forked from Seunghee-Koh/Piggyback-GAN-Pytorch
-
Notifications
You must be signed in to change notification settings - Fork 0
/
pb_cycleGAN.py
285 lines (240 loc) · 11.9 KB
/
pb_cycleGAN.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
# %%
import torch
import torch.nn as nn
import numpy as np
from configs.cycle_GAN_config import CycleGANOptions
import torch.multiprocessing as mp
import torch.distributed as dist
from dataloaders.dataloader import UnalignedDataset
import torch.utils
import os
from models.cycleGAN import CycleGAN
import time
from models.networks import PiggybackConv, PiggybackTransposeConv, load_pb_conv, make_filter_list
import copy
import sys
from pytorch_model_summary import summary as summary_
import pdb
# %%
def train(gpu, opt):
device = torch.device('cuda:{}'.format(gpu)) if gpu>=0 else torch.device('cpu')
if gpu >= 0:
torch.cuda.set_device(gpu)
rank = opt.nr * len(opt.gpu_ids) + gpu
dist.init_process_group(
backend='nccl',
init_method='env://',
world_size=opt.world_size,
rank=rank
)
model = CycleGAN(opt, device)
model = model.to(device)
model = nn.parallel.DistributedDataParallel(model, device_ids=[rank], output_device=rank)
if opt.train_continue:
state_dict = torch.load(opt.ckpt_save_path+'/latest_checkpoint.pt')
model.load_state_dict(state_dict['model'])
opt.start_epoch = state_dict['epoch']
opt.train_continue = False # to prevent load check point in another task
train_dataset = UnalignedDataset(opt)
train_sampler = torch.utils.data.distributed.DistributedSampler(
train_dataset,
num_replicas=opt.world_size,
rank=rank
)
train_loader = torch.utils.data.DataLoader(
dataset=train_dataset,
batch_size=opt.batch_size,
shuffle=False,
num_workers=0,
pin_memory=True,
sampler=train_sampler)
for epoch in range(opt.start_epoch, opt.n_epochs + opt.n_epochs_decay + 1):
epoch_start_time = time.time() # timer for entire epoch
iter_data_time = time.time() # timer for data loading per iteration
print("Length of loader is ",len(train_loader))
for i, data in enumerate(train_loader):
model.module.set_input(data)
model.module.optimize_parameters()
if (i+1) % 50 == 0 and gpu<=0:
print_str = (
f"Task: {opt.task_num} | Epoch: {epoch} | Iter: {i+1} | G_A: {model.module.loss_G_A:.5f} | "
f"G_B: {model.module.loss_G_B:.5f} | cycle_A: {model.module.loss_cycle_A:.5f} | "
f"cycle_B: {model.module.loss_cycle_B:.5f} | idt_A: {model.module.loss_idt_A:.5f} | "
f"idt_B: {model.module.loss_idt_B:.5f} | D_A: {model.module.loss_D_A:.5f} | "
f"D_B: {model.module.loss_D_A:.5f}"
)
print(print_str)
model.module.update_learning_rate()
if gpu<=0:
model.module.save_train_images(epoch)
save_dict = {'model': model.state_dict(),
'epoch': epoch
}
torch.save(save_dict, opt.ckpt_save_path+'/latest_checkpoint.pt')
dist.barrier()
if gpu <= 0:
make_filter_list(model.module.netG_A, opt.netG_A_filter_list, opt.weights_A, opt.task_num)
make_filter_list(model.module.netG_B, opt.netG_B_filter_list, opt.weights_B, opt.task_num)
# print(opt.netG_A_filter_list)
savedict_task = {'netG_A_filter_list':opt.netG_A_filter_list,
'netG_B_filter_list':opt.netG_B_filter_list,
'weights_A':opt.weights_A,
'weights_B':opt.weights_B
}
torch.save(savedict_task, opt.ckpt_save_path+'/filters.pt')
# del netG_A_layer_list
# del netG_B_layer_list
del opt.netG_A_filter_list
del opt.netG_B_filter_list
del opt.weights_A
del opt.weights_B
dist.barrier()
del model
dist.destroy_process_group()
# %%
def test(opt, task_idx):
opt.train = False
# device = torch.device('cpu')
device = torch.device('cuda:{}'.format(gpu)) if gpu>=0 else torch.device('cpu')
model = CycleGAN(opt, device)
model = model.to(device)
model.eval()
test_dataset = UnalignedDataset(opt)
test_loader = torch.utils.data.DataLoader(
dataset=test_dataset,
batch_size=1,
shuffle=False,
num_workers=4,
pin_memory=True)
model.netG_A = load_pb_conv(model.netG_A, opt.netG_A_filter_list, opt.weights_A, task_idx)
model.netG_B = load_pb_conv(model.netG_B, opt.netG_B_filter_list, opt.weights_B, task_idx)
for i, data in enumerate(test_loader):
model.set_input(data)
model.forward()
model.save_test_images(i)
print(f"Task {opt.task_num} : Image {i}")
if i > 50:
break
del model
image_path_list = opt.img_save_path
image_real = 'real_A'
image_fake = 'rec_A'
fid_value = calculate_fid_given_paths(image_path_list, [image_real, image_fake],
50,
True,
2048)
# %%
def main():
opt = CycleGANOptions().parse()
tasks = ['cityscapes', 'maps', 'facades', 'vangogh2photo']
torch.manual_seed(0)
np.random.seed(0)
torch.cuda.manual_seed(0)
torch.cuda.manual_seed_all(0)
if opt.train:
start_task = opt.st_task_idx
end_task = len(tasks)
opt.world_size = len(opt.gpu_ids) * opt.nodes
os.environ['MASTER_ADDR'] = 'localhost'
os.environ['MASTER_PORT'] = '8888'
for task_idx in range(start_task, end_task):
# Create Task folder
opt.task_folder_name = "Task_"+str(task_idx+1)+"_"+tasks[task_idx]+"_"+"cycleGAN"
opt.image_folder_name = "Intermediate_train_images"
if not os.path.exists(os.path.join(opt.checkpoints_dir, opt.task_folder_name, opt.image_folder_name)):
os.makedirs(os.path.join(opt.checkpoints_dir, opt.task_folder_name, opt.image_folder_name))
opt.ckpt_save_path = os.path.join(opt.checkpoints_dir, opt.task_folder_name)
opt.img_save_path = os.path.join(opt.checkpoints_dir, opt.task_folder_name, opt.image_folder_name)
if task_idx == 0:
netG_A_filter_list = []
netG_B_filter_list = []
weights_A = []
weights_B = []
else:
old_task_folder_name = "Task_"+str(task_idx)+"_"+tasks[task_idx-1]+"_"+"cycleGAN"
print("Loading ", os.path.join(opt.checkpoints_dir, old_task_folder_name)+'/filters.pt')
filters = torch.load(os.path.join(opt.checkpoints_dir, old_task_folder_name)+'/filters.pt')
netG_A_filter_list = filters["netG_A_filter_list"]
netG_B_filter_list = filters["netG_B_filter_list"]
weights_A = filters["weights_A"]
weights_B = filters["weights_B"]
opt.netG_A_filter_list = netG_A_filter_list
opt.netG_B_filter_list = netG_B_filter_list
opt.weights_A = weights_A
opt.weights_B = weights_B
opt.dataroot = '../pytorch-CycleGAN-and-pix2pix/datasets/' + tasks[task_idx]
opt.task_num = task_idx+1
opt.task_lambda = 0.25
if opt.taskwise_lambda:
if opt.train_continue:
state_dict = torch.load(opt.ckpt_save_path + '/latest_checkpoint.pt')
opt.task_lambda = state_dict['task_lambda']
else:
from models.lambda_calculators import get_task_lambda
# task_lambdas = [0.125, 0.0625, 0.375, 0.5, 0.75, 0.375]
# opt.task_lambda = task_lambdas[task_idx]
opt.task_lambda = get_task_lambda(opt, 0, task_idx)
print(f"Task{task_idx}: lambda {opt.task_lambda}")
# if task_idx == 1:
# from models.networks import define_G
# from torchsummary import summary
# device='cuda:0'
# netG = define_G(3,3,64,'resnet_6blocks','instance',False,'normal',0.02,2,opt.netG_A_filter_list)
# netG.to(device)
# class Idx():
# def __init__(self):
# self.idx = 0
# def plus(self):
# self.idx += 1
# def make_filter(network, filters, weights, task_num, conv_idx):
# if isinstance(network, PiggybackConv) or isinstance(network, PiggybackTransposeConv):
# network.unc_filt.requires_grad = False
# if task_num == 1:
# filters.append([network.unc_filt.detach().cpu()])
# elif task_num == 2:
# filters[conv_idx.idx].append(network.unc_filt.detach().cpu())
# weights.append([network.weights_mat.detach().cpu()])
# conv_idx.plus()
# #print(f"conv_idx inside function: {conv_idx}")
# else:
# filters[task_num-1][conv_idx.idx].append(network.unc_filt.detach().cpu())
# weights[task_num-1][conv_idx.idx].append(network.weights_mat.detach().cpu())
# conv_idx.plus()
# print(f"conv_idx inter function: {conv_idx.idx}")
# for name, child in network.named_children():
# print(f"conv_idx outside function: {conv_idx.idx}")
# make_filter(child, filters, weights, task_num, conv_idx)
# idx = Idx()
# pdb.set_trace()
# make_filter_list(netG, opt.netG_A_filter_list, opt.weights_A, opt.task_num)
# pdb.set_trace()
mp.spawn(train, nprocs=len(opt.gpu_ids), args=(opt,))
else:
'''
We will load the unconstrained filters and the weights ONLY from the last task.
This is because, after every task we store the unconstrined filter and weight
matrix of that task and all the previous ones. So we will only load from the last one
which will contain everything we need.
'''
print("In Testing mode")
start_task = 0
end_task = len(tasks)
load_filter_path = opt.checkpoints_dir+f"/Task_{len(tasks)}_{tasks[-1]}_cycleGAN/filters.pt"
opt.load_filter_path = load_filter_path
filters = torch.load(opt.load_filter_path)
opt.netG_A_filter_list = filters["netG_A_filter_list"]
opt.netG_B_filter_list = filters["netG_B_filter_list"]
opt.weights_A = filters["weights_A"]
opt.weights_B = filters["weights_B"]
opt.image_folder_name = "Test_images"
for task_idx in range(start_task, end_task):
print(f"Task {task_idx+1}")
opt.task_folder_name = "Task_"+str(task_idx+1)+"_"+tasks[task_idx]+"_"+"cycleGAN"
opt.img_save_path = os.path.join(opt.checkpoints_dir, opt.task_folder_name, opt.image_folder_name)
if not os.path.exists(os.path.join(opt.checkpoints_dir, opt.task_folder_name, opt.image_folder_name)):
os.makedirs(os.path.join(opt.checkpoints_dir, opt.task_folder_name, opt.image_folder_name))
opt.dataroot = '../pytorch-CycleGAN-and-pix2pix/datasets/' + tasks[task_idx]
opt.task_num = task_idx+1
test(opt, task_idx)
if __name__ == "__main__":
main()