-
Notifications
You must be signed in to change notification settings - Fork 0
/
train.py
157 lines (132 loc) · 5.44 KB
/
train.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
import torch
from dataset import LowNormalDataset
from utils import save_checkpoint, load_checkpoint
from torch.utils.data import DataLoader
import torch.nn as nn
import torch.optim as optim
import config
from tqdm import tqdm
from torchvision.utils import save_image
from discriminator_model import Discriminator
from generator_model import Generator
def train_fn(disc_L, disc_N, gen_N, gen_L, loader, opt_disc, opt_gen, l1, mse, d_scaler, g_scaler):
L_reals = 0
L_fakes = 0
N_reals = 0
N_fakes = 0
loop = tqdm(loader, leave=True)
for idx, (normal, low) in enumerate(loop):
normal = normal.to(config.DEVICE)
low = low.to(config.DEVICE)
# Train Discriminators L and N
with torch.cuda.amp.autocast():
fake_low = gen_L(normal)
D_L_real = disc_L(low)
D_L_fake = disc_L(fake_low.detach())
L_reals += D_L_real.mean().item()
L_fakes += D_L_fake.mean().item()
D_L_real_loss = mse(D_L_real, torch.ones_like(D_L_real))
D_L_fake_loss = mse(D_L_fake, torch.zeros_like(D_L_fake))
D_L_loss = D_L_real_loss + D_L_fake_loss
fake_normal = gen_N(low)
D_N_real = disc_N(normal)
D_N_fake = disc_N(fake_normal.detach())
N_reals += D_N_real.mean().item()
N_fakes += D_N_fake.mean().item()
D_N_real_loss = mse(D_N_real, torch.ones_like(D_N_real))
D_N_fake_loss = mse(D_N_fake, torch.zeros_like(D_N_fake))
D_N_loss = D_N_real_loss + D_N_fake_loss
# put it together
D_loss = (D_L_loss + D_N_loss) / 2
opt_disc.zero_grad()
d_scaler.scale(D_loss).backward()
d_scaler.step(opt_disc)
d_scaler.update()
# Train Generators L and N
with torch.cuda.amp.autocast():
# adversarial loss for both generators
D_L_fake = disc_L(fake_low)
D_N_fake = disc_N(fake_normal)
loss_G_L = mse(D_L_fake, torch.ones_like(D_L_fake))
loss_G_N = mse(D_N_fake, torch.ones_like(D_N_fake))
# cycle loss
cycle_normal = gen_N(fake_low)
cycle_low = gen_L(fake_normal)
cycle_normal_loss = l1(normal, cycle_normal)
cycle_low_loss = l1(low, cycle_low)
# identity loss (remove these for efficiency if you set lambda_identity=0)
identity_normal = gen_N(normal)
identity_low = gen_L(low)
identity_normal_loss = l1(normal, identity_normal)
identity_low_loss = l1(low, identity_low)
# add all together
G_loss = (
loss_G_N
+ loss_G_L
+ cycle_normal_loss * config.LAMBDA_CYCLE
+ cycle_low_loss * config.LAMBDA_CYCLE
+ identity_low_loss * config.LAMBDA_IDENTITY
+ identity_normal_loss * config.LAMBDA_IDENTITY
)
opt_gen.zero_grad()
g_scaler.scale(G_loss).backward()
g_scaler.step(opt_gen)
g_scaler.update()
if idx == 200:
# save_image(fake_low * 0.5 + 0.5, f"gen_images/low_fake_{idx}.png")
save_image(fake_normal * 0.5 + 0.5, f"gen_images/normal_fake_{idx}.png")
loop.set_postfix(L_real=L_reals / (idx + 1), L_fake=L_fakes / (idx + 1),
N_real=N_reals / (idx + 1), N_fake=N_fakes / (idx + 1))
def main():
disc_L = Discriminator(in_channels=3).to(config.DEVICE)
disc_N = Discriminator(in_channels=3).to(config.DEVICE)
gen_N = Generator(img_channels=3, num_residuals=9).to(config.DEVICE)
gen_L = Generator(img_channels=3, num_residuals=9).to(config.DEVICE)
opt_disc = optim.Adam(
list(disc_L.parameters()) + list(disc_N.parameters()),
lr=config.LEARNING_RATE,
betas=(0.5, 0.999),
)
opt_gen = optim.Adam(
list(gen_N.parameters()) + list(gen_L.parameters()),
lr=config.LEARNING_RATE,
betas=(0.5, 0.999),
)
L1 = nn.L1Loss()
mse = nn.MSELoss()
if config.LOAD_MODEL:
load_checkpoint(
config.CHECKPOINT_GEN_L, gen_L, opt_gen, config.LEARNING_RATE,
)
load_checkpoint(
config.CHECKPOINT_GEN_N, gen_N, opt_gen, config.LEARNING_RATE,
)
load_checkpoint(
config.CHECKPOINT_CRITIC_L, disc_L, opt_disc, config.LEARNING_RATE,
)
load_checkpoint(
config.CHECKPOINT_CRITIC_N, disc_N, opt_disc, config.LEARNING_RATE,
)
dataset = LowNormalDataset(
root_low=config.TRAIN_DIR + "/low",
root_normal=config.TRAIN_DIR + "/normal",
transform=config.transforms
)
loader = DataLoader(
dataset,
batch_size=config.BATCH_SIZE,
shuffle=False,
num_workers=config.NUM_WORKERS,
pin_memory=True
)
g_scaler = torch.cuda.amp.GradScaler()
d_scaler = torch.cuda.amp.GradScaler()
for epoch in range(config.NUM_EPOCHS):
train_fn(disc_L, disc_N, gen_N, gen_L, loader, opt_disc, opt_gen, L1, mse, d_scaler, g_scaler)
if config.SAVE_MODEL:
save_checkpoint(gen_L, opt_gen, filename=config.CHECKPOINT_GEN_L)
save_checkpoint(gen_N, opt_gen, filename=config.CHECKPOINT_GEN_N)
save_checkpoint(disc_L, opt_disc, filename=config.CHECKPOINT_CRITIC_L)
save_checkpoint(disc_N, opt_disc, filename=config.CHECKPOINT_CRITIC_N)
if __name__ == "__main__":
main()