-
Notifications
You must be signed in to change notification settings - Fork 0
/
hyperparameter.py
45 lines (32 loc) · 1005 Bytes
/
hyperparameter.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
class Hyperparameter:
"""
Parameter for preprocessing, contains wasserstein VAE training.
"""
inst_num = 17
batch_size= 64
epochs = 100
lr = 0.0001
dim_h = 64
n_z = 32
LAMBDA = 1# It used in Wasserstein VAE, it depends on CONLON image normalizing method.
gan_LAMBDA = 1 # It used in Wasserstein VAE, in GAN based penalty.
n_channel = inst_num * 2 # 2 for pattern AE, 34 for CBIR AE
sigma = 1
pattern_length=4
normalize_factor = 6.45873774657 # CONLON velocity channel normalizing factor.
"""
Parameter for AE
"""
AE_batch_size = 32
AE_num_training_updates = 15000
AE_num_hiddens = 128
AE_num_residual_hiddens = 32
AE_num_residual_layers = 2
AE_embedding_dim = 512
AE_learning_rate = 1e-3
"""
Parameter for Graph Embedding Models
"""
GEM_batch_size = 64
GEM_learning_rate = 5e-3
GEM_input_dim = 256 + AE_embedding_dim