginner (the same with the above layers)
-    network = tl.layers.Conv2d(network, n_filter=32, filter_size=(5, 5), strides=(1, 1),
+    network = tl.layers.Conv2d(network, 32, (5, 5), (1, 1),
             act=tf.nn.relu, padding='SAME', name='cnn1')
-    network = tl.layers.MaxPool2d(network, filter_size=(2, 2), strides=(2, 2),
+    network = tl.layers.MaxPool2d(network, (2, 2), (2, 2),
             padding='SAME', name='pool1')
-    network = tl.layers.Conv2d(network, n_filter=64, filter_size=(5, 5), strides=(1, 1),
+    network = tl.layers.Conv2d(network, 64, (5, 5), (1, 1),
             act=tf.nn.relu, padding='SAME', name='cnn2')
-    network = tl.layers.MaxPool2d(network, filter_size=(2, 2), strides=(2, 2),
+    network = tl.layers.MaxPool2d(network, (2, 2), (2, 2),
             padding='SAME', name='pool2')
     ## end of conv
     network = tl.layers.FlattenLayer(network, name='flatten')
     network = tl.layers.DropoutLayer(network, keep=0.5, name='drop1')
-    network = tl.layers.DenseLayer(network, n_units=256,
-                                    act = tf.nn.relu, name='relu1')
+    network = tl.layers.DenseLayer(network, 256, act=tf.nn.relu, name='relu1')
     network = tl.layers.DropoutLayer(network, keep=0.5, name='drop2')
-    network = tl.layers.DenseLayer(network, n_units=10,
-                                    act = tf.identity,
-                                    name='output')
+    network = tl.layers.DenseLayer(network, 10, act=tf.identity, name='output')
 
     y = network.outputs
 

tensorlayer/layers.py

@@ -2489,7 +2489,7 @@ def __init__(
         # n_filter = 32,
         channel_multiplier = 3,
         shape = (3, 3),
-        strides = (1, 1, 1, 1),
+        strides = (1, 1),
         act = None,
         padding='SAME',
         W_init = tf.truncated_normal_initializer(stddev=0.02),
@@ -2500,10 +2500,13 @@ def __init__(
     ):
         Layer.__init__(self, name=name)
         self.inputs = layer.outputs
+
+        if act is None:
+            act = tf.identity
+
         print("  [TL] DepthwiseConv2d %s: shape:%s strides:%s pad:%s act:%s" %
                             (self.name, str(shape), str(strides), padding, act.__name__))
 
-        assert len(strides) == 4, "len(strides) should be 4."
         if act is None:
             act = tf.identity
 
@@ -2515,6 +2518,11 @@ def __init__(
 
         shape = [shape[0], shape[1], pre_channel, channel_multiplier]
 
+        if len(strides) == 2:
+            strides = [1, strides[0], strides[1], 1]
+
+        assert len(strides) == 4, "len(strides) should be 4."            
+
         with tf.variable_scope(name) as vs:
             W = tf.get_variable(name='W_sepconv2d', shape=shape, initializer=W_init, **W_init_args ) # [filter_height, filter_width, in_channels, channel_multiplier]
             if b_init: