Generative Adversarial Networks for Natural Language Processing

``````

In [18]:

import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
import os

``````

State of art weight Initialization strategy

``````

In [33]:

def xavier_init(n_inputs, n_outputs, uniform=True):
"""Set the parameter initialization using the method described.
This method is designed to keep the scale of the gradients roughly the same
in all layers.
Xavier Glorot and Yoshua Bengio (2010):
Understanding the difficulty of training deep feedforward neural
networks. International conference on artificial intelligence and
statistics.
Args:
n_inputs: The number of input nodes into each output.
n_outputs: The number of output nodes for each input.
uniform: If true use a uniform distribution, otherwise use a normal.
Returns:
An initializer.
"""
if uniform:
# 6 was used in the paper.
init_range = tf.sqrt(6.0 / (n_inputs + n_outputs))
return tf.random_uniform_initializer(-init_range, init_range)
else:
# 3 gives us approximately the same limits as above since this repicks
# values greater than 2 standard deviations from the mean.
stddev = tf.sqrt(3.0 / (n_inputs + n_outputs))
return tf.truncated_normal_initializer(stddev=stddev)

``````
``````

In [34]:

'''A recent paper by He, Rang, Zhen and Sun they build on Glorot & Bengio and suggest using 2/size_of_input_neuron
'''
def xavier_init(size):
in_dim = size[0]
#     xavier_stddev = 1. / in_dim
#     xavier_stddev = 2. / in_dim
xavier_stddev = 1. / tf.sqrt(in_dim / 2.)
return tf.random_normal(shape=size, stddev=xavier_stddev)

``````

Discriminator

``````

In [22]:

X = tf.placeholder(tf.float32, shape=[None, 784])

D_W1 = tf.Variable(xavier_init([784, 128]))
D_b1 = tf.Variable(tf.zeros(shape=[128]))

D_W2 = tf.Variable(xavier_init([128, 1]))
D_b2 = tf.Variable(tf.zeros(shape=[1]))

theta_D = [D_W1, D_W2, D_b1, D_b2]

``````
``````

In [ ]:

def discriminator(x):
D_h1 = tf.nn.relu(tf.matmul(x, D_W1) + D_b1)
D_logit = tf.matmul(D_h1, D_W2) + D_b2
D_prob = tf.nn.sigmoid(D_logit)

return D_prob, D_logit

``````

Generator

``````

In [23]:

Z = tf.placeholder(tf.float32, shape=[None, 100])

G_W1 = tf.Variable(xavier_init([100, 128]))
G_b1 = tf.Variable(tf.zeros(shape=[128]))

G_W2 = tf.Variable(xavier_init([128, 784]))
G_b2 = tf.Variable(tf.zeros(shape=[784]))

theta_G = [G_W1, G_W2, G_b1, G_b2]

``````
``````

In [24]:

def sample_Z(m, n):
return np.random.uniform(-1., 1., size=[m, n])

``````
``````

In [25]:

def generator(z):
G_h1 = tf.nn.relu(tf.matmul(z, G_W1) + G_b1)
G_log_prob = tf.matmul(G_h1, G_W2) + G_b2
G_prob = tf.nn.sigmoid(G_log_prob)

return G_prob

``````
``````

In [27]:

def plot(samples):
fig = plt.figure(figsize=(4, 4))
gs = gridspec.GridSpec(4, 4)
gs.update(wspace=0.05, hspace=0.05)

for i, sample in enumerate(samples):
ax = plt.subplot(gs[i])
plt.axis('off')
ax.set_xticklabels([])
ax.set_yticklabels([])
ax.set_aspect('equal')
plt.imshow(sample.reshape(28, 28), cmap='Greys_r')

return fig

``````
``````

In [28]:

G_sample = generator(Z)
D_real, D_logit_real = discriminator(X)
D_fake, D_logit_fake = discriminator(G_sample)

``````
``````

In [29]:

D_loss_real = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=D_logit_real, labels=tf.ones_like(D_logit_real)))
D_loss_fake = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=D_logit_fake, labels=tf.zeros_like(D_logit_fake)))
D_loss = D_loss_real + D_loss_fake
G_loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=D_logit_fake, labels=tf.ones_like(D_logit_fake)))

``````
``````

In [30]:

``````
``````

In [31]:

minibatch_size = 128
Z_dim = 100

sess = tf.Session()
sess.run(tf.global_variables_initializer())

if not os.path.exists('out/'):
os.makedirs('out/')

i = 0

for it in range(1000000):
if it % 1000 == 0:
samples = sess.run(G_sample, feed_dict={Z: sample_Z(16, Z_dim)})

fig = plot(samples)
plt.savefig('out/{}.png'.format(str(i).zfill(3)), bbox_inches='tight')
i += 1
plt.close(fig)

X_mb, _ = mnist.train.next_batch(minibatch_size)

_, D_loss_curr = sess.run([D_solver, D_loss], feed_dict={X: X_mb, Z: sample_Z(minibatch_size, Z_dim)})
_, G_loss_curr = sess.run([G_solver, G_loss], feed_dict={Z: sample_Z(minibatch_size, Z_dim)})

if it % 1000 == 0:
print('Iter: {}'.format(it))
print('D loss: {:.4}'. format(D_loss_curr))
print('G_loss: {:.4}'.format(G_loss_curr))
print()

``````
``````

Extracting ../../MNIST_data\train-images-idx3-ubyte.gz
Extracting ../../MNIST_data\train-labels-idx1-ubyte.gz
Extracting ../../MNIST_data\t10k-images-idx3-ubyte.gz
Extracting ../../MNIST_data\t10k-labels-idx1-ubyte.gz
Iter: 0
D loss: 1.793
G_loss: 1.83

Iter: 1000
D loss: 0.005425
G_loss: 9.048

Iter: 2000
D loss: 0.02739
G_loss: 6.6

Iter: 3000
D loss: 0.03566
G_loss: 5.768

Iter: 4000
D loss: 0.1334
G_loss: 5.255

Iter: 5000
D loss: 0.187
G_loss: 4.526

Iter: 6000
D loss: 0.3584
G_loss: 3.865

Iter: 7000
D loss: 0.3289
G_loss: 4.048

Iter: 8000
D loss: 0.4025
G_loss: 2.979

Iter: 9000
D loss: 0.5652
G_loss: 3.069

Iter: 10000
D loss: 0.5653
G_loss: 3.092

Iter: 11000
D loss: 0.6922
G_loss: 2.194

Iter: 12000
D loss: 0.7456
G_loss: 2.625

Iter: 13000
D loss: 0.548
G_loss: 2.864

Iter: 14000
D loss: 0.5857
G_loss: 2.317

Iter: 15000
D loss: 0.5944
G_loss: 2.065

Iter: 16000
D loss: 0.6969
G_loss: 2.408

Iter: 17000
D loss: 0.5929
G_loss: 2.317

Iter: 18000
D loss: 0.7433
G_loss: 2.224

Iter: 19000
D loss: 0.6149
G_loss: 2.181

Iter: 20000
D loss: 0.6853
G_loss: 1.896

Iter: 21000
D loss: 0.6543
G_loss: 2.018

Iter: 22000
D loss: 0.7029
G_loss: 2.086

Iter: 23000
D loss: 0.7259
G_loss: 2.199

Iter: 24000
D loss: 0.7592
G_loss: 2.145

Iter: 25000
D loss: 0.6902
G_loss: 2.426

Iter: 26000
D loss: 0.5449
G_loss: 2.299

Iter: 27000
D loss: 0.5574
G_loss: 1.886

Iter: 28000
D loss: 0.6411
G_loss: 2.194

Iter: 29000
D loss: 0.7282
G_loss: 2.292

Iter: 30000
D loss: 0.6845
G_loss: 2.092

Iter: 31000
D loss: 0.6813
G_loss: 1.766

Iter: 32000
D loss: 0.7369
G_loss: 2.322

Iter: 33000
D loss: 0.7042
G_loss: 2.478

Iter: 34000
D loss: 0.6658
G_loss: 1.995

Iter: 35000
D loss: 0.7435
G_loss: 2.404

Iter: 36000
D loss: 0.6986
G_loss: 2.427

Iter: 37000
D loss: 0.6231
G_loss: 2.3

Iter: 38000
D loss: 0.5674
G_loss: 2.272

Iter: 39000
D loss: 0.7193
G_loss: 2.466

Iter: 40000
D loss: 0.7521
G_loss: 2.141

Iter: 41000
D loss: 0.6051
G_loss: 2.856

Iter: 42000
D loss: 0.6131
G_loss: 2.263

Iter: 43000
D loss: 0.6125
G_loss: 2.505

Iter: 44000
D loss: 0.6065
G_loss: 2.543

Iter: 45000
D loss: 0.5075
G_loss: 2.408

Iter: 46000
D loss: 0.6732
G_loss: 2.022

Iter: 47000
D loss: 0.7537
G_loss: 2.522

Iter: 48000
D loss: 0.5968
G_loss: 2.576

Iter: 49000
D loss: 0.5731
G_loss: 2.248

Iter: 50000
D loss: 0.5892
G_loss: 2.36

Iter: 51000
D loss: 0.6566
G_loss: 2.054

Iter: 52000
D loss: 0.589
G_loss: 2.229

Iter: 53000
D loss: 0.5453
G_loss: 2.539

Iter: 54000
D loss: 0.6377
G_loss: 2.325

Iter: 55000
D loss: 0.6443
G_loss: 2.201

Iter: 56000
D loss: 0.5691
G_loss: 2.22

Iter: 57000
D loss: 0.602
G_loss: 2.236

Iter: 58000
D loss: 0.6328
G_loss: 2.221

Iter: 59000
D loss: 0.56
G_loss: 2.09

Iter: 60000
D loss: 0.6583
G_loss: 2.434

Iter: 61000
D loss: 0.5936
G_loss: 2.688

Iter: 62000
D loss: 0.6583
G_loss: 2.722

Iter: 63000
D loss: 0.6678
G_loss: 2.077

Iter: 64000
D loss: 0.61
G_loss: 2.343

Iter: 65000
D loss: 0.7147
G_loss: 2.032

Iter: 66000
D loss: 0.556
G_loss: 2.439

Iter: 67000
D loss: 0.502
G_loss: 2.473

Iter: 68000
D loss: 0.6628
G_loss: 2.285

Iter: 69000
D loss: 0.6636
G_loss: 2.367

Iter: 70000
D loss: 0.6008
G_loss: 2.528

Iter: 71000
D loss: 0.5741
G_loss: 2.275

Iter: 72000
D loss: 0.6757
G_loss: 2.364

Iter: 73000
D loss: 0.5951
G_loss: 2.277

Iter: 74000
D loss: 0.6446
G_loss: 2.212

Iter: 75000
D loss: 0.5831
G_loss: 2.549

Iter: 76000
D loss: 0.5225
G_loss: 2.821

Iter: 77000
D loss: 0.5037
G_loss: 2.398

Iter: 78000
D loss: 0.5588
G_loss: 2.748

Iter: 79000
D loss: 0.5139
G_loss: 2.05

Iter: 80000
D loss: 0.6058
G_loss: 2.121

Iter: 81000
D loss: 0.5188
G_loss: 2.151

Iter: 82000
D loss: 0.7628
G_loss: 2.365

Iter: 83000
D loss: 0.5969
G_loss: 2.225

Iter: 84000
D loss: 0.487
G_loss: 2.248

Iter: 85000
D loss: 0.5701
G_loss: 2.347

Iter: 86000
D loss: 0.5312
G_loss: 2.145

Iter: 87000
D loss: 0.6747
G_loss: 2.319

Iter: 88000
D loss: 0.6919
G_loss: 2.565

Iter: 89000
D loss: 0.4909
G_loss: 2.332

Iter: 90000
D loss: 0.6213
G_loss: 2.755

Iter: 91000
D loss: 0.4686
G_loss: 2.596

Iter: 92000
D loss: 0.5075
G_loss: 2.496

Iter: 93000
D loss: 0.6059
G_loss: 2.547

Iter: 94000
D loss: 0.5324
G_loss: 2.446

Iter: 95000
D loss: 0.5849
G_loss: 2.423

Iter: 96000
D loss: 0.5195
G_loss: 2.336

Iter: 97000
D loss: 0.6016
G_loss: 2.444

Iter: 98000
D loss: 0.6684
G_loss: 2.517

Iter: 99000
D loss: 0.5965
G_loss: 2.357

Iter: 100000
D loss: 0.6298
G_loss: 2.45

Iter: 101000
D loss: 0.4503
G_loss: 3.148

Iter: 102000
D loss: 0.5333
G_loss: 2.653

Iter: 103000
D loss: 0.5518
G_loss: 2.494

Iter: 104000
D loss: 0.4912
G_loss: 2.504

Iter: 105000
D loss: 0.579
G_loss: 2.756

Iter: 106000
D loss: 0.4961
G_loss: 2.716

Iter: 107000
D loss: 0.4284
G_loss: 2.555

Iter: 108000
D loss: 0.446
G_loss: 2.709

Iter: 109000
D loss: 0.5174
G_loss: 2.49

Iter: 110000
D loss: 0.5993
G_loss: 2.591

Iter: 111000
D loss: 0.5951
G_loss: 2.631

Iter: 112000
D loss: 0.6414
G_loss: 2.527

Iter: 113000
D loss: 0.4856
G_loss: 2.783

Iter: 114000
D loss: 0.6077
G_loss: 2.436

Iter: 115000
D loss: 0.577
G_loss: 2.643

Iter: 116000
D loss: 0.6172
G_loss: 2.64

Iter: 117000
D loss: 0.5762
G_loss: 2.6

Iter: 118000
D loss: 0.5079
G_loss: 2.45

Iter: 119000
D loss: 0.4511
G_loss: 2.324

Iter: 120000
D loss: 0.3917
G_loss: 2.298

Iter: 121000
D loss: 0.4767
G_loss: 2.362

Iter: 122000
D loss: 0.4181
G_loss: 2.801

Iter: 123000
D loss: 0.4134
G_loss: 2.85

Iter: 124000
D loss: 0.5061
G_loss: 2.628

Iter: 125000
D loss: 0.5514
G_loss: 2.833

Iter: 126000
D loss: 0.5902
G_loss: 2.766

Iter: 127000
D loss: 0.4254
G_loss: 2.622

Iter: 128000
D loss: 0.4781
G_loss: 2.653

Iter: 129000
D loss: 0.5029
G_loss: 2.668

Iter: 130000
D loss: 0.5111
G_loss: 2.386

Iter: 131000
D loss: 0.5245
G_loss: 2.286

Iter: 132000
D loss: 0.4558
G_loss: 2.686

Iter: 133000
D loss: 0.4073
G_loss: 2.605

Iter: 134000
D loss: 0.4317
G_loss: 2.934

Iter: 135000
D loss: 0.508
G_loss: 2.815

Iter: 136000
D loss: 0.5047
G_loss: 3.007

Iter: 137000
D loss: 0.4874
G_loss: 2.655

Iter: 138000
D loss: 0.4133
G_loss: 2.899

Iter: 139000
D loss: 0.5487
G_loss: 2.46

Iter: 140000
D loss: 0.4515
G_loss: 2.761

Iter: 141000
D loss: 0.4463
G_loss: 2.691

Iter: 142000
D loss: 0.5048
G_loss: 2.679

Iter: 143000
D loss: 0.5459
G_loss: 3.019

Iter: 144000
D loss: 0.5126
G_loss: 2.533

Iter: 145000
D loss: 0.5
G_loss: 2.769

Iter: 146000
D loss: 0.4804
G_loss: 2.554

Iter: 147000
D loss: 0.4787
G_loss: 2.755

Iter: 148000
D loss: 0.4578
G_loss: 2.341

Iter: 149000
D loss: 0.4884
G_loss: 2.596

Iter: 150000
D loss: 0.4149
G_loss: 2.711

Iter: 151000
D loss: 0.4372
G_loss: 2.75

Iter: 152000
D loss: 0.3478
G_loss: 2.725

Iter: 153000
D loss: 0.5213
G_loss: 2.754

Iter: 154000
D loss: 0.4596
G_loss: 2.627

Iter: 155000
D loss: 0.5945
G_loss: 2.894

Iter: 156000
D loss: 0.4364
G_loss: 2.734

Iter: 157000
D loss: 0.4627
G_loss: 2.974

Iter: 158000
D loss: 0.5497
G_loss: 2.57

Iter: 159000
D loss: 0.405
G_loss: 2.781

Iter: 160000
D loss: 0.4347
G_loss: 2.784

Iter: 161000
D loss: 0.5982
G_loss: 2.343

Iter: 162000
D loss: 0.4696
G_loss: 2.646

Iter: 163000
D loss: 0.4753
G_loss: 2.665

Iter: 164000
D loss: 0.3958
G_loss: 2.712

Iter: 165000
D loss: 0.4392
G_loss: 2.772

Iter: 166000
D loss: 0.4629
G_loss: 2.808

Iter: 167000
D loss: 0.505
G_loss: 3.164

Iter: 168000
D loss: 0.4861
G_loss: 2.654

Iter: 169000
D loss: 0.4868
G_loss: 3.029

Iter: 170000
D loss: 0.4686
G_loss: 3.031

Iter: 171000
D loss: 0.4126
G_loss: 2.924

Iter: 172000
D loss: 0.4418
G_loss: 3.02

Iter: 173000
D loss: 0.48
G_loss: 2.527

Iter: 174000
D loss: 0.5016
G_loss: 2.885

Iter: 175000
D loss: 0.4959
G_loss: 2.652

Iter: 176000
D loss: 0.4895
G_loss: 3.068

Iter: 177000
D loss: 0.5014
G_loss: 2.874

Iter: 178000
D loss: 0.35
G_loss: 2.54

Iter: 179000
D loss: 0.4125
G_loss: 2.67

Iter: 180000
D loss: 0.4413
G_loss: 2.825

Iter: 181000
D loss: 0.4834
G_loss: 2.992

Iter: 182000
D loss: 0.4504
G_loss: 2.994

Iter: 183000
D loss: 0.4806
G_loss: 2.801

Iter: 184000
D loss: 0.4002
G_loss: 3.03

Iter: 185000
D loss: 0.5614
G_loss: 2.813

Iter: 186000
D loss: 0.4339
G_loss: 2.695

Iter: 187000
D loss: 0.4366
G_loss: 3.178

Iter: 188000
D loss: 0.4254
G_loss: 2.8

Iter: 189000
D loss: 0.4011
G_loss: 2.805

Iter: 190000
D loss: 0.3461
G_loss: 3.089

Iter: 191000
D loss: 0.4336
G_loss: 2.698

Iter: 192000
D loss: 0.381
G_loss: 2.889

Iter: 193000
D loss: 0.381
G_loss: 2.839

Iter: 194000
D loss: 0.4994
G_loss: 2.565

Iter: 195000
D loss: 0.3866
G_loss: 3.228

Iter: 196000
D loss: 0.447
G_loss: 2.54

Iter: 197000
D loss: 0.4347
G_loss: 2.974

Iter: 198000
D loss: 0.4382
G_loss: 2.952

Iter: 199000
D loss: 0.4991
G_loss: 2.791

Iter: 200000
D loss: 0.4671
G_loss: 2.857

Iter: 201000
D loss: 0.4219
G_loss: 3.127

Iter: 202000
D loss: 0.4649
G_loss: 2.819

Iter: 203000
D loss: 0.4194
G_loss: 2.831

Iter: 204000
D loss: 0.3203
G_loss: 3.034

Iter: 205000
D loss: 0.4903
G_loss: 2.609

Iter: 206000
D loss: 0.4918
G_loss: 3.091

Iter: 207000
D loss: 0.3992
G_loss: 3.403

Iter: 208000
D loss: 0.5082
G_loss: 2.679

Iter: 209000
D loss: 0.3978
G_loss: 2.865

Iter: 210000
D loss: 0.4769
G_loss: 3.038

Iter: 211000
D loss: 0.4109
G_loss: 2.454

Iter: 212000
D loss: 0.4438
G_loss: 3.288

Iter: 213000
D loss: 0.382
G_loss: 3.286

Iter: 214000
D loss: 0.3667
G_loss: 3.144

Iter: 215000
D loss: 0.4312
G_loss: 2.92

Iter: 216000
D loss: 0.3487
G_loss: 2.813

Iter: 217000
D loss: 0.3618
G_loss: 2.962

Iter: 218000
D loss: 0.4617
G_loss: 2.901

Iter: 219000
D loss: 0.3277
G_loss: 3.012

Iter: 220000
D loss: 0.3814
G_loss: 3.078

Iter: 221000
D loss: 0.4956
G_loss: 2.749

Iter: 222000
D loss: 0.3789
G_loss: 3.114

Iter: 223000
D loss: 0.3688
G_loss: 3.089

Iter: 224000
D loss: 0.364
G_loss: 3.544

Iter: 225000
D loss: 0.387
G_loss: 2.99

Iter: 226000
D loss: 0.5714
G_loss: 3.223

Iter: 227000
D loss: 0.4255
G_loss: 3.173

Iter: 228000
D loss: 0.3271
G_loss: 2.964

Iter: 229000
D loss: 0.4663
G_loss: 3.241

Iter: 230000
D loss: 0.3565
G_loss: 3.483

Iter: 231000
D loss: 0.4669
G_loss: 3.364

Iter: 232000
D loss: 0.4331
G_loss: 3.078

Iter: 233000
D loss: 0.428
G_loss: 3.239

Iter: 234000
D loss: 0.4664
G_loss: 2.805

Iter: 235000
D loss: 0.5369
G_loss: 2.724

Iter: 236000
D loss: 0.3108
G_loss: 3.1

Iter: 237000
D loss: 0.3192
G_loss: 2.989

Iter: 238000
D loss: 0.4172
G_loss: 2.779

Iter: 239000
D loss: 0.3023
G_loss: 3.009

Iter: 240000
D loss: 0.5114
G_loss: 2.765

Iter: 241000
D loss: 0.4642
G_loss: 2.732

Iter: 242000
D loss: 0.4063
G_loss: 3.368

Iter: 243000
D loss: 0.4061
G_loss: 3.577

Iter: 244000
D loss: 0.3677
G_loss: 2.726

Iter: 245000
D loss: 0.3235
G_loss: 3.061

Iter: 246000
D loss: 0.301
G_loss: 3.029

Iter: 247000
D loss: 0.3593
G_loss: 3.135

Iter: 248000
D loss: 0.293
G_loss: 3.179

Iter: 249000
D loss: 0.3831
G_loss: 3.071

Iter: 250000
D loss: 0.3155
G_loss: 2.937

Iter: 251000
D loss: 0.4003
G_loss: 3.108

Iter: 252000
D loss: 0.3365
G_loss: 3.16

Iter: 253000
D loss: 0.4098
G_loss: 3.086

Iter: 254000
D loss: 0.4333
G_loss: 3.225

Iter: 255000
D loss: 0.333
G_loss: 2.78

Iter: 256000
D loss: 0.4532
G_loss: 3.486

Iter: 257000
D loss: 0.3673
G_loss: 2.995

Iter: 258000
D loss: 0.3818
G_loss: 3.306

Iter: 259000
D loss: 0.356
G_loss: 3.279

Iter: 260000
D loss: 0.3391
G_loss: 3.382

Iter: 261000
D loss: 0.3897
G_loss: 3.427

Iter: 262000
D loss: 0.3828
G_loss: 3.245

Iter: 263000
D loss: 0.3849
G_loss: 2.89

Iter: 264000
D loss: 0.4077
G_loss: 3.417

Iter: 265000
D loss: 0.4809
G_loss: 3.087

Iter: 266000
D loss: 0.3899
G_loss: 2.998

Iter: 267000
D loss: 0.3641
G_loss: 3.366

Iter: 268000
D loss: 0.4158
G_loss: 2.755

Iter: 269000
D loss: 0.2445
G_loss: 2.842

Iter: 270000
D loss: 0.4695
G_loss: 3.269

Iter: 271000
D loss: 0.4519
G_loss: 3.13

Iter: 272000
D loss: 0.2755
G_loss: 3.237

Iter: 273000
D loss: 0.3079
G_loss: 3.524

Iter: 274000
D loss: 0.3466
G_loss: 2.845

Iter: 275000
D loss: 0.3839
G_loss: 3.253

Iter: 276000
D loss: 0.3651
G_loss: 2.925

Iter: 277000
D loss: 0.3907
G_loss: 3.137

Iter: 278000
D loss: 0.4181
G_loss: 3.165

Iter: 279000
D loss: 0.3256
G_loss: 2.948

Iter: 280000
D loss: 0.3061
G_loss: 4.039

Iter: 281000
D loss: 0.305
G_loss: 3.002

Iter: 282000
D loss: 0.4027
G_loss: 3.146

Iter: 283000
D loss: 0.2988
G_loss: 3.43

Iter: 284000
D loss: 0.3738
G_loss: 2.781

Iter: 285000
D loss: 0.4318
G_loss: 3.143

Iter: 286000
D loss: 0.3674
G_loss: 2.801

Iter: 287000
D loss: 0.3956
G_loss: 3.678

Iter: 288000
D loss: 0.3356
G_loss: 2.799

Iter: 289000
D loss: 0.3824
G_loss: 2.962

Iter: 290000
D loss: 0.3466
G_loss: 3.62

Iter: 291000
D loss: 0.3513
G_loss: 3.239

Iter: 292000
D loss: 0.5429
G_loss: 3.389

Iter: 293000
D loss: 0.3016
G_loss: 3.013

Iter: 294000
D loss: 0.3698
G_loss: 2.878

Iter: 295000
D loss: 0.3039
G_loss: 3.466

Iter: 296000
D loss: 0.2872
G_loss: 2.83

Iter: 297000
D loss: 0.3303
G_loss: 3.706

Iter: 298000
D loss: 0.3501
G_loss: 3.022

Iter: 299000
D loss: 0.3423
G_loss: 3.079

Iter: 300000
D loss: 0.4542
G_loss: 3.0

Iter: 301000
D loss: 0.3665
G_loss: 2.957

Iter: 302000
D loss: 0.3652
G_loss: 3.14

Iter: 303000
D loss: 0.3165
G_loss: 3.242

Iter: 304000
D loss: 0.4046
G_loss: 3.293

Iter: 305000
D loss: 0.3454
G_loss: 3.059

Iter: 306000
D loss: 0.3747
G_loss: 3.012

Iter: 307000
D loss: 0.5227
G_loss: 3.177

Iter: 308000
D loss: 0.3386
G_loss: 3.289

Iter: 309000
D loss: 0.2649
G_loss: 3.056

Iter: 310000
D loss: 0.3343
G_loss: 3.018

Iter: 311000
D loss: 0.4001
G_loss: 3.165

Iter: 312000
D loss: 0.4011
G_loss: 3.068

Iter: 313000
D loss: 0.3795
G_loss: 3.351

Iter: 314000
D loss: 0.2492
G_loss: 3.31

Iter: 315000
D loss: 0.3929
G_loss: 3.387

Iter: 316000
D loss: 0.38
G_loss: 2.997

Iter: 317000
D loss: 0.3537
G_loss: 3.136

Iter: 318000
D loss: 0.4273
G_loss: 3.238

Iter: 319000
D loss: 0.2994
G_loss: 3.456

Iter: 320000
D loss: 0.2244
G_loss: 3.332

Iter: 321000
D loss: 0.2255
G_loss: 3.198

Iter: 322000
D loss: 0.3715
G_loss: 3.475

Iter: 323000
D loss: 0.3375
G_loss: 3.011

Iter: 324000
D loss: 0.2819
G_loss: 3.49

Iter: 325000
D loss: 0.3251
G_loss: 3.102

Iter: 326000
D loss: 0.3892
G_loss: 3.256

Iter: 327000
D loss: 0.336
G_loss: 3.006

Iter: 328000
D loss: 0.4611
G_loss: 3.238

Iter: 329000
D loss: 0.2983
G_loss: 3.261

Iter: 330000
D loss: 0.3081
G_loss: 3.04

Iter: 331000
D loss: 0.3848
G_loss: 3.396

Iter: 332000
D loss: 0.2836
G_loss: 2.988

Iter: 333000
D loss: 0.2016
G_loss: 3.377

Iter: 334000
D loss: 0.4176
G_loss: 3.456

Iter: 335000
D loss: 0.3413
G_loss: 3.269

Iter: 336000
D loss: 0.2577
G_loss: 3.358

Iter: 337000
D loss: 0.3024
G_loss: 3.003

Iter: 338000
D loss: 0.3703
G_loss: 3.134

Iter: 339000
D loss: 0.2748
G_loss: 3.308

Iter: 340000
D loss: 0.3149
G_loss: 3.21

Iter: 341000
D loss: 0.3277
G_loss: 3.18

Iter: 342000
D loss: 0.403
G_loss: 3.111

Iter: 343000
D loss: 0.2745
G_loss: 3.472

Iter: 344000
D loss: 0.3921
G_loss: 3.112

Iter: 345000
D loss: 0.2886
G_loss: 3.008

Iter: 346000
D loss: 0.3231
G_loss: 2.888

Iter: 347000
D loss: 0.3216
G_loss: 3.739

Iter: 348000
D loss: 0.3873
G_loss: 3.089

Iter: 349000
D loss: 0.2018
G_loss: 3.001

Iter: 350000
D loss: 0.2804
G_loss: 2.948

Iter: 351000
D loss: 0.2807
G_loss: 3.307

Iter: 352000
D loss: 0.3033
G_loss: 3.028

Iter: 353000
D loss: 0.3058
G_loss: 3.403

Iter: 354000
D loss: 0.3413
G_loss: 2.627

Iter: 355000
D loss: 0.2357
G_loss: 3.314

Iter: 356000
D loss: 0.3531
G_loss: 3.307

Iter: 357000
D loss: 0.4806
G_loss: 3.368

Iter: 358000
D loss: 0.2333
G_loss: 3.305

Iter: 359000
D loss: 0.322
G_loss: 3.706

Iter: 360000
D loss: 0.3056
G_loss: 3.478

Iter: 361000
D loss: 0.3109
G_loss: 3.146

Iter: 362000
D loss: 0.3134
G_loss: 3.193

Iter: 363000
D loss: 0.3746
G_loss: 2.986

Iter: 364000
D loss: 0.3331
G_loss: 3.636

Iter: 365000
D loss: 0.3034
G_loss: 2.914

Iter: 366000
D loss: 0.2778
G_loss: 3.322

Iter: 367000
D loss: 0.3858
G_loss: 2.984

Iter: 368000
D loss: 0.1867
G_loss: 3.19

Iter: 369000
D loss: 0.1791
G_loss: 3.16

Iter: 370000
D loss: 0.3102
G_loss: 3.123

Iter: 371000
D loss: 0.2869
G_loss: 3.276

Iter: 372000
D loss: 0.2736
G_loss: 3.05

Iter: 373000
D loss: 0.2827
G_loss: 3.017

Iter: 374000
D loss: 0.318
G_loss: 3.859

Iter: 375000
D loss: 0.2506
G_loss: 3.327

Iter: 376000
D loss: 0.2926
G_loss: 3.402

Iter: 377000
D loss: 0.3579
G_loss: 3.027

Iter: 378000
D loss: 0.3307
G_loss: 3.286

Iter: 379000
D loss: 0.3569
G_loss: 3.323

Iter: 380000
D loss: 0.3769
G_loss: 3.176

Iter: 381000
D loss: 0.2395
G_loss: 3.081

Iter: 382000
D loss: 0.3862
G_loss: 3.261

Iter: 383000
D loss: 0.3951
G_loss: 3.144

Iter: 384000
D loss: 0.176
G_loss: 3.684

Iter: 385000
D loss: 0.3269
G_loss: 3.029

Iter: 386000
D loss: 0.291
G_loss: 3.464

Iter: 387000
D loss: 0.1834
G_loss: 3.629

Iter: 388000
D loss: 0.1598
G_loss: 3.496

Iter: 389000
D loss: 0.2546
G_loss: 3.611

Iter: 390000
D loss: 0.3678
G_loss: 2.992

Iter: 391000
D loss: 0.1874
G_loss: 3.452

Iter: 392000
D loss: 0.2704
G_loss: 2.993

Iter: 393000
D loss: 0.2211
G_loss: 3.341

Iter: 394000
D loss: 0.3005
G_loss: 3.172

Iter: 395000
D loss: 0.3762
G_loss: 3.89

Iter: 396000
D loss: 0.3607
G_loss: 2.76

Iter: 397000
D loss: 0.3465
G_loss: 3.364

Iter: 398000
D loss: 0.2993
G_loss: 3.222

Iter: 399000
D loss: 0.3102
G_loss: 2.914

Iter: 400000
D loss: 0.2325
G_loss: 3.106

Iter: 401000
D loss: 0.2365
G_loss: 3.179

Iter: 402000
D loss: 0.4196
G_loss: 3.185

Iter: 403000
D loss: 0.3064
G_loss: 2.886

Iter: 404000
D loss: 0.2539
G_loss: 2.939

Iter: 405000
D loss: 0.2577
G_loss: 2.704

Iter: 406000
D loss: 0.3115
G_loss: 2.677

Iter: 407000
D loss: 0.3984
G_loss: 3.005

Iter: 408000
D loss: 0.3035
G_loss: 2.964

Iter: 409000
D loss: 0.2121
G_loss: 3.547

Iter: 410000
D loss: 0.263
G_loss: 3.18

Iter: 411000
D loss: 0.3651
G_loss: 2.82

Iter: 412000
D loss: 0.35
G_loss: 2.883

Iter: 413000
D loss: 0.2087
G_loss: 3.58

Iter: 414000
D loss: 0.2967
G_loss: 3.238

Iter: 415000
D loss: 0.2854
G_loss: 3.156

Iter: 416000
D loss: 0.2597
G_loss: 3.03

Iter: 417000
D loss: 0.4599
G_loss: 3.378

Iter: 418000
D loss: 0.2984
G_loss: 3.017

Iter: 419000
D loss: 0.2905
G_loss: 3.228

Iter: 420000
D loss: 0.2628
G_loss: 3.015

Iter: 421000
D loss: 0.3861
G_loss: 3.387

Iter: 422000
D loss: 0.2089
G_loss: 3.162

Iter: 423000
D loss: 0.3006
G_loss: 3.092

Iter: 424000
D loss: 0.3272
G_loss: 3.545

Iter: 425000
D loss: 0.363
G_loss: 3.811

Iter: 426000
D loss: 0.2468
G_loss: 3.221

Iter: 427000
D loss: 0.3962
G_loss: 2.969

Iter: 428000
D loss: 0.3004
G_loss: 3.143

Iter: 429000
D loss: 0.3746
G_loss: 3.194

Iter: 430000
D loss: 0.333
G_loss: 3.308

Iter: 431000
D loss: 0.2317
G_loss: 3.047

Iter: 432000
D loss: 0.2104
G_loss: 3.217

Iter: 433000
D loss: 0.1962
G_loss: 3.647

Iter: 434000
D loss: 0.3435
G_loss: 3.281

Iter: 435000
D loss: 0.2551
G_loss: 2.988

Iter: 436000
D loss: 0.337
G_loss: 3.029

Iter: 437000
D loss: 0.1609
G_loss: 3.134

Iter: 438000
D loss: 0.2578
G_loss: 2.899

Iter: 439000
D loss: 0.2843
G_loss: 2.702

Iter: 440000
D loss: 0.293
G_loss: 3.033

Iter: 441000
D loss: 0.265
G_loss: 2.597

Iter: 442000
D loss: 0.4579
G_loss: 3.589

Iter: 443000
D loss: 0.2451
G_loss: 3.038

Iter: 444000
D loss: 0.3336
G_loss: 3.208

Iter: 445000
D loss: 0.3487
G_loss: 3.328

Iter: 446000
D loss: 0.5067
G_loss: 3.253

Iter: 447000
D loss: 0.1649
G_loss: 3.214

Iter: 448000
D loss: 0.4033
G_loss: 2.837

Iter: 449000
D loss: 0.2792
G_loss: 3.366

Iter: 450000
D loss: 0.271
G_loss: 3.193

Iter: 451000
D loss: 0.2621
G_loss: 2.704

Iter: 452000
D loss: 0.2495
G_loss: 3.322

Iter: 453000
D loss: 0.3205
G_loss: 2.978

Iter: 454000
D loss: 0.2242
G_loss: 2.857

Iter: 455000
D loss: 0.2252
G_loss: 3.065

Iter: 456000
D loss: 0.1806
G_loss: 3.589

Iter: 457000
D loss: 0.2192
G_loss: 3.043

Iter: 458000
D loss: 0.1946
G_loss: 3.391

Iter: 459000
D loss: 0.2419
G_loss: 3.143

Iter: 460000
D loss: 0.2828
G_loss: 2.794

Iter: 461000
D loss: 0.3025
G_loss: 3.165

Iter: 462000
D loss: 0.2675
G_loss: 3.058

Iter: 463000
D loss: 0.2117
G_loss: 3.261

Iter: 464000
D loss: 0.2925
G_loss: 3.166

Iter: 465000
D loss: 0.246
G_loss: 3.14

Iter: 466000
D loss: 0.2852
G_loss: 2.742

Iter: 467000
D loss: 0.2286
G_loss: 2.91

Iter: 468000
D loss: 0.2499
G_loss: 2.962

Iter: 469000
D loss: 0.3392
G_loss: 3.03

Iter: 470000
D loss: 0.2059
G_loss: 3.146

Iter: 471000
D loss: 0.2649
G_loss: 3.241

Iter: 472000
D loss: 0.3122
G_loss: 3.089

Iter: 473000
D loss: 0.2567
G_loss: 2.906

Iter: 474000
D loss: 0.3858
G_loss: 3.506

Iter: 475000
D loss: 0.2071
G_loss: 2.891

Iter: 476000
D loss: 0.2542
G_loss: 3.071

Iter: 477000
D loss: 0.1394
G_loss: 3.308

Iter: 478000
D loss: 0.2868
G_loss: 2.84

Iter: 479000
D loss: 0.2363
G_loss: 3.231

Iter: 480000
D loss: 0.2698
G_loss: 3.492

Iter: 481000
D loss: 0.2451
G_loss: 3.359

Iter: 482000
D loss: 0.2398
G_loss: 2.967

Iter: 483000
D loss: 0.3568
G_loss: 2.923

Iter: 484000
D loss: 0.2344
G_loss: 3.13

Iter: 485000
D loss: 0.2987
G_loss: 3.311

Iter: 486000
D loss: 0.3293
G_loss: 3.112

Iter: 487000
D loss: 0.2234
G_loss: 2.933

Iter: 488000
D loss: 0.257
G_loss: 2.702

Iter: 489000
D loss: 0.2386
G_loss: 3.006

Iter: 490000
D loss: 0.3294
G_loss: 2.926

Iter: 491000
D loss: 0.214
G_loss: 3.235

Iter: 492000
D loss: 0.1553
G_loss: 3.225

Iter: 493000
D loss: 0.3075
G_loss: 3.016

Iter: 494000
D loss: 0.2062
G_loss: 2.895

Iter: 495000
D loss: 0.2888
G_loss: 3.411

Iter: 496000
D loss: 0.3759
G_loss: 3.295

Iter: 497000
D loss: 0.2379
G_loss: 3.115

Iter: 498000
D loss: 0.2327
G_loss: 3.145

Iter: 499000
D loss: 0.1641
G_loss: 3.333

Iter: 500000
D loss: 0.3464
G_loss: 3.023

Iter: 501000
D loss: 0.351
G_loss: 3.046

Iter: 502000
D loss: 0.1949
G_loss: 3.064

Iter: 503000
D loss: 0.2477
G_loss: 2.805

Iter: 504000
D loss: 0.3087
G_loss: 2.999

Iter: 505000
D loss: 0.2876
G_loss: 3.247

Iter: 506000
D loss: 0.2956
G_loss: 3.54

Iter: 507000
D loss: 0.1908
G_loss: 3.128

Iter: 508000
D loss: 0.2727
G_loss: 3.257

Iter: 509000
D loss: 0.2493
G_loss: 3.336

Iter: 510000
D loss: 0.2053
G_loss: 2.942

Iter: 511000
D loss: 0.2437
G_loss: 3.12

Iter: 512000
D loss: 0.2441
G_loss: 3.72

Iter: 513000
D loss: 0.4195
G_loss: 3.333

Iter: 514000
D loss: 0.2309
G_loss: 3.235

Iter: 515000
D loss: 0.2438
G_loss: 3.027

Iter: 516000
D loss: 0.3388
G_loss: 3.31

Iter: 517000
D loss: 0.2132
G_loss: 3.391

Iter: 518000
D loss: 0.1517
G_loss: 3.278

Iter: 519000
D loss: 0.2011
G_loss: 2.941

Iter: 520000
D loss: 0.1803
G_loss: 3.161

Iter: 521000
D loss: 0.3158
G_loss: 3.324

Iter: 522000
D loss: 0.2949
G_loss: 3.266

Iter: 523000
D loss: 0.2309
G_loss: 3.345

Iter: 524000
D loss: 0.1708
G_loss: 3.379

Iter: 525000
D loss: 0.5243
G_loss: 3.528

Iter: 526000
D loss: 0.2673
G_loss: 3.42

Iter: 527000
D loss: 0.2467
G_loss: 3.324

Iter: 528000
D loss: 0.2718
G_loss: 3.086

Iter: 529000
D loss: 0.3205
G_loss: 3.274

Iter: 530000
D loss: 0.1666
G_loss: 3.284

Iter: 531000
D loss: 0.3151
G_loss: 3.166

Iter: 532000
D loss: 0.3624
G_loss: 3.184

Iter: 533000
D loss: 0.2768
G_loss: 3.106

Iter: 534000
D loss: 0.2611
G_loss: 3.609

Iter: 535000
D loss: 0.2286
G_loss: 3.443

Iter: 536000
D loss: 0.2484
G_loss: 3.019

Iter: 537000
D loss: 0.2028
G_loss: 3.218

Iter: 538000
D loss: 0.2628
G_loss: 3.047

Iter: 539000
D loss: 0.1665
G_loss: 3.296

Iter: 540000
D loss: 0.1748
G_loss: 3.194

Iter: 541000
D loss: 0.2464
G_loss: 3.36

Iter: 542000
D loss: 0.2441
G_loss: 3.142

Iter: 543000
D loss: 0.337
G_loss: 2.824

Iter: 544000
D loss: 0.3385
G_loss: 3.636

Iter: 545000
D loss: 0.345
G_loss: 3.133

Iter: 546000
D loss: 0.1984
G_loss: 3.471

Iter: 547000
D loss: 0.2089
G_loss: 2.936

Iter: 548000
D loss: 0.1675
G_loss: 3.513

Iter: 549000
D loss: 0.2291
G_loss: 3.593

Iter: 550000
D loss: 0.2241
G_loss: 3.013

Iter: 551000
D loss: 0.3379
G_loss: 3.109

Iter: 552000
D loss: 0.1857
G_loss: 3.621

Iter: 553000
D loss: 0.1917
G_loss: 3.526

Iter: 554000
D loss: 0.2165
G_loss: 3.21

Iter: 555000
D loss: 0.2045
G_loss: 3.21

Iter: 556000
D loss: 0.2102
G_loss: 3.147

Iter: 557000
D loss: 0.3712
G_loss: 3.232

Iter: 558000
D loss: 0.1666
G_loss: 3.375

Iter: 559000
D loss: 0.2547
G_loss: 3.285

Iter: 560000
D loss: 0.2582
G_loss: 3.059

Iter: 561000
D loss: 0.1918
G_loss: 3.057

Iter: 562000
D loss: 0.247
G_loss: 3.324

Iter: 563000
D loss: 0.1983
G_loss: 3.418

Iter: 564000
D loss: 0.2402
G_loss: 3.097

Iter: 565000
D loss: 0.2718
G_loss: 3.391

Iter: 566000
D loss: 0.2159
G_loss: 2.893

Iter: 567000
D loss: 0.2415
G_loss: 3.122

Iter: 568000
D loss: 0.2683
G_loss: 2.95

Iter: 569000
D loss: 0.2294
G_loss: 3.171

Iter: 570000
D loss: 0.2483
G_loss: 3.139

Iter: 571000
D loss: 0.3145
G_loss: 2.793

Iter: 572000
D loss: 0.2064
G_loss: 3.281

Iter: 573000
D loss: 0.2039
G_loss: 3.469

Iter: 574000
D loss: 0.1974
G_loss: 3.346

Iter: 575000
D loss: 0.2608
G_loss: 3.043

Iter: 576000
D loss: 0.2977
G_loss: 3.104

Iter: 577000
D loss: 0.1711
G_loss: 3.302

Iter: 578000
D loss: 0.225
G_loss: 3.048

Iter: 579000
D loss: 0.2431
G_loss: 2.886

Iter: 580000
D loss: 0.297
G_loss: 3.187

Iter: 581000
D loss: 0.2901
G_loss: 3.245

Iter: 582000
D loss: 0.2462
G_loss: 2.979

Iter: 583000
D loss: 0.2423
G_loss: 3.284

Iter: 584000
D loss: 0.1745
G_loss: 3.627

Iter: 585000
D loss: 0.387
G_loss: 2.883

Iter: 586000
D loss: 0.2659
G_loss: 3.171

Iter: 587000
D loss: 0.1617
G_loss: 3.538

Iter: 588000
D loss: 0.1601
G_loss: 3.284

Iter: 589000
D loss: 0.1648
G_loss: 3.14

Iter: 590000
D loss: 0.2406
G_loss: 3.588

Iter: 591000
D loss: 0.1768
G_loss: 3.539

Iter: 592000
D loss: 0.2865
G_loss: 2.963

Iter: 593000
D loss: 0.225
G_loss: 3.009

Iter: 594000
D loss: 0.2467
G_loss: 3.68

Iter: 595000
D loss: 0.3592
G_loss: 2.96

Iter: 596000
D loss: 0.2518
G_loss: 2.907

Iter: 597000
D loss: 0.366
G_loss: 3.095

Iter: 598000
D loss: 0.2125
G_loss: 2.963

Iter: 599000
D loss: 0.27
G_loss: 3.025

Iter: 600000
D loss: 0.3191
G_loss: 2.719

Iter: 601000
D loss: 0.2395
G_loss: 3.458

Iter: 602000
D loss: 0.2153
G_loss: 3.033

Iter: 603000
D loss: 0.3231
G_loss: 3.185

Iter: 604000
D loss: 0.2576
G_loss: 3.01

Iter: 605000
D loss: 0.2436
G_loss: 2.884

Iter: 606000
D loss: 0.2317
G_loss: 3.368

Iter: 607000
D loss: 0.2147
G_loss: 3.486

Iter: 608000
D loss: 0.3389
G_loss: 3.043

Iter: 609000
D loss: 0.2741
G_loss: 2.99

Iter: 610000
D loss: 0.2505
G_loss: 3.303

Iter: 611000
D loss: 0.1119
G_loss: 3.158

Iter: 612000
D loss: 0.3302
G_loss: 3.532

Iter: 613000
D loss: 0.1893
G_loss: 3.358

Iter: 614000
D loss: 0.2278
G_loss: 3.221

Iter: 615000
D loss: 0.2425
G_loss: 3.031

Iter: 616000
D loss: 0.2782
G_loss: 3.169

Iter: 617000
D loss: 0.3068
G_loss: 2.93

Iter: 618000
D loss: 0.1983
G_loss: 3.333

Iter: 619000
D loss: 0.2267
G_loss: 3.359

Iter: 620000
D loss: 0.159
G_loss: 3.562

Iter: 621000
D loss: 0.3046
G_loss: 3.14

Iter: 622000
D loss: 0.2135
G_loss: 3.3

Iter: 623000
D loss: 0.1714
G_loss: 3.352

Iter: 624000
D loss: 0.2475
G_loss: 3.278

Iter: 625000
D loss: 0.1457
G_loss: 3.2

Iter: 626000
D loss: 0.2329
G_loss: 3.238

Iter: 627000
D loss: 0.3304
G_loss: 3.297

Iter: 628000
D loss: 0.187
G_loss: 3.164

Iter: 629000
D loss: 0.2419
G_loss: 3.126

Iter: 630000
D loss: 0.27
G_loss: 3.191

Iter: 631000
D loss: 0.4296
G_loss: 3.215

Iter: 632000
D loss: 0.3368
G_loss: 3.206

Iter: 633000
D loss: 0.2355
G_loss: 3.194

Iter: 634000
D loss: 0.2248
G_loss: 3.472

Iter: 635000
D loss: 0.2134
G_loss: 3.23

Iter: 636000
D loss: 0.1967
G_loss: 3.572

Iter: 637000
D loss: 0.3138
G_loss: 3.529

Iter: 638000
D loss: 0.1624
G_loss: 3.115

Iter: 639000
D loss: 0.2297
G_loss: 3.321

Iter: 640000
D loss: 0.2438
G_loss: 3.54

Iter: 641000
D loss: 0.1845
G_loss: 3.163

Iter: 642000
D loss: 0.2813
G_loss: 3.143

Iter: 643000
D loss: 0.2181
G_loss: 3.108

Iter: 644000
D loss: 0.1485
G_loss: 3.234

Iter: 645000
D loss: 0.2695
G_loss: 3.382

Iter: 646000
D loss: 0.2339
G_loss: 3.301

Iter: 647000
D loss: 0.238
G_loss: 3.529

Iter: 648000
D loss: 0.3153
G_loss: 3.489

Iter: 649000
D loss: 0.1465
G_loss: 3.301

Iter: 650000
D loss: 0.2233
G_loss: 3.117

Iter: 651000
D loss: 0.2265
G_loss: 3.019

Iter: 652000
D loss: 0.2159
G_loss: 3.621

Iter: 653000
D loss: 0.3052
G_loss: 3.618

Iter: 654000
D loss: 0.2265
G_loss: 3.777

Iter: 655000
D loss: 0.2284
G_loss: 3.556

Iter: 656000
D loss: 0.2262
G_loss: 3.287

Iter: 657000
D loss: 0.1994
G_loss: 3.69

Iter: 658000
D loss: 0.2243
G_loss: 3.333

Iter: 659000
D loss: 0.2649
G_loss: 3.353

Iter: 660000
D loss: 0.1891
G_loss: 3.572

Iter: 661000
D loss: 0.107
G_loss: 3.751

Iter: 662000
D loss: 0.1923
G_loss: 3.407

Iter: 663000
D loss: 0.1228
G_loss: 3.241

Iter: 664000
D loss: 0.2384
G_loss: 3.175

Iter: 665000
D loss: 0.3018
G_loss: 3.317

Iter: 666000
D loss: 0.2341
G_loss: 3.544

Iter: 667000
D loss: 0.254
G_loss: 2.959

Iter: 668000
D loss: 0.1899
G_loss: 2.876

Iter: 669000
D loss: 0.1461
G_loss: 3.527

Iter: 670000
D loss: 0.2825
G_loss: 3.015

Iter: 671000
D loss: 0.2968
G_loss: 3.517

Iter: 672000
D loss: 0.2804
G_loss: 3.324

Iter: 673000
D loss: 0.2607
G_loss: 3.475

Iter: 674000
D loss: 0.2868
G_loss: 3.442

Iter: 675000
D loss: 0.1854
G_loss: 3.472

Iter: 676000
D loss: 0.1895
G_loss: 3.493

Iter: 677000
D loss: 0.356
G_loss: 3.555

Iter: 678000
D loss: 0.2386
G_loss: 3.205

Iter: 679000
D loss: 0.27
G_loss: 4.77

Iter: 680000
D loss: 0.2295
G_loss: 3.057

Iter: 681000
D loss: 0.2443
G_loss: 2.976

Iter: 682000
D loss: 0.2899
G_loss: 3.2

Iter: 683000
D loss: 0.2999
G_loss: 2.885

Iter: 684000
D loss: 0.2249
G_loss: 3.134

Iter: 685000
D loss: 0.2225
G_loss: 3.224

Iter: 686000
D loss: 0.2511
G_loss: 3.579

Iter: 687000
D loss: 0.2068
G_loss: 3.14

Iter: 688000
D loss: 0.2933
G_loss: 2.9

Iter: 689000
D loss: 0.3127
G_loss: 2.965

Iter: 690000
D loss: 0.1845
G_loss: 3.295

Iter: 691000
D loss: 0.242
G_loss: 3.134

Iter: 692000
D loss: 0.1978
G_loss: 3.655

Iter: 693000
D loss: 0.2482
G_loss: 3.543

Iter: 694000
D loss: 0.2198
G_loss: 3.145

Iter: 695000
D loss: 0.2971
G_loss: 3.295

Iter: 696000
D loss: 0.3052
G_loss: 3.623

Iter: 697000
D loss: 0.307
G_loss: 3.359

Iter: 698000
D loss: 0.1797
G_loss: 3.309

Iter: 699000
D loss: 0.2059
G_loss: 3.317

Iter: 700000
D loss: 0.1643
G_loss: 3.152

Iter: 701000
D loss: 0.1896
G_loss: 3.308

Iter: 702000
D loss: 0.2544
G_loss: 3.455

Iter: 703000
D loss: 0.2362
G_loss: 3.584

Iter: 704000
D loss: 0.2033
G_loss: 3.421

Iter: 705000
D loss: 0.3564
G_loss: 3.217

Iter: 706000
D loss: 0.1822
G_loss: 3.238

Iter: 707000
D loss: 0.1989
G_loss: 3.226

Iter: 708000
D loss: 0.2215
G_loss: 3.258

Iter: 709000
D loss: 0.1599
G_loss: 3.478

Iter: 710000
D loss: 0.225
G_loss: 3.169

Iter: 711000
D loss: 0.1904
G_loss: 3.496

Iter: 712000
D loss: 0.2872
G_loss: 3.124

Iter: 713000
D loss: 0.3396
G_loss: 3.353

Iter: 714000
D loss: 0.272
G_loss: 3.457

Iter: 715000
D loss: 0.1641
G_loss: 3.227

Iter: 716000
D loss: 0.1872
G_loss: 3.175

Iter: 717000
D loss: 0.2204
G_loss: 3.266

Iter: 718000
D loss: 0.1738
G_loss: 3.782

Iter: 719000
D loss: 0.2064
G_loss: 3.548

Iter: 720000
D loss: 0.2378
G_loss: 3.592

Iter: 721000
D loss: 0.2182
G_loss: 3.814

Iter: 722000
D loss: 0.1924
G_loss: 3.297

Iter: 723000
D loss: 0.166
G_loss: 3.466

Iter: 724000
D loss: 0.1371
G_loss: 3.369

Iter: 725000
D loss: 0.2581
G_loss: 3.737

Iter: 726000
D loss: 0.2569
G_loss: 3.233

Iter: 727000
D loss: 0.1518
G_loss: 3.568

Iter: 728000
D loss: 0.2033
G_loss: 3.533

Iter: 729000
D loss: 0.2889
G_loss: 3.524

Iter: 730000
D loss: 0.1676
G_loss: 3.362

Iter: 731000
D loss: 0.2749
G_loss: 3.451

Iter: 732000
D loss: 0.2439
G_loss: 3.217

Iter: 733000
D loss: 0.1883
G_loss: 3.466

Iter: 734000
D loss: 0.2062
G_loss: 3.444

Iter: 735000
D loss: 0.2259
G_loss: 4.156

Iter: 736000
D loss: 0.1723
G_loss: 3.242

Iter: 737000
D loss: 0.1561
G_loss: 3.984

Iter: 738000
D loss: 0.2395
G_loss: 3.614

Iter: 739000
D loss: 0.2414
G_loss: 3.674

Iter: 740000
D loss: 0.2542
G_loss: 3.704

Iter: 741000
D loss: 0.1478
G_loss: 3.147

Iter: 742000
D loss: 0.2395
G_loss: 3.601

Iter: 743000
D loss: 0.134
G_loss: 3.146

Iter: 744000
D loss: 0.2192
G_loss: 3.289

Iter: 745000
D loss: 0.1801
G_loss: 3.337

Iter: 746000
D loss: 0.2186
G_loss: 3.185

Iter: 747000
D loss: 0.2462
G_loss: 3.527

Iter: 748000
D loss: 0.2783
G_loss: 3.758

Iter: 749000
D loss: 0.2387
G_loss: 3.215

Iter: 750000
D loss: 0.1466
G_loss: 3.383

Iter: 751000
D loss: 0.1694
G_loss: 3.285

Iter: 752000
D loss: 0.1787
G_loss: 3.7

Iter: 753000
D loss: 0.1671
G_loss: 3.802

Iter: 754000
D loss: 0.2022
G_loss: 3.387

Iter: 755000
D loss: 0.1398
G_loss: 3.547

Iter: 756000
D loss: 0.235
G_loss: 3.522

Iter: 757000
D loss: 0.19
G_loss: 3.678

Iter: 758000
D loss: 0.1686
G_loss: 3.844

Iter: 759000
D loss: 0.09918
G_loss: 4.142

Iter: 760000
D loss: 0.2729
G_loss: 3.55

Iter: 761000
D loss: 0.1812
G_loss: 3.162

Iter: 762000
D loss: 0.1905
G_loss: 3.677

Iter: 763000
D loss: 0.1788
G_loss: 3.565

Iter: 764000
D loss: 0.2277
G_loss: 3.319

Iter: 765000
D loss: 0.2115
G_loss: 3.591

Iter: 766000
D loss: 0.1696
G_loss: 3.892

Iter: 767000
D loss: 0.2761
G_loss: 3.44

Iter: 768000
D loss: 0.1272
G_loss: 3.624

Iter: 769000
D loss: 0.1941
G_loss: 2.987

Iter: 770000
D loss: 0.151
G_loss: 3.88

Iter: 771000
D loss: 0.2883
G_loss: 3.846

Iter: 772000
D loss: 0.1235
G_loss: 3.671

Iter: 773000
D loss: 0.1935
G_loss: 3.698

Iter: 774000
D loss: 0.3969
G_loss: 3.806

Iter: 775000
D loss: 0.1576
G_loss: 4.208

Iter: 776000
D loss: 0.2489
G_loss: 3.734

Iter: 777000
D loss: 0.3095
G_loss: 3.073

Iter: 778000
D loss: 0.3787
G_loss: 3.777

Iter: 779000
D loss: 0.274
G_loss: 3.777

Iter: 780000
D loss: 0.2544
G_loss: 3.341

Iter: 781000
D loss: 0.1771
G_loss: 3.335

Iter: 782000
D loss: 0.213
G_loss: 3.209

Iter: 783000
D loss: 0.1572
G_loss: 3.501

Iter: 784000
D loss: 0.2181
G_loss: 3.331

Iter: 785000
D loss: 0.1682
G_loss: 4.065

Iter: 786000
D loss: 0.238
G_loss: 3.42

Iter: 787000
D loss: 0.1986
G_loss: 3.4

Iter: 788000
D loss: 0.1913
G_loss: 3.489

Iter: 789000
D loss: 0.3082
G_loss: 4.185

Iter: 790000
D loss: 0.1316
G_loss: 3.714

Iter: 791000
D loss: 0.07559
G_loss: 3.455

Iter: 792000
D loss: 0.2785
G_loss: 3.614

Iter: 793000
D loss: 0.1894
G_loss: 3.543

Iter: 794000
D loss: 0.1549
G_loss: 3.316

Iter: 795000
D loss: 0.2494
G_loss: 3.774

Iter: 796000
D loss: 0.1958
G_loss: 3.667

Iter: 797000
D loss: 0.1536
G_loss: 3.266

Iter: 798000
D loss: 0.2807
G_loss: 3.256

Iter: 799000
D loss: 0.1746
G_loss: 3.68

Iter: 800000
D loss: 0.1295
G_loss: 3.37

Iter: 801000
D loss: 0.1591
G_loss: 3.624

Iter: 802000
D loss: 0.2973
G_loss: 3.763

Iter: 803000
D loss: 0.1612
G_loss: 3.389

Iter: 804000
D loss: 0.1476
G_loss: 4.051

Iter: 805000
D loss: 0.2911
G_loss: 4.136

Iter: 806000
D loss: 0.1111
G_loss: 3.614

Iter: 807000
D loss: 0.1975
G_loss: 3.673

Iter: 808000
D loss: 0.1573
G_loss: 3.491

Iter: 809000
D loss: 0.2062
G_loss: 3.766

Iter: 810000
D loss: 0.2258
G_loss: 3.701

Iter: 811000
D loss: 0.3427
G_loss: 3.349

Iter: 812000
D loss: 0.2079
G_loss: 3.921

Iter: 813000
D loss: 0.1729
G_loss: 3.317

Iter: 814000
D loss: 0.1892
G_loss: 3.953

Iter: 815000
D loss: 0.1558
G_loss: 3.487

Iter: 816000
D loss: 0.2644
G_loss: 3.536

Iter: 817000
D loss: 0.2732
G_loss: 3.644

Iter: 818000
D loss: 0.1557
G_loss: 3.586

Iter: 819000
D loss: 0.2439
G_loss: 3.35

Iter: 820000
D loss: 0.1508
G_loss: 3.216

Iter: 821000
D loss: 0.204
G_loss: 3.246

Iter: 822000
D loss: 0.1285
G_loss: 3.409

Iter: 823000
D loss: 0.1257
G_loss: 3.355

Iter: 824000
D loss: 0.141
G_loss: 3.696

Iter: 825000
D loss: 0.09579
G_loss: 3.964

Iter: 826000
D loss: 0.3356
G_loss: 3.564

Iter: 827000
D loss: 0.1836
G_loss: 3.175

Iter: 828000
D loss: 0.1491
G_loss: 3.423

Iter: 829000
D loss: 0.1526
G_loss: 3.696

Iter: 830000
D loss: 0.2182
G_loss: 3.362

Iter: 831000
D loss: 0.1846
G_loss: 3.226

Iter: 832000
D loss: 0.1698
G_loss: 3.433

Iter: 833000
D loss: 0.2193
G_loss: 3.391

Iter: 834000
D loss: 0.2434
G_loss: 3.494

Iter: 835000
D loss: 0.1417
G_loss: 3.667

Iter: 836000
D loss: 0.2032
G_loss: 3.487

Iter: 837000
D loss: 0.1324
G_loss: 3.922

Iter: 838000
D loss: 0.1301
G_loss: 3.824

Iter: 839000
D loss: 0.2615
G_loss: 3.896

Iter: 840000
D loss: 0.142
G_loss: 3.856

Iter: 841000
D loss: 0.169
G_loss: 3.476

Iter: 842000
D loss: 0.1456
G_loss: 3.089

Iter: 843000
D loss: 0.2677
G_loss: 3.735

Iter: 844000
D loss: 0.2056
G_loss: 3.802

Iter: 845000
D loss: 0.1818
G_loss: 3.466

Iter: 846000
D loss: 0.2538
G_loss: 3.68

Iter: 847000
D loss: 0.1806
G_loss: 3.414

Iter: 848000
D loss: 0.1854
G_loss: 3.592

Iter: 849000
D loss: 0.2321
G_loss: 3.858

Iter: 850000
D loss: 0.1767
G_loss: 3.632

Iter: 851000
D loss: 0.1464
G_loss: 3.194

Iter: 852000
D loss: 0.2123
G_loss: 3.563

Iter: 853000
D loss: 0.1703
G_loss: 3.67

Iter: 854000
D loss: 0.2406
G_loss: 3.562

Iter: 855000
D loss: 0.2378
G_loss: 3.676

Iter: 856000
D loss: 0.2471
G_loss: 3.713

Iter: 857000
D loss: 0.1842
G_loss: 3.476

Iter: 858000
D loss: 0.1431
G_loss: 3.737

Iter: 859000
D loss: 0.2655
G_loss: 3.606

Iter: 860000
D loss: 0.1491
G_loss: 3.919

Iter: 861000
D loss: 0.1807
G_loss: 3.415

Iter: 862000
D loss: 0.1679
G_loss: 3.224

Iter: 863000
D loss: 0.1328
G_loss: 3.443

Iter: 864000
D loss: 0.2518
G_loss: 3.482

Iter: 865000
D loss: 0.1977
G_loss: 3.959

Iter: 866000
D loss: 0.2717
G_loss: 3.818

Iter: 867000
D loss: 0.1366
G_loss: 3.611

Iter: 868000
D loss: 0.09269
G_loss: 3.952

Iter: 869000
D loss: 0.1505
G_loss: 3.743

Iter: 870000
D loss: 0.1921
G_loss: 3.821

Iter: 871000
D loss: 0.1971
G_loss: 4.069

Iter: 872000
D loss: 0.1597
G_loss: 3.582

Iter: 873000
D loss: 0.1971
G_loss: 3.608

Iter: 874000
D loss: 0.1352
G_loss: 3.414

Iter: 875000
D loss: 0.1412
G_loss: 3.667

Iter: 876000
D loss: 0.168
G_loss: 3.918

Iter: 877000
D loss: 0.1916
G_loss: 3.067

Iter: 878000
D loss: 0.1538
G_loss: 3.393

Iter: 879000
D loss: 0.1955
G_loss: 3.461

Iter: 880000
D loss: 0.2431
G_loss: 3.112

Iter: 881000
D loss: 0.2096
G_loss: 3.676

Iter: 882000
D loss: 0.1337
G_loss: 3.645

Iter: 883000
D loss: 0.2114
G_loss: 3.611

Iter: 884000
D loss: 0.3033
G_loss: 3.729

Iter: 885000
D loss: 0.2466
G_loss: 3.669

Iter: 886000
D loss: 0.1785
G_loss: 3.667

Iter: 887000
D loss: 0.183
G_loss: 4.349

Iter: 888000
D loss: 0.3768
G_loss: 3.806

Iter: 889000
D loss: 0.1315
G_loss: 3.505

Iter: 890000
D loss: 0.2957
G_loss: 3.606

Iter: 891000
D loss: 0.1651
G_loss: 3.651

Iter: 892000
D loss: 0.1615
G_loss: 3.357

Iter: 893000
D loss: 0.2094
G_loss: 3.74

Iter: 894000
D loss: 0.1948
G_loss: 3.639

Iter: 895000
D loss: 0.1651
G_loss: 3.319

Iter: 896000
D loss: 0.1348
G_loss: 3.72

Iter: 897000
D loss: 0.3188
G_loss: 3.671

Iter: 898000
D loss: 0.1998
G_loss: 3.386

Iter: 899000
D loss: 0.1165
G_loss: 3.434

Iter: 900000
D loss: 0.137
G_loss: 3.808

Iter: 901000
D loss: 0.1285
G_loss: 3.654

Iter: 902000
D loss: 0.1976
G_loss: 3.602

Iter: 903000
D loss: 0.2083
G_loss: 3.46

Iter: 904000
D loss: 0.1432
G_loss: 3.862

Iter: 905000
D loss: 0.2557
G_loss: 3.554

Iter: 906000
D loss: 0.1612
G_loss: 3.36

Iter: 907000
D loss: 0.2199
G_loss: 3.903

Iter: 908000
D loss: 0.1881
G_loss: 3.798

Iter: 909000
D loss: 0.2579
G_loss: 3.665

Iter: 910000
D loss: 0.2947
G_loss: 4.01

Iter: 911000
D loss: 0.2328
G_loss: 3.988

Iter: 912000
D loss: 0.1583
G_loss: 3.96

Iter: 913000
D loss: 0.2154
G_loss: 3.508

Iter: 914000
D loss: 0.1215
G_loss: 3.595

Iter: 915000
D loss: 0.2063
G_loss: 3.574

Iter: 916000
D loss: 0.2569
G_loss: 4.016

Iter: 917000
D loss: 0.1569
G_loss: 3.749

Iter: 918000
D loss: 0.2388
G_loss: 4.248

Iter: 919000
D loss: 0.1965
G_loss: 3.775

Iter: 920000
D loss: 0.1442
G_loss: 3.736

Iter: 921000
D loss: 0.135
G_loss: 4.064

Iter: 922000
D loss: 0.1318
G_loss: 3.544

Iter: 923000
D loss: 0.1541
G_loss: 3.847

Iter: 924000
D loss: 0.1433
G_loss: 3.869

Iter: 925000
D loss: 0.261
G_loss: 3.489

Iter: 926000
D loss: 0.2182
G_loss: 3.768

Iter: 927000
D loss: 0.2016
G_loss: 3.607

Iter: 928000
D loss: 0.1853
G_loss: 3.767

Iter: 929000
D loss: 0.3451
G_loss: 3.7

Iter: 930000
D loss: 0.1733
G_loss: 3.903

Iter: 931000
D loss: 0.3005
G_loss: 3.953

Iter: 932000
D loss: 0.1109
G_loss: 3.862

Iter: 933000
D loss: 0.1432
G_loss: 3.777

Iter: 934000
D loss: 0.1024
G_loss: 3.496

Iter: 935000
D loss: 0.1932
G_loss: 3.787

Iter: 936000
D loss: 0.1985
G_loss: 3.553

Iter: 937000
D loss: 0.2199
G_loss: 3.708

Iter: 938000
D loss: 0.2405
G_loss: 4.064

Iter: 939000
D loss: 0.1208
G_loss: 3.501

Iter: 940000
D loss: 0.1933
G_loss: 3.712

Iter: 941000
D loss: 0.1117
G_loss: 3.932

Iter: 942000
D loss: 0.116
G_loss: 4.042

Iter: 943000
D loss: 0.1803
G_loss: 4.188

Iter: 944000
D loss: 0.1432
G_loss: 3.43

Iter: 945000
D loss: 0.2393
G_loss: 3.707

Iter: 946000
D loss: 0.09669
G_loss: 4.261

Iter: 947000
D loss: 0.2052
G_loss: 3.949

Iter: 948000
D loss: 0.1569
G_loss: 3.998

Iter: 949000
D loss: 0.2604
G_loss: 3.422

Iter: 950000
D loss: 0.127
G_loss: 3.559

Iter: 951000
D loss: 0.0865
G_loss: 3.712

Iter: 952000
D loss: 0.1853
G_loss: 4.052

Iter: 953000
D loss: 0.1568
G_loss: 4.267

Iter: 954000
D loss: 0.2473
G_loss: 3.848

Iter: 955000
D loss: 0.1935
G_loss: 3.398

Iter: 956000
D loss: 0.1153
G_loss: 4.242

Iter: 957000
D loss: 0.2894
G_loss: 3.541

Iter: 958000
D loss: 0.2526
G_loss: 3.722

Iter: 959000
D loss: 0.1921
G_loss: 3.725

Iter: 960000
D loss: 0.1468
G_loss: 4.32

Iter: 961000
D loss: 0.2058
G_loss: 3.469

Iter: 962000
D loss: 0.2457
G_loss: 3.207

Iter: 963000
D loss: 0.1368
G_loss: 3.243

Iter: 964000
D loss: 0.1626
G_loss: 3.524

Iter: 965000
D loss: 0.2024
G_loss: 3.664

Iter: 966000
D loss: 0.2339
G_loss: 3.703

Iter: 967000
D loss: 0.09176
G_loss: 3.814

Iter: 968000
D loss: 0.1933
G_loss: 3.449

Iter: 969000
D loss: 0.1257
G_loss: 3.649

Iter: 970000
D loss: 0.1197
G_loss: 3.889

Iter: 971000
D loss: 0.09248
G_loss: 4.303

Iter: 972000
D loss: 0.1157
G_loss: 4.018

Iter: 973000
D loss: 0.2459
G_loss: 3.745

Iter: 974000
D loss: 0.1363
G_loss: 3.794

Iter: 975000
D loss: 0.1774
G_loss: 3.961

Iter: 976000
D loss: 0.1306
G_loss: 3.614

Iter: 977000
D loss: 0.1732
G_loss: 3.274

Iter: 978000
D loss: 0.2852
G_loss: 3.889

Iter: 979000
D loss: 0.1105
G_loss: 3.735

Iter: 980000
D loss: 0.274
G_loss: 4.031

Iter: 981000
D loss: 0.1704
G_loss: 3.536

Iter: 982000
D loss: 0.2157
G_loss: 3.391

Iter: 983000
D loss: 0.249
G_loss: 3.563

Iter: 984000
D loss: 0.263
G_loss: 3.959

Iter: 985000
D loss: 0.1704
G_loss: 3.901

Iter: 986000
D loss: 0.2113
G_loss: 4.029

Iter: 987000
D loss: 0.1187
G_loss: 3.559

Iter: 988000
D loss: 0.1795
G_loss: 3.779

Iter: 989000
D loss: 0.2453
G_loss: 3.877

Iter: 990000
D loss: 0.1927
G_loss: 3.835

Iter: 991000
D loss: 0.2889
G_loss: 3.649

Iter: 992000
D loss: 0.2511
G_loss: 3.736

Iter: 993000
D loss: 0.1963
G_loss: 3.518

Iter: 994000
D loss: 0.1559
G_loss: 4.24

Iter: 995000
D loss: 0.1713
G_loss: 3.647

Iter: 996000
D loss: 0.1735
G_loss: 3.445

Iter: 997000
D loss: 0.2216
G_loss: 3.759

Iter: 998000
D loss: 0.1491
G_loss: 4.028

Iter: 999000
D loss: 0.1784
G_loss: 4.136

``````

After checking images in `out` dir we see that the GAN mode collapsed

Fix for that is to let discriminator see ground truth in mini batches

Here's a YouTube of I made from all 1000 images at 100ms delay

``````

In [1]:

from IPython.display import HTML

HTML('<iframe width="560" height="315" src="https://www.youtube.com/embed/ktxhiKhWoEE?rel=0&amp;controls=0&amp;showinfo=0" frameborder="0" allowfullscreen></iframe>')

``````
``````

Out[1]:

``````
``````

In [ ]:

``````