-
Notifications
You must be signed in to change notification settings - Fork 4
/
Copy pathtrain_transformer.py
330 lines (272 loc) · 19.4 KB
/
train_transformer.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
from tqdm import trange, tqdm
import torch
from torch.utils.data import DataLoader
from logger import Logger
from modules.model_transformer import *
from torch.optim.lr_scheduler import MultiStepLR
from sync_batchnorm import DataParallelWithCallback
from frames_dataset_transformer25 import DatasetRepeater
import numpy as np
import math
def adjust_learning_rate(optimizer, itr, base_lr, max_itr, p, is_cosine_decay=False, min_lr=1e-6, warm_up_steps=1000, encoder_lr_ratio=1.0, freeze_params=[]):
if itr < warm_up_steps:
now_lr = min_lr + (base_lr - min_lr) * itr / warm_up_steps
else:
itr = itr - warm_up_steps
max_itr = max_itr - warm_up_steps
if is_cosine_decay:
now_lr = min_lr + (base_lr - min_lr) * (math.cos(math.pi * itr /
(max_itr + 1)) +
1.) * 0.5
else:
now_lr = min_lr + (base_lr - min_lr) * (1 - itr / (max_itr + 1))**p
for param_group in optimizer.param_groups:
if encoder_lr_ratio != 1.0 and "encoder." in param_group["name"]:
param_group['lr'] = (now_lr - min_lr) * encoder_lr_ratio + min_lr
else:
param_group['lr'] = now_lr
for freeze_param in freeze_params:
if freeze_param in param_group["name"]:
param_group['lr'] = 0
param_group['weight_decay'] = 0
break
return now_lr
def train_batch(config, generator, discriminator, kp_detector, audio2kptransformer, checkpoint, log_dir, dataset, device_ids):
train_params = config['train_params']
# optimizer_generator = torch.optim.Adam(generator.parameters(), lr=train_params['lr_generator'], betas=(0.5, 0.999))
# optimizer_discriminator = torch.optim.Adam(discriminator.parameters(), lr=train_params['lr_discriminator'], betas=(0.5, 0.999))
# optimizer_kp_detector = torch.optim.Adam(kp_detector.parameters(), lr=train_params['lr_kp_detector'], betas=(0.5, 0.999))
optimizer_audio2kptransformer = torch.optim.Adam(audio2kptransformer.parameters(), lr=train_params['lr_audio2kptransformer'], betas=(0.5, 0.999))
if checkpoint is not None:
start_epoch = Logger.load_cpk_a2kp(checkpoint, generator, discriminator, kp_detector, audio2kptransformer,
None, None, None, optimizer_audio2kptransformer)
# optimizer_generator, optimizer_discriminator, None, None)
# optimizer_generator, optimizer_discriminator, optimizer_kp_detector, None)
else:
start_epoch = 0
# scheduler_generator = MultiStepLR(optimizer_generator, train_params['epoch_milestones'], gamma=0.1,
# last_epoch=start_epoch - 1)
# scheduler_discriminator = MultiStepLR(optimizer_discriminator, train_params['epoch_milestones'], gamma=0.1,
# last_epoch=start_epoch - 1)
# scheduler_kp_detector = MultiStepLR(optimizer_kp_detector, train_params['epoch_milestones'], gamma=0.1,
# last_epoch=-1 + start_epoch * (train_params['lr_kp_detector'] != 0))
scheduler_audio2kptransformer = MultiStepLR(optimizer_audio2kptransformer, train_params['epoch_milestones'], gamma=0.1,
last_epoch=-1)
# last_epoch=-1 + start_epoch * (train_params['lr_kp_detector'] != 0))
if 'num_repeats' in train_params or train_params['num_repeats'] != 1:
dataset = DatasetRepeater(dataset, train_params['num_repeats'])
dataloader = DataLoader(dataset, batch_size=train_params['batch_size'], shuffle=True, num_workers=16, drop_last=True)
generator_full = GeneratorFullModelBatch(kp_detector, audio2kptransformer, generator, discriminator, train_params, estimate_jacobian=config['model_params']['common_params']['estimate_jacobian'])
discriminator_full = DiscriminatorFullModel(kp_detector, generator, discriminator, train_params)
if torch.cuda.is_available():
generator_full = DataParallelWithCallback(generator_full, device_ids=device_ids)
discriminator_full = DataParallelWithCallback(discriminator_full, device_ids=device_ids)
count = 0
with Logger(log_dir=log_dir, visualizer_params=config['visualizer_params'], checkpoint_freq=train_params['checkpoint_freq']) as logger:
for epoch in trange(start_epoch, train_params['num_epochs']):
for x in tqdm(dataloader):
losses_generator, generated = generator_full(x, train_params['train_with_img'])
loss_values = [val.mean() for val in losses_generator.values()]
loss = sum(loss_values)
loss.backward()
torch.nn.utils.clip_grad_norm_(generator_full.module.audio2kptransformer.parameters(), 0.05)
# optimizer_generator.step()
# optimizer_generator.zero_grad()
# optimizer_kp_detector.step()
# optimizer_kp_detector.zero_grad()
optimizer_audio2kptransformer.step()
optimizer_audio2kptransformer.zero_grad()
if train_params['loss_weights']['generator_gan'] != 0:
# optimizer_discriminator.zero_grad()
losses_discriminator = discriminator_full(x, generated)
loss_values = [val.mean() for val in losses_discriminator.values()]
loss = sum(loss_values)
loss.backward()
# optimizer_discriminator.step()
# optimizer_discriminator.zero_grad()
else:
losses_discriminator = {}
losses_generator.update(losses_discriminator)
losses = {key: value.mean().detach().data.cpu().numpy() for key, value in losses_generator.items()}
count += 1
if count % 20 == 0 :
print(np.array(logger.loss_list).mean(axis=0))
logger.log_iter(losses=losses)
if count % 2000 == 0:
break
# scheduler_generator.step()
# scheduler_discriminator.step()
# scheduler_kp_detector.step()
scheduler_audio2kptransformer.step()
logger.log_epoch(epoch, {'generator': generator,
'discriminator': discriminator,
'kp_detector': kp_detector,
'audio2kptransformer': audio2kptransformer,
# 'optimizer_generator': optimizer_generator,
# 'optimizer_discriminator': optimizer_discriminator,
# 'optimizer_kp_detector': optimizer_kp_detector,
'optimizer_audio2kptransformer': optimizer_audio2kptransformer}, inp=x, out=generated, save_visualize=train_params['train_with_img'])
def train_batch_gen(config, generator, discriminator, kp_detector, audio2kptransformer, checkpoint, log_dir, dataset, device_ids):
train_params = config['train_params']
# optimizer_generator = torch.optim.Adam(generator.parameters(), lr=train_params['lr_generator'], betas=(0.5, 0.999))
# optimizer_discriminator = torch.optim.Adam(discriminator.parameters(), lr=train_params['lr_discriminator'], betas=(0.5, 0.999))
# optimizer_kp_detector = torch.optim.Adam(kp_detector.parameters(), lr=train_params['lr_kp_detector'], betas=(0.5, 0.999))
optimizer_audio2kptransformer = torch.optim.Adam(audio2kptransformer.parameters(), lr=train_params['lr_audio2kptransformer'], betas=(0.5, 0.999))
if checkpoint is not None:
start_epoch = Logger.load_cpk_a2kp_diffgen(checkpoint, generator, discriminator, kp_detector, audio2kptransformer,
None, None, None, optimizer_audio2kptransformer)
# optimizer_generator, optimizer_discriminator, None, None)
# optimizer_generator, optimizer_discriminator, optimizer_kp_detector, None)
else:
start_epoch = 0
# scheduler_generator = MultiStepLR(optimizer_generator, train_params['epoch_milestones'], gamma=0.1,
# last_epoch=start_epoch - 1)
# scheduler_discriminator = MultiStepLR(optimizer_discriminator, train_params['epoch_milestones'], gamma=0.1,
# last_epoch=start_epoch - 1)
# scheduler_kp_detector = MultiStepLR(optimizer_kp_detector, train_params['epoch_milestones'], gamma=0.1,
# last_epoch=-1 + start_epoch * (train_params['lr_kp_detector'] != 0))
scheduler_audio2kptransformer = MultiStepLR(optimizer_audio2kptransformer, train_params['epoch_milestones'], gamma=0.1,
last_epoch=-1)
# last_epoch=-1 + start_epoch * (train_params['lr_kp_detector'] != 0))
if 'num_repeats' in train_params or train_params['num_repeats'] != 1:
dataset = DatasetRepeater(dataset, train_params['num_repeats'])
dataloader = DataLoader(dataset, batch_size=train_params['batch_size'], shuffle=True, num_workers=16, drop_last=True)
generator_full = GeneratorFullModelBatch(kp_detector, audio2kptransformer, generator, discriminator, train_params, estimate_jacobian=config['model_params']['common_params']['estimate_jacobian'])
# discriminator_full = DiscriminatorFullModel(kp_detector, generator, discriminator, train_params)
if torch.cuda.is_available():
generator_full = DataParallelWithCallback(generator_full, device_ids=device_ids)
# discriminator_full = DataParallelWithCallback(discriminator_full, device_ids=device_ids)
count = 0
with Logger(log_dir=log_dir, visualizer_params=config['visualizer_params'], checkpoint_freq=train_params['checkpoint_freq']) as logger:
for epoch in trange(start_epoch, train_params['num_epochs']):
for x in tqdm(dataloader):
losses_generator, generated = generator_full(x, train_params['train_with_img'])
loss_values = [val.mean() for val in losses_generator.values()]
loss = sum(loss_values)
loss.backward()
torch.nn.utils.clip_grad_norm_(generator_full.module.audio2kptransformer.parameters(), 0.05)
# optimizer_generator.step()
# optimizer_generator.zero_grad()
# optimizer_kp_detector.step()
# optimizer_kp_detector.zero_grad()
optimizer_audio2kptransformer.step()
optimizer_audio2kptransformer.zero_grad()
# if train_params['loss_weights']['generator_gan'] != 0:
# # optimizer_discriminator.zero_grad()
# losses_discriminator = discriminator_full(x, generated)
# loss_values = [val.mean() for val in losses_discriminator.values()]
# loss = sum(loss_values)
# loss.backward()
# # optimizer_discriminator.step()
# # optimizer_discriminator.zero_grad()
# else:
# losses_discriminator = {}
# losses_generator.update(losses_discriminator)
losses = {key: value.mean().detach().data.cpu().numpy() for key, value in losses_generator.items()}
count += 1
if count % 100 == 0 :
print(np.array(logger.loss_list).mean(axis=0))
logger.log_iter(losses=losses)
if count % 2000 == 0:
break
# scheduler_generator.step()
# scheduler_discriminator.step()
# scheduler_kp_detector.step()
scheduler_audio2kptransformer.step()
logger.log_epoch(epoch, {'generator': generator,
# 'discriminator': discriminator,
'kp_detector': kp_detector,
'audio2kptransformer': audio2kptransformer,
# 'optimizer_generator': optimizer_generator,
# 'optimizer_discriminator': optimizer_discriminator,
# 'optimizer_kp_detector': optimizer_kp_detector,
'optimizer_audio2kptransformer': optimizer_audio2kptransformer}, inp=x, out=generated, save_visualize=train_params['train_with_img'])
import time
def train_batch_deepprompt_eam3d_sidetuning(config, generator, discriminator, kp_detector, audio2kptransformer, emotionprompt, sidetuning, checkpoint, log_dir, dataset, device_ids):
train_params = config['train_params']
optimizer_generator = torch.optim.Adam(generator.parameters(), lr=train_params['lr_generator'], betas=(0.5, 0.999))
# optimizer_discriminator = torch.optim.Adam(discriminator.parameters(), lr=train_params['lr_discriminator'], betas=(0.5, 0.999))
# optimizer_kp_detector = torch.optim.Adam(kp_detector.parameters(), lr=train_params['lr_kp_detector'], betas=(0.5, 0.999))
optimizer_audio2kptransformer = torch.optim.Adam(audio2kptransformer.parameters(), lr=train_params['lr_audio2kptransformer'], betas=(0.5, 0.999))
optimizer_emotionprompt = torch.optim.Adam(emotionprompt.parameters(), lr=train_params['lr_emotionprompt'], betas=(0.5, 0.999))
optimizer_sidetuning = torch.optim.Adam(sidetuning.parameters(), lr=train_params['lr_sidetuning'], betas=(0.5, 0.999))
if checkpoint is not None:
start_epoch = Logger.load_cpk_a2kp_prompt_sidetuning(checkpoint, generator, discriminator, kp_detector, audio2kptransformer, emotionprompt, sidetuning,
None, None, None, optimizer_audio2kptransformer, optimizer_emotionprompt, optimizer_sidetuning)
# optimizer_generator, optimizer_discriminator, None, None)
# optimizer_generator, optimizer_discriminator, optimizer_kp_detector, None)
else:
start_epoch = 0
# print( train_params['epoch_milestones']) # 180
scheduler_generator = MultiStepLR(optimizer_generator, train_params['epoch_milestones'], gamma=0.1,
last_epoch= - 1)
# scheduler_discriminator = MultiStepLR(optimizer_discriminator, train_params['epoch_milestones'], gamma=0.1,
# last_epoch=start_epoch - 1)
# scheduler_kp_detector = MultiStepLR(optimizer_kp_detector, train_params['epoch_milestones'], gamma=0.1,
# last_epoch=-1 + start_epoch * (train_params['lr_kp_detector'] != 0))
scheduler_audio2kptransformer = MultiStepLR(optimizer_audio2kptransformer, train_params['epoch_milestones'], gamma=0.1,
last_epoch=-1)
scheduler_emotionprompt = MultiStepLR(optimizer_emotionprompt, train_params['epoch_milestones'], gamma=0.1,
last_epoch=-1)
scheduler_sidetuning = MultiStepLR(optimizer_sidetuning, train_params['epoch_milestones'], gamma=0.1,
last_epoch=-1)
# last_epoch=-1 + start_epoch * (train_params['lr_kp_detector'] != 0))
if 'num_repeats' in train_params or train_params['num_repeats'] != 1:
dataset = DatasetRepeater(dataset, train_params['num_repeats'])
dataloader = DataLoader(dataset, batch_size=train_params['batch_size'], shuffle=True, num_workers=16, drop_last=True)
generator_full = GeneratorFullModelBatchDeepPromptSTEAM3D(kp_detector, audio2kptransformer, emotionprompt, sidetuning, generator, discriminator, train_params, estimate_jacobian=config['model_params']['common_params']['estimate_jacobian'])
if torch.cuda.is_available():
generator_full = DataParallelWithCallback(generator_full, device_ids=device_ids)
count = 0
with Logger(log_dir=log_dir, visualizer_params=config['visualizer_params'], checkpoint_freq=train_params['checkpoint_freq']) as logger:
for epoch in trange(start_epoch, train_params['num_epochs']):
time_start = time.time()
for x in tqdm(dataloader):
losses_generator, generated = generator_full(x, train_params['train_with_img'])
loss_values = [val.mean() for val in losses_generator.values()]
loss = sum(loss_values)
loss.backward()
torch.nn.utils.clip_grad_norm_(generator_full.module.audio2kptransformer.parameters(), 0.05)
torch.nn.utils.clip_grad_norm_(generator_full.module.sidetuning.parameters(), 0.05)
optimizer_generator.step()
optimizer_generator.zero_grad()
# optimizer_kp_detector.step()
# optimizer_kp_detector.zero_grad()
optimizer_audio2kptransformer.step()
optimizer_audio2kptransformer.zero_grad()
optimizer_emotionprompt.step()
optimizer_emotionprompt.zero_grad()
optimizer_sidetuning.step()
optimizer_sidetuning.zero_grad()
losses = {key: value.mean().detach().data.cpu().numpy() for key, value in losses_generator.items()}
count += 1
if count % 100 == 0 :
print(np.array(logger.loss_list).mean(axis=0), count%len(dataloader), len(dataloader))
logger.log_iter(losses=losses)
time_now = time.time()
if(time_now - time_start) > 30*60: # 30 min
break
### check learning rate
# print(optimizer_audio2kptransformer.param_groups[0]['lr'])
# adjust_learning_rate(optimizer_emotionprompt, count, train_params['lr_emotionprompt'], 100000, 2)
# print(optimizer_emotionprompt.param_groups[0]['lr'])
# print(scheduler_audio2kptransformer.get_lr())
# print(scheduler_emotionprompt.get_lr())
scheduler_generator.step()
# scheduler_discriminator.step()
# scheduler_kp_detector.step()
scheduler_audio2kptransformer.step()
scheduler_emotionprompt.step()
scheduler_sidetuning.step()
logger.log_epoch(epoch, {'generator': generator,
'discriminator': discriminator,
'kp_detector': kp_detector,
'audio2kptransformer': audio2kptransformer,
'emotionprompt': emotionprompt,
'sidetuning': sidetuning,
'optimizer_generator': optimizer_generator,
# 'optimizer_discriminator': optimizer_discriminator,
# 'optimizer_kp_detector': optimizer_kp_detector,
'optimizer_audio2kptransformer': optimizer_audio2kptransformer,
'optimizer_emotionprompt': optimizer_emotionprompt,
'optimizer_sidetuning': optimizer_sidetuning}, inp=x, out=generated, save_visualize=train_params['train_with_img'])