cGAN_tutorial/setupmodel.py at main · icpac-igad/cGAN_tutorial · GitHub

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
import gc

import tensorflow as tf
from tensorflow import keras
from keras.optimizers import Adam

from model import Deterministic, WGANGP, VAE, generator, discriminator


def setup_model(
    *,
    mode=None,
    arch=None,
    downscaling_steps=None,
    input_channels=None,
    constant_fields=None,
    filters_gen=None,
    filters_disc=None,
    noise_channels=None,
    latent_variables=None,
    padding=None,
    kl_weight=None,
    ensemble_size=None,
    CLtype=None,
    content_loss_weight=None,
    lr_disc=None,
    lr_gen=None
):
    if mode in ("GAN", "VAEGAN"):
        gen_to_use = {
            "normal": generator,
            "forceconv": generator,
            "forceconv-long": generator,
        }[arch]
        disc_to_use = {
            "normal": discriminator,
            "forceconv": discriminator,
            "forceconv-long": discriminator,
        }[arch]
    elif mode == "det":
        gen_to_use = {"normal": generator, "forceconv": generator}[arch]

    if mode == "GAN":
        gen = gen_to_use(
            mode=mode,
            arch=arch,
            downscaling_steps=downscaling_steps,
            input_channels=input_channels,
            constant_fields=constant_fields,
            filters_gen=filters_gen,
            noise_channels=noise_channels,
            padding=padding,
        )
        disc = disc_to_use(
            arch=arch,
            downscaling_steps=downscaling_steps,
            input_channels=input_channels,
            constant_fields=constant_fields,
            filters_disc=filters_disc,
            padding=padding,
        )
        model = WGANGP(
            gen,
            disc,
            mode,
            lr_disc=lr_disc,
            lr_gen=lr_gen,
            ensemble_size=ensemble_size,
            CLtype=CLtype,
            content_loss_weight=content_loss_weight,
        )
    elif mode == "VAEGAN":
        encoder, decoder = gen_to_use(
            mode=mode,
            arch=arch,
            downscaling_steps=downscaling_steps,
            input_channels=input_channels,
            constant_fields=constant_fields,
            filters_gen=filters_gen,
            latent_variables=latent_variables,
            padding=padding,
        )
        disc = disc_to_use(
            arch=arch,
            downscaling_steps=downscaling_steps,
            input_channels=input_channels,
            constant_fields=constant_fields,
            filters_disc=filters_disc,
            padding=padding,
        )
        gen = VAE(encoder, decoder)
        model = WGANGP(
            gen,
            disc,
            mode,
            lr_disc=lr_disc,
            lr_gen=lr_gen,
            kl_weight=kl_weight,
            ensemble_size=ensemble_size,
            CLtype=CLtype,
            content_loss_weight=content_loss_weight,
        )
    elif mode == "det":
        gen = gen_to_use(
            mode=mode,
            arch=arch,
            downscaling_steps=downscaling_steps,
            input_channels=input_channels,
            constant_fields=constant_fields,
            filters_gen=filters_gen,
            padding=padding,
        )
        model = Deterministic(gen, lr=lr_gen, loss="mse", optimizer=Adam)

    gc.collect()
    return model