/
train.py
173 lines (145 loc) · 9.85 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
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
import argparse
import torch
from lib import *
from models.gan_load import build_biggan, build_proggan, build_stylegan2, build_sngan
from torch import nn
def main():
"""PotentialFlow -- Training script.
Options:
===[ Pre-trained GAN Generator (G) ]============================================================================
--gan-type : set pre-trained GAN type
--z-truncation : set latent code sampling truncation parameter. If set, latent codes will be sampled
from a standard Gaussian distribution truncated to the range [-args.z_truncation,
+args.z_truncation]
--biggan-target-classes : set list of classes to use for conditional BigGAN (see BIGGAN_CLASSES in
lib/config.py). E.g., --biggan-target-classes 14 239.
--stylegan2-resolution : set StyleGAN2 generator output images resolution: 256 or 1024 (default: 1024)
--shift-in-w-space : search latent paths in StyleGAN2's W-space (otherwise, look in Z-space)
===[ Support Sets (S) ]=========================================================================================
-K, --num-support-sets : set number of support sets; i.e., number of warping functions -- number of
interpretable paths
-D, --num_support_timesteps : set number of support dipoles per support set
--support-set-lr : set learning rate for learning support sets
===[ Reconstructor (R) ]========================================================================================
--reconstructor-type : set reconstructor network type
--min-shift-magnitude : set minimum shift magnitude
--max-shift-magnitude : set maximum shift magnitude
--reconstructor-lr : set learning rate for reconstructor R optimization
===[ Training ]=================================================================================================
--max-iter : set maximum number of training iterations
--batch-size : set training batch size
--lambda-cls : classification loss weight
--lambda-reg : regression loss weight
--log-freq : set number iterations per log
--ckp-freq : set number iterations per checkpoint model saving
--tensorboard : use TensorBoard
===[ CUDA ]=====================================================================================================
--cuda : use CUDA during training (default)
--no-cuda : do NOT use CUDA during training
================================================================================================================
"""
parser = argparse.ArgumentParser(description="Potential flow training script for pre-trained GANs")
# === Pre-trained GAN Generator (G) ============================================================================== #
parser.add_argument('--gan-type', type=str, choices=GAN_WEIGHTS.keys(), help='set GAN generator model type')
parser.add_argument('--z-truncation', type=float, help="set latent code sampling truncation parameter")
parser.add_argument('--biggan-target-classes', nargs='+', type=int, help="list of classes for conditional BigGAN")
parser.add_argument('--stylegan2-resolution', type=int, default=1024, choices=(256, 1024),
help="StyleGAN2 image resolution")
parser.add_argument('--shift-in-w-space', action='store_true', help="search latent paths in StyleGAN2's W-space")
# === Support Sets (S) ======================================================================== #
parser.add_argument('-K', '--num-support-sets', type=int, help="set number of support sets (potential functions)")
parser.add_argument('-D', '--num-support-timesteps', type=int, help="set number of timesteps per potential")
parser.add_argument('--support-set-lr', type=float, default=1e-4, help="set learning rate")
# === Reconstructor (R) ========================================================================================== #
parser.add_argument('--reconstructor-type', type=str, default='ResNet',
help='set reconstructor network type')
parser.add_argument('--reconstructor-lr', type=float, default=1e-4,
help="set learning rate for reconstructor R optimization")
# === Training =================================================================================================== #
parser.add_argument('--max-iter', type=int, default=100000, help="set maximum number of training iterations")
parser.add_argument('--batch-size', type=int, default=32, help="set batch size")
parser.add_argument('--lambda-cls', type=float, default=1.00, help="classification loss weight")
parser.add_argument('--lambda-reg', type=float, default=1.00, help="regression loss weight")
parser.add_argument('--lambda-pde', type=float, default=1.00, help="regression loss weight")
parser.add_argument('--log-freq', default=10, type=int, help='set number iterations per log')
parser.add_argument('--ckp-freq', default=1000, type=int, help='set number iterations per checkpoint model saving')
parser.add_argument('--tensorboard', action='store_true', help="use tensorboard")
# === CUDA ======================================================================================================= #
parser.add_argument('--cuda', dest='cuda', action='store_true', help="use CUDA during training")
parser.add_argument('--no-cuda', dest='cuda', action='store_false', help="do NOT use CUDA during training")
parser.set_defaults(cuda=True)
# ================================================================================================================ #
# Parse given arguments
args = parser.parse_args()
# Create output dir and save current arguments
exp_dir = create_exp_dir(args)
# CUDA
use_cuda = False
multi_gpu = False
if torch.cuda.is_available():
if args.cuda:
use_cuda = True
torch.set_default_tensor_type('torch.cuda.FloatTensor')
if torch.cuda.device_count() > 1:
multi_gpu = True
else:
print("*** WARNING ***: It looks like you have a CUDA device, but aren't using CUDA.\n"
" Run with --cuda for optimal training speed.")
torch.set_default_tensor_type('torch.FloatTensor')
else:
torch.set_default_tensor_type('torch.FloatTensor')
# Build GAN generator model and load with pre-trained weights
print("#. Build GAN generator model G and load with pre-trained weights...")
print(" \\__GAN type: {}".format(args.gan_type))
if args.gan_type == 'StyleGAN2':
print(" \\__Search for paths in {}-space".format('W' if args.shift_in_w_space else 'Z'))
if args.z_truncation:
print(" \\__Input noise truncation: {}".format(args.z_truncation))
print(" \\__Pre-trained weights: {}".format(
GAN_WEIGHTS[args.gan_type]['weights'][args.stylegan2_resolution] if args.gan_type == 'StyleGAN2' else
GAN_WEIGHTS[args.gan_type]['weights'][GAN_RESOLUTIONS[args.gan_type]]))
# === BigGAN ===
if args.gan_type == 'BigGAN':
G = build_biggan(pretrained_gan_weights=GAN_WEIGHTS[args.gan_type]['weights'][GAN_RESOLUTIONS[args.gan_type]],
target_classes=args.biggan_target_classes)
# === ProgGAN ===
elif args.gan_type == 'ProgGAN':
G = build_proggan(pretrained_gan_weights=GAN_WEIGHTS[args.gan_type]['weights'][GAN_RESOLUTIONS[args.gan_type]])
# === StyleGAN ===
elif args.gan_type == 'StyleGAN2':
G = build_stylegan2(pretrained_gan_weights=GAN_WEIGHTS[args.gan_type]['weights'][args.stylegan2_resolution],
resolution=args.stylegan2_resolution,
shift_in_w_space=args.shift_in_w_space)
# === Spectrally Normalised GAN (SNGAN) ===
else:
G = build_sngan(pretrained_gan_weights=GAN_WEIGHTS[args.gan_type]['weights'][GAN_RESOLUTIONS[args.gan_type]],
gan_type=args.gan_type)
# Build Support Sets model S
print("#. Build Support Sets S...")
print(" \\__Number of Potentials : {}".format(args.num_support_sets))
print(" \\__Number of Timesteps : {}".format(args.num_support_timesteps))
print(" \\__Support Vectors dim : {}".format(G.dim_z))
S = WavePDE(num_support_sets=args.num_support_sets,
num_support_timesteps=args.num_support_timesteps,
support_vectors_dim=G.dim_z)
# For stylegan remove the last activation layer otherwise the changes are too small
if args.gan_type == 'StyleGAN2':
for i in range(S.num_support_sets):
S.MLP_SET[i].activation4 = nn.Identity()
# Count number of trainable parameters
print(" \\__Trainable parameters: {:,}".format(sum(p.numel() for p in S.parameters() if p.requires_grad)))
# Build reconstructor model R
print("#. Build reconstructor model R...")
R = Reconstructor(reconstructor_type=args.reconstructor_type,
dim_index=S.num_support_sets,
dim_time=S.num_support_timesteps,
channels=1 if args.gan_type == 'SNGAN_MNIST' else 3)
# Count number of trainable parameters
print(" \\__Trainable parameters: {:,}".format(sum(p.numel() for p in R.parameters() if p.requires_grad)))
# Set up trainer
print("#. Experiment: {}".format(exp_dir))
trn = Trainer(params=args, exp_dir=exp_dir, use_cuda=use_cuda, multi_gpu=multi_gpu)
# Train
trn.train(generator=G, support_sets=S, reconstructor=R)
if __name__ == '__main__':
main()