def inference(self, x, reuse=False):
print(x.get_shape())
# conv2d arguments = (scope_name, inputs, shape, bias_shape, stride, padding='VALID', wd=0.0, reuse=False, trainable=True, with_w=False)
conv_h0, conv_h0_w, conv_h0_b = mp.conv2d(
'd_conv_h0',
x, [5, 5, x.get_shape()[-1], self.first_conv_dim],
[self.first_conv_dim], [1, 2, 2, 1],
padding='SAME',
reuse=reuse,
with_w=True)
h0 = mp.lrelu(conv_h0)
# conv2d arguments = (scope_name, inputs, shape, bias_shape, stride, padding='VALID', wd=0.0, reuse=False, trainable=True, with_w=False)
conv_h1, conv_h1_w, conv_h1_b = mp.conv2d(
'd_conv_h1',
h0, [5, 5, h0.get_shape()[-1], self.first_conv_dim],
[self.first_conv_dim], [1, 2, 2, 1],
padding='SAME',
reuse=reuse,
with_w=True)
h1 = mp.lrelu(conv_h1)
# linear projection
h2 = mp.linear_project('d_lin_project_h1',
tf.reshape(h1, [self.batch_size, 7 * 7 * 64]),
1,
reuse=reuse)
return tf.nn.sigmoid(h2), h2
def inference(self, z, reuse=False, trainable=True):
with tf.variable_scope("generator") as scope:
if reuse:
scope.reuse_variables()
print("===G")
# linear projection.
z_, h0_w, self.h0_b = mp.linear_project('g_lin_project_h0',
z,
self.first_conv_dim * 8 *
4 * 4,
reuse=reuse,
with_w=True)
# reshape for cnn inputs.
h0 = tf.reshape(z_, [-1, 4, 4, self.first_conv_dim * 8])
# batch norm
h0 = tf.nn.relu(self.g_bn0(h0, trainable=trainable))
# h0 = tf.nn.relu(mp.batch_norm(h0, scope_name='g_bn_h0', reuse=reuse, trainable=trainable))
# deconv1 conv2d_transpose arguments = (scope_name, inputs, shape, output_shape, bias_shape, stride, padding='VALID', reuse=False, trainable=True)
deconv_h1, deconv_h1_w, deconv_h1_b = mp.conv2d_transpose(
'g_deconv_h1',
h0, [5, 5, self.first_conv_dim * 8,
h0.get_shape()[-1]],
[self.batch_size, 8, 8, self.first_conv_dim * 8],
[self.first_conv_dim * 8], [1, 2, 2, 1],
padding='SAME',
reuse=reuse,
with_w=True,
trainable=trainable)
h1 = tf.nn.relu(self.g_bn1(deconv_h1, trainable=trainable))
# h1 = tf.nn.relu(mp.batch_norm(deconv_h1, scope_name='g_bn_h1', reuse=reuse, trainable=trainable))
# deconv2 conv2d_transpose arguments = (scope_name, inputs, shape, output_shape, bias_shape, stride, padding='VALID', reuse=False, trainable=True)
deconv_h2, deconv_h2_w, deconv_h2_b = mp.conv2d_transpose(
'g_deconv_h2',
h1, [5, 5, self.first_conv_dim * 4,
h1.get_shape()[-1]],
[self.batch_size, 16, 16, self.first_conv_dim * 4],
[self.first_conv_dim * 4], [1, 2, 2, 1],
padding='SAME',
reuse=reuse,
with_w=True,
trainable=trainable)
print(deconv_h2.get_shape())
h2 = tf.nn.relu(self.g_bn2(deconv_h2, trainable=trainable))
# h2 = tf.nn.relu(mp.batch_norm(deconv_h2, scope_name='g_bn_h2', reuse=reuse, trainable=trainable))
# 3rd
deconv_h3, deconv_h3_w, deconv_h3_b = mp.conv2d_transpose(
'g_deconv_h3',
h2, [5, 5, 3, h2.get_shape()[-1]],
[self.batch_size, 32, 32, 3], [3], [1, 2, 2, 1],
padding='SAME',
reuse=reuse,
with_w=True,
trainable=trainable)
print(deconv_h3.get_shape())
return tf.nn.tanh(deconv_h3)
def inference(self, x, reuse=False, trainable=True):
with tf.variable_scope("discriminator") as scope:
if reuse:
scope.reuse_variables()
print("===D")
print(x.get_shape())
# conv2d arguments = (scope_name, inputs, shape, bias_shape, stride, padding='VALID', wd=0.0, reuse=False, trainable=True, with_w=False)
conv_h0, conv_h0_w, conv_h0_b = mp.conv2d(
'd_conv_h0',
x, [5, 5, x.get_shape()[-1], self.first_conv_dim],
[self.first_conv_dim], [1, 2, 2, 1],
padding='SAME',
reuse=reuse,
with_w=True,
trainable=trainable)
h0 = mp.lrelu(conv_h0)
print(h0.get_shape())
# conv2d arguments = (scope_name, inputs, shape, bias_shape, stride, padding='VALID', wd=0.0, reuse=False, trainable=True, with_w=False)
conv_h1, conv_h1_w, conv_h1_b = mp.conv2d(
'd_conv_h1',
h0, [5, 5, h0.get_shape()[-1], self.first_conv_dim * 2],
[self.first_conv_dim * 2], [1, 2, 2, 1],
padding='SAME',
reuse=reuse,
with_w=True,
trainable=trainable)
#h1 = mp.lrelu(conv_h1)
h1 = mp.lrelu(self.d_bn1(conv_h1, trainable=trainable))
# h1 = mp.lrelu(mp.batch_norm(conv_h1, scope_name='d_bn_h1', reuse=reuse, trainable=trainable))
# 3rd
conv_h2, conv_h2_w, conv_h2_b = mp.conv2d(
'd_conv_h2',
h1, [5, 5, h1.get_shape()[-1], self.first_conv_dim * 4],
[self.first_conv_dim * 4], [1, 2, 2, 1],
padding='SAME',
reuse=reuse,
with_w=True,
trainable=trainable)
#h2 = mp.lrelu(conv_h2)
h2 = mp.lrelu(self.d_bn2(conv_h2, trainable=trainable))
# h2 = mp.lrelu(mp.batch_norm(conv_h2, scope_name='d_bn_h2', reuse=reuse, trainable=trainable))
print(h2.get_shape())
# linear projection (skip h3)
h3 = mp.linear_project('d_lin_project_h3',
tf.reshape(h2, [self.batch_size, -1]),
1,
reuse=reuse)
return tf.nn.sigmoid(h3), h3
def inference(self, z, reuse=False):
# linear projection.
z_, h0_w, self.h0_b = mp.linear_project('g_lin_project_h0',
z,
self.first_conv_dim * 8 * 7 *
7,
reuse=reuse,
with_w=True)
# reshape for cnn inputs.
h0 = tf.reshape(z_, [-1, 7, 7, self.first_conv_dim * 8])
# batch norm
h0 = tf.nn.relu(mp.batch_norm(h0, scope_name='g_bn_h0', reuse=reuse))
# deconv1 conv2d_transpose arguments = (scope_name, inputs, shape, output_shape, bias_shape, stride, padding='VALID', reuse=False, trainable=True)
deconv_h1, deconv_h1_w, deconv_h1_b = mp.conv2d_transpose(
'g_deconv_h1',
h0, [5, 5, self.first_conv_dim * 4,
h0.get_shape()[-1]],
[self.batch_size, 14, 14, self.first_conv_dim * 4],
[self.first_conv_dim * 4], [1, 2, 2, 1],
padding='SAME',
reuse=reuse,
with_w=True)
h1 = tf.nn.relu(
mp.batch_norm(deconv_h1, scope_name='g_bn_h1', reuse=reuse))
# deconv2 conv2d_transpose arguments = (scope_name, inputs, shape, output_shape, bias_shape, stride, padding='VALID', reuse=False, trainable=True)
deconv_h2, deconv_h2_w, deconv_h2_b = mp.conv2d_transpose(
'g_deconv_h2',
h1, [5, 5, 1, h1.get_shape()[-1]], [self.batch_size, 28, 28, 1],
[1], [1, 2, 2, 1],
padding='SAME',
reuse=reuse,
with_w=True)
print(deconv_h2.get_shape())
return tf.nn.tanh(deconv_h2)