-
Notifications
You must be signed in to change notification settings - Fork 2
/
generate.py
416 lines (354 loc) · 18.9 KB
/
generate.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
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
"""Generate random images using the techniques described in the paper
"Elucidating the Design Space of Diffusion-Based Generative Models"."""
import os
import re
import click
import tqdm
import pickle
import json
import numpy as np
import torch
import PIL.Image
import dnnlib
from torch_utils import misc
from torch.nn.functional import interpolate
from torch_utils import distributed as dist
from torchvision.utils import save_image
from einops import rearrange
from training.datasets.dataset import WindowedDataset
from torchvision.transforms.functional import gaussian_blur
#----------------------------------------------------------------------------
# Deterministic EDM sampler.
def deterministic_edm_sampler(
net, latents, class_labels=None,
num_steps=18, sigma_min=0.002, sigma_max=80, rho=7
):
# Adjust noise levels based on what's supported by the network.
sigma_min = max(sigma_min, net.sigma_min)
sigma_max = min(sigma_max, net.sigma_max)
# Time step discretization.
step_indices = torch.arange(num_steps, dtype=torch.float64, device=latents.device)
t_steps = (sigma_max ** (1 / rho) + step_indices / (num_steps - 1) * (sigma_min ** (1 / rho) - sigma_max ** (1 / rho))) ** rho
t_steps = torch.cat([net.round_sigma(t_steps), torch.zeros_like(t_steps[:1])]) # t_N = 0
# Main sampling loop.
x_next = latents.to(torch.float64) * t_steps[0]
for i, (t_cur, t_next) in enumerate(zip(t_steps[:-1], t_steps[1:])): # 0, ..., N-1
x_cur = x_next
x_hat = x_cur
t_hat = t_cur
# Euler step.
denoised = net(x_hat, t_hat, class_labels).to(torch.float64)
d_cur = (x_hat - denoised) / t_hat
x_next = x_hat + (t_next - t_hat) * d_cur
# Apply 2nd order correction.
if i < num_steps - 1:
denoised = net(x_next, t_next, class_labels).to(torch.float64)
d_prime = (x_next - denoised) / t_next
x_next = x_hat + (t_next - t_hat) * (0.5 * d_cur + 0.5 * d_prime)
return x_next
def deterministic_ablation_sampler(
net, latents, class_labels=None,
num_steps=18, sigma_min=None, sigma_max=None, rho=7,
solver='heun', discretization='edm', schedule='linear', scaling='none',
epsilon_s=1e-3, C_1=0.001, C_2=0.008, M=1000, alpha=1
):
assert solver in ['euler', 'heun']
assert discretization in ['vp', 've', 'iddpm', 'edm']
assert schedule in ['vp', 've', 'linear']
assert scaling in ['vp', 'none']
# Helper functions for VP & VE noise level schedules.
vp_sigma = lambda beta_d, beta_min: lambda t: (np.e ** (0.5 * beta_d * (t ** 2) + beta_min * t) - 1) ** 0.5
vp_sigma_deriv = lambda beta_d, beta_min: lambda t: 0.5 * (beta_min + beta_d * t) * (sigma(t) + 1 / sigma(t))
vp_sigma_inv = lambda beta_d, beta_min: lambda sigma: ((beta_min ** 2 + 2 * beta_d * (sigma ** 2 + 1).log()).sqrt() - beta_min) / beta_d
ve_sigma = lambda t: t.sqrt()
ve_sigma_deriv = lambda t: 0.5 / t.sqrt()
ve_sigma_inv = lambda sigma: sigma ** 2
# Select default noise level range based on the specified time step discretization.
if sigma_min is None:
vp_def = vp_sigma(beta_d=19.9, beta_min=0.1)(t=epsilon_s)
sigma_min = {'vp': vp_def, 've': 0.02, 'iddpm': 0.002, 'edm': 0.002}[discretization]
if sigma_max is None:
vp_def = vp_sigma(beta_d=19.9, beta_min=0.1)(t=1)
sigma_max = {'vp': vp_def, 've': 100, 'iddpm': 81, 'edm': 80}[discretization]
# Adjust noise levels based on what's supported by the network.
sigma_min = max(sigma_min, net.sigma_min)
sigma_max = min(sigma_max, net.sigma_max)
# Compute corresponding betas for VP.
vp_beta_d = 2 * (np.log(sigma_min ** 2 + 1) / epsilon_s - np.log(sigma_max ** 2 + 1)) / (epsilon_s - 1)
vp_beta_min = np.log(sigma_max ** 2 + 1) - 0.5 * vp_beta_d
# Define time steps in terms of noise level.
step_indices = torch.arange(num_steps, dtype=torch.float64, device=latents.device)
if discretization == 'vp':
orig_t_steps = 1 + step_indices / (num_steps - 1) * (epsilon_s - 1)
sigma_steps = vp_sigma(vp_beta_d, vp_beta_min)(orig_t_steps)
elif discretization == 've':
orig_t_steps = (sigma_max ** 2) * ((sigma_min ** 2 / sigma_max ** 2) ** (step_indices / (num_steps - 1)))
sigma_steps = ve_sigma(orig_t_steps)
elif discretization == 'iddpm':
u = torch.zeros(M + 1, dtype=torch.float64, device=latents.device)
alpha_bar = lambda j: (0.5 * np.pi * j / M / (C_2 + 1)).sin() ** 2
for j in torch.arange(M, 0, -1, device=latents.device): # M, ..., 1
u[j - 1] = ((u[j] ** 2 + 1) / (alpha_bar(j - 1) / alpha_bar(j)).clip(min=C_1) - 1).sqrt()
u_filtered = u[torch.logical_and(u >= sigma_min, u <= sigma_max)]
sigma_steps = u_filtered[((len(u_filtered) - 1) / (num_steps - 1) * step_indices).round().to(torch.int64)]
else:
assert discretization == 'edm'
sigma_steps = (sigma_max ** (1 / rho) + step_indices / (num_steps - 1) * (sigma_min ** (1 / rho) - sigma_max ** (1 / rho))) ** rho
# Define noise level schedule.
if schedule == 'vp':
sigma = vp_sigma(vp_beta_d, vp_beta_min)
sigma_deriv = vp_sigma_deriv(vp_beta_d, vp_beta_min)
sigma_inv = vp_sigma_inv(vp_beta_d, vp_beta_min)
elif schedule == 've':
sigma = ve_sigma
sigma_deriv = ve_sigma_deriv
sigma_inv = ve_sigma_inv
else:
assert schedule == 'linear'
sigma = lambda t: t
sigma_deriv = lambda t: 1
sigma_inv = lambda sigma: sigma
# Define scaling schedule.
if scaling == 'vp':
s = lambda t: 1 / (1 + sigma(t) ** 2).sqrt()
s_deriv = lambda t: -sigma(t) * sigma_deriv(t) * (s(t) ** 3)
else:
assert scaling == 'none'
s = lambda t: 1
s_deriv = lambda t: 0
# Compute final time steps based on the corresponding noise levels.
t_steps = sigma_inv(net.round_sigma(sigma_steps))
t_steps = torch.cat([t_steps, torch.zeros_like(t_steps[:1])]) # t_N = 0
# Main sampling loop.
t_next = t_steps[0]
x_next = latents.to(torch.float64) * (sigma(t_next) * s(t_next))
for i, (t_cur, t_next) in enumerate(zip(t_steps[:-1], t_steps[1:])): # 0, ..., N-1
x_cur = x_next
x_hat = x_cur
t_hat = t_cur
# Euler step.
h = t_next - t_hat
denoised = net(x_hat / s(t_hat), sigma(t_hat), class_labels).to(torch.float64)
d_cur = (sigma_deriv(t_hat) / sigma(t_hat) + s_deriv(t_hat) / s(t_hat)) * x_hat - sigma_deriv(t_hat) * s(t_hat) / sigma(t_hat) * denoised
x_prime = x_hat + alpha * h * d_cur
t_prime = t_hat + alpha * h
# Apply 2nd order correction.
if solver == 'euler' or i == num_steps - 1:
x_next = x_hat + h * d_cur
else:
assert solver == 'heun'
denoised = net(x_prime / s(t_prime), sigma(t_prime), class_labels).to(torch.float64)
d_prime = (sigma_deriv(t_prime) / sigma(t_prime) + s_deriv(t_prime) / s(t_prime)) * x_prime - sigma_deriv(t_prime) * s(t_prime) / sigma(t_prime) * denoised
x_next = x_hat + h * ((1 - 1 / (2 * alpha)) * d_cur + 1 / (2 * alpha) * d_prime)
return x_next
#----------------------------------------------------------------------------
# Wrapper for torch.Generator that allows specifying a different random seed
# for each sample in a minibatch.
class StackedRandomGenerator:
def __init__(self, device, seeds):
super().__init__()
self.generators = [torch.Generator(device).manual_seed(int(seed) % (1 << 32)) for seed in seeds]
def randn(self, size, **kwargs):
assert size[0] == len(self.generators)
return torch.stack([torch.randn(size[1:], generator=gen, **kwargs) for gen in self.generators])
def randn_like(self, input):
return self.randn(input.shape, dtype=input.dtype, layout=input.layout, device=input.device)
def randint(self, *args, size, **kwargs):
assert size[0] == len(self.generators)
return torch.stack([torch.randint(*args, size=size[1:], generator=gen, **kwargs) for gen in self.generators])
#----------------------------------------------------------------------------
# Parse a comma separated list of numbers or ranges and return a list of ints.
# Example: '1,2,5-10' returns [1, 2, 5, 6, 7, 8, 9, 10]
def parse_int_list(s):
if isinstance(s, list): return s
ranges = []
range_re = re.compile(r'^(\d+)-(\d+)$')
for p in s.split(','):
m = range_re.match(p)
if m:
ranges.extend(range(int(m.group(1)), int(m.group(2))+1))
else:
ranges.append(int(p))
return ranges
#----------------------------------------------------------------------------
@click.command()
@click.option('--network', 'network_pkl', help='Network pickle filename', metavar='PATH|URL', type=str, required=True)
@click.option("--reload_network", help="If set, do not use network code pickled in checkpoint", is_flag=True)
@click.option("--resolution", help="Desired resolution of noise (and therefore generated images", type=int, default=None)
@click.option('--outfile', help='Where to save the output images', metavar='DIR', type=str, required=True)
@click.option('--subdirs', help='Create subdirectory for every 1000 seeds', is_flag=True)
# The number of forecasts (x's) we generate per x_t
@click.option('--examples_per_t', metavar='INT', type=click.IntRange(min=1), default=64, show_default=True)
# The number of timesteps y_t we consider, for t = {1, ..., t_max}.
@click.option('--t_max', help='Number of timesteps (examples) to generate in total', metavar='INT', type=click.IntRange(min=1), default=2)
# Batch size for generation.
@click.option('--batch_size', help='Batch size for generation', metavar='INT', type=click.IntRange(min=1), default=32)
@click.option('--num_workers', help='Number of workers for data loader', metavar='INT', type=click.IntRange(min=0), default=0)
#@click.option('--noise_kwargs', type=str, default="{}")
@click.option('--rbf_scale', help="RBF scale", metavar='INT', type=click.FloatRange(min=0, min_open=True), default=None)
@click.option('--steps', 'num_steps', help='Number of sampling steps', metavar='INT', type=click.IntRange(min=1), default=18, show_default=True)
@click.option('--sigma_min', help='Lowest noise level [default: varies]', metavar='FLOAT', type=click.FloatRange(min=0, min_open=True), default=0.0002)
@click.option('--sigma_max', help='Highest noise level [default: varies]', metavar='FLOAT', type=click.FloatRange(min=0, min_open=True))
@click.option('--rho', help='Time step exponent', metavar='FLOAT', type=click.FloatRange(min=0, min_open=True), default=7, show_default=True)
@click.option('--solver', help='Ablate ODE solver', metavar='euler|heun', type=click.Choice(['euler', 'heun']))
@click.option('--disc', 'discretization', help='Ablate time step discretization {t_i}', metavar='vp|ve|iddpm|edm', type=click.Choice(['vp', 've', 'iddpm', 'edm']))
@click.option('--schedule', help='Ablate noise schedule sigma(t)', metavar='vp|ve|linear', type=click.Choice(['vp', 've', 'linear']))
@click.option('--scaling', help='Ablate signal scaling s(t)', metavar='vp|none', type=click.Choice(['vp', 'none']))
def main(network_pkl,
reload_network,
resolution,
outfile,
subdirs,
examples_per_t,
t_max,
batch_size,
num_workers,
#noise_kwargs,
device=torch.device('cuda'),
**sampler_kwargs):
"""Generate random images using the techniques described in the paper
"Elucidating the Design Space of Diffusion-Based Generative Models".
Examples:
\b
# Generate 64 images and save them as out/*.png
python generate.py --outdir=out --seeds=0-63 --batch=64 \\
--network=https://nvlabs-fi-cdn.nvidia.com/edm/pretrained/edm-cifar10-32x32-cond-vp.pkl
"""
if (t_max*examples_per_t) % batch_size != 0:
raise ValueError("t_max * examples_per_t must be evenly divisible by batch_size!" + \
" values are {} * {}, batch_size = {}".format(t_max,examples_per_t,batch_size))
dist.init()
# Load dataset because we need to be able to sample y's to condition on.
exp_dir = os.path.dirname(network_pkl)
config = dnnlib.EasyDict(json.loads(
open(os.path.join(exp_dir, "training_options.json"), "r").read()
))
dist.print0('Loading dataset...')
dataset_obj = dnnlib.util.construct_class_by_name(**config.dataset_kwargs) # subclass of training.dataset.Dataset
dist.print0('Windowing dataset...')
dataset_obj = WindowedDataset(dataset_obj, window_size=config.window_size)
# Load network.
if reload_network:
# If this is set, do NOT load the network code from the pickle. Reconstruct
# the network from the actual current code and only load in the weights.
# This should be set if you've made post-hoc changes to the network code
# but are loading in weights corresponding to an older version.
dist.print0('Constructing network...')
interface_kwargs = dict(
img_resolution=dataset_obj.resolution,
img_channels=dataset_obj.num_channels,
label_dim=dataset_obj.label_dim
)
net = dnnlib.util.construct_class_by_name(**config.network_kwargs, **interface_kwargs) # subclass of torch.nn.Module
net.eval().requires_grad_(False).to(device)
dist.print0(f'Loading network, load weights from inside "{network_pkl}"...')
with dnnlib.util.open_url(network_pkl, verbose=(dist.get_rank() == 0)) as f:
net_weights = pickle.load(f)['ema'].to(device).state_dict()
net.load_state_dict(net_weights)
else:
dist.print0(f'Loading network from inside "{network_pkl}"...')
with dnnlib.util.open_url(network_pkl, verbose=(dist.get_rank() == 0)) as f:
net = pickle.load(f)['ema'].to(device)
dist.print0("Sampler kwargs: {}".format(sampler_kwargs))
dataset_sampler = misc.InfiniteSampler(
dataset=dataset_obj,
rank=dist.get_rank(),
num_replicas=dist.get_world_size(),
shuffle=False,
seed=0 # TODO make it an arg
)
data_loader_kwargs = dnnlib.EasyDict(
pin_memory=True,
num_workers=num_workers,
prefetch_factor=2 # what is this?
)
dataset_iterator = iter(
torch.utils.data.DataLoader(
dataset=dataset_obj,
sampler=dataset_sampler,
batch_size=t_max, # only return `t_max` images
**data_loader_kwargs
)
)
#noise_kwargs = json.loads(noise_kwargs)
dist.print0("Loading noise sampler...")
noise_sampler_kwargs = dnnlib.EasyDict(config.sampler_kwargs)
noise_sampler_kwargs.n_in = dataset_obj.num_channels
noise_sampler_kwargs.device = device
if resolution is not None:
noise_sampler_kwargs.Ln1 = resolution
noise_sampler_kwargs.Ln2 = resolution
# We can override arguments in the noise_sampler at generation time,
# for instance if we want to increase the resolution or change the
# smoothness of the noise.
"""
if len(noise_kwargs.keys()) > 0:
for key in noise_kwargs.keys():
if key in noise_sampler_kwargs:
noise_sampler_kwargs[key] = noise_kwargs[key]
dist.print0(f' noise_sampler: override {key}={noise_kwargs[key]} ...')
else:
raise ValueError(f'Unknown key for noise_sampler: "{key}"')
"""
noise_sampler = dnnlib.util.construct_class_by_name(**noise_sampler_kwargs)
# Pick latents and labels.
#rnd = StackedRandomGenerator(device, np.arange(0, examples_per_t).tolist())
# shape: (t_max, nc, h, w) and (t_max, w, nc, h, w)
images_real_, class_labels_ = next(dataset_iterator)
# t = timestep, ws = window size, nc = num channels
class_labels = rearrange(class_labels_, 't ws nc h w -> t (ws nc) h w')
class_labels = rearrange(class_labels, 't N h w -> t 1 N h w').\
repeat(1, examples_per_t, 1, 1, 1)
class_labels = rearrange(class_labels, 't rep N h w -> (t rep) N h w')
#images_real = images_real.view(-1, *tuple(images_real.shape[2:]))
class_labels = class_labels.to(device)
# batch_size = the number of conditioning images
# TODO parallelise this
buf_samples = []
N_total = class_labels.size(0)
n_iters = int(np.ceil(N_total / batch_size))
for j in range(n_iters):
dist.print0("Processing batch: {} / {} ...".format(j+1, n_iters))
this_slice = slice(j*batch_size, (j+1)*batch_size)
this_class_labels = class_labels[this_slice]
this_latents = noise_sampler.sample(this_class_labels.size(0)).to(device)
if this_class_labels.size(-1) != this_latents.size(-1):
# If we're doing super-resolution
dist.print0(f' `this_class_label` and `latents` spatial dim mismatch: {this_class_labels.size(-1)} and {this_latents.size(-1)}, upscaling `this_class_label`...')
this_class_labels = interpolate(
this_class_labels,
(this_latents.size(-2), this_latents.size(-1)),
mode='bilinear'
)
# Generate images.
sampler_kwargs = {key: value for key, value in sampler_kwargs.items() if value is not None}
have_ablation_kwargs = any(x in sampler_kwargs for x in ['solver', 'discretization', 'schedule', 'scaling'])
sampler_fn = deterministic_ablation_sampler if have_ablation_kwargs else deterministic_edm_sampler
samples = sampler_fn(net, this_latents, this_class_labels, **sampler_kwargs)
print(" samples min={}, max={}".format(samples.min(), samples.max()))
samples_torch = ((samples*0.5 + 0.5)).cpu()
buf_samples.append(samples_torch)
# shape = (t_max*examples_per_t, ch_x, h, w)
buf_samples = torch.cat(buf_samples, dim=0)
# shape = (t_max, examples_per_t, ch_x, h, w)
buf_samples = buf_samples.reshape(
t_max, examples_per_t, *tuple(buf_samples.shape[1:])
)
# shape = (t_max, ch_x, h, w)
images_real_ = (images_real_*0.5 + 0.5)
# shape = (t_max, ch_y, h, w)
class_labels_ = (class_labels_*0.5 + 0.5)
outdir = os.path.dirname(outfile)
if not os.path.exists(outdir):
os.makedirs(outdir)
dist.print0("Saving to: {}".format(outfile))
torch.save(
dict(gen=buf_samples, x=images_real_, y=class_labels_, metadata={}),
outfile
)
# Done.
dist.print0('Done.')
#----------------------------------------------------------------------------
if __name__ == "__main__":
main()
#----------------------------------------------------------------------------