Spaces:
Runtime error
Runtime error
Mehdi Cherti
commited on
Commit
•
27911d6
1
Parent(s):
ee8f9c5
simplify
Browse files- clip_encoder.py +1 -1
- run.py → model_configs.py +36 -93
- test_ddgan.py +83 -157
clip_encoder.py
CHANGED
@@ -16,7 +16,7 @@ class CLIPEncoder(nn.Module):
|
|
16 |
self.model, _, _ = open_clip.create_model_and_transforms(model, pretrained=pretrained)
|
17 |
self.output_size = self.model.transformer.width
|
18 |
|
19 |
-
def forward(self, texts, return_only_pooled=
|
20 |
device = next(self.parameters()).device
|
21 |
toks = open_clip.tokenize(texts).to(device)
|
22 |
x = self.model.token_embedding(toks) # [batch_size, n_ctx, d_model]
|
|
|
16 |
self.model, _, _ = open_clip.create_model_and_transforms(model, pretrained=pretrained)
|
17 |
self.output_size = self.model.transformer.width
|
18 |
|
19 |
+
def forward(self, texts, return_only_pooled=False):
|
20 |
device = next(self.parameters()).device
|
21 |
toks = open_clip.tokenize(texts).to(device)
|
22 |
x = self.model.token_embedding(toks) # [batch_size, n_ctx, d_model]
|
run.py → model_configs.py
RENAMED
@@ -10,30 +10,41 @@ def base():
|
|
10 |
"n": 8,
|
11 |
},
|
12 |
"model":{
|
13 |
-
"dataset"
|
14 |
-
"
|
15 |
-
"
|
16 |
"num_channels": 3,
|
|
|
|
|
|
|
|
|
17 |
"num_channels_dae": 128,
|
18 |
-
"
|
19 |
-
"
|
20 |
"num_res_blocks": 2,
|
21 |
-
"
|
22 |
-
"
|
23 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
24 |
"embedding_type": "positional",
|
25 |
-
"
|
26 |
-
"
|
27 |
-
"
|
|
|
|
|
28 |
"z_emb_dim": 256,
|
29 |
-
"
|
30 |
-
"
|
31 |
-
"
|
32 |
-
"
|
33 |
-
|
34 |
-
"masked_mean": "",
|
35 |
-
"resume": "",
|
36 |
-
},
|
37 |
}
|
38 |
def ddgan_cc12m_v2():
|
39 |
cfg = base()
|
@@ -72,7 +83,7 @@ def ddgan_cc12m_v11():
|
|
72 |
cfg = base()
|
73 |
cfg['model']['text_encoder'] = "google/t5-v1_1-large"
|
74 |
cfg['model']['classifier_free_guidance_proba'] = 0.2
|
75 |
-
cfg['model']['cross_attention'] =
|
76 |
return cfg
|
77 |
|
78 |
def ddgan_cc12m_v12():
|
@@ -102,7 +113,7 @@ def ddgan_cifar10_cond17():
|
|
102 |
cfg['model']['image_size'] = 32
|
103 |
cfg['model']['classifier_free_guidance_proba'] = 0.2
|
104 |
cfg['model']['ch_mult'] = "1 2 2 2"
|
105 |
-
cfg['model']['cross_attention'] =
|
106 |
cfg['model']['dataset'] = "cifar10"
|
107 |
cfg['model']['n_mlp'] = 4
|
108 |
return cfg
|
@@ -276,7 +287,7 @@ def ddgan_ddb_v7():
|
|
276 |
|
277 |
def ddgan_ddb_v9():
|
278 |
cfg = ddgan_ddb_v3()
|
279 |
-
cfg['model']['attn_resolutions'] =
|
280 |
return cfg
|
281 |
|
282 |
def ddgan_laion_aesthetic_v15():
|
@@ -313,6 +324,7 @@ models = [
|
|
313 |
ddgan_cc12m_v13, # T5-XL + cross attention + classifier free guidance + random_resized_crop_v1 + cond attn
|
314 |
ddgan_cc12m_v14, # T5-XL + cross attention + classifier free guidance + random_resized_crop_v1 + 300M model
|
315 |
ddgan_cc12m_v15, # fine-tune v11 with --mismatch_loss and --grad_penalty_cond
|
|
|
316 |
ddgan_laion_aesthetic_v1, # like ddgan_cc12m_v11 but fine-tuned on laion aesthetic
|
317 |
ddgan_laion_aesthetic_v2, # like ddgan_laion_aesthetic_v1 but trained from scratch with the new cross attn discr
|
318 |
ddgan_laion_aesthetic_v3, # like ddgan_laion_aesthetic_v1 but trained from scratch with T5-XL (continue from 23aug with mismatch and grad penalty and random_resized_crop_v1)
|
@@ -352,76 +364,7 @@ models = [
|
|
352 |
ddgan_ddb_v12,
|
353 |
]
|
354 |
|
355 |
-
def
|
356 |
for model in models:
|
357 |
if model.__name__ == model_name:
|
358 |
-
return model()
|
359 |
-
|
360 |
-
|
361 |
-
def test(model_name, *, cond_text="", batch_size:int=None, epoch:int=None, guidance_scale:float=0, fid=False, real_img_dir="", q=0.0, seed=0, nb_images_for_fid=0, scale_factor_h=1, scale_factor_w=1, compute_clip_score=False, eval_name="", scale_method="convolutional", compute_image_reward=False):
|
362 |
-
|
363 |
-
cfg = get_model(model_name)
|
364 |
-
model = cfg['model']
|
365 |
-
if epoch is None:
|
366 |
-
paths = glob('./saved_info/dd_gan/{}/{}/netG_*.pth'.format(model["dataset"], model_name))
|
367 |
-
epoch = max(
|
368 |
-
[int(os.path.basename(path).replace(".pth", "").split("_")[1]) for path in paths]
|
369 |
-
)
|
370 |
-
args = {}
|
371 |
-
args['exp'] = model_name
|
372 |
-
args['image_size'] = model['image_size']
|
373 |
-
args['seed'] = seed
|
374 |
-
args['num_channels'] = model['num_channels']
|
375 |
-
args['dataset'] = model['dataset']
|
376 |
-
args['num_channels_dae'] = model['num_channels_dae']
|
377 |
-
args['ch_mult'] = model['ch_mult']
|
378 |
-
args['num_timesteps'] = model['num_timesteps']
|
379 |
-
args['num_res_blocks'] = model['num_res_blocks']
|
380 |
-
args['batch_size'] = model['batch_size'] if batch_size is None else batch_size
|
381 |
-
args['epoch'] = epoch
|
382 |
-
args['cond_text'] = f'"{cond_text}"'
|
383 |
-
args['text_encoder'] = model.get("text_encoder")
|
384 |
-
args['cross_attention'] = model.get("cross_attention")
|
385 |
-
args['guidance_scale'] = guidance_scale
|
386 |
-
args['masked_mean'] = model.get("masked_mean")
|
387 |
-
args['dynamic_thresholding_quantile'] = q
|
388 |
-
args['scale_factor_h'] = scale_factor_h
|
389 |
-
args['scale_factor_w'] = scale_factor_w
|
390 |
-
args['n_mlp'] = model.get("n_mlp")
|
391 |
-
args['scale_method'] = scale_method
|
392 |
-
args['attn_resolutions'] = model.get("attn_resolutions", "16")
|
393 |
-
if fid:
|
394 |
-
args['compute_fid'] = ''
|
395 |
-
args['real_img_dir'] = real_img_dir
|
396 |
-
args['nb_images_for_fid'] = nb_images_for_fid
|
397 |
-
if compute_clip_score:
|
398 |
-
args['compute_clip_score'] = ""
|
399 |
-
|
400 |
-
if compute_image_reward:
|
401 |
-
args['compute_image_reward'] = ""
|
402 |
-
if eval_name:
|
403 |
-
args["eval_name"] = eval_name
|
404 |
-
cmd = "python -u test_ddgan.py " + " ".join(f"--{k} {v}" for k, v in args.items() if v is not None)
|
405 |
-
print(cmd)
|
406 |
-
call(cmd, shell=True)
|
407 |
-
|
408 |
-
def eval_results(model_name):
|
409 |
-
import pandas as pd
|
410 |
-
rows = []
|
411 |
-
cfg = get_model(model_name)
|
412 |
-
model = cfg['model']
|
413 |
-
paths = glob('./saved_info/dd_gan/{}/{}/fid*.json'.format(model["dataset"], model_name))
|
414 |
-
for path in paths:
|
415 |
-
with open(path, "r") as fd:
|
416 |
-
data = json.load(fd)
|
417 |
-
row = {}
|
418 |
-
row['fid'] = data['fid']
|
419 |
-
row['epoch'] = data['epoch_id']
|
420 |
-
rows.append(row)
|
421 |
-
out = './saved_info/dd_gan/{}/{}/fid.csv'.format(model["dataset"], model_name)
|
422 |
-
df = pd.DataFrame(rows)
|
423 |
-
df.to_csv(out, index=False)
|
424 |
-
|
425 |
-
if __name__ == "__main__":
|
426 |
-
from clize import run
|
427 |
-
run([test, eval_results])
|
|
|
10 |
"n": 8,
|
11 |
},
|
12 |
"model":{
|
13 |
+
"dataset": "wds",
|
14 |
+
"seed": 0,
|
15 |
+
"cross_attention": False,
|
16 |
"num_channels": 3,
|
17 |
+
"centered": True,
|
18 |
+
"use_geometric": False,
|
19 |
+
"beta_min": 0.1,
|
20 |
+
"beta_max": 20.0,
|
21 |
"num_channels_dae": 128,
|
22 |
+
"n_mlp": 3,
|
23 |
+
"ch_mult": [1, 1, 2, 2, 4, 4],
|
24 |
"num_res_blocks": 2,
|
25 |
+
"attn_resolutions": (16,),
|
26 |
+
"dropout": 0.0,
|
27 |
+
"resamp_with_conv": True,
|
28 |
+
"conditional": True,
|
29 |
+
"fir": True,
|
30 |
+
"fir_kernel": [1, 3, 3, 1],
|
31 |
+
"skip_rescale": True,
|
32 |
+
"resblock_type": "biggan",
|
33 |
+
"progressive": "none",
|
34 |
+
"progressive_input": "residual",
|
35 |
+
"progressive_combine": "sum",
|
36 |
"embedding_type": "positional",
|
37 |
+
"fourier_scale": 16.0,
|
38 |
+
"not_use_tanh": False,
|
39 |
+
"image_size": 256,
|
40 |
+
"nz": 100,
|
41 |
+
"num_timesteps": 4,
|
42 |
"z_emb_dim": 256,
|
43 |
+
"t_emb_dim": 256,
|
44 |
+
"text_encoder": "google/t5-v1_1-base",
|
45 |
+
"masked_mean": True,
|
46 |
+
"cross_attention_block": "basic",
|
47 |
+
}
|
|
|
|
|
|
|
48 |
}
|
49 |
def ddgan_cc12m_v2():
|
50 |
cfg = base()
|
|
|
83 |
cfg = base()
|
84 |
cfg['model']['text_encoder'] = "google/t5-v1_1-large"
|
85 |
cfg['model']['classifier_free_guidance_proba'] = 0.2
|
86 |
+
cfg['model']['cross_attention'] = True
|
87 |
return cfg
|
88 |
|
89 |
def ddgan_cc12m_v12():
|
|
|
113 |
cfg['model']['image_size'] = 32
|
114 |
cfg['model']['classifier_free_guidance_proba'] = 0.2
|
115 |
cfg['model']['ch_mult'] = "1 2 2 2"
|
116 |
+
cfg['model']['cross_attention'] = True
|
117 |
cfg['model']['dataset'] = "cifar10"
|
118 |
cfg['model']['n_mlp'] = 4
|
119 |
return cfg
|
|
|
287 |
|
288 |
def ddgan_ddb_v9():
|
289 |
cfg = ddgan_ddb_v3()
|
290 |
+
cfg['model']['attn_resolutions'] = [4, 8, 16, 32]
|
291 |
return cfg
|
292 |
|
293 |
def ddgan_laion_aesthetic_v15():
|
|
|
324 |
ddgan_cc12m_v13, # T5-XL + cross attention + classifier free guidance + random_resized_crop_v1 + cond attn
|
325 |
ddgan_cc12m_v14, # T5-XL + cross attention + classifier free guidance + random_resized_crop_v1 + 300M model
|
326 |
ddgan_cc12m_v15, # fine-tune v11 with --mismatch_loss and --grad_penalty_cond
|
327 |
+
|
328 |
ddgan_laion_aesthetic_v1, # like ddgan_cc12m_v11 but fine-tuned on laion aesthetic
|
329 |
ddgan_laion_aesthetic_v2, # like ddgan_laion_aesthetic_v1 but trained from scratch with the new cross attn discr
|
330 |
ddgan_laion_aesthetic_v3, # like ddgan_laion_aesthetic_v1 but trained from scratch with T5-XL (continue from 23aug with mismatch and grad penalty and random_resized_crop_v1)
|
|
|
364 |
ddgan_ddb_v12,
|
365 |
]
|
366 |
|
367 |
+
def get_model_config(model_name):
|
368 |
for model in models:
|
369 |
if model.__name__ == model_name:
|
370 |
+
return model()['model']
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
test_ddgan.py
CHANGED
@@ -11,8 +11,32 @@ import time
|
|
11 |
import os
|
12 |
import json
|
13 |
import torchvision
|
|
|
|
|
14 |
from score_sde.models.ncsnpp_generator_adagn import NCSNpp
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
15 |
from encoder import build_encoder
|
|
|
|
|
|
|
16 |
|
17 |
#%% Diffusion coefficients
|
18 |
def var_func_vp(t, beta_min, beta_max):
|
@@ -138,6 +162,12 @@ def sample_from_model(coefficients, generator, n_time, x_init, T, opt, cond=None
|
|
138 |
return x
|
139 |
|
140 |
|
|
|
|
|
|
|
|
|
|
|
|
|
141 |
def sample_from_model_classifier_free_guidance(coefficients, generator, n_time, x_init, T, opt, text_encoder, cond=None, guidance_scale=0):
|
142 |
x = x_init
|
143 |
null = text_encoder([""] * len(x_init), return_only_pooled=False)
|
@@ -353,106 +383,84 @@ def get_fold_unfold(x, kernel_size, stride, split_input_params, uf=1, df=1): #
|
|
353 |
|
354 |
return fold, unfold, normalization, weighting
|
355 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
356 |
|
357 |
|
358 |
#%%
|
359 |
-
def sample_and_test(args):
|
360 |
-
torch.manual_seed(args.seed)
|
361 |
|
362 |
-
device = 'cuda:0'
|
363 |
-
text_encoder =build_encoder(name=args.text_encoder, masked_mean=args.masked_mean).to(device)
|
364 |
-
args.cond_size = text_encoder.output_size
|
365 |
-
if args.dataset == 'cifar10':
|
366 |
-
real_img_dir = 'pytorch_fid/cifar10_train_stat.npy'
|
367 |
-
elif args.dataset == 'celeba_256':
|
368 |
-
real_img_dir = 'pytorch_fid/celeba_256_stat.npy'
|
369 |
-
elif args.dataset == 'lsun':
|
370 |
-
real_img_dir = 'pytorch_fid/lsun_church_stat.npy'
|
371 |
-
else:
|
372 |
-
real_img_dir = args.real_img_dir
|
373 |
-
|
374 |
-
to_range_0_1 = lambda x: (x + 1.) / 2.
|
375 |
|
376 |
-
|
377 |
-
|
378 |
-
|
|
|
379 |
if args.epoch_id == -1:
|
380 |
epochs = range(1000)
|
381 |
else:
|
382 |
epochs = [args.epoch_id]
|
383 |
if args.compute_image_reward:
|
384 |
-
import ImageReward as RM
|
385 |
#image_reward = RM.load("ImageReward-v1.0", download_root=".").to(device)
|
386 |
image_reward = RM.load("ImageReward.pt", download_root=".").to(device)
|
387 |
-
|
388 |
for epoch in epochs:
|
389 |
args.epoch_id = epoch
|
390 |
-
|
391 |
-
|
|
|
|
|
392 |
if not os.path.exists(path):
|
393 |
continue
|
394 |
if not os.path.exists(next_next_path):
|
395 |
break
|
396 |
print("PATH", path)
|
397 |
-
|
398 |
-
#if not os.path.exists(next_path):
|
399 |
-
# print(f"STOP at {epoch}")
|
400 |
-
# break
|
401 |
-
try:
|
402 |
-
ckpt = torch.load(path, map_location=device)
|
403 |
-
except Exception:
|
404 |
-
continue
|
405 |
suffix = '_' + args.eval_name if args.eval_name else ""
|
406 |
-
dest = './saved_info/dd_gan/{}/{}/eval_{}{}.json'.format(
|
407 |
if (args.compute_fid or args.compute_clip_score or args.compute_image_reward) and os.path.exists(dest):
|
408 |
continue
|
409 |
-
print("
|
410 |
-
|
411 |
-
|
412 |
-
|
413 |
-
if key.startswith("module"):
|
414 |
-
ckpt[key[7:]] = ckpt.pop(key)
|
415 |
-
netG.load_state_dict(ckpt)
|
416 |
-
netG.eval()
|
417 |
-
|
418 |
-
|
419 |
-
T = get_time_schedule(args, device)
|
420 |
-
|
421 |
-
pos_coeff = Posterior_Coefficients(args, device)
|
422 |
-
|
423 |
-
|
424 |
-
save_dir = "./generated_samples/{}".format(args.dataset)
|
425 |
|
426 |
if not os.path.exists(save_dir):
|
427 |
os.makedirs(save_dir)
|
428 |
|
429 |
|
430 |
if args.compute_fid or args.compute_clip_score or args.compute_image_reward:
|
431 |
-
|
432 |
-
from pytorch_fid.fid_score import calculate_activation_statistics, calculate_fid_given_paths, ImagePathDataset, compute_statistics_of_path, calculate_frechet_distance
|
433 |
-
from pytorch_fid.inception import InceptionV3
|
434 |
-
import random
|
435 |
random.seed(args.seed)
|
436 |
texts = open(args.cond_text).readlines()
|
437 |
texts = [t.strip() for t in texts]
|
438 |
if args.nb_images_for_fid:
|
439 |
random.shuffle(texts)
|
440 |
texts = texts[0:args.nb_images_for_fid]
|
441 |
-
#iters_needed = len(texts) // args.batch_size
|
442 |
-
#texts = list(map(lambda s:s.strip(), texts))
|
443 |
-
#ntimes = max(30000 // len(texts), 1)
|
444 |
-
#texts = texts * ntimes
|
445 |
print("Text size:", len(texts))
|
446 |
-
#print("Iters:", iters_needed)
|
447 |
i = 0
|
448 |
-
|
449 |
if args.compute_fid:
|
450 |
dims = 2048
|
451 |
block_idx = InceptionV3.BLOCK_INDEX_BY_DIM[dims]
|
452 |
inceptionv3 = InceptionV3([block_idx]).to(device)
|
453 |
|
454 |
if args.compute_clip_score:
|
455 |
-
import clip
|
456 |
CLIP_MEAN = [0.48145466, 0.4578275, 0.40821073]
|
457 |
CLIP_STD = [0.26862954, 0.26130258, 0.27577711]
|
458 |
clip_model, preprocess = clip.load(args.clip_model, device)
|
@@ -481,14 +489,14 @@ def sample_and_test(args):
|
|
481 |
for b in range(0, len(texts), args.batch_size):
|
482 |
text = texts[b:b+args.batch_size]
|
483 |
with torch.no_grad():
|
484 |
-
cond = text_encoder(text
|
485 |
bs = len(text)
|
486 |
t0 = time.time()
|
487 |
-
x_t_1 = torch.randn(bs,
|
488 |
if args.guidance_scale:
|
489 |
fake_sample = sample_from_model_classifier_free_guidance(pos_coeff, netG, args.num_timesteps, x_t_1,T, args, text_encoder, cond=cond, guidance_scale=args.guidance_scale)
|
490 |
else:
|
491 |
-
fake_sample =
|
492 |
fake_sample = to_range_0_1(fake_sample)
|
493 |
|
494 |
if args.compute_fid:
|
@@ -513,8 +521,8 @@ def sample_and_test(args):
|
|
513 |
clip_scores.append(((imf * txtf).sum(dim=1)).cpu())
|
514 |
|
515 |
if args.compute_image_reward:
|
516 |
-
for k,
|
517 |
-
img =
|
518 |
img = img * 255
|
519 |
img = img.astype(np.uint8)
|
520 |
text_k = text[k]
|
@@ -542,7 +550,8 @@ def sample_and_test(args):
|
|
542 |
with open(dest, "w") as fd:
|
543 |
json.dump(results, fd)
|
544 |
print(results)
|
545 |
-
else:
|
|
|
546 |
if args.cond_text.endswith(".txt"):
|
547 |
texts = open(args.cond_text).readlines()
|
548 |
texts = [t.strip() for t in texts]
|
@@ -550,7 +559,6 @@ def sample_and_test(args):
|
|
550 |
texts = [args.cond_text] * args.batch_size
|
551 |
clip_guidance = False
|
552 |
if clip_guidance:
|
553 |
-
from clip_encoder import CLIPImageEncoder
|
554 |
cond = text_encoder(texts, return_only_pooled=False)
|
555 |
clip_image_model = CLIPImageEncoder().to(device)
|
556 |
x_t_1 = torch.randn(len(texts), args.num_channels,args.image_size*args.scale_factor_h, args.image_size*args.scale_factor_w).to(device)
|
@@ -559,14 +567,14 @@ def sample_and_test(args):
|
|
559 |
torchvision.utils.save_image(fake_sample, './samples_{}.jpg'.format(args.dataset))
|
560 |
|
561 |
else:
|
562 |
-
cond = text_encoder(texts
|
563 |
-
x_t_1 = torch.randn(len(texts),
|
564 |
t0 = time.time()
|
565 |
if args.guidance_scale:
|
566 |
if args.scale_factor_h > 1 or args.scale_factor_w > 1:
|
567 |
if args.scale_method == "convolutional":
|
568 |
split_input_params = {
|
569 |
-
"ks": (
|
570 |
"stride": (150, 150),
|
571 |
"clip_max_tie_weight": 0.5,
|
572 |
"clip_min_tie_weight": 0.01,
|
@@ -583,22 +591,17 @@ def sample_and_test(args):
|
|
583 |
else:
|
584 |
fake_sample = sample_from_model_classifier_free_guidance(pos_coeff, netG, args.num_timesteps, x_t_1,T, args, text_encoder, cond=cond, guidance_scale=args.guidance_scale)
|
585 |
else:
|
586 |
-
fake_sample =
|
587 |
|
588 |
print(time.time() - t0)
|
589 |
fake_sample = to_range_0_1(fake_sample)
|
590 |
-
torchvision.utils.save_image(fake_sample, '
|
591 |
-
|
592 |
-
|
593 |
-
|
594 |
-
|
595 |
-
|
596 |
-
|
597 |
|
598 |
if __name__ == '__main__':
|
599 |
parser = argparse.ArgumentParser('ddgan parameters')
|
600 |
-
parser.add_argument('--
|
601 |
-
|
|
|
602 |
parser.add_argument('--compute_fid', action='store_true', default=False,
|
603 |
help='whether or not compute FID')
|
604 |
parser.add_argument('--compute_clip_score', action='store_true', default=False,
|
@@ -608,92 +611,15 @@ if __name__ == '__main__':
|
|
608 |
|
609 |
parser.add_argument('--clip_model', type=str,default="ViT-L/14")
|
610 |
parser.add_argument('--eval_name', type=str,default="")
|
611 |
-
|
612 |
-
parser.add_argument('--epoch_id', type=int,default=1000)
|
613 |
parser.add_argument('--guidance_scale', type=float,default=0)
|
614 |
parser.add_argument('--dynamic_thresholding_quantile', type=float,default=0)
|
615 |
-
parser.add_argument('--cond_text', type=str,default="
|
616 |
parser.add_argument('--scale_factor_h', type=int,default=1)
|
617 |
parser.add_argument('--scale_factor_w', type=int,default=1)
|
618 |
parser.add_argument('--scale_method', type=str,default="convolutional")
|
619 |
-
|
620 |
-
parser.add_argument('--cross_attention', action='store_true',default=False)
|
621 |
-
|
622 |
-
|
623 |
-
parser.add_argument('--num_channels', type=int, default=3,
|
624 |
-
help='channel of image')
|
625 |
-
parser.add_argument('--centered', action='store_false', default=True,
|
626 |
-
help='-1,1 scale')
|
627 |
-
parser.add_argument('--use_geometric', action='store_true',default=False)
|
628 |
-
parser.add_argument('--beta_min', type=float, default= 0.1,
|
629 |
-
help='beta_min for diffusion')
|
630 |
-
parser.add_argument('--beta_max', type=float, default=20.,
|
631 |
-
help='beta_max for diffusion')
|
632 |
-
|
633 |
-
|
634 |
-
parser.add_argument('--num_channels_dae', type=int, default=128,
|
635 |
-
help='number of initial channels in denosing model')
|
636 |
-
parser.add_argument('--n_mlp', type=int, default=3,
|
637 |
-
help='number of mlp layers for z')
|
638 |
-
parser.add_argument('--ch_mult', nargs='+', type=int,
|
639 |
-
help='channel multiplier')
|
640 |
-
|
641 |
-
parser.add_argument('--num_res_blocks', type=int, default=2,
|
642 |
-
help='number of resnet blocks per scale')
|
643 |
-
parser.add_argument('--attn_resolutions', default=(16,), nargs='+', type=int,
|
644 |
-
help='resolution of applying attention')
|
645 |
-
parser.add_argument('--dropout', type=float, default=0.,
|
646 |
-
help='drop-out rate')
|
647 |
-
parser.add_argument('--resamp_with_conv', action='store_false', default=True,
|
648 |
-
help='always up/down sampling with conv')
|
649 |
-
parser.add_argument('--conditional', action='store_false', default=True,
|
650 |
-
help='noise conditional')
|
651 |
-
parser.add_argument('--fir', action='store_false', default=True,
|
652 |
-
help='FIR')
|
653 |
-
parser.add_argument('--fir_kernel', default=[1, 3, 3, 1],
|
654 |
-
help='FIR kernel')
|
655 |
-
parser.add_argument('--skip_rescale', action='store_false', default=True,
|
656 |
-
help='skip rescale')
|
657 |
-
parser.add_argument('--resblock_type', default='biggan',
|
658 |
-
help='tyle of resnet block, choice in biggan and ddpm')
|
659 |
-
parser.add_argument('--progressive', type=str, default='none', choices=['none', 'output_skip', 'residual'],
|
660 |
-
help='progressive type for output')
|
661 |
-
parser.add_argument('--progressive_input', type=str, default='residual', choices=['none', 'input_skip', 'residual'],
|
662 |
-
help='progressive type for input')
|
663 |
-
parser.add_argument('--progressive_combine', type=str, default='sum', choices=['sum', 'cat'],
|
664 |
-
help='progressive combine method.')
|
665 |
-
|
666 |
-
parser.add_argument('--embedding_type', type=str, default='positional', choices=['positional', 'fourier'],
|
667 |
-
help='type of time embedding')
|
668 |
-
parser.add_argument('--fourier_scale', type=float, default=16.,
|
669 |
-
help='scale of fourier transform')
|
670 |
-
parser.add_argument('--not_use_tanh', action='store_true',default=False)
|
671 |
-
|
672 |
-
#geenrator and training
|
673 |
-
parser.add_argument('--exp', default='experiment_cifar_default', help='name of experiment')
|
674 |
-
parser.add_argument('--real_img_dir', default='./pytorch_fid/cifar10_train_stat.npy', help='directory to real images for FID computation')
|
675 |
-
|
676 |
-
parser.add_argument('--dataset', default='cifar10', help='name of dataset')
|
677 |
-
parser.add_argument('--image_size', type=int, default=32,
|
678 |
-
help='size of image')
|
679 |
-
|
680 |
-
parser.add_argument('--nz', type=int, default=100)
|
681 |
-
parser.add_argument('--num_timesteps', type=int, default=4)
|
682 |
-
|
683 |
-
|
684 |
-
parser.add_argument('--z_emb_dim', type=int, default=256)
|
685 |
-
parser.add_argument('--t_emb_dim', type=int, default=256)
|
686 |
-
parser.add_argument('--batch_size', type=int, default=200, help='sample generating batch size')
|
687 |
-
parser.add_argument('--text_encoder', type=str, default="google/t5-v1_1-base")
|
688 |
-
parser.add_argument('--masked_mean', action='store_true',default=False)
|
689 |
parser.add_argument('--nb_images_for_fid', type=int, default=0)
|
690 |
-
|
691 |
-
|
692 |
-
|
693 |
-
|
694 |
-
|
695 |
args = parser.parse_args()
|
696 |
-
|
697 |
sample_and_test(args)
|
698 |
|
699 |
|
|
|
11 |
import os
|
12 |
import json
|
13 |
import torchvision
|
14 |
+
import random
|
15 |
+
|
16 |
from score_sde.models.ncsnpp_generator_adagn import NCSNpp
|
17 |
+
from torch.nn.functional import adaptive_avg_pool2d
|
18 |
+
|
19 |
+
try:
|
20 |
+
from pytorch_fid.fid_score import calculate_activation_statistics, calculate_fid_given_paths, ImagePathDataset, compute_statistics_of_path, calculate_frechet_distance
|
21 |
+
from pytorch_fid.inception import InceptionV3
|
22 |
+
except ImportError:
|
23 |
+
pass
|
24 |
+
|
25 |
+
try:
|
26 |
+
import ImageReward as RM
|
27 |
+
except ImportError:
|
28 |
+
pass
|
29 |
+
|
30 |
+
|
31 |
+
try:
|
32 |
+
import clip
|
33 |
+
except ImportError:
|
34 |
+
pass
|
35 |
+
|
36 |
from encoder import build_encoder
|
37 |
+
from clip_encoder import CLIPImageEncoder
|
38 |
+
|
39 |
+
from model_configs import get_model_config
|
40 |
|
41 |
#%% Diffusion coefficients
|
42 |
def var_func_vp(t, beta_min, beta_max):
|
|
|
162 |
return x
|
163 |
|
164 |
|
165 |
+
def sample(generator, x_init, cond=None):
|
166 |
+
return sample_from_model(
|
167 |
+
generator.pos_coeff, generator, n_time=generator.config.num_timesteps, x_init=x_init,
|
168 |
+
T=generator.time_schedule, opt=generator.config, cond=cond
|
169 |
+
)
|
170 |
+
|
171 |
def sample_from_model_classifier_free_guidance(coefficients, generator, n_time, x_init, T, opt, text_encoder, cond=None, guidance_scale=0):
|
172 |
x = x_init
|
173 |
null = text_encoder([""] * len(x_init), return_only_pooled=False)
|
|
|
383 |
|
384 |
return fold, unfold, normalization, weighting
|
385 |
|
386 |
+
class ObjectFromDict:
|
387 |
+
def __init__(self, d):
|
388 |
+
self.__dict__ = d
|
389 |
+
|
390 |
+
def load_model(config, path, device="cpu"):
|
391 |
+
config = ObjectFromDict(config)
|
392 |
+
text_encoder = build_encoder(name=config.text_encoder, masked_mean=config.masked_mean)
|
393 |
+
config.cond_size = text_encoder.output_size
|
394 |
+
netG = NCSNpp(config)
|
395 |
+
ckpt = torch.load(path, map_location="cpu")
|
396 |
+
for key in list(ckpt.keys()):
|
397 |
+
if key.startswith("module"):
|
398 |
+
ckpt[key[7:]] = ckpt.pop(key)
|
399 |
+
netG.load_state_dict(ckpt)
|
400 |
+
netG.eval()
|
401 |
+
netG.pos_coeff = Posterior_Coefficients(config, device)
|
402 |
+
netG.text_encoder = text_encoder
|
403 |
+
netG.config = config
|
404 |
+
netG.time_schedule = get_time_schedule(config, device)
|
405 |
+
netG = netG.to(device)
|
406 |
+
return netG
|
407 |
|
408 |
|
409 |
#%%
|
|
|
|
|
410 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
411 |
|
412 |
+
def sample_and_test(args):
|
413 |
+
torch.manual_seed(args.seed)
|
414 |
+
device = 'cuda' if torch.cuda.is_available() else 'cpu'
|
415 |
+
to_range_0_1 = lambda x: (x + 1.) / 2.
|
416 |
if args.epoch_id == -1:
|
417 |
epochs = range(1000)
|
418 |
else:
|
419 |
epochs = [args.epoch_id]
|
420 |
if args.compute_image_reward:
|
|
|
421 |
#image_reward = RM.load("ImageReward-v1.0", download_root=".").to(device)
|
422 |
image_reward = RM.load("ImageReward.pt", download_root=".").to(device)
|
423 |
+
cfg = get_model_config(args.name)
|
424 |
for epoch in epochs:
|
425 |
args.epoch_id = epoch
|
426 |
+
|
427 |
+
path = './saved_info/dd_gan/{}/{}/netG_{}.pth'.format(cfg['dataset'], args.name, args.epoch_id)
|
428 |
+
next_next_path = './saved_info/dd_gan/{}/{}/netG_{}.pth'.format(cfg['dataset'], args.name, args.epoch_id+2)
|
429 |
+
print(path)
|
430 |
if not os.path.exists(path):
|
431 |
continue
|
432 |
if not os.path.exists(next_next_path):
|
433 |
break
|
434 |
print("PATH", path)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
435 |
suffix = '_' + args.eval_name if args.eval_name else ""
|
436 |
+
dest = './saved_info/dd_gan/{}/{}/eval_{}{}.json'.format(cfg['dataset'],'ddgan', args.epoch_id, suffix)
|
437 |
if (args.compute_fid or args.compute_clip_score or args.compute_image_reward) and os.path.exists(dest):
|
438 |
continue
|
439 |
+
print("Load epoch", args.epoch_id, "checkpoint")
|
440 |
+
|
441 |
+
netG = load_model(cfg, path, device=device)
|
442 |
+
save_dir = "./generated_samples/{}".format(cfg['dataset'])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
443 |
|
444 |
if not os.path.exists(save_dir):
|
445 |
os.makedirs(save_dir)
|
446 |
|
447 |
|
448 |
if args.compute_fid or args.compute_clip_score or args.compute_image_reward:
|
449 |
+
# Evaluate
|
|
|
|
|
|
|
450 |
random.seed(args.seed)
|
451 |
texts = open(args.cond_text).readlines()
|
452 |
texts = [t.strip() for t in texts]
|
453 |
if args.nb_images_for_fid:
|
454 |
random.shuffle(texts)
|
455 |
texts = texts[0:args.nb_images_for_fid]
|
|
|
|
|
|
|
|
|
456 |
print("Text size:", len(texts))
|
|
|
457 |
i = 0
|
|
|
458 |
if args.compute_fid:
|
459 |
dims = 2048
|
460 |
block_idx = InceptionV3.BLOCK_INDEX_BY_DIM[dims]
|
461 |
inceptionv3 = InceptionV3([block_idx]).to(device)
|
462 |
|
463 |
if args.compute_clip_score:
|
|
|
464 |
CLIP_MEAN = [0.48145466, 0.4578275, 0.40821073]
|
465 |
CLIP_STD = [0.26862954, 0.26130258, 0.27577711]
|
466 |
clip_model, preprocess = clip.load(args.clip_model, device)
|
|
|
489 |
for b in range(0, len(texts), args.batch_size):
|
490 |
text = texts[b:b+args.batch_size]
|
491 |
with torch.no_grad():
|
492 |
+
cond = netG.text_encoder(text)
|
493 |
bs = len(text)
|
494 |
t0 = time.time()
|
495 |
+
x_t_1 = torch.randn(bs, cfg['num_channels'], cfg['image_size'], cfg['image_size']).to(device)
|
496 |
if args.guidance_scale:
|
497 |
fake_sample = sample_from_model_classifier_free_guidance(pos_coeff, netG, args.num_timesteps, x_t_1,T, args, text_encoder, cond=cond, guidance_scale=args.guidance_scale)
|
498 |
else:
|
499 |
+
fake_sample = sample(generator=model, x_init=x_init, cond=cond)
|
500 |
fake_sample = to_range_0_1(fake_sample)
|
501 |
|
502 |
if args.compute_fid:
|
|
|
521 |
clip_scores.append(((imf * txtf).sum(dim=1)).cpu())
|
522 |
|
523 |
if args.compute_image_reward:
|
524 |
+
for k, img in enumerate(fake_sample):
|
525 |
+
img = img.cpu().numpy().transpose(1,2,0)
|
526 |
img = img * 255
|
527 |
img = img.astype(np.uint8)
|
528 |
text_k = text[k]
|
|
|
550 |
with open(dest, "w") as fd:
|
551 |
json.dump(results, fd)
|
552 |
print(results)
|
553 |
+
else:
|
554 |
+
# just generate some samples
|
555 |
if args.cond_text.endswith(".txt"):
|
556 |
texts = open(args.cond_text).readlines()
|
557 |
texts = [t.strip() for t in texts]
|
|
|
559 |
texts = [args.cond_text] * args.batch_size
|
560 |
clip_guidance = False
|
561 |
if clip_guidance:
|
|
|
562 |
cond = text_encoder(texts, return_only_pooled=False)
|
563 |
clip_image_model = CLIPImageEncoder().to(device)
|
564 |
x_t_1 = torch.randn(len(texts), args.num_channels,args.image_size*args.scale_factor_h, args.image_size*args.scale_factor_w).to(device)
|
|
|
567 |
torchvision.utils.save_image(fake_sample, './samples_{}.jpg'.format(args.dataset))
|
568 |
|
569 |
else:
|
570 |
+
cond = netG.text_encoder(texts)
|
571 |
+
x_t_1 = torch.randn(len(texts), cfg['num_channels'], cfg['image_size'] * args.scale_factor_h, cfg['image_size'] * args.scale_factor_w).to(device)
|
572 |
t0 = time.time()
|
573 |
if args.guidance_scale:
|
574 |
if args.scale_factor_h > 1 or args.scale_factor_w > 1:
|
575 |
if args.scale_method == "convolutional":
|
576 |
split_input_params = {
|
577 |
+
"ks": (cfg['image_size'], cfg['image_size']),
|
578 |
"stride": (150, 150),
|
579 |
"clip_max_tie_weight": 0.5,
|
580 |
"clip_min_tie_weight": 0.01,
|
|
|
591 |
else:
|
592 |
fake_sample = sample_from_model_classifier_free_guidance(pos_coeff, netG, args.num_timesteps, x_t_1,T, args, text_encoder, cond=cond, guidance_scale=args.guidance_scale)
|
593 |
else:
|
594 |
+
fake_sample = sample(generator=netG, x_init=x_t_1, cond=cond)
|
595 |
|
596 |
print(time.time() - t0)
|
597 |
fake_sample = to_range_0_1(fake_sample)
|
598 |
+
torchvision.utils.save_image(fake_sample, 'samples.jpg')
|
|
|
|
|
|
|
|
|
|
|
|
|
599 |
|
600 |
if __name__ == '__main__':
|
601 |
parser = argparse.ArgumentParser('ddgan parameters')
|
602 |
+
parser.add_argument('--name', type=str, default="", help="model config name")
|
603 |
+
parser.add_argument('--batch_size', type=int, default=16)
|
604 |
+
parser.add_argument('--seed', type=int, default=1024, help='seed used for initialization')
|
605 |
parser.add_argument('--compute_fid', action='store_true', default=False,
|
606 |
help='whether or not compute FID')
|
607 |
parser.add_argument('--compute_clip_score', action='store_true', default=False,
|
|
|
611 |
|
612 |
parser.add_argument('--clip_model', type=str,default="ViT-L/14")
|
613 |
parser.add_argument('--eval_name', type=str,default="")
|
614 |
+
parser.add_argument('--epoch_id', type=int,default=-1)
|
|
|
615 |
parser.add_argument('--guidance_scale', type=float,default=0)
|
616 |
parser.add_argument('--dynamic_thresholding_quantile', type=float,default=0)
|
617 |
+
parser.add_argument('--cond_text', type=str,default="a chair in the form of an avocado")
|
618 |
parser.add_argument('--scale_factor_h', type=int,default=1)
|
619 |
parser.add_argument('--scale_factor_w', type=int,default=1)
|
620 |
parser.add_argument('--scale_method', type=str,default="convolutional")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
621 |
parser.add_argument('--nb_images_for_fid', type=int, default=0)
|
|
|
|
|
|
|
|
|
|
|
622 |
args = parser.parse_args()
|
|
|
623 |
sample_and_test(args)
|
624 |
|
625 |
|