Hugging Face's logo

Phospy
/
test

TensorBoard
Model card Files Metrics Community
Phospy commited on
Commit
222dab9
·
1 Parent(s): 1d28f68

vit_model_kotae

Browse files
Files changed (11) hide show
  1. kotae/D_0.pth +3 -0
  2. kotae/D_22.pth +3 -0
  3. kotae/D_32.pth +3 -0
  4. kotae/D_43.pth +3 -0
  5. kotae/G_0.pth +3 -0
  6. kotae/G_22.pth +3 -0
  7. kotae/G_32.pth +3 -0
  8. kotae/G_43.pth +3 -0
  9. kotae/config.json +98 -0
  10. kotae/lightning_logs/version_0/events.out.tfevents.1683215349.3b0d6373a930.9895.0 +3 -0
  11. kotae/lightning_logs/version_0/hparams.yaml +83 -0
kotae/D_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:635be5c3409aaf3eec4135a1f5a771595683f3a6461ffc5bdea43441e50269a9
3
+ size 187018591
kotae/D_22.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:99d81f7c1f28ce3cecd611f87327a4e26349268aaed0fb454ecaa48ba6af37e8
3
+ size 561098249
kotae/D_32.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c8d1201f90cb9896a0ab47859c3fdbf3e797d37da84bb97b0d7d5339ac5c62c6
3
+ size 561098249
kotae/D_43.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:50875dc6f5edf9c75d8f05db4e5fb66d7a0ff7de94e5d4e31715301f1637c251
3
+ size 561098249
kotae/G_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:20a327c54e5731bed377bd38404bc32ab98e66a1b2777b0af4cc034d4d6914b0
3
+ size 180628517
kotae/G_22.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:840bfd89cfd62ac6d44265d16786f8947b8ce466dd76819f50a75784e965ce9d
3
+ size 542789469
kotae/G_32.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2527f559e025de1582ec2fd78f13582c857d4c64f8b75627744bc1e2fdc46c26
3
+ size 542789469
kotae/G_43.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:42dc63a0c195438e1b7905ed2f37d6fc56c944b2f4545e9d16a2b4a9f619b29f
3
+ size 542789469
kotae/config.json ADDED
@@ -0,0 +1,98 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "train": {
3
+ "log_interval": 200,
4
+ "eval_interval": 800,
5
+ "seed": 1234,
6
+ "epochs": 10000,
7
+ "learning_rate": 0.0001,
8
+ "betas": [
9
+ 0.8,
10
+ 0.99
11
+ ],
12
+ "eps": 1e-09,
13
+ "batch_size": 16,
14
+ "fp16_run": false,
15
+ "bf16_run": false,
16
+ "lr_decay": 0.999875,
17
+ "segment_size": 10240,
18
+ "init_lr_ratio": 1,
19
+ "warmup_epochs": 0,
20
+ "c_mel": 45,
21
+ "c_kl": 1.0,
22
+ "use_sr": true,
23
+ "max_speclen": 512,
24
+ "port": "8001",
25
+ "keep_ckpts": 3,
26
+ "num_workers": 4,
27
+ "log_version": 0,
28
+ "ckpt_name_by_step": false,
29
+ "accumulate_grad_batches": 1
30
+ },
31
+ "data": {
32
+ "training_files": "filelists/44k/train.txt",
33
+ "validation_files": "filelists/44k/val.txt",
34
+ "max_wav_value": 32768.0,
35
+ "sampling_rate": 44100,
36
+ "filter_length": 2048,
37
+ "hop_length": 512,
38
+ "win_length": 2048,
39
+ "n_mel_channels": 80,
40
+ "mel_fmin": 0.0,
41
+ "mel_fmax": 22050
42
+ },
43
+ "model": {
44
+ "inter_channels": 192,
45
+ "hidden_channels": 192,
46
+ "filter_channels": 768,
47
+ "n_heads": 2,
48
+ "n_layers": 6,
49
+ "kernel_size": 3,
50
+ "p_dropout": 0.1,
51
+ "resblock": "1",
52
+ "resblock_kernel_sizes": [
53
+ 3,
54
+ 7,
55
+ 11
56
+ ],
57
+ "resblock_dilation_sizes": [
58
+ [
59
+ 1,
60
+ 3,
61
+ 5
62
+ ],
63
+ [
64
+ 1,
65
+ 3,
66
+ 5
67
+ ],
68
+ [
69
+ 1,
70
+ 3,
71
+ 5
72
+ ]
73
+ ],
74
+ "upsample_rates": [
75
+ 8,
76
+ 8,
77
+ 2,
78
+ 2,
79
+ 2
80
+ ],
81
+ "upsample_initial_channel": 512,
82
+ "upsample_kernel_sizes": [
83
+ 16,
84
+ 16,
85
+ 4,
86
+ 4,
87
+ 4
88
+ ],
89
+ "n_layers_q": 3,
90
+ "use_spectral_norm": false,
91
+ "gin_channels": 256,
92
+ "ssl_dim": 256,
93
+ "n_speakers": 200
94
+ },
95
+ "spk": {
96
+ "vn": 0
97
+ }
98
+ }
kotae/lightning_logs/version_0/events.out.tfevents.1683215349.3b0d6373a930.9895.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e033eb72f0fd3c2c56df2924f791ba144217ba9f07ac1902667005b362fd92dc
3
+ size 18515786
kotae/lightning_logs/version_0/hparams.yaml ADDED
@@ -0,0 +1,83 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ data: !!python/object:so_vits_svc_fork.hparams.HParams
2
+ filter_length: 2048
3
+ hop_length: 512
4
+ max_wav_value: 32768.0
5
+ mel_fmax: 22050
6
+ mel_fmin: 0.0
7
+ n_mel_channels: 80
8
+ sampling_rate: 44100
9
+ training_files: filelists/44k/train.txt
10
+ validation_files: filelists/44k/val.txt
11
+ win_length: 2048
12
+ model: !!python/object:so_vits_svc_fork.hparams.HParams
13
+ filter_channels: 768
14
+ gin_channels: 256
15
+ hidden_channels: 192
16
+ inter_channels: 192
17
+ kernel_size: 3
18
+ n_heads: 2
19
+ n_layers: 6
20
+ n_layers_q: 3
21
+ n_speakers: 200
22
+ p_dropout: 0.1
23
+ resblock: '1'
24
+ resblock_dilation_sizes:
25
+ - - 1
26
+ - 3
27
+ - 5
28
+ - - 1
29
+ - 3
30
+ - 5
31
+ - - 1
32
+ - 3
33
+ - 5
34
+ resblock_kernel_sizes:
35
+ - 3
36
+ - 7
37
+ - 11
38
+ ssl_dim: 256
39
+ upsample_initial_channel: 512
40
+ upsample_kernel_sizes:
41
+ - 16
42
+ - 16
43
+ - 4
44
+ - 4
45
+ - 4
46
+ upsample_rates:
47
+ - 8
48
+ - 8
49
+ - 2
50
+ - 2
51
+ - 2
52
+ use_spectral_norm: false
53
+ model_dir: /content/test_output
54
+ reset_optimizer: false
55
+ spk: !!python/object:so_vits_svc_fork.hparams.HParams
56
+ vn: 0
57
+ train: !!python/object:so_vits_svc_fork.hparams.HParams
58
+ accumulate_grad_batches: 1
59
+ batch_size: 16
60
+ betas:
61
+ - 0.8
62
+ - 0.99
63
+ bf16_run: false
64
+ c_kl: 1.0
65
+ c_mel: 45
66
+ ckpt_name_by_step: false
67
+ epochs: 10000
68
+ eps: 1.0e-09
69
+ eval_interval: 800
70
+ fp16_run: false
71
+ init_lr_ratio: 1
72
+ keep_ckpts: 3
73
+ learning_rate: 0.0001
74
+ log_interval: 200
75
+ log_version: 0
76
+ lr_decay: 0.999875
77
+ max_speclen: 512
78
+ num_workers: 4
79
+ port: '8001'
80
+ seed: 1234
81
+ segment_size: 10240
82
+ use_sr: true
83
+ warmup_epochs: 0