File size: 25,515 Bytes
e526878
2e9bf0c
e526878
 
 
 
 
 
 
 
 
 
2e9bf0c
 
 
 
e526878
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2e9bf0c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e526878
2e9bf0c
 
 
e526878
 
2e9bf0c
e526878
2e9bf0c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e526878
2e9bf0c
 
e526878
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2e9bf0c
e526878
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2e9bf0c
e526878
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2e9bf0c
 
 
 
 
 
e526878
2e9bf0c
 
 
 
 
 
 
 
 
 
 
e526878
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2e9bf0c
 
 
e526878
2e9bf0c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e526878
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2e9bf0c
 
 
 
 
 
 
e526878
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2e9bf0c
e526878
 
 
 
 
 
 
 
 
 
 
 
 
2e9bf0c
e526878
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2e9bf0c
 
 
 
 
 
 
 
 
 
 
e526878
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2e9bf0c
 
 
e526878
 
 
 
 
 
2e9bf0c
 
e526878
2e9bf0c
 
 
 
 
e526878
2e9bf0c
 
 
 
 
 
 
e526878
 
 
 
 
 
 
 
 
 
 
2e9bf0c
 
e526878
 
 
 
 
 
 
 
 
 
2e9bf0c
 
e526878
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2e9bf0c
 
 
 
 
e526878
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2e9bf0c
 
 
e526878
 
 
 
 
 
 
2e9bf0c
 
 
e526878
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2e9bf0c
e526878
 
 
 
 
 
 
 
 
2e9bf0c
 
e526878
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2e9bf0c
 
e526878
 
 
 
 
 
 
2e9bf0c
e526878
 
2e9bf0c
e526878
 
 
 
 
 
 
 
 
2e9bf0c
 
e526878
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
# flake8: noqa: E402
import platform
import os
import torch
from torch.nn import functional as F
from torch.utils.data import DataLoader
from torch.utils.tensorboard import SummaryWriter
import torch.distributed as dist
from torch.nn.parallel import DistributedDataParallel as DDP
from torch.cuda.amp import autocast, GradScaler
from tqdm import tqdm
import logging
from config import config
import argparse
import datetime
import gc

logging.getLogger("numba").setLevel(logging.WARNING)
import commons
import utils
from data_utils import (
    TextAudioSpeakerLoader,
    TextAudioSpeakerCollate,
    DistributedBucketSampler,
)
from models import (
    SynthesizerTrn,
    MultiPeriodDiscriminator,
    DurationDiscriminator,
)
from losses import generator_loss, discriminator_loss, feature_loss, kl_loss
from mel_processing import mel_spectrogram_torch, spec_to_mel_torch
from text.symbols import symbols

torch.backends.cuda.matmul.allow_tf32 = True
torch.backends.cudnn.allow_tf32 = (
    True  # If encontered training problem,please try to disable TF32.
)
torch.set_float32_matmul_precision("medium")
torch.backends.cuda.sdp_kernel("flash")
torch.backends.cuda.enable_flash_sdp(True)
torch.backends.cuda.enable_mem_efficient_sdp(
    True
)  # Not available if torch version is lower than 2.0
torch.backends.cuda.enable_math_sdp(True)
global_step = 0


def run():
    # 环境变量解析
    envs = config.train_ms_config.env
    for env_name, env_value in envs.items():
        if env_name not in os.environ.keys():
            print("加载config中的配置{}".format(str(env_value)))
            os.environ[env_name] = str(env_value)
    print(
        "加载环境变量 \nMASTER_ADDR: {},\nMASTER_PORT: {},\nWORLD_SIZE: {},\nRANK: {},\nLOCAL_RANK: {}".format(
            os.environ["MASTER_ADDR"],
            os.environ["MASTER_PORT"],
            os.environ["WORLD_SIZE"],
            os.environ["RANK"],
            os.environ["LOCAL_RANK"],
        )
    )

    backend = "nccl"
    if platform.system() == "Windows":
        backend = "gloo"  # If Windows,switch to gloo backend.
    dist.init_process_group(
        backend=backend,
        init_method="env://",
        timeout=datetime.timedelta(seconds=300),
    )  # Use torchrun instead of mp.spawn
    rank = dist.get_rank()
    local_rank = int(os.environ["LOCAL_RANK"])
    n_gpus = dist.get_world_size()

    # 命令行/config.yml配置解析
    # hps = utils.get_hparams()
    parser = argparse.ArgumentParser()
    # 非必要不建议使用命令行配置,请使用config.yml文件
    parser.add_argument(
        "-c",
        "--config",
        type=str,
        default=config.train_ms_config.config_path,
        help="JSON file for configuration",
    )

    parser.add_argument(
        "-m",
        "--model",
        type=str,
        help="数据集文件夹路径,请注意,数据不再默认放在/logs文件夹下。如果需要用命令行配置,请声明相对于根目录的路径",
        default=config.dataset_path,
    )
    args = parser.parse_args()
    model_dir = os.path.join(args.model, config.train_ms_config.model)
    if not os.path.exists(model_dir):
        os.makedirs(model_dir)
    hps = utils.get_hparams_from_file(args.config)
    hps.model_dir = model_dir
    # 比较路径是否相同
    if os.path.realpath(args.config) != os.path.realpath(
        config.train_ms_config.config_path
    ):
        with open(args.config, "r", encoding="utf-8") as f:
            data = f.read()
        with open(config.train_ms_config.config_path, "w", encoding="utf-8") as f:
            f.write(data)

    torch.manual_seed(hps.train.seed)
    torch.cuda.set_device(local_rank)

    global global_step
    if rank == 0:
        logger = utils.get_logger(hps.model_dir)
        logger.info(hps)
        utils.check_git_hash(hps.model_dir)
        writer = SummaryWriter(log_dir=hps.model_dir)
        writer_eval = SummaryWriter(log_dir=os.path.join(hps.model_dir, "eval"))
    train_dataset = TextAudioSpeakerLoader(hps.data.training_files, hps.data)
    train_sampler = DistributedBucketSampler(
        train_dataset,
        hps.train.batch_size,
        [32, 300, 400, 500, 600, 700, 800, 900, 1000],
        num_replicas=n_gpus,
        rank=rank,
        shuffle=True,
    )
    collate_fn = TextAudioSpeakerCollate()
    train_loader = DataLoader(
        train_dataset,
        num_workers=min(config.train_ms_config.num_workers, os.cpu_count() - 1),
        shuffle=False,
        pin_memory=True,
        collate_fn=collate_fn,
        batch_sampler=train_sampler,
        persistent_workers=True,
        prefetch_factor=4,
    )  # DataLoader config could be adjusted.
    if rank == 0:
        eval_dataset = TextAudioSpeakerLoader(hps.data.validation_files, hps.data)
        eval_loader = DataLoader(
            eval_dataset,
            num_workers=0,
            shuffle=False,
            batch_size=1,
            pin_memory=True,
            drop_last=False,
            collate_fn=collate_fn,
        )
    if (
        "use_noise_scaled_mas" in hps.model.keys()
        and hps.model.use_noise_scaled_mas is True
    ):
        print("Using noise scaled MAS for VITS2")
        mas_noise_scale_initial = 0.01
        noise_scale_delta = 2e-6
    else:
        print("Using normal MAS for VITS1")
        mas_noise_scale_initial = 0.0
        noise_scale_delta = 0.0
    if (
        "use_duration_discriminator" in hps.model.keys()
        and hps.model.use_duration_discriminator is True
    ):
        print("Using duration discriminator for VITS2")
        net_dur_disc = DurationDiscriminator(
            hps.model.hidden_channels,
            hps.model.hidden_channels,
            3,
            0.1,
            gin_channels=hps.model.gin_channels if hps.data.n_speakers != 0 else 0,
        ).cuda(local_rank)
    if (
        "use_spk_conditioned_encoder" in hps.model.keys()
        and hps.model.use_spk_conditioned_encoder is True
    ):
        if hps.data.n_speakers == 0:
            raise ValueError(
                "n_speakers must be > 0 when using spk conditioned encoder to train multi-speaker model"
            )
    else:
        print("Using normal encoder for VITS1")

    net_g = SynthesizerTrn(
        len(symbols),
        hps.data.filter_length // 2 + 1,
        hps.train.segment_size // hps.data.hop_length,
        n_speakers=hps.data.n_speakers,
        mas_noise_scale_initial=mas_noise_scale_initial,
        noise_scale_delta=noise_scale_delta,
        **hps.model,
    ).cuda(local_rank)

    if getattr(hps.train, "freeze_ZH_bert", False):
        print("Freezing ZH bert encoder !!!")
        for param in net_g.enc_p.bert_proj.parameters():
            param.requires_grad = False

    if getattr(hps.train, "freeze_EN_bert", False):
        print("Freezing EN bert encoder !!!")
        for param in net_g.enc_p.en_bert_proj.parameters():
            param.requires_grad = False

    if getattr(hps.train, "freeze_JP_bert", False):
        print("Freezing JP bert encoder !!!")
        for param in net_g.enc_p.ja_bert_proj.parameters():
            param.requires_grad = False

    net_d = MultiPeriodDiscriminator(hps.model.use_spectral_norm).cuda(local_rank)
    optim_g = torch.optim.AdamW(
        filter(lambda p: p.requires_grad, net_g.parameters()),
        hps.train.learning_rate,
        betas=hps.train.betas,
        eps=hps.train.eps,
    )
    optim_d = torch.optim.AdamW(
        net_d.parameters(),
        hps.train.learning_rate,
        betas=hps.train.betas,
        eps=hps.train.eps,
    )
    if net_dur_disc is not None:
        optim_dur_disc = torch.optim.AdamW(
            net_dur_disc.parameters(),
            hps.train.learning_rate,
            betas=hps.train.betas,
            eps=hps.train.eps,
        )
    else:
        optim_dur_disc = None
    net_g = DDP(net_g, device_ids=[local_rank], bucket_cap_mb=512)
    net_d = DDP(net_d, device_ids=[local_rank], bucket_cap_mb=512)
    dur_resume_lr = None
    if net_dur_disc is not None:
        net_dur_disc = DDP(
            net_dur_disc,
            device_ids=[local_rank],
            find_unused_parameters=True,
            bucket_cap_mb=512,
        )

    # 下载底模
    if config.train_ms_config.base["use_base_model"]:
        utils.download_checkpoint(
            hps.model_dir,
            config.train_ms_config.base,
            token=config.openi_token,
            mirror=config.mirror,
        )

    try:
        if net_dur_disc is not None:
            _, _, dur_resume_lr, epoch_str = utils.load_checkpoint(
                utils.latest_checkpoint_path(hps.model_dir, "DUR_*.pth"),
                net_dur_disc,
                optim_dur_disc,
                skip_optimizer=hps.train.skip_optimizer
                if "skip_optimizer" in hps.train
                else True,
            )
            _, optim_g, g_resume_lr, epoch_str = utils.load_checkpoint(
                utils.latest_checkpoint_path(hps.model_dir, "G_*.pth"),
                net_g,
                optim_g,
                skip_optimizer=hps.train.skip_optimizer
                if "skip_optimizer" in hps.train
                else True,
            )
            _, optim_d, d_resume_lr, epoch_str = utils.load_checkpoint(
                utils.latest_checkpoint_path(hps.model_dir, "D_*.pth"),
                net_d,
                optim_d,
                skip_optimizer=hps.train.skip_optimizer
                if "skip_optimizer" in hps.train
                else True,
            )
            if not optim_g.param_groups[0].get("initial_lr"):
                optim_g.param_groups[0]["initial_lr"] = g_resume_lr
            if not optim_d.param_groups[0].get("initial_lr"):
                optim_d.param_groups[0]["initial_lr"] = d_resume_lr
            if not optim_dur_disc.param_groups[0].get("initial_lr"):
                optim_dur_disc.param_groups[0]["initial_lr"] = dur_resume_lr

        epoch_str = max(epoch_str, 1)
        # global_step = (epoch_str - 1) * len(train_loader)
        global_step = int(
            utils.get_steps(utils.latest_checkpoint_path(hps.model_dir, "G_*.pth"))
        )
        print(
            f"******************检测到模型存在,epoch为 {epoch_str},gloabl step为 {global_step}*********************"
        )
    except Exception as e:
        print(e)
        epoch_str = 1
        global_step = 0

    scheduler_g = torch.optim.lr_scheduler.ExponentialLR(
        optim_g, gamma=hps.train.lr_decay, last_epoch=epoch_str - 2
    )
    scheduler_d = torch.optim.lr_scheduler.ExponentialLR(
        optim_d, gamma=hps.train.lr_decay, last_epoch=epoch_str - 2
    )
    if net_dur_disc is not None:
        if not optim_dur_disc.param_groups[0].get("initial_lr"):
            optim_dur_disc.param_groups[0]["initial_lr"] = dur_resume_lr
        scheduler_dur_disc = torch.optim.lr_scheduler.ExponentialLR(
            optim_dur_disc, gamma=hps.train.lr_decay, last_epoch=epoch_str - 2
        )
    else:
        scheduler_dur_disc = None
    scaler = GradScaler(enabled=hps.train.fp16_run)

    for epoch in range(epoch_str, hps.train.epochs + 1):
        if rank == 0:
            train_and_evaluate(
                rank,
                local_rank,
                epoch,
                hps,
                [net_g, net_d, net_dur_disc],
                [optim_g, optim_d, optim_dur_disc],
                [scheduler_g, scheduler_d, scheduler_dur_disc],
                scaler,
                [train_loader, eval_loader],
                logger,
                [writer, writer_eval],
            )
        else:
            train_and_evaluate(
                rank,
                local_rank,
                epoch,
                hps,
                [net_g, net_d, net_dur_disc],
                [optim_g, optim_d, optim_dur_disc],
                [scheduler_g, scheduler_d, scheduler_dur_disc],
                scaler,
                [train_loader, None],
                None,
                None,
            )
        scheduler_g.step()
        scheduler_d.step()
        if net_dur_disc is not None:
            scheduler_dur_disc.step()


def train_and_evaluate(
    rank,
    local_rank,
    epoch,
    hps,
    nets,
    optims,
    schedulers,
    scaler,
    loaders,
    logger,
    writers,
):
    net_g, net_d, net_dur_disc = nets
    optim_g, optim_d, optim_dur_disc = optims
    scheduler_g, scheduler_d, scheduler_dur_disc = schedulers
    train_loader, eval_loader = loaders
    if writers is not None:
        writer, writer_eval = writers

    train_loader.batch_sampler.set_epoch(epoch)
    global global_step

    net_g.train()
    net_d.train()
    if net_dur_disc is not None:
        net_dur_disc.train()
    for batch_idx, (
        x,
        x_lengths,
        spec,
        spec_lengths,
        y,
        y_lengths,
        speakers,
        tone,
        language,
        bert,
        ja_bert,
        en_bert,
        emo,
    ) in enumerate(tqdm(train_loader)):
        if net_g.module.use_noise_scaled_mas:
            current_mas_noise_scale = (
                net_g.module.mas_noise_scale_initial
                - net_g.module.noise_scale_delta * global_step
            )
            net_g.module.current_mas_noise_scale = max(current_mas_noise_scale, 0.0)
        x, x_lengths = x.cuda(local_rank, non_blocking=True), x_lengths.cuda(
            local_rank, non_blocking=True
        )
        spec, spec_lengths = spec.cuda(
            local_rank, non_blocking=True
        ), spec_lengths.cuda(local_rank, non_blocking=True)
        y, y_lengths = y.cuda(local_rank, non_blocking=True), y_lengths.cuda(
            local_rank, non_blocking=True
        )
        speakers = speakers.cuda(local_rank, non_blocking=True)
        tone = tone.cuda(local_rank, non_blocking=True)
        language = language.cuda(local_rank, non_blocking=True)
        bert = bert.cuda(local_rank, non_blocking=True)
        ja_bert = ja_bert.cuda(local_rank, non_blocking=True)
        en_bert = en_bert.cuda(local_rank, non_blocking=True)
        emo = emo.cuda(local_rank, non_blocking=True)

        with autocast(enabled=hps.train.fp16_run):
            (
                y_hat,
                l_length,
                attn,
                ids_slice,
                x_mask,
                z_mask,
                (z, z_p, m_p, logs_p, m_q, logs_q),
                (hidden_x, logw, logw_),
                g,
                loss_commit,
            ) = net_g(
                x,
                x_lengths,
                spec,
                spec_lengths,
                speakers,
                tone,
                language,
                bert,
                ja_bert,
                en_bert,
                emo,
            )
            mel = spec_to_mel_torch(
                spec,
                hps.data.filter_length,
                hps.data.n_mel_channels,
                hps.data.sampling_rate,
                hps.data.mel_fmin,
                hps.data.mel_fmax,
            )
            y_mel = commons.slice_segments(
                mel, ids_slice, hps.train.segment_size // hps.data.hop_length
            )
            y_hat_mel = mel_spectrogram_torch(
                y_hat.squeeze(1),
                hps.data.filter_length,
                hps.data.n_mel_channels,
                hps.data.sampling_rate,
                hps.data.hop_length,
                hps.data.win_length,
                hps.data.mel_fmin,
                hps.data.mel_fmax,
            )

            y = commons.slice_segments(
                y, ids_slice * hps.data.hop_length, hps.train.segment_size
            )  # slice

            # Discriminator
            y_d_hat_r, y_d_hat_g, _, _ = net_d(y, y_hat.detach())
            with autocast(enabled=False):
                loss_disc, losses_disc_r, losses_disc_g = discriminator_loss(
                    y_d_hat_r, y_d_hat_g
                )
                loss_disc_all = loss_disc
            if net_dur_disc is not None:
                y_dur_hat_r, y_dur_hat_g = net_dur_disc(
                    hidden_x.detach(),
                    x_mask.detach(),
                    logw.detach(),
                    logw_.detach(),
                    g.detach(),
                )
                with autocast(enabled=False):
                    # TODO: I think need to mean using the mask, but for now, just mean all
                    (
                        loss_dur_disc,
                        losses_dur_disc_r,
                        losses_dur_disc_g,
                    ) = discriminator_loss(y_dur_hat_r, y_dur_hat_g)
                    loss_dur_disc_all = loss_dur_disc
                optim_dur_disc.zero_grad()
                scaler.scale(loss_dur_disc_all).backward()
                scaler.unscale_(optim_dur_disc)
                commons.clip_grad_value_(net_dur_disc.parameters(), None)
                scaler.step(optim_dur_disc)

        optim_d.zero_grad()
        scaler.scale(loss_disc_all).backward()
        scaler.unscale_(optim_d)
        grad_norm_d = commons.clip_grad_value_(net_d.parameters(), None)
        scaler.step(optim_d)

        with autocast(enabled=hps.train.fp16_run):
            # Generator
            y_d_hat_r, y_d_hat_g, fmap_r, fmap_g = net_d(y, y_hat)
            if net_dur_disc is not None:
                y_dur_hat_r, y_dur_hat_g = net_dur_disc(
                    hidden_x, x_mask, logw, logw_, g
                )
            with autocast(enabled=False):
                loss_dur = torch.sum(l_length.float())
                loss_mel = F.l1_loss(y_mel, y_hat_mel) * hps.train.c_mel
                loss_kl = kl_loss(z_p, logs_q, m_p, logs_p, z_mask) * hps.train.c_kl

                loss_fm = feature_loss(fmap_r, fmap_g)
                loss_gen, losses_gen = generator_loss(y_d_hat_g)
                loss_gen_all = (
                    loss_gen + loss_fm + loss_mel + loss_dur + loss_kl + loss_commit
                )
                if net_dur_disc is not None:
                    loss_dur_gen, losses_dur_gen = generator_loss(y_dur_hat_g)
                    loss_gen_all += loss_dur_gen
        optim_g.zero_grad()
        scaler.scale(loss_gen_all).backward()
        scaler.unscale_(optim_g)
        grad_norm_g = commons.clip_grad_value_(net_g.parameters(), None)
        scaler.step(optim_g)
        scaler.update()

        if rank == 0:
            if global_step % hps.train.log_interval == 0:
                lr = optim_g.param_groups[0]["lr"]
                losses = [loss_disc, loss_gen, loss_fm, loss_mel, loss_dur, loss_kl]
                logger.info(
                    "Train Epoch: {} [{:.0f}%]".format(
                        epoch, 100.0 * batch_idx / len(train_loader)
                    )
                )
                logger.info([x.item() for x in losses] + [global_step, lr])

                scalar_dict = {
                    "loss/g/total": loss_gen_all,
                    "loss/d/total": loss_disc_all,
                    "learning_rate": lr,
                    "grad_norm_d": grad_norm_d,
                    "grad_norm_g": grad_norm_g,
                }
                scalar_dict.update(
                    {
                        "loss/g/fm": loss_fm,
                        "loss/g/mel": loss_mel,
                        "loss/g/dur": loss_dur,
                        "loss/g/kl": loss_kl,
                    }
                )
                scalar_dict.update(
                    {"loss/g/{}".format(i): v for i, v in enumerate(losses_gen)}
                )
                scalar_dict.update(
                    {"loss/d_r/{}".format(i): v for i, v in enumerate(losses_disc_r)}
                )
                scalar_dict.update(
                    {"loss/d_g/{}".format(i): v for i, v in enumerate(losses_disc_g)}
                )

                image_dict = {
                    "slice/mel_org": utils.plot_spectrogram_to_numpy(
                        y_mel[0].data.cpu().numpy()
                    ),
                    "slice/mel_gen": utils.plot_spectrogram_to_numpy(
                        y_hat_mel[0].data.cpu().numpy()
                    ),
                    "all/mel": utils.plot_spectrogram_to_numpy(
                        mel[0].data.cpu().numpy()
                    ),
                    "all/attn": utils.plot_alignment_to_numpy(
                        attn[0, 0].data.cpu().numpy()
                    ),
                }
                utils.summarize(
                    writer=writer,
                    global_step=global_step,
                    images=image_dict,
                    scalars=scalar_dict,
                )

            if global_step % hps.train.eval_interval == 0:
                evaluate(hps, net_g, eval_loader, writer_eval)
                utils.save_checkpoint(
                    net_g,
                    optim_g,
                    hps.train.learning_rate,
                    epoch,
                    os.path.join(hps.model_dir, "G_{}.pth".format(global_step)),
                )
                utils.save_checkpoint(
                    net_d,
                    optim_d,
                    hps.train.learning_rate,
                    epoch,
                    os.path.join(hps.model_dir, "D_{}.pth".format(global_step)),
                )
                if net_dur_disc is not None:
                    utils.save_checkpoint(
                        net_dur_disc,
                        optim_dur_disc,
                        hps.train.learning_rate,
                        epoch,
                        os.path.join(hps.model_dir, "DUR_{}.pth".format(global_step)),
                    )
                keep_ckpts = config.train_ms_config.keep_ckpts
                if keep_ckpts > 0:
                    utils.clean_checkpoints(
                        path_to_models=hps.model_dir,
                        n_ckpts_to_keep=keep_ckpts,
                        sort_by_time=True,
                    )

        global_step += 1

    gc.collect()
    torch.cuda.empty_cache()
    if rank == 0:
        logger.info("====> Epoch: {}".format(epoch))


def evaluate(hps, generator, eval_loader, writer_eval):
    generator.eval()
    image_dict = {}
    audio_dict = {}
    print("Evaluating ...")
    with torch.no_grad():
        for batch_idx, (
            x,
            x_lengths,
            spec,
            spec_lengths,
            y,
            y_lengths,
            speakers,
            tone,
            language,
            bert,
            ja_bert,
            en_bert,
            emo,
        ) in enumerate(eval_loader):
            x, x_lengths = x.cuda(), x_lengths.cuda()
            spec, spec_lengths = spec.cuda(), spec_lengths.cuda()
            y, y_lengths = y.cuda(), y_lengths.cuda()
            speakers = speakers.cuda()
            bert = bert.cuda()
            ja_bert = ja_bert.cuda()
            en_bert = en_bert.cuda()
            tone = tone.cuda()
            language = language.cuda()
            emo = emo.cuda()
            for use_sdp in [True, False]:
                y_hat, attn, mask, *_ = generator.module.infer(
                    x,
                    x_lengths,
                    speakers,
                    tone,
                    language,
                    bert,
                    ja_bert,
                    en_bert,
                    emo,
                    y=spec,
                    max_len=1000,
                    sdp_ratio=0.0 if not use_sdp else 1.0,
                )
                y_hat_lengths = mask.sum([1, 2]).long() * hps.data.hop_length

                mel = spec_to_mel_torch(
                    spec,
                    hps.data.filter_length,
                    hps.data.n_mel_channels,
                    hps.data.sampling_rate,
                    hps.data.mel_fmin,
                    hps.data.mel_fmax,
                )
                y_hat_mel = mel_spectrogram_torch(
                    y_hat.squeeze(1).float(),
                    hps.data.filter_length,
                    hps.data.n_mel_channels,
                    hps.data.sampling_rate,
                    hps.data.hop_length,
                    hps.data.win_length,
                    hps.data.mel_fmin,
                    hps.data.mel_fmax,
                )
                image_dict.update(
                    {
                        f"gen/mel_{batch_idx}": utils.plot_spectrogram_to_numpy(
                            y_hat_mel[0].cpu().numpy()
                        )
                    }
                )
                audio_dict.update(
                    {
                        f"gen/audio_{batch_idx}_{use_sdp}": y_hat[
                            0, :, : y_hat_lengths[0]
                        ]
                    }
                )
                image_dict.update(
                    {
                        f"gt/mel_{batch_idx}": utils.plot_spectrogram_to_numpy(
                            mel[0].cpu().numpy()
                        )
                    }
                )
                audio_dict.update({f"gt/audio_{batch_idx}": y[0, :, : y_lengths[0]]})

    utils.summarize(
        writer=writer_eval,
        global_step=global_step,
        images=image_dict,
        audios=audio_dict,
        audio_sampling_rate=hps.data.sampling_rate,
    )
    generator.train()


if __name__ == "__main__":
    run()