-
Notifications
You must be signed in to change notification settings - Fork 0
/
config.yaml
72 lines (57 loc) · 2.2 KB
/
config.yaml
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
train: # Data flow generation 학습을 위한 configuration
# Basic settings
gpu_id: "0 1" # String 형태
start_epoch: 0
num_epochs: 1000
num_workers: 0 # 양수로 두면 multi-process data loading
save_summary_steps: 10
cur_iter: 0 # 현재까지 학습한 iteration --> domainess 설정과 tensorboard logging에 쓰임
total_iter: 0 # 학습해야하는 total iteration --> domainess 설정에서 사용
# Directorys
data_root_dir: "./datasets" # Target, Source 폴더가 있는 directory
checkpoint_dir: "./model/checkpoint" # Chcekpoint를 저장할 directory
restore_dir: False # Fill if restoring needed
output_dir: "./outputs"
# Related to dataset
S_nc: 3
T_nc: 3
S_H: 563 # Aspect ratio를 유지하면서 width가 1024가 되도록 하는 height --> Dataloader에 사용
T_H: 512
fixed_pair: False
# Hyperparameters
lr: 0.00002
beta: 0.5 # Adam optimizer hyperparameter
lambda_cycle: 10
lambda_ident: 10
batch_size: 2
# Architectures
ngf: 64
ndf: 32
nef: 32
nlatent: 16
use_dropout: False
n_res_blocks: 9
infer: # Data flow를 generation하기 위한 configuration
gpu_id: "0 1"
batch_size: 1
num_workers: 0 # 양수로 두면 multi-process data loading
# Directories
data_root_dir: "./datasets" # Target, Source 폴더가 있는 directory
infer_dir : "./datasets/GTA5" # Inference할 데이터가 있는 directory
checkpoint_dir: "./model/checkpoint"
DLOW_dir: "./datasets/DLOW"
DLOW_list_dir: "./datasets/Source_list"
# Related to dataset
S_nc: 3
T_nc: 3
S_H: 563 # Aspect ratio를 유지하면서 width가 1024가 되도록 하는 height
T_H: 512
fixed_pair: False
# Architectures
ngf: 64
ndf: 32
nef: 32
nlatent: 16
use_dropout: False
n_res_blocks: 9
test: