lamooon commited on
Commit
e8f806f
Β·
verified Β·
1 Parent(s): 6a522f7

Upload logs/rmvpe_ft/rmvpe_ft_v3_20260329_222405.log with huggingface_hub

Browse files
logs/rmvpe_ft/rmvpe_ft_v3_20260329_222405.log ADDED
@@ -0,0 +1,121 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0
  outputs/rmvpe_ft_v3/best.pt : 14%|β–ˆβ–Ž | 50.2MB / 368MB 
 
 
1
  outputs/rmvpe_ft_v3/best.pt : 14%|β–ˆβ–Ž | 50.2MB / 368MB 
 
 
2
  outputs/rmvpe_ft_v3/best.pt : 30%|β–ˆβ–ˆβ–‰ | 109MB / 368MB 
 
 
3
  outputs/rmvpe_ft_v3/best.pt : 46%|β–ˆβ–ˆβ–ˆβ–ˆβ–Œ | 168MB / 368MB 
 
 
4
  outputs/rmvpe_ft_v3/best.pt : 61%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ– | 226MB / 368MB 
 
 
5
  outputs/rmvpe_ft_v3/best.pt : 77%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‹ | 285MB / 368MB 
 
 
6
  outputs/rmvpe_ft_v3/best.pt : 96%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–Œ| 352MB / 368MB 
 
 
7
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‰| 368MB / 368MB 
 
 
8
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆοΏ½οΏ½β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‰| 368MB / 368MB 
 
 
9
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‰| 368MB / 368MB 
 
 
10
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‰| 368MB / 368MB 
 
 
11
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‰| 368MB / 368MB 
 
 
12
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‰| 368MB / 368MB 
 
 
13
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‰| 368MB / 368MB 
 
 
14
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
 
 
 
15
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
 
 
16
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
 
 
17
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
 
 
18
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB
 
 
 
 
 
 
 
 
19
  outputs/rmvpe_ft_v3/last.pt : 16%|β–ˆβ–Œ | 58.6MB / 368MB 
 
 
20
  outputs/rmvpe_ft_v3/last.pt : 16%|β–ˆβ–Œ | 58.6MB / 368MB 
 
 
21
  outputs/rmvpe_ft_v3/last.pt : 34%|β–ˆβ–ˆβ–ˆβ– | 126MB / 368MB 
 
 
22
  outputs/rmvpe_ft_v3/last.pt : 50%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆ | 185MB / 368MB 
 
 
23
  outputs/rmvpe_ft_v3/last.pt : 66%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–Œ | 243MB / 368MB 
 
 
24
  outputs/rmvpe_ft_v3/last.pt : 84%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ– | 310MB / 368MB 
 
 
25
  outputs/rmvpe_ft_v3/last.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
 
 
26
  outputs/rmvpe_ft_v3/last.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
 
 
27
  outputs/rmvpe_ft_v3/last.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
 
 
28
  outputs/rmvpe_ft_v3/last.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
 
 
29
  outputs/rmvpe_ft_v3/last.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
 
 
30
  outputs/rmvpe_ft_v3/last.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB
 
1
+ === Fine-tune RMVPE β€” Sun Mar 29 22:24:05 HKT 2026 ===
2
+
3
+ === Run 1/5 β€” Sun Mar 29 22:24:05 HKT 2026 ===
4
+ /userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/resampy/filters.py:50: UserWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html. The pkg_resources package is slated for removal as early as 2025-11-30. Refrain from using this package or pin to Setuptools<81.
5
+ import pkg_resources
6
+ /userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/finetune_rmvpe.py:557: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead.
7
+ scaler = GradScaler(enabled=args.fp16 and device.type == 'cuda')
8
+ Datasets: ['mir_st500', 'medleydb']
9
+ Train: 318 tracks, Val: 39 tracks
10
+ Smoke test: audio=torch.Size([40960]), label=torch.Size([257, 360]), label_max=0.9686, label_sum=625.3
11
+ RMVPE E2E verified at /userhome/cs/u3571824/fyp/muse/backend/ml_models/prod/weights/rmvpe.pt (368MB)
12
+ Loading pretrained RMVPE from /userhome/cs/u3571824/fyp/muse/backend/ml_models/prod/weights/rmvpe.pt
13
+ RMVPE: 92.0M params
14
+ Frozen 230/417 parameter groups (encoder + intermediate)
15
+
16
+ === Fine-tuning RMVPE: 30 epochs, lr=5e-05 ===
17
+ /userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/finetune_rmvpe.py:586: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead.
18
+ with cuda_autocast(enabled=args.fp16 and device.type == 'cuda'):
19
+ Epoch 1/30 (6s) lr_head=4.99e-05 lr_enc=9.97e-07 | loss=0.1911 | val_RPA=0.5991
20
+ β˜… New best RPA: 0.5991
21
+ Epoch 2/30 (5s) lr_head=4.95e-05 lr_enc=9.89e-07 | loss=0.1896 | val_RPA=0.5957
22
+ Epoch 3/30 (5s) lr_head=4.88e-05 lr_enc=9.76e-07 | loss=0.1896 | val_RPA=0.5966
23
+ Epoch 4/30 (5s) lr_head=4.78e-05 lr_enc=9.57e-07 | loss=0.1888 | val_RPA=0.5932
24
+ Epoch 5/30 (5s) lr_head=4.67e-05 lr_enc=9.33e-07 | loss=0.1890 | val_RPA=0.5900
25
+ Epoch 6/30 (5s) lr_head=4.52e-05 lr_enc=9.05e-07 | loss=0.1885 | val_RPA=0.5929
26
+ Epoch 7/30 (5s) lr_head=4.36e-05 lr_enc=8.72e-07 | loss=0.1866 | val_RPA=0.5996
27
+ β˜… New best RPA: 0.5996
28
+ Epoch 8/30 (5s) lr_head=4.17e-05 lr_enc=8.35e-07 | loss=0.1874 | val_RPA=0.5992
29
+ Epoch 9/30 (5s) lr_head=3.97e-05 lr_enc=7.94e-07 | loss=0.1869 | val_RPA=0.5967
30
+ Epoch 10/30 (5s) lr_head=3.75e-05 lr_enc=7.50e-07 | loss=0.1858 | val_RPA=0.6113
31
+ β˜… New best RPA: 0.6113
32
+ Epoch 11/30 (5s) lr_head=3.52e-05 lr_enc=7.03e-07 | loss=0.1851 | val_RPA=0.6185
33
+ β˜… New best RPA: 0.6185
34
+ Epoch 12/30 (5s) lr_head=3.27e-05 lr_enc=6.55e-07 | loss=0.1854 | val_RPA=0.6253
35
+ β˜… New best RPA: 0.6253
36
+ Epoch 13/30 (5s) lr_head=3.02e-05 lr_enc=6.04e-07 | loss=0.1850 | val_RPA=0.6220
37
+ Epoch 14/30 (5s) lr_head=2.76e-05 lr_enc=5.52e-07 | loss=0.1849 | val_RPA=0.6218
38
+ Epoch 15/30 (5s) lr_head=2.50e-05 lr_enc=5.00e-07 | loss=0.1853 | val_RPA=0.6181
39
+ Epoch 16/30 (5s) lr_head=2.24e-05 lr_enc=4.48e-07 | loss=0.1851 | val_RPA=0.6149
40
+ Epoch 17/30 (5s) lr_head=1.98e-05 lr_enc=3.96e-07 | loss=0.1840 | val_RPA=0.6157
41
+ Epoch 18/30 (5s) lr_head=1.73e-05 lr_enc=3.45e-07 | loss=0.1848 | val_RPA=0.6157
42
+ Epoch 19/30 (5s) lr_head=1.48e-05 lr_enc=2.97e-07 | loss=0.1851 | val_RPA=0.6157
43
+ Epoch 20/30 (5s) lr_head=1.25e-05 lr_enc=2.50e-07 | loss=0.1844 | val_RPA=0.6211
44
+ Early stopping at epoch 20
45
+
46
+ Fine-tuning complete. Best RPA: 0.6253
47
+ Training finished at Sun Mar 29 22:26:35 HKT 2026
48
+
49
+ === Upload to HuggingFace ===
50
+ Uploading best.pt...
51
+
52
+
53
+
54
+
55
  outputs/rmvpe_ft_v3/best.pt : 14%|β–ˆβ–Ž | 50.2MB / 368MB 
56
+
57
+
58
  outputs/rmvpe_ft_v3/best.pt : 14%|β–ˆβ–Ž | 50.2MB / 368MB 
59
+
60
+
61
  outputs/rmvpe_ft_v3/best.pt : 30%|β–ˆβ–ˆβ–‰ | 109MB / 368MB 
62
+
63
+
64
  outputs/rmvpe_ft_v3/best.pt : 46%|β–ˆβ–ˆβ–ˆβ–ˆβ–Œ | 168MB / 368MB 
65
+
66
+
67
  outputs/rmvpe_ft_v3/best.pt : 61%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ– | 226MB / 368MB 
68
+
69
+
70
  outputs/rmvpe_ft_v3/best.pt : 77%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‹ | 285MB / 368MB 
71
+
72
+
73
  outputs/rmvpe_ft_v3/best.pt : 96%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–Œ| 352MB / 368MB 
74
+
75
+
76
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‰| 368MB / 368MB 
77
+
78
+
79
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆοΏ½οΏ½β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‰| 368MB / 368MB 
80
+
81
+
82
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‰| 368MB / 368MB 
83
+
84
+
85
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‰| 368MB / 368MB 
86
+
87
+
88
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‰| 368MB / 368MB 
89
+
90
+
91
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‰| 368MB / 368MB 
92
+
93
+
94
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‰| 368MB / 368MB 
95
+
96
+
97
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
98
+
99
+
100
+
101
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
102
+
103
+
104
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
105
+
106
+
107
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
108
+
109
+
110
  outputs/rmvpe_ft_v3/best.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB
111
+ Done: checkpoints/rmvpe_ft_v3/best.pt
112
+ Uploading training_log.jsonl...
113
+ Done: checkpoints/rmvpe_ft_v3/training_log.jsonl
114
+ Uploading last.pt...
115
+
116
+
117
+
118
+
119
  outputs/rmvpe_ft_v3/last.pt : 16%|β–ˆβ–Œ | 58.6MB / 368MB 
120
+
121
+
122
  outputs/rmvpe_ft_v3/last.pt : 16%|β–ˆβ–Œ | 58.6MB / 368MB 
123
+
124
+
125
  outputs/rmvpe_ft_v3/last.pt : 34%|β–ˆβ–ˆβ–ˆβ– | 126MB / 368MB 
126
+
127
+
128
  outputs/rmvpe_ft_v3/last.pt : 50%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆ | 185MB / 368MB 
129
+
130
+
131
  outputs/rmvpe_ft_v3/last.pt : 66%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–Œ | 243MB / 368MB 
132
+
133
+
134
  outputs/rmvpe_ft_v3/last.pt : 84%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ– | 310MB / 368MB 
135
+
136
+
137
  outputs/rmvpe_ft_v3/last.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
138
+
139
+
140
  outputs/rmvpe_ft_v3/last.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
141
+
142
+
143
  outputs/rmvpe_ft_v3/last.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
144
+
145
+
146
  outputs/rmvpe_ft_v3/last.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
147
+
148
+
149
  outputs/rmvpe_ft_v3/last.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB 
150
+
151
+
152
  outputs/rmvpe_ft_v3/last.pt : 100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 368MB / 368MB