-
Notifications
You must be signed in to change notification settings - Fork 0
/
zeroshotgan.py
324 lines (248 loc) · 19.1 KB
/
zeroshotgan.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
import sys
import sys
import sys
import inspect
import traceback
import os
import pdb
import time
import shutil
import argparse
# # # MNIST
# parser = argparse.ArgumentParser(description='Tensorflow Gan Models')
# parser.add_argument('--global_exp_dir', type=str, default='./experimentsDAEGAN40', help='Directory to put the experiments.')
# parser.add_argument('--exp_dir_postfix', type=str, default='', help='Directory to put the experiment postfix.')
# parser.add_argument('--dataset_dir', type=str, default='../dataset/dataset_both2/scripted/*.npz', help='Directory of data.')
# parser.add_argument('--restore_dir', type=str, default='/7cb1611a0a584269a019dc8342aee213/checkpoint/', help='Directory of restore experiment.')
# parser.add_argument('--restore', type=bool, default=False, help='Restore model.')
# parser.add_argument('--analyticKL', type=bool, default=True, help='Type of KL divergence to use.')
# parser.add_argument('--transformedQ', type=bool, default=False, help='Use posterior Transform.')
# parser.add_argument('--epochs', type=int, default=100000, help='Number of epochs to train.')
# parser.add_argument('--batch_size', type=int, default=50, help='Input batch size for training.')
# parser.add_argument('--time_steps', type=int, default=1, help='Number of timesteps')
# parser.add_argument('--hierarchy_rate', type=int, default=1, help='Number of timesteps')
# parser.add_argument('--optimizer_class', type=str, default='Adam', help='Optimizer type.')
# parser.add_argument('--learning_rate', type=float, default=0.0001, help='Initial learning rate.')
# parser.add_argument('--momentum', type=float, default=0.9, help='Initial momentum.')
# parser.add_argument('--weight_decay', type=float, default=0, help='Initial weight decay.')
# parser.add_argument('--initial_temp', type=float, default=1, help='Initial temperature for KL divergence.')
# parser.add_argument('--max_step_temp', type=float, default=15000, help='Starting step for temp=1.')
# parser.add_argument('--cuda', action='store_true', default=False, help='enables CUDA training')
# parser.add_argument('--seed', type=int, default=1, help='random seed')
# parser.add_argument('--log_interval', type=int, default=100, help='how many batches to wait before logging training status')
# parser.add_argument('--n_encoder', type=int, default=400, help='n_encoder.')
# parser.add_argument('--n_decoder', type=int, default=400, help='n_decoder.')
# parser.add_argument('--n_context', type=int, default=1, help='n_context.')
# parser.add_argument('--n_state', type=int, default=1, help='n_state.')
# parser.add_argument('--n_latent', type=int, default=100, help='n_latent.')
# global_args = parser.parse_args()
# global_args.curr_epoch = 1
# from datasetLoaders.MnistLoader import DataLoader
# data_loader = DataLoader(batch_size = global_args.batch_size, time_steps = global_args.time_steps)
# # # # #############################################################################################################################
# # # # TOY
# parser = argparse.ArgumentParser(description='Tensorflow Gan Models')
# parser.add_argument('--global_exp_dir', type=str, default='./experimentsGanTests25', help='Directory to put the experiments.')
# parser.add_argument('--exp_dir_postfix', type=str, default='', help='Directory to put the experiment postfix.')
# parser.add_argument('--dataset_dir', type=str, default='../dataset/dataset_both2/scripted/*.npz', help='Directory of data.')
# parser.add_argument('--restore_dir', type=str, default='/7cb1611a0a584269a019dc8342aee213/checkpoint/', help='Directory of restore experiment.')
# parser.add_argument('--restore', type=bool, default=False, help='Restore model.')
# parser.add_argument('--analyticKL', type=bool, default=True, help='Type of KL divergence to use.')
# parser.add_argument('--transformedQ', type=bool, default=True, help='Use posterior Transform.')
# parser.add_argument('--epochs', type=int, default=100000, help='Number of epochs to train.')
# parser.add_argument('--batch_size', type=int, default=50, help='Input batch size for training.')
# parser.add_argument('--time_steps', type=int, default=1, help='Number of timesteps')
# parser.add_argument('--hierarchy_rate', type=int, default=1, help='Number of timesteps')
# parser.add_argument('--optimizer_class', type=str, default='Adam', help='Optimizer type.')
# parser.add_argument('--learning_rate', type=float, default=0.0001, help='Initial learning rate.')
# parser.add_argument('--momentum', type=float, default=0.9, help='Initial momentum.')
# parser.add_argument('--weight_decay', type=float, default=0, help='Initial weight decay.')
# parser.add_argument('--initial_temp', type=float, default=0.1, help='Initial temperature for KL divergence.')
# parser.add_argument('--max_step_temp', type=float, default=15000, help='Starting step for temp=1.')
# parser.add_argument('--cuda', action='store_true', default=False, help='enables CUDA training')
# parser.add_argument('--seed', type=int, default=1, help='random seed')
# parser.add_argument('--log_interval', type=int, default=100, help='how many batches to wait before logging training status')
# parser.add_argument('--n_encoder', type=int, default=100, help='n_encoder.')
# parser.add_argument('--n_decoder', type=int, default=100, help='n_decoder.')
# parser.add_argument('--n_context', type=int, default=1, help='n_context.')
# parser.add_argument('--n_state', type=int, default=1, help='n_state.')
# parser.add_argument('--n_latent', type=int, default=2, help='n_latent.')
# global_args = parser.parse_args()
# global_args.curr_epoch = 1
# # from datasetLoaders.ToyDataLoader import DataLoader
# # data_loader = DataLoader(batch_size = global_args.batch_size, time_steps = global_args.time_steps)
# from datasetLoaders.RandomManifoldDataLoader import DataLoader
# data_loader = DataLoader(batch_size = global_args.batch_size, time_steps = global_args.time_steps)
##############################################################################################################################
# # CNN FEATURES
parser = argparse.ArgumentParser(description='Tensorflow Gan Models')
parser.add_argument('--global_exp_dir', type=str, default='./experimentsZERO', help='Directory to put the experiments.')
parser.add_argument('--exp_dir_postfix', type=str, default='', help='Directory to put the experiment postfix.')
parser.add_argument('--dataset_dir', type=str, default='../dataset/dataset_both2/scripted/*.npz', help='Directory of data.')
parser.add_argument('--restore_dir', type=str, default='/7cb1611a0a584269a019dc8342aee213/checkpoint/', help='Directory of restore experiment.')
parser.add_argument('--restore', type=bool, default=False, help='Restore model.')
parser.add_argument('--analyticKL', type=bool, default=True, help='Type of KL divergence to use.')
parser.add_argument('--transformedQ', type=bool, default=False, help='Use posterior Transform.')
parser.add_argument('--epochs', type=int, default=100000, help='Number of epochs to train.')
parser.add_argument('--batch_size', type=int, default=30, help='Input batch size for training.')
parser.add_argument('--time_steps', type=int, default=1, help='Number of timesteps')
parser.add_argument('--hierarchy_rate', type=int, default=1, help='Number of timesteps')
parser.add_argument('--optimizer_class', type=str, default='Adam', help='Optimizer type.')
parser.add_argument('--learning_rate', type=float, default=0.0001, help='Initial learning rate.')
parser.add_argument('--momentum', type=float, default=0.9, help='Initial momentum.')
parser.add_argument('--weight_decay', type=float, default=0, help='Initial weight decay.')
parser.add_argument('--initial_temp', type=float, default=1, help='Initial temperature for KL divergence.')
parser.add_argument('--max_step_temp', type=float, default=15000, help='Starting step for temp=1.')
parser.add_argument('--cuda', action='store_true', default=False, help='enables CUDA training')
parser.add_argument('--seed', type=int, default=1, help='random seed')
parser.add_argument('--log_interval', type=int, default=100, help='how many batches to wait before logging training status')
parser.add_argument('--n_encoder', type=int, default=400, help='n_encoder.')
parser.add_argument('--n_decoder', type=int, default=400, help='n_decoder.')
parser.add_argument('--n_context', type=int, default=1, help='n_context.')
parser.add_argument('--n_state', type=int, default=1, help='n_state.')
parser.add_argument('--n_latent', type=int, default=200, help='n_latent.')
global_args = parser.parse_args()
global_args.curr_epoch = 1
from datasetLoaders.FeatureAttributeLoader import DataLoader
data_loader = DataLoader(batch_size = global_args.batch_size, time_steps = global_args.time_steps)
from models.ZEROSHOTGAN.Model import Model
import distributions
import helper
import random
import numpy as np
import tensorflow as tf
random.seed(global_args.seed)
np.random.seed(global_args.seed)
tf.set_random_seed(global_args.seed)
global_args.exp_dir = helper.get_exp_dir(global_args)
os.environ['CUDA_VISIBLE_DEVICES'] = '0'
print("os.environ['CUDA_VISIBLE_DEVICES'], ", os.environ['CUDA_VISIBLE_DEVICES'])
print("TENSORBOARD: Linux:\npython -m tensorflow.tensorboard --logdir=model1:"+\
os.path.realpath(global_args.exp_dir)+" --port="+str(20000+int(global_args.exp_dir[-4:-1], 16))+" &")
print("TENSORBOARD: Mac:\nhttp://0.0.0.0:"+str(20000+int(global_args.exp_dir[-4:-1], 16)))
print("\n\n\n")
# shutil.copyfile('./models/SLVM.py', global_args.exp_dir+'SLVM.py')
# shutil.copyfile('./models/ModelGTM.py', global_args.exp_dir+'ModelGTM.py')
_, _, batch = next(data_loader)
with tf.Graph().as_default():
tf.set_random_seed(global_args.seed)
model = Model(vars(global_args))
global_step = tf.Variable(0.0, name='global_step', trainable=False)
with tf.variable_scope("training"):
tf.set_random_seed(global_args.seed)
additional_inputs_tf = tf.placeholder(tf.float32, [1])
batch_tf, input_dict_func = helper.tf_batch_and_input_dict(batch, additional_inputs_tf)
train_outs_dict, test_outs_dict = model.inference(batch_tf, additional_inputs_tf)
generative_dict = model.generative_model(batch_tf)
inference_obs_dist = model.obs_dist
discriminator_vars = [v for v in tf.trainable_variables() if 'Discriminator' in v.name]
generator_vars = [v for v in tf.trainable_variables() if 'Discriminator' not in v.name]
# Weight clipping
discriminator_vars_flat_concat = tf.concat([tf.reshape(e, [-1]) for e in discriminator_vars], axis=0)
max_abs_discriminator_vars = tf.reduce_max(tf.abs(discriminator_vars_flat_concat))
clip_op_list = []
for e in discriminator_vars:
clip_op_list.append(tf.assign(e, tf.clip_by_value(e, -0.01, 0.01)))
if global_args.optimizer_class == 'RmsProp':
train_generator_step_tf = tf.train.RMSPropOptimizer(learning_rate=global_args.learning_rate,
momentum=0.9).minimize(train_outs_dict['generator_cost'], var_list=generator_vars, global_step=global_step)
train_discriminator_step_tf = tf.train.RMSPropOptimizer(learning_rate=global_args.learning_rate,
momentum=0.9).minimize(train_outs_dict['discriminator_cost'], var_list=discriminator_vars, global_step=global_step)
elif global_args.optimizer_class == 'Adam':
train_generator_step_tf = tf.train.AdamOptimizer(learning_rate=0.0001,
beta1=0.5, beta2=0.999, epsilon=1e-08).minimize(train_outs_dict['generator_cost'], var_list=generator_vars, global_step=global_step)
train_discriminator_step_tf = tf.train.AdamOptimizer(learning_rate=0.0001,
beta1=0.5, beta2=0.999, epsilon=1e-08).minimize(train_outs_dict['discriminator_cost'], var_list=discriminator_vars, global_step=global_step)
helper.variable_summaries(train_outs_dict['generator_cost'], '/generator_cost')
helper.variable_summaries(train_outs_dict['discriminator_cost'], '/discriminator_cost')
init = tf.global_variables_initializer()
saver = tf.train.Saver()
sess = tf.InteractiveSession()
merged_summaries = tf.summary.merge_all()
summary_writer = tf.summary.FileWriter(global_args.exp_dir+'/summaries', sess.graph)
sess.run(init)
if global_args.restore:
print("=> Loading checkpoint: '{}'".format(global_args.global_exp_dir+global_args.restore_dir))
try:
helper.load_checkpoint(saver, sess, global_args.global_exp_dir+global_args.restore_dir)
print("=> Loaded checkpoint: '{}'".format(global_args.global_exp_dir+global_args.restore_dir))
except: print("=> FAILED to load checkpoint: '{}'".format(global_args.global_exp_dir+global_args.restore_dir))
def train(epoch):
data_loader.train()
train_gen_loss_accum, train_dis_loss_accum, train_likelihood_accum, train_kl_accum, batch_size_accum = 0, 0, 0, 0, 0
start = time.time();
for batch_idx, curr_batch_size, batch in data_loader:
disc_train_step_np = sess.run([train_discriminator_step_tf], feed_dict = input_dict_func(batch, np.asarray([0,])))
if batch_idx % 5 !=0: continue
gen_train_step_np, generator_cost_np, discriminator_cost_np = \
sess.run([train_generator_step_tf, train_outs_dict['generator_cost'], train_outs_dict['discriminator_cost']],
feed_dict = input_dict_func(batch, np.asarray([0,])))
max_discriminator_weight = sess.run(max_abs_discriminator_vars)
train_gen_loss_accum += curr_batch_size*generator_cost_np
train_dis_loss_accum += curr_batch_size*discriminator_cost_np
batch_size_accum += curr_batch_size
if batch_idx % global_args.log_interval == 0:
end = time.time();
print('Train: Epoch {} [{:7d} ()]\tGenerator Cost: {:.6f}\tDiscriminator Cost: {:.6f}\tTime: {:.3f}, Max disc weight {:.6f}'.format(
epoch, batch_idx * curr_batch_size, generator_cost_np, discriminator_cost_np, (end - start), max_discriminator_weight))
with open(global_args.exp_dir+"training_traces.txt", "a") as text_file:
text_file.write(str(generator_cost_np) + ', ' + str(discriminator_cost_np) + '\n')
start = time.time()
summary_str = sess.run(merged_summaries, feed_dict = input_dict_func(batch, np.asarray([0,])))
summary_writer.add_summary(summary_str, (tf.train.global_step(sess, global_step)))
checkpoint_time = 1
if data_loader.__module__ == 'datasetLoaders.RandomManifoldDataLoader' or data_loader.__module__ == 'datasetLoaders.ToyDataLoader':
checkpoint_time = 20
if epoch % checkpoint_time == 0:
print('====> Average Train: Epoch: {}\tGenerator Cost: {:.6f}\tDiscriminator Cost: {:.6f}'.format(
epoch, train_gen_loss_accum/batch_size_accum, train_dis_loss_accum/batch_size_accum))
# helper.draw_bar_plot(rate_similarity_gen_np[:,0,0], y_min_max = [0,1], save_dir=global_args.exp_dir+'Visualization/inversion_weight/', postfix='inversion_weight'+str(epoch))
# helper.draw_bar_plot(effective_z_cost_np[:,0,0], thres = [np.mean(effective_z_cost_np), np.max(effective_z_cost_np)], save_dir=global_args.exp_dir+'Visualization/inversion_cost/', postfix='inversion_cost'+str(epoch))
# helper.draw_bar_plot(disc_cost_gen_np[:,0,0], thres = [0, 0], save_dir=global_args.exp_dir+'Visualization/disc_cost/', postfix='disc_cost'+str(epoch))
# if data_loader.__module__ == 'datasetLoaders.RandomManifoldDataLoader' or data_loader.__module__ == 'datasetLoaders.ToyDataLoader':
# helper.visualize_datasets(sess, input_dict_func(batch), data_loader.dataset, generative_dict['obs_sample_out'], generative_dict['latent_sample_out'],
# save_dir=global_args.exp_dir+'Visualization/', postfix=str(epoch))
# xmin, xmax, ymin, ymax, X_dense, Y_dense = -3.5, 3.5, -3.5, 3.5, 250, 250
# xlist = np.linspace(xmin, xmax, X_dense)
# ylist = np.linspace(ymin, ymax, Y_dense)
# X, Y = np.meshgrid(xlist, ylist)
# XY = np.concatenate([X.reshape(-1,1), Y.reshape(-1,1)], axis=1)
# batch['observed']['data']['flat'] = XY[:, np.newaxis, :]
# disc_cost_real_np = sess.run(train_outs_dict['critic_real'], feed_dict = input_dict_func(batch, np.asarray([0,])))
# f = np.reshape(disc_cost_real_np[:,0,0], [Y_dense, X_dense])
# helper.plot_ffs(X, Y, f, save_dir=global_args.exp_dir+'Visualization/discriminator_function/', postfix='discriminator_function'+str(epoch))
# else:
# distributions.visualizeProductDistribution(sess, input_dict_func(batch), batch, inference_obs_dist, generative_dict['obs_dist'],
# save_dir=global_args.exp_dir+'Visualization/Train/', postfix='train_'+str(epoch))
checkpoint_path1 = global_args.exp_dir+'checkpoint/'
checkpoint_path2 = global_args.exp_dir+'checkpoint2/'
print('====> Saving checkpoint. Epoch: ', epoch); start_tmp = time.time()
helper.save_checkpoint(saver, sess, global_step, checkpoint_path1)
end_tmp = time.time(); print('Checkpoint path: '+checkpoint_path1+' ====> It took: ', end_tmp - start_tmp)
if epoch % 60 == 0:
print('====> Saving checkpoint backup. Epoch: ', epoch); start_tmp = time.time()
helper.save_checkpoint(saver, sess, global_step, checkpoint_path2)
end_tmp = time.time(); print('Checkpoint path: '+checkpoint_path2+' ====> It took: ', end_tmp - start_tmp)
def test(epoch):
data_loader.eval()
test_gen_loss_accum, test_dis_loss_accum, test_likelihood_accum, test_kl_accum, batch_size_accum = 0, 0, 0, 0, 0
start = time.time()
for batch_idx, curr_batch_size, batch in data_loader:
test_generator_cost_np, test_discriminator_cost_np = sess.run([test_outs_dict['generator_cost'], test_outs_dict['discriminator_cost']],
feed_dict = input_dict_func(batch, np.asarray([0,])))
test_gen_loss_accum += curr_batch_size*test_generator_cost_np
test_dis_loss_accum += curr_batch_size*test_discriminator_cost_np
batch_size_accum += curr_batch_size
end = time.time();
print('====> Average Test: Epoch {}\tGenerator Cost: {:.6f}\tDiscriminator Cost: {:.6f}\tTime: {:.3f}'.format(
epoch, test_gen_loss_accum/batch_size_accum, test_dis_loss_accum/batch_size_accum, (end - start)))
with open(global_args.exp_dir+"test_traces.txt", "a") as text_file:
text_file.write(str(test_gen_loss_accum/batch_size_accum) + ', ' + str(test_dis_loss_accum/batch_size_accum) + '\n')
# distributions.visualizeProductDistribution(sess, input_dict_func(batch), batch, inference_obs_dist, generative_dict['obs_dist'],
# save_dir = global_args.exp_dir+'Visualization/Test/', postfix='test_'+str(epoch))
print('Starting training.')
while global_args.curr_epoch < global_args.epochs + 1:
train(global_args.curr_epoch)
if global_args.curr_epoch % 20 == 0:
test(global_args.curr_epoch)
global_args.curr_epoch += 1