-
Notifications
You must be signed in to change notification settings - Fork 5
/
sample_conditional.py
256 lines (219 loc) · 8.14 KB
/
sample_conditional.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
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
import copy
import math
from functools import partial
from inspect import isfunction
from pathlib import Path
import numpy as np
import torch
import torch.nn.functional as F
from einops import rearrange
from PIL import Image
from torch import einsum, nn, optim
from torch.nn import functional as F
from torch.optim import Adam
from torch.utils import data
from torchvision import transforms, utils
from tqdm import tqdm
from dataset import create_loader
from model import select_model
from utils.util import set_seed
import argparse
import os
import os.path as osp
import numpy as np
import torch as th
import torch.distributed as dist
from model import select_model
from model.set_diffusion import dist_util, logger
from model.set_diffusion.script_util import (
NUM_CLASSES,
add_dict_to_argparser,
args_to_dict,
create_model_and_diffusion,
model_and_diffusion_defaults,
)
from utils.path import set_folder
from utils.util import count_params, set_seed
DIR = set_folder()
def main():
args = create_argparser().parse_args()
print(args)
# dct = vars(args)
# for k in sorted(dct):
# print(k, dct[k])
# print()
# dist_util.setup_dist()
logger.configure(
dir=DIR,
mode="-".join(["sampling-conditional", args.mode_conditional_sampling]),
args=args,
tag="",
)
logger.log("creating model and diffusion...")
model = select_model(args)(args)
print(count_params(model))
#print(args.model_path)
_path = list(args.model_path.split("/"))
for j in range(len(_path)):
_p = list(_path[j].split("_"))
_p = [ i for i in _p if i not in ["", None, "None", "none"] ]
_path[j] = "_".join(_p)
#_path = list(args.model_path.split("_"))
#_path = [ i for i in _path if i not in ["", None, "None", "none"] ]
model_path = "/".join(_path)
#print(model_path)
model.load_state_dict(
dist_util.load_state_dict(osp.join(DIR, model_path), map_location="cpu")
)
model.to(args.device)
if args.use_fp16:
model.convert_to_fp16()
model.eval()
logger.log("creating data loader...")
args.transfer=False
if args.transfer:
if args.image_size == 32:
args.dataset = "minimagenet"
else:
args.dataset = "cub"
if args.mode_conditional_sampling == "out-distro":
loader = create_loader(args, split="test", shuffle=True)
else:
loader = create_loader(args, split="train", shuffle=True)
logger.log("conditional sampling...")
all_images = []
all_labels = []
all_conditioning_images = []
while len(all_images) * args.batch_size < args.num_samples * args.k:
with torch.no_grad():
try:
x_set = next(loader)
except StopIteration:
loader = iter(loader)
x_set = next(loader)
x_set = x_set.to(args.device)
if args.model == "ddpm":
c = None
else:
c = model.sample_conditional(x_set, args.sample_size)["c"]
c = c.unsqueeze(1) # attention here
c = torch.repeat_interleave(c, args.k * args.sample_size, dim=1)
c = c.view(-1, c.size(-2), c.size(-1))
print(c.size())
# c_list = []
# for i in range(x_set.shape[1]):
# x_set_tmp = x_set.clone()
# x_set_tmp[:, i] = 0
# # build c
# out = model.sample_conditional(x_set_tmp, args.sample_size, 1)
# c_tmp = out["c"]
# c_list.append(c_tmp.unsqueeze(1))
# del x_set_tmp
# th.cuda.empty_cache()
# c = th.cat(c_list, dim=1)
# c = c.view(-1, c.size(-2), c.size(-1))
model_kwargs = {}
if args.class_cond:
classes = th.randint(
low=0, high=NUM_CLASSES, size=(args.batch_size,), device=dist_util.dev()
)
model_kwargs["y"] = classes
sample_fn = (
model.diffusion.p_sample_loop
if not args.use_ddim
else model.diffusion.ddim_sample_loop
)
try:
sample = sample_fn(
model.generative_model,
(
args.batch_size * args.sample_size * args.k,
args.in_channels,
args.image_size,
args.image_size,
),
c=c,
clip_denoised=args.clip_denoised,
model_kwargs=model_kwargs,
)
# [-1, 1] ---> [0, 2] ---> [0, 255]
sample = ((sample + 1) * 127.5).clamp(0, 255).to(th.uint8)
sample = sample.permute(0, 2, 3, 1)
sample = sample.contiguous()
# fix this
# gathered_samples = [th.zeros_like(sample) for _ in range(dist.get_world_size())]
# dist.all_gather(gathered_samples, sample) # gather not supported with NCCL
# all_images.extend([sample.cpu().numpy() for sample in gathered_samples])
all_images.extend([sample.cpu().numpy()])
x_set = x_set.view(-1, args.in_channels, args.image_size, args.image_size)
x_set = ((x_set + 1) * 127.5).clamp(0, 255).to(th.uint8)
x_set = x_set.permute(0, 2, 3, 1)
x_set = x_set.contiguous()
# normalize x_set
all_conditioning_images.extend([x_set.cpu().numpy()])
if args.class_cond:
# gathered_labels = [
# th.zeros_like(classes) for _ in range(dist.get_world_size())
# ]
# dist.all_gather(gathered_labels, classes)
# all_labels.extend([labels.cpu().numpy() for labels in gathered_labels])
all_labels.extend([classes.cpu().numpy()])
logger.log(f"created {len(all_images) * args.batch_size} samples")
# problem with the last batch in the loader
except RuntimeError:
continue
arr = np.concatenate(all_images, axis=0)
arr = arr[: args.num_samples]
arr_cond = np.concatenate(all_conditioning_images, axis=0)
arr_cond = arr_cond[: args.num_samples]
if args.class_cond:
label_arr = np.concatenate(all_labels, axis=0)
label_arr = label_arr[: args.num_samples]
# if dist.get_rank() == 0:
shape_str = "x".join([str(x) for x in arr.shape])
if args.transfer:
out_path = os.path.join(logger.get_dir(), f"full_samples_conditional_{shape_str}_{args.mode_conditional_sampling}_{args.sample_size}_transfer_{args.dataset}.npz")
else:
out_path = os.path.join(logger.get_dir(), f"full_samples_conditional_{shape_str}_{args.mode_conditional_sampling}_{args.sample_size}.npz")
logger.log(f"saving to {out_path}")
if args.class_cond:
np.savez(out_path, arr, label_arr)
else:
np.savez(out_path, arr, arr_cond)
# dist.barrier()
logger.log("sampling complete")
def create_argparser():
defaults = dict(
clip_denoised=True,
num_samples=10000,
batch_size=32,
batch_size_eval=32,
use_ddim=False,
model_path="",
k=1, # multiplier for conditional samples
model="vfsddpm",
dataset="cifar100",
pool='cls', # mean, mean_patch
image_size=32,
sample_size=5,
patch_size=8,
hdim=256,
in_channels=3,
encoder_mode="vit",
context_channels=256,
num_classes=1,
mode_context="deterministic",
mode_conditional_sampling="out-distro",
mode_conditioning='bias', # film, lag,
augment=False,
device="cuda",
data_dir="/home/gigi/ns_data",
transfer=False,
)
defaults.update(model_and_diffusion_defaults())
parser = argparse.ArgumentParser()
add_dict_to_argparser(parser, defaults)
return parser
if __name__ == "__main__":
s = set_seed(0)
main()