-
Notifications
You must be signed in to change notification settings - Fork 0
/
detaux.py
168 lines (136 loc) · 5.9 KB
/
detaux.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
#!/usr/bin/env python
# -*- coding: utf-8 -*-
import os
import wandb
from datetime import datetime
from argparse import ArgumentParser
from pytorch_lightning.loggers import WandbLogger
from pytorch_lightning import Trainer, seed_everything
from pytorch_lightning.callbacks import ModelCheckpoint
from models import conv_models
from models.resnet_ae import get_configs
from models.dis_large_model import LargeModel
from models.dis_basic_model import BasicModel
from data.dis_datasets import dataset_names
def train(args):
print(f'Running with arguments: {args}')
# Seeds for reproducibility.
seed = args.seed
seed_everything(seed, workers=True)
# Create run_name.
dt_string = datetime.now().strftime('%d-%m-%Y-%H-%M')
run_name = f'DIS_{dataset_names[args.n_dataset]}_BS{args.batch_size}_E{args.max_epochs}_NLFactors{args.n_latent_factors}_LatentDim{args.latentdim}'
if args.k > 0:
run_name += f'_k{args.k}'
else:
run_name += '_kall'
if args.occlusion:
run_name += f'_occlusion'
if args.noise:
run_name += f'_noise'
if args.crop:
run_name += f'_crop'
if args.observation >= 0:
run_name += f'_obs{args.observation}'
if args.oracle_probability:
run_name += f'_op{args.oracle_probability}'
if args.random_item:
run_name += f'_randomitem'
if args.full_random:
run_name += f'_fullrandom'
# Additional information in the run name.
if args.add_info:
run_name += f'_{args.add_info}'
if args.forced:
assert args.observation >= 0 and args.oracle_probability
run_name += f'_forced'
run_name += f'_{dt_string}'
# TODO: Check if this is correct.
if args.ckpt:
run_name = os.path.basename(args.ckpt)[:-5]
if args.wandb:
wandb_logger = WandbLogger(
project="",
name=run_name,
reinit=True,
entity=''
)
# Load autoencoder architecture.
if args.conv_model == 'resnetae':
resnet_configs, _ = get_configs("resnet18")
encoder_args = [resnet_configs, args.latentdim]
decoder_args = [resnet_configs[::-1], args.latentdim]
else:
encoder_args = [args.latentdim, args.in_dim, 64, args.batch_size, False]
decoder_args = [args.latentdim, args.in_dim, 64, False]
# 8 = ImageWoof, 13 = CUB, 14 = MEDIC, 15 = Cars, 16 = Pets, 17 = CIFAR, 18 = SVHN.
if args.n_dataset in [8, 13, 14, 15, 16, 17, 18]:
# AE w/o FC bottleneck.
model = LargeModel(
hparams=args,
encoder=conv_models[args.conv_model][0](*encoder_args, final_fc=False),
decoder=conv_models[args.conv_model][1](*decoder_args, initial_fc=False)
)
else:
# AE w/ FC bottleneck.
model = BasicModel(
hparams=args,
encoder=conv_models[args.conv_model][0](*encoder_args),
decoder=conv_models[args.conv_model][1](*decoder_args)
)
checkpoint_callback = ModelCheckpoint(
dirpath=args.save_dir,
filename=run_name + "_{epoch}",
monitor="val_loss",
mode="min",
save_last=True,
save_top_k=1,
verbose=True
)
checkpoint_callback.CHECKPOINT_NAME_LAST = f"last_{run_name}"
if args.wandb:
# Without 16 bit precision.
trainer = Trainer.from_argparse_args(args, devices=[args.gpu_idx], accelerator="gpu", logger=wandb_logger, callbacks=[checkpoint_callback], deterministic=True)
elif args.save_dir:
# Without 16 bit precision.
trainer = Trainer.from_argparse_args(args, devices=[args.gpu_idx], accelerator="gpu", callbacks=[checkpoint_callback], deterministic=True)
else:
trainer = Trainer.from_argparse_args(args, devices=[args.gpu_idx], accelerator="gpu", deterministic=True)
if not args.ckpt:
trainer.fit(model)
output = trainer.predict(ckpt_path=trainer.checkpoint_callback.last_model_path)
else:
output = trainer.predict(model, ckpt_path=args.ckpt)
print("Done!")
if __name__ == "__main__":
parser = ArgumentParser(description='Detaux: Disentanglement.')
parser = Trainer.add_argparse_args(parser)
parser.set_defaults(max_epochs=-1)
parser.set_defaults(check_val_every_n_epoch=1)
parser = LargeModel.add_model_specific_args(parser)
parser.add_argument('--seed', type=int, help='seed for randoms', default=1234)
parser.add_argument('--dataset_path', type=str)
parser.add_argument('--warmup_c', action='store_true')
parser.add_argument('--latattn', action='store_true')
parser.add_argument('--contrast', action='store_true')
parser.add_argument('--dataset_fraction', type=float, default=1.0)
parser.add_argument('--save_dir', type=str)
parser.add_argument('--gpu_idx', type=int, default=0)
parser.add_argument('--wandb_tags', type=str, nargs='*', default=[])
parser.add_argument('--conv_model', type=str, choices=list(conv_models.keys()), default='simpleconv')
parser.add_argument('--loss', type=str, default='mse')
parser.add_argument('--add_info', type=str)
parser.add_argument('--occlusion', action='store_true')
parser.add_argument('--noise', action='store_true')
parser.add_argument('--crop', action='store_true')
parser.add_argument('--full_random', action='store_true')
parser.add_argument('--observation', type=int, default=-1)
parser.add_argument('--oracle_probability', type=float)
parser.add_argument('--random_item', action='store_true')
parser.add_argument('--outpath', type=str, default='latent_spaces')
parser.add_argument('--ckpt', type=str, help='if given, skip training and try testing with this checkpoint')
# Store false for sweep (tmp).
parser.add_argument('--forced', action='store_true', help='if True, use the forced disentanglement model with the modified oracle')
parser.add_argument('--wandb', action='store_true')
args = parser.parse_args()
train(args)