HopeLiang commited on
Commit
68abc5c
·
verified ·
1 Parent(s): 42df963

Upload SSL Layer MOS checkpoints (part 2)

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. results_bvcc/w2v2_base/layer0/run2/model.pt +3 -0
  2. results_bvcc/w2v2_base/layer0/run2/model_best.pt +3 -0
  3. results_bvcc/w2v2_base/layer0/run2/train.log +163 -0
  4. results_bvcc/w2v2_base/layer0/run3/config.gin +31 -0
  5. results_bvcc/w2v2_base/layer0/run3/model.pt +3 -0
  6. results_bvcc/w2v2_base/layer0/run3/model_best.pt +3 -0
  7. results_bvcc/w2v2_base/layer0/run3/train.log +163 -0
  8. results_bvcc/w2v2_base/layer0/run4/config.gin +31 -0
  9. results_bvcc/w2v2_base/layer0/run4/model.pt +3 -0
  10. results_bvcc/w2v2_base/layer0/run4/model_best.pt +3 -0
  11. results_bvcc/w2v2_base/layer0/run4/train.log +163 -0
  12. results_bvcc/w2v2_base/layer1/run0/config.gin +31 -0
  13. results_bvcc/w2v2_base/layer1/run0/model.pt +3 -0
  14. results_bvcc/w2v2_base/layer1/run0/model_best.pt +3 -0
  15. results_bvcc/w2v2_base/layer1/run0/train.log +163 -0
  16. results_bvcc/w2v2_base/layer1/run1/config.gin +31 -0
  17. results_bvcc/w2v2_base/layer1/run1/model.pt +3 -0
  18. results_bvcc/w2v2_base/layer1/run1/model_best.pt +3 -0
  19. results_bvcc/w2v2_base/layer1/run1/train.log +163 -0
  20. results_bvcc/w2v2_base/layer1/run2/config.gin +31 -0
  21. results_bvcc/w2v2_base/layer1/run2/model.pt +3 -0
  22. results_bvcc/w2v2_base/layer1/run2/model_best.pt +3 -0
  23. results_bvcc/w2v2_base/layer1/run2/train.log +163 -0
  24. results_bvcc/w2v2_base/layer1/run3/config.gin +31 -0
  25. results_bvcc/w2v2_base/layer1/run3/model.pt +3 -0
  26. results_bvcc/w2v2_base/layer1/run3/model_best.pt +3 -0
  27. results_bvcc/w2v2_base/layer1/run3/train.log +163 -0
  28. results_bvcc/w2v2_base/layer1/run4/config.gin +31 -0
  29. results_bvcc/w2v2_base/layer1/run4/model.pt +3 -0
  30. results_bvcc/w2v2_base/layer1/run4/model_best.pt +3 -0
  31. results_bvcc/w2v2_base/layer1/run4/train.log +163 -0
  32. results_bvcc/w2v2_base/layer10/run0/config.gin +31 -0
  33. results_bvcc/w2v2_base/layer10/run0/model.pt +3 -0
  34. results_bvcc/w2v2_base/layer10/run0/model_best.pt +3 -0
  35. results_bvcc/w2v2_base/layer10/run0/train.log +163 -0
  36. results_bvcc/w2v2_base/layer10/run1/config.gin +31 -0
  37. results_bvcc/w2v2_base/layer10/run1/model.pt +3 -0
  38. results_bvcc/w2v2_base/layer10/run1/model_best.pt +3 -0
  39. results_bvcc/w2v2_base/layer10/run1/train.log +163 -0
  40. results_bvcc/w2v2_base/layer10/run2/config.gin +31 -0
  41. results_bvcc/w2v2_base/layer10/run2/model.pt +3 -0
  42. results_bvcc/w2v2_base/layer10/run2/model_best.pt +3 -0
  43. results_bvcc/w2v2_base/layer10/run2/train.log +163 -0
  44. results_bvcc/w2v2_base/layer10/run3/config.gin +31 -0
  45. results_bvcc/w2v2_base/layer10/run3/model.pt +3 -0
  46. results_bvcc/w2v2_base/layer10/run3/model_best.pt +3 -0
  47. results_bvcc/w2v2_base/layer10/run3/train.log +163 -0
  48. results_bvcc/w2v2_base/layer10/run4/config.gin +31 -0
  49. results_bvcc/w2v2_base/layer10/run4/model.pt +3 -0
  50. results_bvcc/w2v2_base/layer10/run4/model_best.pt +3 -0
results_bvcc/w2v2_base/layer0/run2/model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:39072446dd19190b27a6518cbaded35e7f226989b4b6b1c0c861f8e2983921c1
3
+ size 2747979
results_bvcc/w2v2_base/layer0/run2/model_best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2479c85d5a78aafa1167c7cbbb43de37648403333e991cb561c4fecdb8a68427
3
+ size 2749702
results_bvcc/w2v2_base/layer0/run2/train.log ADDED
@@ -0,0 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ INFO:root:Num train speech clips: 4974
2
+ INFO:root:Num val speech clips: 1066
3
+ INFO:root:Num test speech clips: 1066
4
+ INFO:root:Device=cuda
5
+ INFO:root:Average loss=0.5249429885775615
6
+ INFO:root:
7
+ [Valid][0][UTT][ MSE = 0.4273 | LCC = 0.7562 | SRCC = 0.7566 ]
8
+ INFO:root:
9
+ [Valid][0][SYS][ MSE = 0.3347 | LCC = 0.8364 | SRCC = 0.8395 ]
10
+ INFO:root:Average loss=0.30377670434805065
11
+ INFO:root:
12
+ [Valid][1][UTT][ MSE = 0.3053 | LCC = 0.8010 | SRCC = 0.7986 ]
13
+ INFO:root:
14
+ [Valid][1][SYS][ MSE = 0.1880 | LCC = 0.8753 | SRCC = 0.8653 ]
15
+ INFO:root:Average loss=0.2591542744865784
16
+ INFO:root:
17
+ [Valid][2][UTT][ MSE = 0.2888 | LCC = 0.8165 | SRCC = 0.8144 ]
18
+ INFO:root:
19
+ [Valid][2][SYS][ MSE = 0.1695 | LCC = 0.8919 | SRCC = 0.8861 ]
20
+ INFO:root:Average loss=0.23760433877125764
21
+ INFO:root:
22
+ [Valid][3][UTT][ MSE = 0.2803 | LCC = 0.8266 | SRCC = 0.8273 ]
23
+ INFO:root:
24
+ [Valid][3][SYS][ MSE = 0.1720 | LCC = 0.9016 | SRCC = 0.9013 ]
25
+ INFO:root:Average loss=0.21500673259680086
26
+ INFO:root:
27
+ [Valid][4][UTT][ MSE = 0.2706 | LCC = 0.8306 | SRCC = 0.8277 ]
28
+ INFO:root:
29
+ [Valid][4][SYS][ MSE = 0.1630 | LCC = 0.9061 | SRCC = 0.8987 ]
30
+ INFO:root:Average loss=0.2158757595297618
31
+ INFO:root:
32
+ [Valid][5][UTT][ MSE = 0.2761 | LCC = 0.8284 | SRCC = 0.8302 ]
33
+ INFO:root:
34
+ [Valid][5][SYS][ MSE = 0.1438 | LCC = 0.9066 | SRCC = 0.9067 ]
35
+ INFO:root:Average loss=0.21459559991191596
36
+ INFO:root:
37
+ [Valid][6][UTT][ MSE = 0.2816 | LCC = 0.8356 | SRCC = 0.8333 ]
38
+ INFO:root:
39
+ [Valid][6][SYS][ MSE = 0.1471 | LCC = 0.9174 | SRCC = 0.9117 ]
40
+ INFO:root:Average loss=0.1934651486002482
41
+ INFO:root:
42
+ [Valid][7][UTT][ MSE = 0.2609 | LCC = 0.8387 | SRCC = 0.8361 ]
43
+ INFO:root:
44
+ [Valid][7][SYS][ MSE = 0.1405 | LCC = 0.9237 | SRCC = 0.9188 ]
45
+ INFO:root:Average loss=0.17381173763901758
46
+ INFO:root:
47
+ [Valid][8][UTT][ MSE = 0.2560 | LCC = 0.8451 | SRCC = 0.8417 ]
48
+ INFO:root:
49
+ [Valid][8][SYS][ MSE = 0.1467 | LCC = 0.9229 | SRCC = 0.9171 ]
50
+ INFO:root:Average loss=0.17481609118672517
51
+ INFO:root:
52
+ [Valid][9][UTT][ MSE = 0.2483 | LCC = 0.8449 | SRCC = 0.8430 ]
53
+ INFO:root:
54
+ [Valid][9][SYS][ MSE = 0.1314 | LCC = 0.9251 | SRCC = 0.9197 ]
55
+ INFO:root:Average loss=0.17029272182247576
56
+ INFO:root:
57
+ [Valid][10][UTT][ MSE = 0.2478 | LCC = 0.8429 | SRCC = 0.8405 ]
58
+ INFO:root:
59
+ [Valid][10][SYS][ MSE = 0.1252 | LCC = 0.9212 | SRCC = 0.9163 ]
60
+ INFO:root:Average loss=0.1518475871819716
61
+ INFO:root:
62
+ [Valid][11][UTT][ MSE = 0.2435 | LCC = 0.8453 | SRCC = 0.8433 ]
63
+ INFO:root:
64
+ [Valid][11][SYS][ MSE = 0.1191 | LCC = 0.9256 | SRCC = 0.9191 ]
65
+ INFO:root:Average loss=0.1431034279939456
66
+ INFO:root:
67
+ [Valid][12][UTT][ MSE = 0.2419 | LCC = 0.8462 | SRCC = 0.8440 ]
68
+ INFO:root:
69
+ [Valid][12][SYS][ MSE = 0.1183 | LCC = 0.9252 | SRCC = 0.9196 ]
70
+ INFO:root:Average loss=0.13872973315226725
71
+ INFO:root:
72
+ [Valid][13][UTT][ MSE = 0.2482 | LCC = 0.8422 | SRCC = 0.8420 ]
73
+ INFO:root:
74
+ [Valid][13][SYS][ MSE = 0.1253 | LCC = 0.9246 | SRCC = 0.9228 ]
75
+ INFO:root:Average loss=0.1254938454964222
76
+ INFO:root:
77
+ [Valid][14][UTT][ MSE = 0.2690 | LCC = 0.8387 | SRCC = 0.8416 ]
78
+ INFO:root:
79
+ [Valid][14][SYS][ MSE = 0.1587 | LCC = 0.9203 | SRCC = 0.9232 ]
80
+ INFO:root:Average loss=0.12176823959900783
81
+ INFO:root:
82
+ [Valid][15][UTT][ MSE = 0.2515 | LCC = 0.8395 | SRCC = 0.8416 ]
83
+ INFO:root:
84
+ [Valid][15][SYS][ MSE = 0.1324 | LCC = 0.9226 | SRCC = 0.9240 ]
85
+ INFO:root:Average loss=0.11692235389581093
86
+ INFO:root:
87
+ [Valid][16][UTT][ MSE = 0.2548 | LCC = 0.8403 | SRCC = 0.8428 ]
88
+ INFO:root:
89
+ [Valid][16][SYS][ MSE = 0.1451 | LCC = 0.9205 | SRCC = 0.9239 ]
90
+ INFO:root:Average loss=0.11832601949572563
91
+ INFO:root:
92
+ [Valid][17][UTT][ MSE = 0.2401 | LCC = 0.8479 | SRCC = 0.8469 ]
93
+ INFO:root:
94
+ [Valid][17][SYS][ MSE = 0.1205 | LCC = 0.9271 | SRCC = 0.9238 ]
95
+ INFO:root:Average loss=0.10400467728957152
96
+ INFO:root:
97
+ [Valid][18][UTT][ MSE = 0.2378 | LCC = 0.8469 | SRCC = 0.8464 ]
98
+ INFO:root:
99
+ [Valid][18][SYS][ MSE = 0.1086 | LCC = 0.9307 | SRCC = 0.9273 ]
100
+ INFO:root:Average loss=0.10339826460068043
101
+ INFO:root:
102
+ [Valid][19][UTT][ MSE = 0.2356 | LCC = 0.8478 | SRCC = 0.8469 ]
103
+ INFO:root:
104
+ [Valid][19][SYS][ MSE = 0.1147 | LCC = 0.9298 | SRCC = 0.9276 ]
105
+ INFO:root:Average loss=0.10017874765281494
106
+ INFO:root:
107
+ [Valid][20][UTT][ MSE = 0.2394 | LCC = 0.8518 | SRCC = 0.8508 ]
108
+ INFO:root:
109
+ [Valid][20][SYS][ MSE = 0.1305 | LCC = 0.9318 | SRCC = 0.9281 ]
110
+ INFO:root:Average loss=0.09424310903518628
111
+ INFO:root:
112
+ [Valid][21][UTT][ MSE = 0.2368 | LCC = 0.8460 | SRCC = 0.8457 ]
113
+ INFO:root:
114
+ [Valid][21][SYS][ MSE = 0.1091 | LCC = 0.9294 | SRCC = 0.9279 ]
115
+ INFO:root:Average loss=0.09064795965185532
116
+ INFO:root:
117
+ [Valid][22][UTT][ MSE = 0.2336 | LCC = 0.8543 | SRCC = 0.8507 ]
118
+ INFO:root:
119
+ [Valid][22][SYS][ MSE = 0.1214 | LCC = 0.9338 | SRCC = 0.9255 ]
120
+ INFO:root:Average loss=0.09291106758591457
121
+ INFO:root:
122
+ [Valid][23][UTT][ MSE = 0.2494 | LCC = 0.8492 | SRCC = 0.8489 ]
123
+ INFO:root:
124
+ [Valid][23][SYS][ MSE = 0.1401 | LCC = 0.9328 | SRCC = 0.9335 ]
125
+ INFO:root:Average loss=0.0867785691546324
126
+ INFO:root:
127
+ [Valid][24][UTT][ MSE = 0.2381 | LCC = 0.8477 | SRCC = 0.8477 ]
128
+ INFO:root:
129
+ [Valid][24][SYS][ MSE = 0.1199 | LCC = 0.9325 | SRCC = 0.9303 ]
130
+ INFO:root:Average loss=0.08516144384749424
131
+ INFO:root:
132
+ [Valid][25][UTT][ MSE = 0.2410 | LCC = 0.8479 | SRCC = 0.8472 ]
133
+ INFO:root:
134
+ [Valid][25][SYS][ MSE = 0.1080 | LCC = 0.9384 | SRCC = 0.9383 ]
135
+ INFO:root:Average loss=0.08252089599577281
136
+ INFO:root:
137
+ [Valid][26][UTT][ MSE = 0.2298 | LCC = 0.8537 | SRCC = 0.8518 ]
138
+ INFO:root:
139
+ [Valid][26][SYS][ MSE = 0.1116 | LCC = 0.9307 | SRCC = 0.9273 ]
140
+ INFO:root:Average loss=0.0788705332730061
141
+ INFO:root:
142
+ [Valid][27][UTT][ MSE = 0.2392 | LCC = 0.8526 | SRCC = 0.8505 ]
143
+ INFO:root:
144
+ [Valid][27][SYS][ MSE = 0.1313 | LCC = 0.9333 | SRCC = 0.9303 ]
145
+ INFO:root:Average loss=0.07371497244980091
146
+ INFO:root:
147
+ [Valid][28][UTT][ MSE = 0.2288 | LCC = 0.8548 | SRCC = 0.8532 ]
148
+ INFO:root:
149
+ [Valid][28][SYS][ MSE = 0.1076 | LCC = 0.9363 | SRCC = 0.9329 ]
150
+ INFO:root:Average loss=0.06961088651456894
151
+ INFO:root:
152
+ [Valid][29][UTT][ MSE = 0.2316 | LCC = 0.8551 | SRCC = 0.8536 ]
153
+ INFO:root:
154
+ [Valid][29][SYS][ MSE = 0.1062 | LCC = 0.9368 | SRCC = 0.9344 ]
155
+ INFO:root:Average loss=0.07108233381922428
156
+ INFO:root:
157
+ [Valid][30][UTT][ MSE = 0.2239 | LCC = 0.8575 | SRCC = 0.8542 ]
158
+ INFO:root:
159
+ [Valid][30][SYS][ MSE = 0.1046 | LCC = 0.9353 | SRCC = 0.9271 ]
160
+ INFO:root:
161
+ [Test][31][UTT][ MSE = 0.2422 | LCC = 0.8542 | SRCC = 0.8538 ]
162
+ INFO:root:
163
+ [Test][31][SYS][ MSE = 0.1504 | LCC = 0.9074 | SRCC = 0.9035 ]
results_bvcc/w2v2_base/layer0/run3/config.gin ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Training config on BVCC dataset.
2
+
3
+ # ------------------------------
4
+ # Architecture
5
+ # ------------------------------
6
+ ProjectionHead:
7
+ conv_channels = (32, 32)
8
+ dense_neurons = (64, 32, 1)
9
+ use_poolings = (True, True)
10
+ kernel_size = 5
11
+ pool_size = 5
12
+
13
+ # ------------------------------
14
+ # Dataset
15
+ # ------------------------------
16
+ Bvcc:
17
+ debug = False
18
+
19
+ get_dataloader:
20
+ num_workers = 12
21
+ shuffle = True
22
+
23
+ # ------------------------------
24
+ # Training loop
25
+ # ------------------------------
26
+ TrainingLoop:
27
+ num_epochs = 30
28
+ learning_rate = 1e-4
29
+ dataset_cls = @Bvcc
30
+ weight_decay = 0.0
31
+ batch_size_train = 64
results_bvcc/w2v2_base/layer0/run3/model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc4a982063b4ffb202b6ac254c3799359012b89e76327055d54f1c9b1ec838d5
3
+ size 2747979
results_bvcc/w2v2_base/layer0/run3/model_best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f7621b5451fa3fb81d41231986683d3e6643a5009211efbbae03ed650229ec95
3
+ size 2749702
results_bvcc/w2v2_base/layer0/run3/train.log ADDED
@@ -0,0 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ INFO:root:Num train speech clips: 4974
2
+ INFO:root:Num val speech clips: 1066
3
+ INFO:root:Num test speech clips: 1066
4
+ INFO:root:Device=cuda
5
+ INFO:root:Average loss=0.563079174894553
6
+ INFO:root:
7
+ [Valid][0][UTT][ MSE = 0.4337 | LCC = 0.7487 | SRCC = 0.7508 ]
8
+ INFO:root:
9
+ [Valid][0][SYS][ MSE = 0.2867 | LCC = 0.8306 | SRCC = 0.8267 ]
10
+ INFO:root:Average loss=0.33041678082484466
11
+ INFO:root:
12
+ [Valid][1][UTT][ MSE = 0.3267 | LCC = 0.7845 | SRCC = 0.7847 ]
13
+ INFO:root:
14
+ [Valid][1][SYS][ MSE = 0.1826 | LCC = 0.8734 | SRCC = 0.8705 ]
15
+ INFO:root:Average loss=0.27019520084827375
16
+ INFO:root:
17
+ [Valid][2][UTT][ MSE = 0.3007 | LCC = 0.8091 | SRCC = 0.8075 ]
18
+ INFO:root:
19
+ [Valid][2][SYS][ MSE = 0.1818 | LCC = 0.8906 | SRCC = 0.8862 ]
20
+ INFO:root:Average loss=0.2408250344869418
21
+ INFO:root:
22
+ [Valid][3][UTT][ MSE = 0.2780 | LCC = 0.8216 | SRCC = 0.8216 ]
23
+ INFO:root:
24
+ [Valid][3][SYS][ MSE = 0.1474 | LCC = 0.9006 | SRCC = 0.8963 ]
25
+ INFO:root:Average loss=0.2187499035245333
26
+ INFO:root:
27
+ [Valid][4][UTT][ MSE = 0.2767 | LCC = 0.8258 | SRCC = 0.8234 ]
28
+ INFO:root:
29
+ [Valid][4][SYS][ MSE = 0.1388 | LCC = 0.9111 | SRCC = 0.9036 ]
30
+ INFO:root:Average loss=0.2035780111566568
31
+ INFO:root:
32
+ [Valid][5][UTT][ MSE = 0.3005 | LCC = 0.8413 | SRCC = 0.8387 ]
33
+ INFO:root:
34
+ [Valid][5][SYS][ MSE = 0.1667 | LCC = 0.9103 | SRCC = 0.9020 ]
35
+ INFO:root:Average loss=0.19972144229671893
36
+ INFO:root:
37
+ [Valid][6][UTT][ MSE = 0.2593 | LCC = 0.8409 | SRCC = 0.8403 ]
38
+ INFO:root:
39
+ [Valid][6][SYS][ MSE = 0.1387 | LCC = 0.9159 | SRCC = 0.9150 ]
40
+ INFO:root:Average loss=0.18598864466334
41
+ INFO:root:
42
+ [Valid][7][UTT][ MSE = 0.2857 | LCC = 0.8395 | SRCC = 0.8397 ]
43
+ INFO:root:
44
+ [Valid][7][SYS][ MSE = 0.1415 | LCC = 0.9165 | SRCC = 0.9147 ]
45
+ INFO:root:Average loss=0.17202930343456757
46
+ INFO:root:
47
+ [Valid][8][UTT][ MSE = 0.2496 | LCC = 0.8409 | SRCC = 0.8413 ]
48
+ INFO:root:
49
+ [Valid][8][SYS][ MSE = 0.1294 | LCC = 0.9146 | SRCC = 0.9129 ]
50
+ INFO:root:Average loss=0.1724835539666506
51
+ INFO:root:
52
+ [Valid][9][UTT][ MSE = 0.2536 | LCC = 0.8402 | SRCC = 0.8403 ]
53
+ INFO:root:
54
+ [Valid][9][SYS][ MSE = 0.1263 | LCC = 0.9170 | SRCC = 0.9169 ]
55
+ INFO:root:Average loss=0.15639462551245323
56
+ INFO:root:
57
+ [Valid][10][UTT][ MSE = 0.2747 | LCC = 0.8431 | SRCC = 0.8417 ]
58
+ INFO:root:
59
+ [Valid][10][SYS][ MSE = 0.1342 | LCC = 0.9183 | SRCC = 0.9163 ]
60
+ INFO:root:Average loss=0.146468396179187
61
+ INFO:root:
62
+ [Valid][11][UTT][ MSE = 0.2483 | LCC = 0.8401 | SRCC = 0.8411 ]
63
+ INFO:root:
64
+ [Valid][11][SYS][ MSE = 0.1328 | LCC = 0.9172 | SRCC = 0.9202 ]
65
+ INFO:root:Average loss=0.1407206617295742
66
+ INFO:root:
67
+ [Valid][12][UTT][ MSE = 0.2554 | LCC = 0.8406 | SRCC = 0.8417 ]
68
+ INFO:root:
69
+ [Valid][12][SYS][ MSE = 0.1231 | LCC = 0.9119 | SRCC = 0.9137 ]
70
+ INFO:root:Average loss=0.1333160166365978
71
+ INFO:root:
72
+ [Valid][13][UTT][ MSE = 0.2423 | LCC = 0.8498 | SRCC = 0.8501 ]
73
+ INFO:root:
74
+ [Valid][13][SYS][ MSE = 0.1120 | LCC = 0.9260 | SRCC = 0.9263 ]
75
+ INFO:root:Average loss=0.1278129292604251
76
+ INFO:root:
77
+ [Valid][14][UTT][ MSE = 0.2507 | LCC = 0.8486 | SRCC = 0.8471 ]
78
+ INFO:root:
79
+ [Valid][14][SYS][ MSE = 0.1249 | LCC = 0.9207 | SRCC = 0.9179 ]
80
+ INFO:root:Average loss=0.11769445154529351
81
+ INFO:root:
82
+ [Valid][15][UTT][ MSE = 0.2547 | LCC = 0.8402 | SRCC = 0.8384 ]
83
+ INFO:root:
84
+ [Valid][15][SYS][ MSE = 0.1506 | LCC = 0.9111 | SRCC = 0.9063 ]
85
+ INFO:root:Average loss=0.11952666766368426
86
+ INFO:root:
87
+ [Valid][16][UTT][ MSE = 0.2770 | LCC = 0.8528 | SRCC = 0.8521 ]
88
+ INFO:root:
89
+ [Valid][16][SYS][ MSE = 0.1419 | LCC = 0.9216 | SRCC = 0.9163 ]
90
+ INFO:root:Average loss=0.1103925523467553
91
+ INFO:root:
92
+ [Valid][17][UTT][ MSE = 0.2377 | LCC = 0.8482 | SRCC = 0.8477 ]
93
+ INFO:root:
94
+ [Valid][17][SYS][ MSE = 0.1205 | LCC = 0.9235 | SRCC = 0.9231 ]
95
+ INFO:root:Average loss=0.1063495807063121
96
+ INFO:root:
97
+ [Valid][18][UTT][ MSE = 0.2468 | LCC = 0.8488 | SRCC = 0.8468 ]
98
+ INFO:root:
99
+ [Valid][18][SYS][ MSE = 0.1251 | LCC = 0.9173 | SRCC = 0.9115 ]
100
+ INFO:root:Average loss=0.09886931179043575
101
+ INFO:root:
102
+ [Valid][19][UTT][ MSE = 0.2427 | LCC = 0.8461 | SRCC = 0.8453 ]
103
+ INFO:root:
104
+ [Valid][19][SYS][ MSE = 0.1137 | LCC = 0.9232 | SRCC = 0.9202 ]
105
+ INFO:root:Average loss=0.10060006232024768
106
+ INFO:root:
107
+ [Valid][20][UTT][ MSE = 0.2573 | LCC = 0.8453 | SRCC = 0.8445 ]
108
+ INFO:root:
109
+ [Valid][20][SYS][ MSE = 0.1289 | LCC = 0.9172 | SRCC = 0.9168 ]
110
+ INFO:root:Average loss=0.09440205828883709
111
+ INFO:root:
112
+ [Valid][21][UTT][ MSE = 0.2490 | LCC = 0.8523 | SRCC = 0.8508 ]
113
+ INFO:root:
114
+ [Valid][21][SYS][ MSE = 0.1207 | LCC = 0.9225 | SRCC = 0.9174 ]
115
+ INFO:root:Average loss=0.08925941424109997
116
+ INFO:root:
117
+ [Valid][22][UTT][ MSE = 0.2360 | LCC = 0.8490 | SRCC = 0.8479 ]
118
+ INFO:root:
119
+ [Valid][22][SYS][ MSE = 0.1294 | LCC = 0.9212 | SRCC = 0.9222 ]
120
+ INFO:root:Average loss=0.10100043240265968
121
+ INFO:root:
122
+ [Valid][23][UTT][ MSE = 0.2638 | LCC = 0.8526 | SRCC = 0.8519 ]
123
+ INFO:root:
124
+ [Valid][23][SYS][ MSE = 0.1328 | LCC = 0.9226 | SRCC = 0.9184 ]
125
+ INFO:root:Average loss=0.08822186515690425
126
+ INFO:root:
127
+ [Valid][24][UTT][ MSE = 0.2478 | LCC = 0.8504 | SRCC = 0.8506 ]
128
+ INFO:root:
129
+ [Valid][24][SYS][ MSE = 0.1212 | LCC = 0.9195 | SRCC = 0.9163 ]
130
+ INFO:root:Average loss=0.08835151099050656
131
+ INFO:root:
132
+ [Valid][25][UTT][ MSE = 0.2535 | LCC = 0.8521 | SRCC = 0.8513 ]
133
+ INFO:root:
134
+ [Valid][25][SYS][ MSE = 0.1201 | LCC = 0.9272 | SRCC = 0.9232 ]
135
+ INFO:root:Average loss=0.08405076368496968
136
+ INFO:root:
137
+ [Valid][26][UTT][ MSE = 0.2364 | LCC = 0.8467 | SRCC = 0.8469 ]
138
+ INFO:root:
139
+ [Valid][26][SYS][ MSE = 0.1268 | LCC = 0.9179 | SRCC = 0.9190 ]
140
+ INFO:root:Average loss=0.07794641779783444
141
+ INFO:root:
142
+ [Valid][27][UTT][ MSE = 0.2393 | LCC = 0.8504 | SRCC = 0.8501 ]
143
+ INFO:root:
144
+ [Valid][27][SYS][ MSE = 0.1070 | LCC = 0.9272 | SRCC = 0.9239 ]
145
+ INFO:root:Average loss=0.07662971938649814
146
+ INFO:root:
147
+ [Valid][28][UTT][ MSE = 0.2442 | LCC = 0.8480 | SRCC = 0.8488 ]
148
+ INFO:root:
149
+ [Valid][28][SYS][ MSE = 0.1081 | LCC = 0.9257 | SRCC = 0.9264 ]
150
+ INFO:root:Average loss=0.07344848729479007
151
+ INFO:root:
152
+ [Valid][29][UTT][ MSE = 0.2342 | LCC = 0.8485 | SRCC = 0.8484 ]
153
+ INFO:root:
154
+ [Valid][29][SYS][ MSE = 0.1131 | LCC = 0.9235 | SRCC = 0.9243 ]
155
+ INFO:root:Average loss=0.07641311434025948
156
+ INFO:root:
157
+ [Valid][30][UTT][ MSE = 0.2306 | LCC = 0.8542 | SRCC = 0.8536 ]
158
+ INFO:root:
159
+ [Valid][30][SYS][ MSE = 0.1115 | LCC = 0.9230 | SRCC = 0.9198 ]
160
+ INFO:root:
161
+ [Test][31][UTT][ MSE = 0.2582 | LCC = 0.8525 | SRCC = 0.8507 ]
162
+ INFO:root:
163
+ [Test][31][SYS][ MSE = 0.1560 | LCC = 0.9164 | SRCC = 0.9162 ]
results_bvcc/w2v2_base/layer0/run4/config.gin ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Training config on BVCC dataset.
2
+
3
+ # ------------------------------
4
+ # Architecture
5
+ # ------------------------------
6
+ ProjectionHead:
7
+ conv_channels = (32, 32)
8
+ dense_neurons = (64, 32, 1)
9
+ use_poolings = (True, True)
10
+ kernel_size = 5
11
+ pool_size = 5
12
+
13
+ # ------------------------------
14
+ # Dataset
15
+ # ------------------------------
16
+ Bvcc:
17
+ debug = False
18
+
19
+ get_dataloader:
20
+ num_workers = 12
21
+ shuffle = True
22
+
23
+ # ------------------------------
24
+ # Training loop
25
+ # ------------------------------
26
+ TrainingLoop:
27
+ num_epochs = 30
28
+ learning_rate = 1e-4
29
+ dataset_cls = @Bvcc
30
+ weight_decay = 0.0
31
+ batch_size_train = 64
results_bvcc/w2v2_base/layer0/run4/model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3b892c4bc00d64d85ad907801d74ae2ed51bae41731b27c3287caa565c085a67
3
+ size 2747979
results_bvcc/w2v2_base/layer0/run4/model_best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c5ac614f976d7e5c1af693df380c3438a8015b518b2035d789cb7abb4e5070d2
3
+ size 2749702
results_bvcc/w2v2_base/layer0/run4/train.log ADDED
@@ -0,0 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ INFO:root:Num train speech clips: 4974
2
+ INFO:root:Num val speech clips: 1066
3
+ INFO:root:Num test speech clips: 1066
4
+ INFO:root:Device=cuda
5
+ INFO:root:Average loss=0.5348865794829833
6
+ INFO:root:
7
+ [Valid][0][UTT][ MSE = 0.3870 | LCC = 0.7508 | SRCC = 0.7568 ]
8
+ INFO:root:
9
+ [Valid][0][SYS][ MSE = 0.2511 | LCC = 0.8399 | SRCC = 0.8412 ]
10
+ INFO:root:Average loss=0.30528190120672566
11
+ INFO:root:
12
+ [Valid][1][UTT][ MSE = 0.3300 | LCC = 0.7935 | SRCC = 0.7971 ]
13
+ INFO:root:
14
+ [Valid][1][SYS][ MSE = 0.1870 | LCC = 0.8798 | SRCC = 0.8809 ]
15
+ INFO:root:Average loss=0.27072139905813414
16
+ INFO:root:
17
+ [Valid][2][UTT][ MSE = 0.3061 | LCC = 0.8173 | SRCC = 0.8150 ]
18
+ INFO:root:
19
+ [Valid][2][SYS][ MSE = 0.1760 | LCC = 0.8958 | SRCC = 0.8839 ]
20
+ INFO:root:Average loss=0.24719594438106585
21
+ INFO:root:
22
+ [Valid][3][UTT][ MSE = 0.2898 | LCC = 0.8216 | SRCC = 0.8233 ]
23
+ INFO:root:
24
+ [Valid][3][SYS][ MSE = 0.1825 | LCC = 0.9025 | SRCC = 0.9002 ]
25
+ INFO:root:Average loss=0.2230509876822814
26
+ INFO:root:
27
+ [Valid][4][UTT][ MSE = 0.2595 | LCC = 0.8320 | SRCC = 0.8328 ]
28
+ INFO:root:
29
+ [Valid][4][SYS][ MSE = 0.1384 | LCC = 0.9110 | SRCC = 0.9070 ]
30
+ INFO:root:Average loss=0.21045147875944772
31
+ INFO:root:
32
+ [Valid][5][UTT][ MSE = 0.2662 | LCC = 0.8380 | SRCC = 0.8368 ]
33
+ INFO:root:
34
+ [Valid][5][SYS][ MSE = 0.1260 | LCC = 0.9239 | SRCC = 0.9178 ]
35
+ INFO:root:Average loss=0.19907851985249764
36
+ INFO:root:
37
+ [Valid][6][UTT][ MSE = 0.2598 | LCC = 0.8359 | SRCC = 0.8356 ]
38
+ INFO:root:
39
+ [Valid][6][SYS][ MSE = 0.1449 | LCC = 0.9130 | SRCC = 0.9083 ]
40
+ INFO:root:Average loss=0.18605596562608695
41
+ INFO:root:
42
+ [Valid][7][UTT][ MSE = 0.2725 | LCC = 0.8359 | SRCC = 0.8348 ]
43
+ INFO:root:
44
+ [Valid][7][SYS][ MSE = 0.1601 | LCC = 0.9201 | SRCC = 0.9151 ]
45
+ INFO:root:Average loss=0.17120693194178435
46
+ INFO:root:
47
+ [Valid][8][UTT][ MSE = 0.2497 | LCC = 0.8443 | SRCC = 0.8440 ]
48
+ INFO:root:
49
+ [Valid][8][SYS][ MSE = 0.1271 | LCC = 0.9212 | SRCC = 0.9187 ]
50
+ INFO:root:Average loss=0.1676345711144117
51
+ INFO:root:
52
+ [Valid][9][UTT][ MSE = 0.2577 | LCC = 0.8432 | SRCC = 0.8428 ]
53
+ INFO:root:
54
+ [Valid][9][SYS][ MSE = 0.1316 | LCC = 0.9158 | SRCC = 0.9108 ]
55
+ INFO:root:Average loss=0.1591709982126187
56
+ INFO:root:
57
+ [Valid][10][UTT][ MSE = 0.2489 | LCC = 0.8445 | SRCC = 0.8425 ]
58
+ INFO:root:
59
+ [Valid][10][SYS][ MSE = 0.1471 | LCC = 0.9161 | SRCC = 0.9117 ]
60
+ INFO:root:Average loss=0.1513063931503357
61
+ INFO:root:
62
+ [Valid][11][UTT][ MSE = 0.2455 | LCC = 0.8420 | SRCC = 0.8415 ]
63
+ INFO:root:
64
+ [Valid][11][SYS][ MSE = 0.1280 | LCC = 0.9207 | SRCC = 0.9172 ]
65
+ INFO:root:Average loss=0.1426447409276779
66
+ INFO:root:
67
+ [Valid][12][UTT][ MSE = 0.2440 | LCC = 0.8489 | SRCC = 0.8470 ]
68
+ INFO:root:
69
+ [Valid][12][SYS][ MSE = 0.1391 | LCC = 0.9214 | SRCC = 0.9171 ]
70
+ INFO:root:Average loss=0.13574676148784467
71
+ INFO:root:
72
+ [Valid][13][UTT][ MSE = 0.2714 | LCC = 0.8429 | SRCC = 0.8428 ]
73
+ INFO:root:
74
+ [Valid][13][SYS][ MSE = 0.1724 | LCC = 0.9209 | SRCC = 0.9193 ]
75
+ INFO:root:Average loss=0.12894271820401534
76
+ INFO:root:
77
+ [Valid][14][UTT][ MSE = 0.2481 | LCC = 0.8496 | SRCC = 0.8490 ]
78
+ INFO:root:
79
+ [Valid][14][SYS][ MSE = 0.1459 | LCC = 0.9257 | SRCC = 0.9230 ]
80
+ INFO:root:Average loss=0.12140956731178822
81
+ INFO:root:
82
+ [Valid][15][UTT][ MSE = 0.2455 | LCC = 0.8519 | SRCC = 0.8506 ]
83
+ INFO:root:
84
+ [Valid][15][SYS][ MSE = 0.1155 | LCC = 0.9284 | SRCC = 0.9241 ]
85
+ INFO:root:Average loss=0.1175903591972131
86
+ INFO:root:
87
+ [Valid][16][UTT][ MSE = 0.2432 | LCC = 0.8484 | SRCC = 0.8483 ]
88
+ INFO:root:
89
+ [Valid][16][SYS][ MSE = 0.1351 | LCC = 0.9256 | SRCC = 0.9244 ]
90
+ INFO:root:Average loss=0.12394394878393565
91
+ INFO:root:
92
+ [Valid][17][UTT][ MSE = 0.2360 | LCC = 0.8479 | SRCC = 0.8470 ]
93
+ INFO:root:
94
+ [Valid][17][SYS][ MSE = 0.1200 | LCC = 0.9205 | SRCC = 0.9185 ]
95
+ INFO:root:Average loss=0.10359991466005643
96
+ INFO:root:
97
+ [Valid][18][UTT][ MSE = 0.2292 | LCC = 0.8530 | SRCC = 0.8525 ]
98
+ INFO:root:
99
+ [Valid][18][SYS][ MSE = 0.1115 | LCC = 0.9263 | SRCC = 0.9236 ]
100
+ INFO:root:Average loss=0.10143001272510259
101
+ INFO:root:
102
+ [Valid][19][UTT][ MSE = 0.2323 | LCC = 0.8522 | SRCC = 0.8506 ]
103
+ INFO:root:
104
+ [Valid][19][SYS][ MSE = 0.1221 | LCC = 0.9221 | SRCC = 0.9172 ]
105
+ INFO:root:Average loss=0.09811931160780099
106
+ INFO:root:
107
+ [Valid][20][UTT][ MSE = 0.2316 | LCC = 0.8523 | SRCC = 0.8503 ]
108
+ INFO:root:
109
+ [Valid][20][SYS][ MSE = 0.1170 | LCC = 0.9257 | SRCC = 0.9210 ]
110
+ INFO:root:Average loss=0.10091435054364876
111
+ INFO:root:
112
+ [Valid][21][UTT][ MSE = 0.2368 | LCC = 0.8521 | SRCC = 0.8509 ]
113
+ INFO:root:
114
+ [Valid][21][SYS][ MSE = 0.1222 | LCC = 0.9238 | SRCC = 0.9222 ]
115
+ INFO:root:Average loss=0.09523546003187314
116
+ INFO:root:
117
+ [Valid][22][UTT][ MSE = 0.2291 | LCC = 0.8537 | SRCC = 0.8521 ]
118
+ INFO:root:
119
+ [Valid][22][SYS][ MSE = 0.1136 | LCC = 0.9258 | SRCC = 0.9201 ]
120
+ INFO:root:Average loss=0.09089563381022368
121
+ INFO:root:
122
+ [Valid][23][UTT][ MSE = 0.2247 | LCC = 0.8544 | SRCC = 0.8542 ]
123
+ INFO:root:
124
+ [Valid][23][SYS][ MSE = 0.1146 | LCC = 0.9247 | SRCC = 0.9249 ]
125
+ INFO:root:Average loss=0.08332412919172874
126
+ INFO:root:
127
+ [Valid][24][UTT][ MSE = 0.2285 | LCC = 0.8525 | SRCC = 0.8506 ]
128
+ INFO:root:
129
+ [Valid][24][SYS][ MSE = 0.1080 | LCC = 0.9294 | SRCC = 0.9245 ]
130
+ INFO:root:Average loss=0.08792871752610573
131
+ INFO:root:
132
+ [Valid][25][UTT][ MSE = 0.2277 | LCC = 0.8567 | SRCC = 0.8546 ]
133
+ INFO:root:
134
+ [Valid][25][SYS][ MSE = 0.1191 | LCC = 0.9283 | SRCC = 0.9248 ]
135
+ INFO:root:Average loss=0.08099713615882091
136
+ INFO:root:
137
+ [Valid][26][UTT][ MSE = 0.2318 | LCC = 0.8537 | SRCC = 0.8520 ]
138
+ INFO:root:
139
+ [Valid][26][SYS][ MSE = 0.1259 | LCC = 0.9279 | SRCC = 0.9247 ]
140
+ INFO:root:Average loss=0.08232074331205624
141
+ INFO:root:
142
+ [Valid][27][UTT][ MSE = 0.2305 | LCC = 0.8555 | SRCC = 0.8539 ]
143
+ INFO:root:
144
+ [Valid][27][SYS][ MSE = 0.1268 | LCC = 0.9290 | SRCC = 0.9274 ]
145
+ INFO:root:Average loss=0.07562345275894189
146
+ INFO:root:
147
+ [Valid][28][UTT][ MSE = 0.2202 | LCC = 0.8610 | SRCC = 0.8587 ]
148
+ INFO:root:
149
+ [Valid][28][SYS][ MSE = 0.1060 | LCC = 0.9394 | SRCC = 0.9350 ]
150
+ INFO:root:Average loss=0.07162571507386672
151
+ INFO:root:
152
+ [Valid][29][UTT][ MSE = 0.2280 | LCC = 0.8543 | SRCC = 0.8536 ]
153
+ INFO:root:
154
+ [Valid][29][SYS][ MSE = 0.1162 | LCC = 0.9325 | SRCC = 0.9302 ]
155
+ INFO:root:Average loss=0.0684171619419104
156
+ INFO:root:
157
+ [Valid][30][UTT][ MSE = 0.2380 | LCC = 0.8528 | SRCC = 0.8505 ]
158
+ INFO:root:
159
+ [Valid][30][SYS][ MSE = 0.1342 | LCC = 0.9304 | SRCC = 0.9282 ]
160
+ INFO:root:
161
+ [Test][31][UTT][ MSE = 0.2461 | LCC = 0.8471 | SRCC = 0.8464 ]
162
+ INFO:root:
163
+ [Test][31][SYS][ MSE = 0.1242 | LCC = 0.9184 | SRCC = 0.9143 ]
results_bvcc/w2v2_base/layer1/run0/config.gin ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Training config on BVCC dataset.
2
+
3
+ # ------------------------------
4
+ # Architecture
5
+ # ------------------------------
6
+ ProjectionHead:
7
+ conv_channels = (32, 32)
8
+ dense_neurons = (64, 32, 1)
9
+ use_poolings = (True, True)
10
+ kernel_size = 5
11
+ pool_size = 5
12
+
13
+ # ------------------------------
14
+ # Dataset
15
+ # ------------------------------
16
+ Bvcc:
17
+ debug = False
18
+
19
+ get_dataloader:
20
+ num_workers = 12
21
+ shuffle = True
22
+
23
+ # ------------------------------
24
+ # Training loop
25
+ # ------------------------------
26
+ TrainingLoop:
27
+ num_epochs = 30
28
+ learning_rate = 1e-4
29
+ dataset_cls = @Bvcc
30
+ weight_decay = 0.0
31
+ batch_size_train = 64
results_bvcc/w2v2_base/layer1/run0/model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d74b445d94921e01df770e0fca2317b4575a18bcffa59b43f6efcaa870061321
3
+ size 2747979
results_bvcc/w2v2_base/layer1/run0/model_best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:82325465fff9a7757bdaa487f00ec8bfa4d39ab17fbebef504c32e9a15e747ae
3
+ size 2749702
results_bvcc/w2v2_base/layer1/run0/train.log ADDED
@@ -0,0 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ INFO:root:Num train speech clips: 4974
2
+ INFO:root:Num val speech clips: 1066
3
+ INFO:root:Num test speech clips: 1066
4
+ INFO:root:Device=cuda
5
+ INFO:root:Average loss=0.5303507426228279
6
+ INFO:root:
7
+ [Valid][0][UTT][ MSE = 0.3587 | LCC = 0.7716 | SRCC = 0.7740 ]
8
+ INFO:root:
9
+ [Valid][0][SYS][ MSE = 0.2443 | LCC = 0.8536 | SRCC = 0.8533 ]
10
+ INFO:root:Average loss=0.273792809400803
11
+ INFO:root:
12
+ [Valid][1][UTT][ MSE = 0.2932 | LCC = 0.8067 | SRCC = 0.8129 ]
13
+ INFO:root:
14
+ [Valid][1][SYS][ MSE = 0.1614 | LCC = 0.8929 | SRCC = 0.9015 ]
15
+ INFO:root:Average loss=0.2344587573256248
16
+ INFO:root:
17
+ [Valid][2][UTT][ MSE = 0.2788 | LCC = 0.8335 | SRCC = 0.8347 ]
18
+ INFO:root:
19
+ [Valid][2][SYS][ MSE = 0.1712 | LCC = 0.9109 | SRCC = 0.9159 ]
20
+ INFO:root:Average loss=0.21123353793070868
21
+ INFO:root:
22
+ [Valid][3][UTT][ MSE = 0.2564 | LCC = 0.8466 | SRCC = 0.8460 ]
23
+ INFO:root:
24
+ [Valid][3][SYS][ MSE = 0.1470 | LCC = 0.9214 | SRCC = 0.9252 ]
25
+ INFO:root:Average loss=0.19582427608279082
26
+ INFO:root:
27
+ [Valid][4][UTT][ MSE = 0.2817 | LCC = 0.8441 | SRCC = 0.8424 ]
28
+ INFO:root:
29
+ [Valid][4][SYS][ MSE = 0.1735 | LCC = 0.9215 | SRCC = 0.9235 ]
30
+ INFO:root:Average loss=0.18156705643886176
31
+ INFO:root:
32
+ [Valid][5][UTT][ MSE = 0.2363 | LCC = 0.8535 | SRCC = 0.8534 ]
33
+ INFO:root:
34
+ [Valid][5][SYS][ MSE = 0.1150 | LCC = 0.9334 | SRCC = 0.9386 ]
35
+ INFO:root:Average loss=0.17155700033673874
36
+ INFO:root:
37
+ [Valid][6][UTT][ MSE = 0.2387 | LCC = 0.8533 | SRCC = 0.8524 ]
38
+ INFO:root:
39
+ [Valid][6][SYS][ MSE = 0.1198 | LCC = 0.9340 | SRCC = 0.9375 ]
40
+ INFO:root:Average loss=0.1611969740344928
41
+ INFO:root:
42
+ [Valid][7][UTT][ MSE = 0.2486 | LCC = 0.8581 | SRCC = 0.8560 ]
43
+ INFO:root:
44
+ [Valid][7][SYS][ MSE = 0.1405 | LCC = 0.9351 | SRCC = 0.9363 ]
45
+ INFO:root:Average loss=0.14487831963178438
46
+ INFO:root:
47
+ [Valid][8][UTT][ MSE = 0.2628 | LCC = 0.8577 | SRCC = 0.8555 ]
48
+ INFO:root:
49
+ [Valid][8][SYS][ MSE = 0.1570 | LCC = 0.9384 | SRCC = 0.9398 ]
50
+ INFO:root:Average loss=0.14753570140172273
51
+ INFO:root:
52
+ [Valid][9][UTT][ MSE = 0.2528 | LCC = 0.8597 | SRCC = 0.8595 ]
53
+ INFO:root:
54
+ [Valid][9][SYS][ MSE = 0.1479 | LCC = 0.9385 | SRCC = 0.9427 ]
55
+ INFO:root:Average loss=0.13028067197555152
56
+ INFO:root:
57
+ [Valid][10][UTT][ MSE = 0.2273 | LCC = 0.8649 | SRCC = 0.8641 ]
58
+ INFO:root:
59
+ [Valid][10][SYS][ MSE = 0.1176 | LCC = 0.9425 | SRCC = 0.9470 ]
60
+ INFO:root:Average loss=0.11848474179322903
61
+ INFO:root:
62
+ [Valid][11][UTT][ MSE = 0.2503 | LCC = 0.8592 | SRCC = 0.8575 ]
63
+ INFO:root:
64
+ [Valid][11][SYS][ MSE = 0.1492 | LCC = 0.9336 | SRCC = 0.9342 ]
65
+ INFO:root:Average loss=0.11848746966093014
66
+ INFO:root:
67
+ [Valid][12][UTT][ MSE = 0.2657 | LCC = 0.8569 | SRCC = 0.8577 ]
68
+ INFO:root:
69
+ [Valid][12][SYS][ MSE = 0.1625 | LCC = 0.9322 | SRCC = 0.9380 ]
70
+ INFO:root:Average loss=0.10621509194756165
71
+ INFO:root:
72
+ [Valid][13][UTT][ MSE = 0.2611 | LCC = 0.8605 | SRCC = 0.8571 ]
73
+ INFO:root:
74
+ [Valid][13][SYS][ MSE = 0.1713 | LCC = 0.9290 | SRCC = 0.9259 ]
75
+ INFO:root:Average loss=0.0977262431421341
76
+ INFO:root:
77
+ [Valid][14][UTT][ MSE = 0.2265 | LCC = 0.8617 | SRCC = 0.8613 ]
78
+ INFO:root:
79
+ [Valid][14][SYS][ MSE = 0.1169 | LCC = 0.9364 | SRCC = 0.9382 ]
80
+ INFO:root:Average loss=0.09514909762984668
81
+ INFO:root:
82
+ [Valid][15][UTT][ MSE = 0.2794 | LCC = 0.8582 | SRCC = 0.8558 ]
83
+ INFO:root:
84
+ [Valid][15][SYS][ MSE = 0.1889 | LCC = 0.9303 | SRCC = 0.9292 ]
85
+ INFO:root:Average loss=0.09148985472245094
86
+ INFO:root:
87
+ [Valid][16][UTT][ MSE = 0.2399 | LCC = 0.8646 | SRCC = 0.8634 ]
88
+ INFO:root:
89
+ [Valid][16][SYS][ MSE = 0.1305 | LCC = 0.9424 | SRCC = 0.9436 ]
90
+ INFO:root:Average loss=0.09739843593576016
91
+ INFO:root:
92
+ [Valid][17][UTT][ MSE = 0.2148 | LCC = 0.8666 | SRCC = 0.8647 ]
93
+ INFO:root:
94
+ [Valid][17][SYS][ MSE = 0.1077 | LCC = 0.9397 | SRCC = 0.9388 ]
95
+ INFO:root:Average loss=0.0810291952907275
96
+ INFO:root:
97
+ [Valid][18][UTT][ MSE = 0.2443 | LCC = 0.8612 | SRCC = 0.8598 ]
98
+ INFO:root:
99
+ [Valid][18][SYS][ MSE = 0.1418 | LCC = 0.9395 | SRCC = 0.9395 ]
100
+ INFO:root:Average loss=0.0809667952454243
101
+ INFO:root:
102
+ [Valid][19][UTT][ MSE = 0.2668 | LCC = 0.8560 | SRCC = 0.8556 ]
103
+ INFO:root:
104
+ [Valid][19][SYS][ MSE = 0.1587 | LCC = 0.9368 | SRCC = 0.9400 ]
105
+ INFO:root:Average loss=0.07862621292662926
106
+ INFO:root:
107
+ [Valid][20][UTT][ MSE = 0.2336 | LCC = 0.8662 | SRCC = 0.8630 ]
108
+ INFO:root:
109
+ [Valid][20][SYS][ MSE = 0.1369 | LCC = 0.9354 | SRCC = 0.9315 ]
110
+ INFO:root:Average loss=0.07469513434439133
111
+ INFO:root:
112
+ [Valid][21][UTT][ MSE = 0.2299 | LCC = 0.8662 | SRCC = 0.8643 ]
113
+ INFO:root:
114
+ [Valid][21][SYS][ MSE = 0.1314 | LCC = 0.9385 | SRCC = 0.9377 ]
115
+ INFO:root:Average loss=0.07152971806816566
116
+ INFO:root:
117
+ [Valid][22][UTT][ MSE = 0.2245 | LCC = 0.8649 | SRCC = 0.8630 ]
118
+ INFO:root:
119
+ [Valid][22][SYS][ MSE = 0.1187 | LCC = 0.9403 | SRCC = 0.9398 ]
120
+ INFO:root:Average loss=0.06999625805287789
121
+ INFO:root:
122
+ [Valid][23][UTT][ MSE = 0.2443 | LCC = 0.8652 | SRCC = 0.8650 ]
123
+ INFO:root:
124
+ [Valid][23][SYS][ MSE = 0.1456 | LCC = 0.9432 | SRCC = 0.9463 ]
125
+ INFO:root:Average loss=0.07724339954363994
126
+ INFO:root:
127
+ [Valid][24][UTT][ MSE = 0.2206 | LCC = 0.8648 | SRCC = 0.8642 ]
128
+ INFO:root:
129
+ [Valid][24][SYS][ MSE = 0.1147 | LCC = 0.9401 | SRCC = 0.9422 ]
130
+ INFO:root:Average loss=0.0698593703026955
131
+ INFO:root:
132
+ [Valid][25][UTT][ MSE = 0.2379 | LCC = 0.8625 | SRCC = 0.8616 ]
133
+ INFO:root:
134
+ [Valid][25][SYS][ MSE = 0.1334 | LCC = 0.9405 | SRCC = 0.9405 ]
135
+ INFO:root:Average loss=0.06205379327711386
136
+ INFO:root:
137
+ [Valid][26][UTT][ MSE = 0.2316 | LCC = 0.8685 | SRCC = 0.8676 ]
138
+ INFO:root:
139
+ [Valid][26][SYS][ MSE = 0.1330 | LCC = 0.9434 | SRCC = 0.9436 ]
140
+ INFO:root:Average loss=0.061733570570747055
141
+ INFO:root:
142
+ [Valid][27][UTT][ MSE = 0.2087 | LCC = 0.8708 | SRCC = 0.8679 ]
143
+ INFO:root:
144
+ [Valid][27][SYS][ MSE = 0.0991 | LCC = 0.9410 | SRCC = 0.9373 ]
145
+ INFO:root:Average loss=0.061150606554479174
146
+ INFO:root:
147
+ [Valid][28][UTT][ MSE = 0.2310 | LCC = 0.8676 | SRCC = 0.8664 ]
148
+ INFO:root:
149
+ [Valid][28][SYS][ MSE = 0.1337 | LCC = 0.9417 | SRCC = 0.9434 ]
150
+ INFO:root:Average loss=0.057036525975817286
151
+ INFO:root:
152
+ [Valid][29][UTT][ MSE = 0.2331 | LCC = 0.8676 | SRCC = 0.8664 ]
153
+ INFO:root:
154
+ [Valid][29][SYS][ MSE = 0.1354 | LCC = 0.9419 | SRCC = 0.9426 ]
155
+ INFO:root:Average loss=0.05510546892690353
156
+ INFO:root:
157
+ [Valid][30][UTT][ MSE = 0.2237 | LCC = 0.8658 | SRCC = 0.8650 ]
158
+ INFO:root:
159
+ [Valid][30][SYS][ MSE = 0.1233 | LCC = 0.9387 | SRCC = 0.9414 ]
160
+ INFO:root:
161
+ [Test][31][UTT][ MSE = 0.2377 | LCC = 0.8587 | SRCC = 0.8551 ]
162
+ INFO:root:
163
+ [Test][31][SYS][ MSE = 0.1383 | LCC = 0.9224 | SRCC = 0.9203 ]
results_bvcc/w2v2_base/layer1/run1/config.gin ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Training config on BVCC dataset.
2
+
3
+ # ------------------------------
4
+ # Architecture
5
+ # ------------------------------
6
+ ProjectionHead:
7
+ conv_channels = (32, 32)
8
+ dense_neurons = (64, 32, 1)
9
+ use_poolings = (True, True)
10
+ kernel_size = 5
11
+ pool_size = 5
12
+
13
+ # ------------------------------
14
+ # Dataset
15
+ # ------------------------------
16
+ Bvcc:
17
+ debug = False
18
+
19
+ get_dataloader:
20
+ num_workers = 12
21
+ shuffle = True
22
+
23
+ # ------------------------------
24
+ # Training loop
25
+ # ------------------------------
26
+ TrainingLoop:
27
+ num_epochs = 30
28
+ learning_rate = 1e-4
29
+ dataset_cls = @Bvcc
30
+ weight_decay = 0.0
31
+ batch_size_train = 64
results_bvcc/w2v2_base/layer1/run1/model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9557e734a10a18329b6b88241bd5b295f63f75b0530ae47b0ebf1e55a289c366
3
+ size 2747979
results_bvcc/w2v2_base/layer1/run1/model_best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6707611934c02690335c62cde1fef5d4c5205e932bb7e973f9f62aac74c83690
3
+ size 2749702
results_bvcc/w2v2_base/layer1/run1/train.log ADDED
@@ -0,0 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ INFO:root:Num train speech clips: 4974
2
+ INFO:root:Num val speech clips: 1066
3
+ INFO:root:Num test speech clips: 1066
4
+ INFO:root:Device=cuda
5
+ INFO:root:Average loss=0.49480017905051893
6
+ INFO:root:
7
+ [Valid][0][UTT][ MSE = 0.3707 | LCC = 0.7667 | SRCC = 0.7704 ]
8
+ INFO:root:
9
+ [Valid][0][SYS][ MSE = 0.2482 | LCC = 0.8397 | SRCC = 0.8452 ]
10
+ INFO:root:Average loss=0.27845622771061385
11
+ INFO:root:
12
+ [Valid][1][UTT][ MSE = 0.2963 | LCC = 0.8141 | SRCC = 0.8191 ]
13
+ INFO:root:
14
+ [Valid][1][SYS][ MSE = 0.1968 | LCC = 0.8806 | SRCC = 0.8873 ]
15
+ INFO:root:Average loss=0.2325418186493409
16
+ INFO:root:
17
+ [Valid][2][UTT][ MSE = 0.2657 | LCC = 0.8388 | SRCC = 0.8393 ]
18
+ INFO:root:
19
+ [Valid][2][SYS][ MSE = 0.1636 | LCC = 0.9002 | SRCC = 0.9034 ]
20
+ INFO:root:Average loss=0.21036842312568274
21
+ INFO:root:
22
+ [Valid][3][UTT][ MSE = 0.2611 | LCC = 0.8452 | SRCC = 0.8457 ]
23
+ INFO:root:
24
+ [Valid][3][SYS][ MSE = 0.1507 | LCC = 0.9157 | SRCC = 0.9200 ]
25
+ INFO:root:Average loss=0.19588429203782326
26
+ INFO:root:
27
+ [Valid][4][UTT][ MSE = 0.2576 | LCC = 0.8492 | SRCC = 0.8464 ]
28
+ INFO:root:
29
+ [Valid][4][SYS][ MSE = 0.1489 | LCC = 0.9182 | SRCC = 0.9138 ]
30
+ INFO:root:Average loss=0.18050869554281235
31
+ INFO:root:
32
+ [Valid][5][UTT][ MSE = 0.2392 | LCC = 0.8511 | SRCC = 0.8496 ]
33
+ INFO:root:
34
+ [Valid][5][SYS][ MSE = 0.1167 | LCC = 0.9244 | SRCC = 0.9236 ]
35
+ INFO:root:Average loss=0.1697243203719457
36
+ INFO:root:
37
+ [Valid][6][UTT][ MSE = 0.2389 | LCC = 0.8544 | SRCC = 0.8517 ]
38
+ INFO:root:
39
+ [Valid][6][SYS][ MSE = 0.1242 | LCC = 0.9268 | SRCC = 0.9240 ]
40
+ INFO:root:Average loss=0.15253915876532212
41
+ INFO:root:
42
+ [Valid][7][UTT][ MSE = 0.2330 | LCC = 0.8590 | SRCC = 0.8571 ]
43
+ INFO:root:
44
+ [Valid][7][SYS][ MSE = 0.1104 | LCC = 0.9312 | SRCC = 0.9303 ]
45
+ INFO:root:Average loss=0.14838704428611657
46
+ INFO:root:
47
+ [Valid][8][UTT][ MSE = 0.2413 | LCC = 0.8550 | SRCC = 0.8575 ]
48
+ INFO:root:
49
+ [Valid][8][SYS][ MSE = 0.1363 | LCC = 0.9283 | SRCC = 0.9348 ]
50
+ INFO:root:Average loss=0.1364711202107943
51
+ INFO:root:
52
+ [Valid][9][UTT][ MSE = 0.2387 | LCC = 0.8608 | SRCC = 0.8591 ]
53
+ INFO:root:
54
+ [Valid][9][SYS][ MSE = 0.1134 | LCC = 0.9396 | SRCC = 0.9387 ]
55
+ INFO:root:Average loss=0.12884573237254068
56
+ INFO:root:
57
+ [Valid][10][UTT][ MSE = 0.2213 | LCC = 0.8658 | SRCC = 0.8639 ]
58
+ INFO:root:
59
+ [Valid][10][SYS][ MSE = 0.1090 | LCC = 0.9361 | SRCC = 0.9352 ]
60
+ INFO:root:Average loss=0.13734690319651213
61
+ INFO:root:
62
+ [Valid][11][UTT][ MSE = 0.2510 | LCC = 0.8653 | SRCC = 0.8640 ]
63
+ INFO:root:
64
+ [Valid][11][SYS][ MSE = 0.1196 | LCC = 0.9394 | SRCC = 0.9383 ]
65
+ INFO:root:Average loss=0.11653993694254985
66
+ INFO:root:
67
+ [Valid][12][UTT][ MSE = 0.2371 | LCC = 0.8585 | SRCC = 0.8580 ]
68
+ INFO:root:
69
+ [Valid][12][SYS][ MSE = 0.1264 | LCC = 0.9351 | SRCC = 0.9386 ]
70
+ INFO:root:Average loss=0.10805719918929614
71
+ INFO:root:
72
+ [Valid][13][UTT][ MSE = 0.2204 | LCC = 0.8631 | SRCC = 0.8619 ]
73
+ INFO:root:
74
+ [Valid][13][SYS][ MSE = 0.0972 | LCC = 0.9399 | SRCC = 0.9394 ]
75
+ INFO:root:Average loss=0.10435576784687164
76
+ INFO:root:
77
+ [Valid][14][UTT][ MSE = 0.2360 | LCC = 0.8607 | SRCC = 0.8607 ]
78
+ INFO:root:
79
+ [Valid][14][SYS][ MSE = 0.1297 | LCC = 0.9367 | SRCC = 0.9399 ]
80
+ INFO:root:Average loss=0.09637260752228591
81
+ INFO:root:
82
+ [Valid][15][UTT][ MSE = 0.2703 | LCC = 0.8568 | SRCC = 0.8589 ]
83
+ INFO:root:
84
+ [Valid][15][SYS][ MSE = 0.1738 | LCC = 0.9327 | SRCC = 0.9391 ]
85
+ INFO:root:Average loss=0.10165326985029075
86
+ INFO:root:
87
+ [Valid][16][UTT][ MSE = 0.2279 | LCC = 0.8598 | SRCC = 0.8575 ]
88
+ INFO:root:
89
+ [Valid][16][SYS][ MSE = 0.1126 | LCC = 0.9366 | SRCC = 0.9355 ]
90
+ INFO:root:Average loss=0.08950381109920832
91
+ INFO:root:
92
+ [Valid][17][UTT][ MSE = 0.2257 | LCC = 0.8576 | SRCC = 0.8556 ]
93
+ INFO:root:
94
+ [Valid][17][SYS][ MSE = 0.1056 | LCC = 0.9355 | SRCC = 0.9357 ]
95
+ INFO:root:Average loss=0.08791988768065587
96
+ INFO:root:
97
+ [Valid][18][UTT][ MSE = 0.2214 | LCC = 0.8607 | SRCC = 0.8588 ]
98
+ INFO:root:
99
+ [Valid][18][SYS][ MSE = 0.1032 | LCC = 0.9386 | SRCC = 0.9374 ]
100
+ INFO:root:Average loss=0.0842550451365801
101
+ INFO:root:
102
+ [Valid][19][UTT][ MSE = 0.2271 | LCC = 0.8623 | SRCC = 0.8610 ]
103
+ INFO:root:
104
+ [Valid][19][SYS][ MSE = 0.1003 | LCC = 0.9398 | SRCC = 0.9383 ]
105
+ INFO:root:Average loss=0.07734839694622235
106
+ INFO:root:
107
+ [Valid][20][UTT][ MSE = 0.2193 | LCC = 0.8599 | SRCC = 0.8585 ]
108
+ INFO:root:
109
+ [Valid][20][SYS][ MSE = 0.0937 | LCC = 0.9395 | SRCC = 0.9408 ]
110
+ INFO:root:Average loss=0.07589360111607955
111
+ INFO:root:
112
+ [Valid][21][UTT][ MSE = 0.2313 | LCC = 0.8581 | SRCC = 0.8577 ]
113
+ INFO:root:
114
+ [Valid][21][SYS][ MSE = 0.1258 | LCC = 0.9336 | SRCC = 0.9396 ]
115
+ INFO:root:Average loss=0.07119909267968093
116
+ INFO:root:
117
+ [Valid][22][UTT][ MSE = 0.2225 | LCC = 0.8595 | SRCC = 0.8595 ]
118
+ INFO:root:
119
+ [Valid][22][SYS][ MSE = 0.1092 | LCC = 0.9375 | SRCC = 0.9424 ]
120
+ INFO:root:Average loss=0.06933691834983154
121
+ INFO:root:
122
+ [Valid][23][UTT][ MSE = 0.2205 | LCC = 0.8625 | SRCC = 0.8595 ]
123
+ INFO:root:
124
+ [Valid][23][SYS][ MSE = 0.1082 | LCC = 0.9367 | SRCC = 0.9368 ]
125
+ INFO:root:Average loss=0.06987673411957729
126
+ INFO:root:
127
+ [Valid][24][UTT][ MSE = 0.2165 | LCC = 0.8619 | SRCC = 0.8612 ]
128
+ INFO:root:
129
+ [Valid][24][SYS][ MSE = 0.1056 | LCC = 0.9356 | SRCC = 0.9410 ]
130
+ INFO:root:Average loss=0.0654885101203735
131
+ INFO:root:
132
+ [Valid][25][UTT][ MSE = 0.2249 | LCC = 0.8634 | SRCC = 0.8615 ]
133
+ INFO:root:
134
+ [Valid][25][SYS][ MSE = 0.1198 | LCC = 0.9369 | SRCC = 0.9392 ]
135
+ INFO:root:Average loss=0.06392660608085302
136
+ INFO:root:
137
+ [Valid][26][UTT][ MSE = 0.2273 | LCC = 0.8618 | SRCC = 0.8602 ]
138
+ INFO:root:
139
+ [Valid][26][SYS][ MSE = 0.1213 | LCC = 0.9369 | SRCC = 0.9393 ]
140
+ INFO:root:Average loss=0.060871114858832114
141
+ INFO:root:
142
+ [Valid][27][UTT][ MSE = 0.2281 | LCC = 0.8603 | SRCC = 0.8594 ]
143
+ INFO:root:
144
+ [Valid][27][SYS][ MSE = 0.1161 | LCC = 0.9407 | SRCC = 0.9429 ]
145
+ INFO:root:Average loss=0.06481427379334584
146
+ INFO:root:
147
+ [Valid][28][UTT][ MSE = 0.2182 | LCC = 0.8616 | SRCC = 0.8591 ]
148
+ INFO:root:
149
+ [Valid][28][SYS][ MSE = 0.0974 | LCC = 0.9408 | SRCC = 0.9411 ]
150
+ INFO:root:Average loss=0.05660455673933029
151
+ INFO:root:
152
+ [Valid][29][UTT][ MSE = 0.2210 | LCC = 0.8649 | SRCC = 0.8615 ]
153
+ INFO:root:
154
+ [Valid][29][SYS][ MSE = 0.1056 | LCC = 0.9422 | SRCC = 0.9392 ]
155
+ INFO:root:Average loss=0.05522246692234125
156
+ INFO:root:
157
+ [Valid][30][UTT][ MSE = 0.2146 | LCC = 0.8651 | SRCC = 0.8636 ]
158
+ INFO:root:
159
+ [Valid][30][SYS][ MSE = 0.0934 | LCC = 0.9415 | SRCC = 0.9418 ]
160
+ INFO:root:
161
+ [Test][31][UTT][ MSE = 0.2336 | LCC = 0.8647 | SRCC = 0.8619 ]
162
+ INFO:root:
163
+ [Test][31][SYS][ MSE = 0.1565 | LCC = 0.9132 | SRCC = 0.9094 ]
results_bvcc/w2v2_base/layer1/run2/config.gin ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Training config on BVCC dataset.
2
+
3
+ # ------------------------------
4
+ # Architecture
5
+ # ------------------------------
6
+ ProjectionHead:
7
+ conv_channels = (32, 32)
8
+ dense_neurons = (64, 32, 1)
9
+ use_poolings = (True, True)
10
+ kernel_size = 5
11
+ pool_size = 5
12
+
13
+ # ------------------------------
14
+ # Dataset
15
+ # ------------------------------
16
+ Bvcc:
17
+ debug = False
18
+
19
+ get_dataloader:
20
+ num_workers = 12
21
+ shuffle = True
22
+
23
+ # ------------------------------
24
+ # Training loop
25
+ # ------------------------------
26
+ TrainingLoop:
27
+ num_epochs = 30
28
+ learning_rate = 1e-4
29
+ dataset_cls = @Bvcc
30
+ weight_decay = 0.0
31
+ batch_size_train = 64
results_bvcc/w2v2_base/layer1/run2/model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:05ebbc29c7d329bc1ee97f50f3096a73b785e22268be34c95185bffd8b68c7dd
3
+ size 2747979
results_bvcc/w2v2_base/layer1/run2/model_best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f573a61f31325e8566236456c2318a88fb7ab56c2c619bd6185de7c66325c33a
3
+ size 2749702
results_bvcc/w2v2_base/layer1/run2/train.log ADDED
@@ -0,0 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ INFO:root:Num train speech clips: 4974
2
+ INFO:root:Num val speech clips: 1066
3
+ INFO:root:Num test speech clips: 1066
4
+ INFO:root:Device=cuda
5
+ INFO:root:Average loss=0.5799432725478442
6
+ INFO:root:
7
+ [Valid][0][UTT][ MSE = 0.3725 | LCC = 0.7729 | SRCC = 0.7788 ]
8
+ INFO:root:
9
+ [Valid][0][SYS][ MSE = 0.2333 | LCC = 0.8492 | SRCC = 0.8574 ]
10
+ INFO:root:Average loss=0.2807462093157646
11
+ INFO:root:
12
+ [Valid][1][UTT][ MSE = 0.2828 | LCC = 0.8240 | SRCC = 0.8240 ]
13
+ INFO:root:
14
+ [Valid][1][SYS][ MSE = 0.1628 | LCC = 0.9012 | SRCC = 0.9030 ]
15
+ INFO:root:Average loss=0.23757183398955908
16
+ INFO:root:
17
+ [Valid][2][UTT][ MSE = 0.2619 | LCC = 0.8395 | SRCC = 0.8398 ]
18
+ INFO:root:
19
+ [Valid][2][SYS][ MSE = 0.1473 | LCC = 0.9143 | SRCC = 0.9177 ]
20
+ INFO:root:Average loss=0.21484326590330172
21
+ INFO:root:
22
+ [Valid][3][UTT][ MSE = 0.2650 | LCC = 0.8500 | SRCC = 0.8505 ]
23
+ INFO:root:
24
+ [Valid][3][SYS][ MSE = 0.1563 | LCC = 0.9237 | SRCC = 0.9270 ]
25
+ INFO:root:Average loss=0.20002879679967195
26
+ INFO:root:
27
+ [Valid][4][UTT][ MSE = 0.2447 | LCC = 0.8565 | SRCC = 0.8558 ]
28
+ INFO:root:
29
+ [Valid][4][SYS][ MSE = 0.1377 | LCC = 0.9289 | SRCC = 0.9316 ]
30
+ INFO:root:Average loss=0.18346711955009362
31
+ INFO:root:
32
+ [Valid][5][UTT][ MSE = 0.2279 | LCC = 0.8599 | SRCC = 0.8584 ]
33
+ INFO:root:
34
+ [Valid][5][SYS][ MSE = 0.1060 | LCC = 0.9367 | SRCC = 0.9382 ]
35
+ INFO:root:Average loss=0.1678030132674254
36
+ INFO:root:
37
+ [Valid][6][UTT][ MSE = 0.2292 | LCC = 0.8561 | SRCC = 0.8585 ]
38
+ INFO:root:
39
+ [Valid][6][SYS][ MSE = 0.1143 | LCC = 0.9311 | SRCC = 0.9361 ]
40
+ INFO:root:Average loss=0.15920545428227156
41
+ INFO:root:
42
+ [Valid][7][UTT][ MSE = 0.2251 | LCC = 0.8621 | SRCC = 0.8593 ]
43
+ INFO:root:
44
+ [Valid][7][SYS][ MSE = 0.1069 | LCC = 0.9379 | SRCC = 0.9357 ]
45
+ INFO:root:Average loss=0.14479620401293802
46
+ INFO:root:
47
+ [Valid][8][UTT][ MSE = 0.2212 | LCC = 0.8621 | SRCC = 0.8615 ]
48
+ INFO:root:
49
+ [Valid][8][SYS][ MSE = 0.1146 | LCC = 0.9312 | SRCC = 0.9337 ]
50
+ INFO:root:Average loss=0.13811541081239015
51
+ INFO:root:
52
+ [Valid][9][UTT][ MSE = 0.2240 | LCC = 0.8674 | SRCC = 0.8632 ]
53
+ INFO:root:
54
+ [Valid][9][SYS][ MSE = 0.1123 | LCC = 0.9409 | SRCC = 0.9389 ]
55
+ INFO:root:Average loss=0.13562335083499932
56
+ INFO:root:
57
+ [Valid][10][UTT][ MSE = 0.2191 | LCC = 0.8687 | SRCC = 0.8656 ]
58
+ INFO:root:
59
+ [Valid][10][SYS][ MSE = 0.1104 | LCC = 0.9393 | SRCC = 0.9387 ]
60
+ INFO:root:Average loss=0.1258917586543621
61
+ INFO:root:
62
+ [Valid][11][UTT][ MSE = 0.2320 | LCC = 0.8611 | SRCC = 0.8588 ]
63
+ INFO:root:
64
+ [Valid][11][SYS][ MSE = 0.1264 | LCC = 0.9345 | SRCC = 0.9313 ]
65
+ INFO:root:Average loss=0.12100212400158246
66
+ INFO:root:
67
+ [Valid][12][UTT][ MSE = 0.2333 | LCC = 0.8706 | SRCC = 0.8672 ]
68
+ INFO:root:
69
+ [Valid][12][SYS][ MSE = 0.1101 | LCC = 0.9452 | SRCC = 0.9416 ]
70
+ INFO:root:Average loss=0.11105435886062108
71
+ INFO:root:
72
+ [Valid][13][UTT][ MSE = 0.2303 | LCC = 0.8628 | SRCC = 0.8612 ]
73
+ INFO:root:
74
+ [Valid][13][SYS][ MSE = 0.1234 | LCC = 0.9392 | SRCC = 0.9412 ]
75
+ INFO:root:Average loss=0.10584970210225154
76
+ INFO:root:
77
+ [Valid][14][UTT][ MSE = 0.2299 | LCC = 0.8686 | SRCC = 0.8641 ]
78
+ INFO:root:
79
+ [Valid][14][SYS][ MSE = 0.1114 | LCC = 0.9407 | SRCC = 0.9340 ]
80
+ INFO:root:Average loss=0.1046583229341568
81
+ INFO:root:
82
+ [Valid][15][UTT][ MSE = 0.2171 | LCC = 0.8656 | SRCC = 0.8622 ]
83
+ INFO:root:
84
+ [Valid][15][SYS][ MSE = 0.1020 | LCC = 0.9402 | SRCC = 0.9382 ]
85
+ INFO:root:Average loss=0.09526471244410062
86
+ INFO:root:
87
+ [Valid][16][UTT][ MSE = 0.2188 | LCC = 0.8669 | SRCC = 0.8638 ]
88
+ INFO:root:
89
+ [Valid][16][SYS][ MSE = 0.1085 | LCC = 0.9419 | SRCC = 0.9415 ]
90
+ INFO:root:Average loss=0.09406070654781964
91
+ INFO:root:
92
+ [Valid][17][UTT][ MSE = 0.2255 | LCC = 0.8654 | SRCC = 0.8640 ]
93
+ INFO:root:
94
+ [Valid][17][SYS][ MSE = 0.1177 | LCC = 0.9418 | SRCC = 0.9440 ]
95
+ INFO:root:Average loss=0.08924258161240663
96
+ INFO:root:
97
+ [Valid][18][UTT][ MSE = 0.2167 | LCC = 0.8688 | SRCC = 0.8649 ]
98
+ INFO:root:
99
+ [Valid][18][SYS][ MSE = 0.0992 | LCC = 0.9429 | SRCC = 0.9380 ]
100
+ INFO:root:Average loss=0.08068299785447426
101
+ INFO:root:
102
+ [Valid][19][UTT][ MSE = 0.2161 | LCC = 0.8686 | SRCC = 0.8646 ]
103
+ INFO:root:
104
+ [Valid][19][SYS][ MSE = 0.1112 | LCC = 0.9383 | SRCC = 0.9338 ]
105
+ INFO:root:Average loss=0.08465328506934337
106
+ INFO:root:
107
+ [Valid][20][UTT][ MSE = 0.2141 | LCC = 0.8689 | SRCC = 0.8651 ]
108
+ INFO:root:
109
+ [Valid][20][SYS][ MSE = 0.1057 | LCC = 0.9440 | SRCC = 0.9409 ]
110
+ INFO:root:Average loss=0.07797851294088058
111
+ INFO:root:
112
+ [Valid][21][UTT][ MSE = 0.2213 | LCC = 0.8704 | SRCC = 0.8664 ]
113
+ INFO:root:
114
+ [Valid][21][SYS][ MSE = 0.1045 | LCC = 0.9446 | SRCC = 0.9388 ]
115
+ INFO:root:Average loss=0.07383806950961933
116
+ INFO:root:
117
+ [Valid][22][UTT][ MSE = 0.2054 | LCC = 0.8721 | SRCC = 0.8696 ]
118
+ INFO:root:
119
+ [Valid][22][SYS][ MSE = 0.0898 | LCC = 0.9471 | SRCC = 0.9455 ]
120
+ INFO:root:Average loss=0.06801195805653548
121
+ INFO:root:
122
+ [Valid][23][UTT][ MSE = 0.2108 | LCC = 0.8702 | SRCC = 0.8671 ]
123
+ INFO:root:
124
+ [Valid][23][SYS][ MSE = 0.1030 | LCC = 0.9449 | SRCC = 0.9438 ]
125
+ INFO:root:Average loss=0.06952695314509746
126
+ INFO:root:
127
+ [Valid][24][UTT][ MSE = 0.2072 | LCC = 0.8740 | SRCC = 0.8719 ]
128
+ INFO:root:
129
+ [Valid][24][SYS][ MSE = 0.0980 | LCC = 0.9481 | SRCC = 0.9477 ]
130
+ INFO:root:Average loss=0.06444193609058857
131
+ INFO:root:
132
+ [Valid][25][UTT][ MSE = 0.2194 | LCC = 0.8664 | SRCC = 0.8645 ]
133
+ INFO:root:
134
+ [Valid][25][SYS][ MSE = 0.1171 | LCC = 0.9404 | SRCC = 0.9416 ]
135
+ INFO:root:Average loss=0.06203614681577071
136
+ INFO:root:
137
+ [Valid][26][UTT][ MSE = 0.2200 | LCC = 0.8691 | SRCC = 0.8654 ]
138
+ INFO:root:
139
+ [Valid][26][SYS][ MSE = 0.0975 | LCC = 0.9463 | SRCC = 0.9413 ]
140
+ INFO:root:Average loss=0.05990246645151041
141
+ INFO:root:
142
+ [Valid][27][UTT][ MSE = 0.2192 | LCC = 0.8689 | SRCC = 0.8656 ]
143
+ INFO:root:
144
+ [Valid][27][SYS][ MSE = 0.1137 | LCC = 0.9417 | SRCC = 0.9394 ]
145
+ INFO:root:Average loss=0.060253849921700284
146
+ INFO:root:
147
+ [Valid][28][UTT][ MSE = 0.2255 | LCC = 0.8686 | SRCC = 0.8647 ]
148
+ INFO:root:
149
+ [Valid][28][SYS][ MSE = 0.1256 | LCC = 0.9428 | SRCC = 0.9410 ]
150
+ INFO:root:Average loss=0.057994768262291566
151
+ INFO:root:
152
+ [Valid][29][UTT][ MSE = 0.2181 | LCC = 0.8666 | SRCC = 0.8649 ]
153
+ INFO:root:
154
+ [Valid][29][SYS][ MSE = 0.1148 | LCC = 0.9399 | SRCC = 0.9399 ]
155
+ INFO:root:Average loss=0.05246185893431688
156
+ INFO:root:
157
+ [Valid][30][UTT][ MSE = 0.2124 | LCC = 0.8670 | SRCC = 0.8639 ]
158
+ INFO:root:
159
+ [Valid][30][SYS][ MSE = 0.1001 | LCC = 0.9424 | SRCC = 0.9408 ]
160
+ INFO:root:
161
+ [Test][31][UTT][ MSE = 0.2362 | LCC = 0.8620 | SRCC = 0.8610 ]
162
+ INFO:root:
163
+ [Test][31][SYS][ MSE = 0.1418 | LCC = 0.9148 | SRCC = 0.9130 ]
results_bvcc/w2v2_base/layer1/run3/config.gin ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Training config on BVCC dataset.
2
+
3
+ # ------------------------------
4
+ # Architecture
5
+ # ------------------------------
6
+ ProjectionHead:
7
+ conv_channels = (32, 32)
8
+ dense_neurons = (64, 32, 1)
9
+ use_poolings = (True, True)
10
+ kernel_size = 5
11
+ pool_size = 5
12
+
13
+ # ------------------------------
14
+ # Dataset
15
+ # ------------------------------
16
+ Bvcc:
17
+ debug = False
18
+
19
+ get_dataloader:
20
+ num_workers = 12
21
+ shuffle = True
22
+
23
+ # ------------------------------
24
+ # Training loop
25
+ # ------------------------------
26
+ TrainingLoop:
27
+ num_epochs = 30
28
+ learning_rate = 1e-4
29
+ dataset_cls = @Bvcc
30
+ weight_decay = 0.0
31
+ batch_size_train = 64
results_bvcc/w2v2_base/layer1/run3/model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:346cbada120b6a68fa4def54564ec2d0494a6dd82cdfd147b8e54b0d8823730a
3
+ size 2747979
results_bvcc/w2v2_base/layer1/run3/model_best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:88c82d0e77fb4254ac1593d055299fa629b2f0f064e853c1b6feb98fb96b3754
3
+ size 2749702
results_bvcc/w2v2_base/layer1/run3/train.log ADDED
@@ -0,0 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ INFO:root:Num train speech clips: 4974
2
+ INFO:root:Num val speech clips: 1066
3
+ INFO:root:Num test speech clips: 1066
4
+ INFO:root:Device=cuda
5
+ INFO:root:Average loss=0.46871751355819213
6
+ INFO:root:
7
+ [Valid][0][UTT][ MSE = 0.3610 | LCC = 0.7796 | SRCC = 0.7904 ]
8
+ INFO:root:
9
+ [Valid][0][SYS][ MSE = 0.2642 | LCC = 0.8517 | SRCC = 0.8624 ]
10
+ INFO:root:Average loss=0.27671181295926756
11
+ INFO:root:
12
+ [Valid][1][UTT][ MSE = 0.2842 | LCC = 0.8195 | SRCC = 0.8230 ]
13
+ INFO:root:
14
+ [Valid][1][SYS][ MSE = 0.1693 | LCC = 0.8907 | SRCC = 0.8959 ]
15
+ INFO:root:Average loss=0.23440087357392678
16
+ INFO:root:
17
+ [Valid][2][UTT][ MSE = 0.2656 | LCC = 0.8478 | SRCC = 0.8472 ]
18
+ INFO:root:
19
+ [Valid][2][SYS][ MSE = 0.1619 | LCC = 0.9165 | SRCC = 0.9190 ]
20
+ INFO:root:Average loss=0.2157593487929075
21
+ INFO:root:
22
+ [Valid][3][UTT][ MSE = 0.2531 | LCC = 0.8485 | SRCC = 0.8497 ]
23
+ INFO:root:
24
+ [Valid][3][SYS][ MSE = 0.1396 | LCC = 0.9237 | SRCC = 0.9290 ]
25
+ INFO:root:Average loss=0.19040684105876166
26
+ INFO:root:
27
+ [Valid][4][UTT][ MSE = 0.2466 | LCC = 0.8519 | SRCC = 0.8542 ]
28
+ INFO:root:
29
+ [Valid][4][SYS][ MSE = 0.1470 | LCC = 0.9214 | SRCC = 0.9277 ]
30
+ INFO:root:Average loss=0.18986343038387787
31
+ INFO:root:
32
+ [Valid][5][UTT][ MSE = 0.2325 | LCC = 0.8606 | SRCC = 0.8614 ]
33
+ INFO:root:
34
+ [Valid][5][SYS][ MSE = 0.1093 | LCC = 0.9342 | SRCC = 0.9375 ]
35
+ INFO:root:Average loss=0.16716703103903013
36
+ INFO:root:
37
+ [Valid][6][UTT][ MSE = 0.2298 | LCC = 0.8626 | SRCC = 0.8629 ]
38
+ INFO:root:
39
+ [Valid][6][SYS][ MSE = 0.1131 | LCC = 0.9310 | SRCC = 0.9332 ]
40
+ INFO:root:Average loss=0.15025553078605577
41
+ INFO:root:
42
+ [Valid][7][UTT][ MSE = 0.2353 | LCC = 0.8638 | SRCC = 0.8606 ]
43
+ INFO:root:
44
+ [Valid][7][SYS][ MSE = 0.1247 | LCC = 0.9331 | SRCC = 0.9293 ]
45
+ INFO:root:Average loss=0.14692852034782752
46
+ INFO:root:
47
+ [Valid][8][UTT][ MSE = 0.2680 | LCC = 0.8597 | SRCC = 0.8601 ]
48
+ INFO:root:
49
+ [Valid][8][SYS][ MSE = 0.1739 | LCC = 0.9308 | SRCC = 0.9349 ]
50
+ INFO:root:Average loss=0.13832987290926468
51
+ INFO:root:
52
+ [Valid][9][UTT][ MSE = 0.2308 | LCC = 0.8610 | SRCC = 0.8612 ]
53
+ INFO:root:
54
+ [Valid][9][SYS][ MSE = 0.1122 | LCC = 0.9356 | SRCC = 0.9380 ]
55
+ INFO:root:Average loss=0.12529149489142957
56
+ INFO:root:
57
+ [Valid][10][UTT][ MSE = 0.2126 | LCC = 0.8691 | SRCC = 0.8693 ]
58
+ INFO:root:
59
+ [Valid][10][SYS][ MSE = 0.1073 | LCC = 0.9379 | SRCC = 0.9403 ]
60
+ INFO:root:Average loss=0.12105241398780774
61
+ INFO:root:
62
+ [Valid][11][UTT][ MSE = 0.2209 | LCC = 0.8679 | SRCC = 0.8677 ]
63
+ INFO:root:
64
+ [Valid][11][SYS][ MSE = 0.1138 | LCC = 0.9410 | SRCC = 0.9420 ]
65
+ INFO:root:Average loss=0.11520517350007327
66
+ INFO:root:
67
+ [Valid][12][UTT][ MSE = 0.2131 | LCC = 0.8694 | SRCC = 0.8693 ]
68
+ INFO:root:
69
+ [Valid][12][SYS][ MSE = 0.1143 | LCC = 0.9357 | SRCC = 0.9355 ]
70
+ INFO:root:Average loss=0.10659859214837734
71
+ INFO:root:
72
+ [Valid][13][UTT][ MSE = 0.2217 | LCC = 0.8688 | SRCC = 0.8675 ]
73
+ INFO:root:
74
+ [Valid][13][SYS][ MSE = 0.1192 | LCC = 0.9406 | SRCC = 0.9401 ]
75
+ INFO:root:Average loss=0.10968755309780438
76
+ INFO:root:
77
+ [Valid][14][UTT][ MSE = 0.2105 | LCC = 0.8711 | SRCC = 0.8689 ]
78
+ INFO:root:
79
+ [Valid][14][SYS][ MSE = 0.1039 | LCC = 0.9395 | SRCC = 0.9362 ]
80
+ INFO:root:Average loss=0.10449104106579071
81
+ INFO:root:
82
+ [Valid][15][UTT][ MSE = 0.2135 | LCC = 0.8707 | SRCC = 0.8703 ]
83
+ INFO:root:
84
+ [Valid][15][SYS][ MSE = 0.1221 | LCC = 0.9335 | SRCC = 0.9342 ]
85
+ INFO:root:Average loss=0.0918716401912463
86
+ INFO:root:
87
+ [Valid][16][UTT][ MSE = 0.2455 | LCC = 0.8659 | SRCC = 0.8691 ]
88
+ INFO:root:
89
+ [Valid][16][SYS][ MSE = 0.1538 | LCC = 0.9333 | SRCC = 0.9413 ]
90
+ INFO:root:Average loss=0.0942793258298666
91
+ INFO:root:
92
+ [Valid][17][UTT][ MSE = 0.2137 | LCC = 0.8720 | SRCC = 0.8689 ]
93
+ INFO:root:
94
+ [Valid][17][SYS][ MSE = 0.1149 | LCC = 0.9356 | SRCC = 0.9312 ]
95
+ INFO:root:Average loss=0.08522394886956765
96
+ INFO:root:
97
+ [Valid][18][UTT][ MSE = 0.2344 | LCC = 0.8721 | SRCC = 0.8702 ]
98
+ INFO:root:
99
+ [Valid][18][SYS][ MSE = 0.1425 | LCC = 0.9390 | SRCC = 0.9406 ]
100
+ INFO:root:Average loss=0.08684245191323452
101
+ INFO:root:
102
+ [Valid][19][UTT][ MSE = 0.2264 | LCC = 0.8685 | SRCC = 0.8688 ]
103
+ INFO:root:
104
+ [Valid][19][SYS][ MSE = 0.1366 | LCC = 0.9331 | SRCC = 0.9367 ]
105
+ INFO:root:Average loss=0.07973687728055012
106
+ INFO:root:
107
+ [Valid][20][UTT][ MSE = 0.2107 | LCC = 0.8706 | SRCC = 0.8691 ]
108
+ INFO:root:
109
+ [Valid][20][SYS][ MSE = 0.1134 | LCC = 0.9364 | SRCC = 0.9359 ]
110
+ INFO:root:Average loss=0.07529464426139991
111
+ INFO:root:
112
+ [Valid][21][UTT][ MSE = 0.2128 | LCC = 0.8692 | SRCC = 0.8677 ]
113
+ INFO:root:
114
+ [Valid][21][SYS][ MSE = 0.1149 | LCC = 0.9344 | SRCC = 0.9338 ]
115
+ INFO:root:Average loss=0.07538936134332265
116
+ INFO:root:
117
+ [Valid][22][UTT][ MSE = 0.2157 | LCC = 0.8723 | SRCC = 0.8697 ]
118
+ INFO:root:
119
+ [Valid][22][SYS][ MSE = 0.1235 | LCC = 0.9354 | SRCC = 0.9309 ]
120
+ INFO:root:Average loss=0.07405028368035953
121
+ INFO:root:
122
+ [Valid][23][UTT][ MSE = 0.2426 | LCC = 0.8700 | SRCC = 0.8697 ]
123
+ INFO:root:
124
+ [Valid][23][SYS][ MSE = 0.1588 | LCC = 0.9342 | SRCC = 0.9355 ]
125
+ INFO:root:Average loss=0.06866880921790233
126
+ INFO:root:
127
+ [Valid][24][UTT][ MSE = 0.2103 | LCC = 0.8714 | SRCC = 0.8695 ]
128
+ INFO:root:
129
+ [Valid][24][SYS][ MSE = 0.1108 | LCC = 0.9367 | SRCC = 0.9354 ]
130
+ INFO:root:Average loss=0.0702828078602369
131
+ INFO:root:
132
+ [Valid][25][UTT][ MSE = 0.2039 | LCC = 0.8734 | SRCC = 0.8705 ]
133
+ INFO:root:
134
+ [Valid][25][SYS][ MSE = 0.1016 | LCC = 0.9377 | SRCC = 0.9323 ]
135
+ INFO:root:Average loss=0.06925542175005643
136
+ INFO:root:
137
+ [Valid][26][UTT][ MSE = 0.2048 | LCC = 0.8728 | SRCC = 0.8711 ]
138
+ INFO:root:
139
+ [Valid][26][SYS][ MSE = 0.0996 | LCC = 0.9356 | SRCC = 0.9303 ]
140
+ INFO:root:Average loss=0.06134156978283173
141
+ INFO:root:
142
+ [Valid][27][UTT][ MSE = 0.2206 | LCC = 0.8677 | SRCC = 0.8649 ]
143
+ INFO:root:
144
+ [Valid][27][SYS][ MSE = 0.1209 | LCC = 0.9371 | SRCC = 0.9334 ]
145
+ INFO:root:Average loss=0.058670274125268825
146
+ INFO:root:
147
+ [Valid][28][UTT][ MSE = 0.2128 | LCC = 0.8692 | SRCC = 0.8667 ]
148
+ INFO:root:
149
+ [Valid][28][SYS][ MSE = 0.1019 | LCC = 0.9381 | SRCC = 0.9330 ]
150
+ INFO:root:Average loss=0.05878604814792291
151
+ INFO:root:
152
+ [Valid][29][UTT][ MSE = 0.2116 | LCC = 0.8700 | SRCC = 0.8661 ]
153
+ INFO:root:
154
+ [Valid][29][SYS][ MSE = 0.1094 | LCC = 0.9363 | SRCC = 0.9316 ]
155
+ INFO:root:Average loss=0.055968338528122656
156
+ INFO:root:
157
+ [Valid][30][UTT][ MSE = 0.2131 | LCC = 0.8720 | SRCC = 0.8678 ]
158
+ INFO:root:
159
+ [Valid][30][SYS][ MSE = 0.1195 | LCC = 0.9362 | SRCC = 0.9300 ]
160
+ INFO:root:
161
+ [Test][31][UTT][ MSE = 0.2366 | LCC = 0.8587 | SRCC = 0.8593 ]
162
+ INFO:root:
163
+ [Test][31][SYS][ MSE = 0.1257 | LCC = 0.9256 | SRCC = 0.9279 ]
results_bvcc/w2v2_base/layer1/run4/config.gin ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Training config on BVCC dataset.
2
+
3
+ # ------------------------------
4
+ # Architecture
5
+ # ------------------------------
6
+ ProjectionHead:
7
+ conv_channels = (32, 32)
8
+ dense_neurons = (64, 32, 1)
9
+ use_poolings = (True, True)
10
+ kernel_size = 5
11
+ pool_size = 5
12
+
13
+ # ------------------------------
14
+ # Dataset
15
+ # ------------------------------
16
+ Bvcc:
17
+ debug = False
18
+
19
+ get_dataloader:
20
+ num_workers = 12
21
+ shuffle = True
22
+
23
+ # ------------------------------
24
+ # Training loop
25
+ # ------------------------------
26
+ TrainingLoop:
27
+ num_epochs = 30
28
+ learning_rate = 1e-4
29
+ dataset_cls = @Bvcc
30
+ weight_decay = 0.0
31
+ batch_size_train = 64
results_bvcc/w2v2_base/layer1/run4/model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:47e12b81faf37172040cce1a322306ec6c1fc2972e1f5316d571952d17534f46
3
+ size 2747979
results_bvcc/w2v2_base/layer1/run4/model_best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:10b7f6630f6e8cd9082f912d75b22a7d2699b33fd9d4ffa7be02629920573271
3
+ size 2749702
results_bvcc/w2v2_base/layer1/run4/train.log ADDED
@@ -0,0 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ INFO:root:Num train speech clips: 4974
2
+ INFO:root:Num val speech clips: 1066
3
+ INFO:root:Num test speech clips: 1066
4
+ INFO:root:Device=cuda
5
+ INFO:root:Average loss=0.5067888184999808
6
+ INFO:root:
7
+ [Valid][0][UTT][ MSE = 0.3787 | LCC = 0.7676 | SRCC = 0.7716 ]
8
+ INFO:root:
9
+ [Valid][0][SYS][ MSE = 0.2580 | LCC = 0.8547 | SRCC = 0.8602 ]
10
+ INFO:root:Average loss=0.28694182959122533
11
+ INFO:root:
12
+ [Valid][1][UTT][ MSE = 0.3054 | LCC = 0.8105 | SRCC = 0.8177 ]
13
+ INFO:root:
14
+ [Valid][1][SYS][ MSE = 0.1863 | LCC = 0.8930 | SRCC = 0.9010 ]
15
+ INFO:root:Average loss=0.23920302990919504
16
+ INFO:root:
17
+ [Valid][2][UTT][ MSE = 0.2741 | LCC = 0.8386 | SRCC = 0.8393 ]
18
+ INFO:root:
19
+ [Valid][2][SYS][ MSE = 0.1467 | LCC = 0.9210 | SRCC = 0.9235 ]
20
+ INFO:root:Average loss=0.21217203235779053
21
+ INFO:root:
22
+ [Valid][3][UTT][ MSE = 0.2616 | LCC = 0.8402 | SRCC = 0.8408 ]
23
+ INFO:root:
24
+ [Valid][3][SYS][ MSE = 0.1357 | LCC = 0.9121 | SRCC = 0.9131 ]
25
+ INFO:root:Average loss=0.19956507419164365
26
+ INFO:root:
27
+ [Valid][4][UTT][ MSE = 0.2505 | LCC = 0.8506 | SRCC = 0.8512 ]
28
+ INFO:root:
29
+ [Valid][4][SYS][ MSE = 0.1288 | LCC = 0.9267 | SRCC = 0.9300 ]
30
+ INFO:root:Average loss=0.18455939013988543
31
+ INFO:root:
32
+ [Valid][5][UTT][ MSE = 0.2528 | LCC = 0.8536 | SRCC = 0.8531 ]
33
+ INFO:root:
34
+ [Valid][5][SYS][ MSE = 0.1231 | LCC = 0.9293 | SRCC = 0.9300 ]
35
+ INFO:root:Average loss=0.16596660868097574
36
+ INFO:root:
37
+ [Valid][6][UTT][ MSE = 0.2391 | LCC = 0.8592 | SRCC = 0.8570 ]
38
+ INFO:root:
39
+ [Valid][6][SYS][ MSE = 0.1131 | LCC = 0.9329 | SRCC = 0.9302 ]
40
+ INFO:root:Average loss=0.15587880462408066
41
+ INFO:root:
42
+ [Valid][7][UTT][ MSE = 0.2430 | LCC = 0.8558 | SRCC = 0.8537 ]
43
+ INFO:root:
44
+ [Valid][7][SYS][ MSE = 0.1304 | LCC = 0.9287 | SRCC = 0.9277 ]
45
+ INFO:root:Average loss=0.14549451025250631
46
+ INFO:root:
47
+ [Valid][8][UTT][ MSE = 0.2537 | LCC = 0.8568 | SRCC = 0.8572 ]
48
+ INFO:root:
49
+ [Valid][8][SYS][ MSE = 0.1547 | LCC = 0.9273 | SRCC = 0.9312 ]
50
+ INFO:root:Average loss=0.14089427315271819
51
+ INFO:root:
52
+ [Valid][9][UTT][ MSE = 0.2450 | LCC = 0.8516 | SRCC = 0.8518 ]
53
+ INFO:root:
54
+ [Valid][9][SYS][ MSE = 0.1301 | LCC = 0.9300 | SRCC = 0.9345 ]
55
+ INFO:root:Average loss=0.1309406508046847
56
+ INFO:root:
57
+ [Valid][10][UTT][ MSE = 0.2291 | LCC = 0.8593 | SRCC = 0.8578 ]
58
+ INFO:root:
59
+ [Valid][10][SYS][ MSE = 0.1124 | LCC = 0.9358 | SRCC = 0.9358 ]
60
+ INFO:root:Average loss=0.12291629870350544
61
+ INFO:root:
62
+ [Valid][11][UTT][ MSE = 0.2327 | LCC = 0.8623 | SRCC = 0.8601 ]
63
+ INFO:root:
64
+ [Valid][11][SYS][ MSE = 0.1056 | LCC = 0.9389 | SRCC = 0.9350 ]
65
+ INFO:root:Average loss=0.1144241709739734
66
+ INFO:root:
67
+ [Valid][12][UTT][ MSE = 0.2246 | LCC = 0.8619 | SRCC = 0.8598 ]
68
+ INFO:root:
69
+ [Valid][12][SYS][ MSE = 0.1116 | LCC = 0.9373 | SRCC = 0.9369 ]
70
+ INFO:root:Average loss=0.10850251150818971
71
+ INFO:root:
72
+ [Valid][13][UTT][ MSE = 0.2227 | LCC = 0.8616 | SRCC = 0.8594 ]
73
+ INFO:root:
74
+ [Valid][13][SYS][ MSE = 0.1055 | LCC = 0.9379 | SRCC = 0.9376 ]
75
+ INFO:root:Average loss=0.10566861149019156
76
+ INFO:root:
77
+ [Valid][14][UTT][ MSE = 0.2212 | LCC = 0.8608 | SRCC = 0.8593 ]
78
+ INFO:root:
79
+ [Valid][14][SYS][ MSE = 0.1054 | LCC = 0.9364 | SRCC = 0.9392 ]
80
+ INFO:root:Average loss=0.10105009959676327
81
+ INFO:root:
82
+ [Valid][15][UTT][ MSE = 0.2293 | LCC = 0.8554 | SRCC = 0.8543 ]
83
+ INFO:root:
84
+ [Valid][15][SYS][ MSE = 0.1115 | LCC = 0.9358 | SRCC = 0.9401 ]
85
+ INFO:root:Average loss=0.09794910175678058
86
+ INFO:root:
87
+ [Valid][16][UTT][ MSE = 0.2247 | LCC = 0.8605 | SRCC = 0.8583 ]
88
+ INFO:root:
89
+ [Valid][16][SYS][ MSE = 0.1109 | LCC = 0.9350 | SRCC = 0.9357 ]
90
+ INFO:root:Average loss=0.0921257061836047
91
+ INFO:root:
92
+ [Valid][17][UTT][ MSE = 0.2215 | LCC = 0.8593 | SRCC = 0.8574 ]
93
+ INFO:root:
94
+ [Valid][17][SYS][ MSE = 0.0989 | LCC = 0.9380 | SRCC = 0.9397 ]
95
+ INFO:root:Average loss=0.08327317982912064
96
+ INFO:root:
97
+ [Valid][18][UTT][ MSE = 0.2166 | LCC = 0.8626 | SRCC = 0.8613 ]
98
+ INFO:root:
99
+ [Valid][18][SYS][ MSE = 0.0964 | LCC = 0.9357 | SRCC = 0.9363 ]
100
+ INFO:root:Average loss=0.09554384051798245
101
+ INFO:root:
102
+ [Valid][19][UTT][ MSE = 0.2440 | LCC = 0.8582 | SRCC = 0.8562 ]
103
+ INFO:root:
104
+ [Valid][19][SYS][ MSE = 0.1127 | LCC = 0.9306 | SRCC = 0.9285 ]
105
+ INFO:root:Average loss=0.0855937705685695
106
+ INFO:root:
107
+ [Valid][20][UTT][ MSE = 0.2158 | LCC = 0.8632 | SRCC = 0.8613 ]
108
+ INFO:root:
109
+ [Valid][20][SYS][ MSE = 0.0965 | LCC = 0.9388 | SRCC = 0.9368 ]
110
+ INFO:root:Average loss=0.07785610897609821
111
+ INFO:root:
112
+ [Valid][21][UTT][ MSE = 0.2248 | LCC = 0.8624 | SRCC = 0.8590 ]
113
+ INFO:root:
114
+ [Valid][21][SYS][ MSE = 0.1069 | LCC = 0.9421 | SRCC = 0.9383 ]
115
+ INFO:root:Average loss=0.07071381960159694
116
+ INFO:root:
117
+ [Valid][22][UTT][ MSE = 0.2162 | LCC = 0.8618 | SRCC = 0.8584 ]
118
+ INFO:root:
119
+ [Valid][22][SYS][ MSE = 0.0971 | LCC = 0.9382 | SRCC = 0.9361 ]
120
+ INFO:root:Average loss=0.0724777755064842
121
+ INFO:root:
122
+ [Valid][23][UTT][ MSE = 0.2325 | LCC = 0.8600 | SRCC = 0.8571 ]
123
+ INFO:root:
124
+ [Valid][23][SYS][ MSE = 0.1164 | LCC = 0.9416 | SRCC = 0.9406 ]
125
+ INFO:root:Average loss=0.06505997803730842
126
+ INFO:root:
127
+ [Valid][24][UTT][ MSE = 0.2212 | LCC = 0.8638 | SRCC = 0.8613 ]
128
+ INFO:root:
129
+ [Valid][24][SYS][ MSE = 0.0981 | LCC = 0.9424 | SRCC = 0.9394 ]
130
+ INFO:root:Average loss=0.0653465433189502
131
+ INFO:root:
132
+ [Valid][25][UTT][ MSE = 0.2362 | LCC = 0.8579 | SRCC = 0.8555 ]
133
+ INFO:root:
134
+ [Valid][25][SYS][ MSE = 0.1219 | LCC = 0.9391 | SRCC = 0.9373 ]
135
+ INFO:root:Average loss=0.06491684961395386
136
+ INFO:root:
137
+ [Valid][26][UTT][ MSE = 0.2331 | LCC = 0.8584 | SRCC = 0.8556 ]
138
+ INFO:root:
139
+ [Valid][26][SYS][ MSE = 0.1030 | LCC = 0.9373 | SRCC = 0.9331 ]
140
+ INFO:root:Average loss=0.0655498340821419
141
+ INFO:root:
142
+ [Valid][27][UTT][ MSE = 0.2169 | LCC = 0.8609 | SRCC = 0.8594 ]
143
+ INFO:root:
144
+ [Valid][27][SYS][ MSE = 0.0975 | LCC = 0.9371 | SRCC = 0.9358 ]
145
+ INFO:root:Average loss=0.05822523506597067
146
+ INFO:root:
147
+ [Valid][28][UTT][ MSE = 0.2225 | LCC = 0.8625 | SRCC = 0.8595 ]
148
+ INFO:root:
149
+ [Valid][28][SYS][ MSE = 0.1022 | LCC = 0.9419 | SRCC = 0.9394 ]
150
+ INFO:root:Average loss=0.06262223728192158
151
+ INFO:root:
152
+ [Valid][29][UTT][ MSE = 0.2209 | LCC = 0.8590 | SRCC = 0.8579 ]
153
+ INFO:root:
154
+ [Valid][29][SYS][ MSE = 0.0941 | LCC = 0.9341 | SRCC = 0.9334 ]
155
+ INFO:root:Average loss=0.061830226809550554
156
+ INFO:root:
157
+ [Valid][30][UTT][ MSE = 0.2208 | LCC = 0.8628 | SRCC = 0.8607 ]
158
+ INFO:root:
159
+ [Valid][30][SYS][ MSE = 0.0992 | LCC = 0.9370 | SRCC = 0.9355 ]
160
+ INFO:root:
161
+ [Test][31][UTT][ MSE = 0.2542 | LCC = 0.8541 | SRCC = 0.8542 ]
162
+ INFO:root:
163
+ [Test][31][SYS][ MSE = 0.1683 | LCC = 0.9123 | SRCC = 0.9148 ]
results_bvcc/w2v2_base/layer10/run0/config.gin ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Training config on BVCC dataset.
2
+
3
+ # ------------------------------
4
+ # Architecture
5
+ # ------------------------------
6
+ ProjectionHead:
7
+ conv_channels = (32, 32)
8
+ dense_neurons = (64, 32, 1)
9
+ use_poolings = (True, True)
10
+ kernel_size = 5
11
+ pool_size = 5
12
+
13
+ # ------------------------------
14
+ # Dataset
15
+ # ------------------------------
16
+ Bvcc:
17
+ debug = False
18
+
19
+ get_dataloader:
20
+ num_workers = 12
21
+ shuffle = True
22
+
23
+ # ------------------------------
24
+ # Training loop
25
+ # ------------------------------
26
+ TrainingLoop:
27
+ num_epochs = 30
28
+ learning_rate = 1e-4
29
+ dataset_cls = @Bvcc
30
+ weight_decay = 0.0
31
+ batch_size_train = 64
results_bvcc/w2v2_base/layer10/run0/model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:34c1a63ec83a1430692c77ecf5066f37ea69f3e10b10914faf5dfa5611595741
3
+ size 2747979
results_bvcc/w2v2_base/layer10/run0/model_best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1a2c5e2c1fc190799f3088ee8b609abb21a34b65276a7504346f38a469024ab1
3
+ size 2749702
results_bvcc/w2v2_base/layer10/run0/train.log ADDED
@@ -0,0 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ INFO:root:Num train speech clips: 4974
2
+ INFO:root:Num val speech clips: 1066
3
+ INFO:root:Num test speech clips: 1066
4
+ INFO:root:Device=cuda
5
+ INFO:root:Average loss=0.6340566608004081
6
+ INFO:root:
7
+ [Valid][0][UTT][ MSE = 0.4106 | LCC = 0.7514 | SRCC = 0.7600 ]
8
+ INFO:root:
9
+ [Valid][0][SYS][ MSE = 0.3009 | LCC = 0.8284 | SRCC = 0.8318 ]
10
+ INFO:root:Average loss=0.3847933440254285
11
+ INFO:root:
12
+ [Valid][1][UTT][ MSE = 0.3723 | LCC = 0.7784 | SRCC = 0.7856 ]
13
+ INFO:root:
14
+ [Valid][1][SYS][ MSE = 0.2704 | LCC = 0.8500 | SRCC = 0.8501 ]
15
+ INFO:root:Average loss=0.30829281417223126
16
+ INFO:root:
17
+ [Valid][2][UTT][ MSE = 0.3684 | LCC = 0.7963 | SRCC = 0.7985 ]
18
+ INFO:root:
19
+ [Valid][2][SYS][ MSE = 0.2706 | LCC = 0.8699 | SRCC = 0.8680 ]
20
+ INFO:root:Average loss=0.31271022462692016
21
+ INFO:root:
22
+ [Valid][3][UTT][ MSE = 0.3253 | LCC = 0.8152 | SRCC = 0.8146 ]
23
+ INFO:root:
24
+ [Valid][3][SYS][ MSE = 0.1964 | LCC = 0.8825 | SRCC = 0.8798 ]
25
+ INFO:root:Average loss=0.302239019710284
26
+ INFO:root:
27
+ [Valid][4][UTT][ MSE = 0.3614 | LCC = 0.8170 | SRCC = 0.8186 ]
28
+ INFO:root:
29
+ [Valid][4][SYS][ MSE = 0.2740 | LCC = 0.8858 | SRCC = 0.8825 ]
30
+ INFO:root:Average loss=0.3026625733727064
31
+ INFO:root:
32
+ [Valid][5][UTT][ MSE = 0.3027 | LCC = 0.8235 | SRCC = 0.8254 ]
33
+ INFO:root:
34
+ [Valid][5][SYS][ MSE = 0.2049 | LCC = 0.8904 | SRCC = 0.8884 ]
35
+ INFO:root:Average loss=0.27604097318954957
36
+ INFO:root:
37
+ [Valid][6][UTT][ MSE = 0.3244 | LCC = 0.8330 | SRCC = 0.8298 ]
38
+ INFO:root:
39
+ [Valid][6][SYS][ MSE = 0.2295 | LCC = 0.8938 | SRCC = 0.8855 ]
40
+ INFO:root:Average loss=0.26417100658783543
41
+ INFO:root:
42
+ [Valid][7][UTT][ MSE = 0.3696 | LCC = 0.8294 | SRCC = 0.8307 ]
43
+ INFO:root:
44
+ [Valid][7][SYS][ MSE = 0.2218 | LCC = 0.8951 | SRCC = 0.8915 ]
45
+ INFO:root:Average loss=0.2875201409825912
46
+ INFO:root:
47
+ [Valid][8][UTT][ MSE = 0.3298 | LCC = 0.8324 | SRCC = 0.8360 ]
48
+ INFO:root:
49
+ [Valid][8][SYS][ MSE = 0.2464 | LCC = 0.8945 | SRCC = 0.8941 ]
50
+ INFO:root:Average loss=0.24525141200193992
51
+ INFO:root:
52
+ [Valid][9][UTT][ MSE = 0.2832 | LCC = 0.8442 | SRCC = 0.8422 ]
53
+ INFO:root:
54
+ [Valid][9][SYS][ MSE = 0.1743 | LCC = 0.9060 | SRCC = 0.9009 ]
55
+ INFO:root:Average loss=0.2364092094775958
56
+ INFO:root:
57
+ [Valid][10][UTT][ MSE = 0.3405 | LCC = 0.8430 | SRCC = 0.8429 ]
58
+ INFO:root:
59
+ [Valid][10][SYS][ MSE = 0.2558 | LCC = 0.9111 | SRCC = 0.9070 ]
60
+ INFO:root:Average loss=0.25019912135142547
61
+ INFO:root:
62
+ [Valid][11][UTT][ MSE = 0.2921 | LCC = 0.8451 | SRCC = 0.8446 ]
63
+ INFO:root:
64
+ [Valid][11][SYS][ MSE = 0.1648 | LCC = 0.9099 | SRCC = 0.9045 ]
65
+ INFO:root:Average loss=0.27862557539573085
66
+ INFO:root:
67
+ [Valid][12][UTT][ MSE = 0.2895 | LCC = 0.8432 | SRCC = 0.8413 ]
68
+ INFO:root:
69
+ [Valid][12][SYS][ MSE = 0.1873 | LCC = 0.9077 | SRCC = 0.9023 ]
70
+ INFO:root:Average loss=0.23084685531182167
71
+ INFO:root:
72
+ [Valid][13][UTT][ MSE = 0.2827 | LCC = 0.8465 | SRCC = 0.8464 ]
73
+ INFO:root:
74
+ [Valid][13][SYS][ MSE = 0.1910 | LCC = 0.9125 | SRCC = 0.9102 ]
75
+ INFO:root:Average loss=0.22213796965586832
76
+ INFO:root:
77
+ [Valid][14][UTT][ MSE = 0.3125 | LCC = 0.8457 | SRCC = 0.8463 ]
78
+ INFO:root:
79
+ [Valid][14][SYS][ MSE = 0.2268 | LCC = 0.9111 | SRCC = 0.9076 ]
80
+ INFO:root:Average loss=0.21562972512000647
81
+ INFO:root:
82
+ [Valid][15][UTT][ MSE = 0.2733 | LCC = 0.8510 | SRCC = 0.8493 ]
83
+ INFO:root:
84
+ [Valid][15][SYS][ MSE = 0.1689 | LCC = 0.9194 | SRCC = 0.9141 ]
85
+ INFO:root:Average loss=0.20544885232662544
86
+ INFO:root:
87
+ [Valid][16][UTT][ MSE = 0.2582 | LCC = 0.8532 | SRCC = 0.8531 ]
88
+ INFO:root:
89
+ [Valid][16][SYS][ MSE = 0.1508 | LCC = 0.9153 | SRCC = 0.9110 ]
90
+ INFO:root:Average loss=0.19289059908344194
91
+ INFO:root:
92
+ [Valid][17][UTT][ MSE = 0.2610 | LCC = 0.8567 | SRCC = 0.8573 ]
93
+ INFO:root:
94
+ [Valid][17][SYS][ MSE = 0.1694 | LCC = 0.9142 | SRCC = 0.9118 ]
95
+ INFO:root:Average loss=0.1999117200000164
96
+ INFO:root:
97
+ [Valid][18][UTT][ MSE = 0.3502 | LCC = 0.8483 | SRCC = 0.8489 ]
98
+ INFO:root:
99
+ [Valid][18][SYS][ MSE = 0.2647 | LCC = 0.9167 | SRCC = 0.9136 ]
100
+ INFO:root:Average loss=0.21447260677814484
101
+ INFO:root:
102
+ [Valid][19][UTT][ MSE = 0.2960 | LCC = 0.8518 | SRCC = 0.8516 ]
103
+ INFO:root:
104
+ [Valid][19][SYS][ MSE = 0.1730 | LCC = 0.9164 | SRCC = 0.9078 ]
105
+ INFO:root:Average loss=0.20319852691430312
106
+ INFO:root:
107
+ [Valid][20][UTT][ MSE = 0.2693 | LCC = 0.8538 | SRCC = 0.8543 ]
108
+ INFO:root:
109
+ [Valid][20][SYS][ MSE = 0.1738 | LCC = 0.9145 | SRCC = 0.9124 ]
110
+ INFO:root:Average loss=0.21281205280086932
111
+ INFO:root:
112
+ [Valid][21][UTT][ MSE = 0.2741 | LCC = 0.8537 | SRCC = 0.8529 ]
113
+ INFO:root:
114
+ [Valid][21][SYS][ MSE = 0.1787 | LCC = 0.9133 | SRCC = 0.9075 ]
115
+ INFO:root:Average loss=0.222677637082644
116
+ INFO:root:
117
+ [Valid][22][UTT][ MSE = 0.2668 | LCC = 0.8554 | SRCC = 0.8565 ]
118
+ INFO:root:
119
+ [Valid][22][SYS][ MSE = 0.1624 | LCC = 0.9186 | SRCC = 0.9168 ]
120
+ INFO:root:Average loss=0.18749733603535554
121
+ INFO:root:
122
+ [Valid][23][UTT][ MSE = 0.2773 | LCC = 0.8512 | SRCC = 0.8497 ]
123
+ INFO:root:
124
+ [Valid][23][SYS][ MSE = 0.1769 | LCC = 0.9171 | SRCC = 0.9100 ]
125
+ INFO:root:Average loss=0.21209082150688538
126
+ INFO:root:
127
+ [Valid][24][UTT][ MSE = 0.2507 | LCC = 0.8562 | SRCC = 0.8569 ]
128
+ INFO:root:
129
+ [Valid][24][SYS][ MSE = 0.1468 | LCC = 0.9200 | SRCC = 0.9166 ]
130
+ INFO:root:Average loss=0.17649115985020614
131
+ INFO:root:
132
+ [Valid][25][UTT][ MSE = 0.2750 | LCC = 0.8531 | SRCC = 0.8521 ]
133
+ INFO:root:
134
+ [Valid][25][SYS][ MSE = 0.1826 | LCC = 0.9181 | SRCC = 0.9133 ]
135
+ INFO:root:Average loss=0.1752329561381768
136
+ INFO:root:
137
+ [Valid][26][UTT][ MSE = 0.2695 | LCC = 0.8559 | SRCC = 0.8585 ]
138
+ INFO:root:
139
+ [Valid][26][SYS][ MSE = 0.1472 | LCC = 0.9228 | SRCC = 0.9195 ]
140
+ INFO:root:Average loss=0.18399202670806494
141
+ INFO:root:
142
+ [Valid][27][UTT][ MSE = 0.2607 | LCC = 0.8577 | SRCC = 0.8576 ]
143
+ INFO:root:
144
+ [Valid][27][SYS][ MSE = 0.1629 | LCC = 0.9278 | SRCC = 0.9257 ]
145
+ INFO:root:Average loss=0.1969409548701384
146
+ INFO:root:
147
+ [Valid][28][UTT][ MSE = 0.3153 | LCC = 0.8558 | SRCC = 0.8557 ]
148
+ INFO:root:
149
+ [Valid][28][SYS][ MSE = 0.2016 | LCC = 0.9231 | SRCC = 0.9166 ]
150
+ INFO:root:Average loss=0.16963752693472764
151
+ INFO:root:
152
+ [Valid][29][UTT][ MSE = 0.2594 | LCC = 0.8517 | SRCC = 0.8516 ]
153
+ INFO:root:
154
+ [Valid][29][SYS][ MSE = 0.1607 | LCC = 0.9219 | SRCC = 0.9184 ]
155
+ INFO:root:Average loss=0.15732398762916908
156
+ INFO:root:
157
+ [Valid][30][UTT][ MSE = 0.2864 | LCC = 0.8545 | SRCC = 0.8538 ]
158
+ INFO:root:
159
+ [Valid][30][SYS][ MSE = 0.1970 | LCC = 0.9199 | SRCC = 0.9155 ]
160
+ INFO:root:
161
+ [Test][31][UTT][ MSE = 0.2476 | LCC = 0.8550 | SRCC = 0.8542 ]
162
+ INFO:root:
163
+ [Test][31][SYS][ MSE = 0.1155 | LCC = 0.9270 | SRCC = 0.9283 ]
results_bvcc/w2v2_base/layer10/run1/config.gin ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Training config on BVCC dataset.
2
+
3
+ # ------------------------------
4
+ # Architecture
5
+ # ------------------------------
6
+ ProjectionHead:
7
+ conv_channels = (32, 32)
8
+ dense_neurons = (64, 32, 1)
9
+ use_poolings = (True, True)
10
+ kernel_size = 5
11
+ pool_size = 5
12
+
13
+ # ------------------------------
14
+ # Dataset
15
+ # ------------------------------
16
+ Bvcc:
17
+ debug = False
18
+
19
+ get_dataloader:
20
+ num_workers = 12
21
+ shuffle = True
22
+
23
+ # ------------------------------
24
+ # Training loop
25
+ # ------------------------------
26
+ TrainingLoop:
27
+ num_epochs = 30
28
+ learning_rate = 1e-4
29
+ dataset_cls = @Bvcc
30
+ weight_decay = 0.0
31
+ batch_size_train = 64
results_bvcc/w2v2_base/layer10/run1/model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:598dc0012f60cdfc06d5253279853e5fc3e9ab77df774cfa9e1e3ed384fce21b
3
+ size 2747979
results_bvcc/w2v2_base/layer10/run1/model_best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:306949bbbe0fb3d5003ae9772e2142cba578554ebc0cb3f80988304a6e3fdbc6
3
+ size 2749702
results_bvcc/w2v2_base/layer10/run1/train.log ADDED
@@ -0,0 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ INFO:root:Num train speech clips: 4974
2
+ INFO:root:Num val speech clips: 1066
3
+ INFO:root:Num test speech clips: 1066
4
+ INFO:root:Device=cuda
5
+ INFO:root:Average loss=0.7796100844175388
6
+ INFO:root:
7
+ [Valid][0][UTT][ MSE = 0.6900 | LCC = 0.6318 | SRCC = 0.6236 ]
8
+ INFO:root:
9
+ [Valid][0][SYS][ MSE = 0.5890 | LCC = 0.7349 | SRCC = 0.7272 ]
10
+ INFO:root:Average loss=0.43842379137491566
11
+ INFO:root:
12
+ [Valid][1][UTT][ MSE = 0.4299 | LCC = 0.7606 | SRCC = 0.7773 ]
13
+ INFO:root:
14
+ [Valid][1][SYS][ MSE = 0.3243 | LCC = 0.8389 | SRCC = 0.8462 ]
15
+ INFO:root:Average loss=0.33721469266292375
16
+ INFO:root:
17
+ [Valid][2][UTT][ MSE = 0.3480 | LCC = 0.7960 | SRCC = 0.7999 ]
18
+ INFO:root:
19
+ [Valid][2][SYS][ MSE = 0.2464 | LCC = 0.8636 | SRCC = 0.8620 ]
20
+ INFO:root:Average loss=0.3431662714634186
21
+ INFO:root:
22
+ [Valid][3][UTT][ MSE = 0.3195 | LCC = 0.8054 | SRCC = 0.8097 ]
23
+ INFO:root:
24
+ [Valid][3][SYS][ MSE = 0.2148 | LCC = 0.8697 | SRCC = 0.8674 ]
25
+ INFO:root:Average loss=0.30712308906591856
26
+ INFO:root:
27
+ [Valid][4][UTT][ MSE = 0.3144 | LCC = 0.8185 | SRCC = 0.8199 ]
28
+ INFO:root:
29
+ [Valid][4][SYS][ MSE = 0.1908 | LCC = 0.8827 | SRCC = 0.8823 ]
30
+ INFO:root:Average loss=0.28962959884068906
31
+ INFO:root:
32
+ [Valid][5][UTT][ MSE = 0.3502 | LCC = 0.8183 | SRCC = 0.8205 ]
33
+ INFO:root:
34
+ [Valid][5][SYS][ MSE = 0.2134 | LCC = 0.8749 | SRCC = 0.8772 ]
35
+ INFO:root:Average loss=0.28130568888707036
36
+ INFO:root:
37
+ [Valid][6][UTT][ MSE = 0.3032 | LCC = 0.8288 | SRCC = 0.8286 ]
38
+ INFO:root:
39
+ [Valid][6][SYS][ MSE = 0.1951 | LCC = 0.8926 | SRCC = 0.8887 ]
40
+ INFO:root:Average loss=0.2795686335899891
41
+ INFO:root:
42
+ [Valid][7][UTT][ MSE = 0.3075 | LCC = 0.8311 | SRCC = 0.8295 ]
43
+ INFO:root:
44
+ [Valid][7][SYS][ MSE = 0.2042 | LCC = 0.8928 | SRCC = 0.8890 ]
45
+ INFO:root:Average loss=0.25468874741823244
46
+ INFO:root:
47
+ [Valid][8][UTT][ MSE = 0.3200 | LCC = 0.8369 | SRCC = 0.8390 ]
48
+ INFO:root:
49
+ [Valid][8][SYS][ MSE = 0.1840 | LCC = 0.9005 | SRCC = 0.9005 ]
50
+ INFO:root:Average loss=0.2632343329680272
51
+ INFO:root:
52
+ [Valid][9][UTT][ MSE = 0.3394 | LCC = 0.8340 | SRCC = 0.8335 ]
53
+ INFO:root:
54
+ [Valid][9][SYS][ MSE = 0.2438 | LCC = 0.8982 | SRCC = 0.8920 ]
55
+ INFO:root:Average loss=0.2561314589320085
56
+ INFO:root:
57
+ [Valid][10][UTT][ MSE = 0.2806 | LCC = 0.8409 | SRCC = 0.8399 ]
58
+ INFO:root:
59
+ [Valid][10][SYS][ MSE = 0.1698 | LCC = 0.9036 | SRCC = 0.8982 ]
60
+ INFO:root:Average loss=0.2590295391587111
61
+ INFO:root:
62
+ [Valid][11][UTT][ MSE = 0.3220 | LCC = 0.8449 | SRCC = 0.8445 ]
63
+ INFO:root:
64
+ [Valid][11][SYS][ MSE = 0.1911 | LCC = 0.9085 | SRCC = 0.9055 ]
65
+ INFO:root:Average loss=0.24746226251889497
66
+ INFO:root:
67
+ [Valid][12][UTT][ MSE = 0.2911 | LCC = 0.8403 | SRCC = 0.8417 ]
68
+ INFO:root:
69
+ [Valid][12][SYS][ MSE = 0.1961 | LCC = 0.9034 | SRCC = 0.9010 ]
70
+ INFO:root:Average loss=0.2464088023855136
71
+ INFO:root:
72
+ [Valid][13][UTT][ MSE = 0.2756 | LCC = 0.8400 | SRCC = 0.8411 ]
73
+ INFO:root:
74
+ [Valid][13][SYS][ MSE = 0.1819 | LCC = 0.9006 | SRCC = 0.8981 ]
75
+ INFO:root:Average loss=0.2302155883457416
76
+ INFO:root:
77
+ [Valid][14][UTT][ MSE = 0.2737 | LCC = 0.8459 | SRCC = 0.8459 ]
78
+ INFO:root:
79
+ [Valid][14][SYS][ MSE = 0.1546 | LCC = 0.9050 | SRCC = 0.9027 ]
80
+ INFO:root:Average loss=0.2331328264031655
81
+ INFO:root:
82
+ [Valid][15][UTT][ MSE = 0.2586 | LCC = 0.8418 | SRCC = 0.8408 ]
83
+ INFO:root:
84
+ [Valid][15][SYS][ MSE = 0.1446 | LCC = 0.9063 | SRCC = 0.9006 ]
85
+ INFO:root:Average loss=0.21601504622361598
86
+ INFO:root:
87
+ [Valid][16][UTT][ MSE = 0.2791 | LCC = 0.8456 | SRCC = 0.8441 ]
88
+ INFO:root:
89
+ [Valid][16][SYS][ MSE = 0.1559 | LCC = 0.9105 | SRCC = 0.9069 ]
90
+ INFO:root:Average loss=0.21259737740724516
91
+ INFO:root:
92
+ [Valid][17][UTT][ MSE = 0.3048 | LCC = 0.8435 | SRCC = 0.8427 ]
93
+ INFO:root:
94
+ [Valid][17][SYS][ MSE = 0.2081 | LCC = 0.9063 | SRCC = 0.9020 ]
95
+ INFO:root:Average loss=0.2054756549306405
96
+ INFO:root:
97
+ [Valid][18][UTT][ MSE = 0.2682 | LCC = 0.8466 | SRCC = 0.8461 ]
98
+ INFO:root:
99
+ [Valid][18][SYS][ MSE = 0.1679 | LCC = 0.9086 | SRCC = 0.9032 ]
100
+ INFO:root:Average loss=0.19966396890007532
101
+ INFO:root:
102
+ [Valid][19][UTT][ MSE = 0.2665 | LCC = 0.8492 | SRCC = 0.8481 ]
103
+ INFO:root:
104
+ [Valid][19][SYS][ MSE = 0.1388 | LCC = 0.9182 | SRCC = 0.9130 ]
105
+ INFO:root:Average loss=0.20426208172471094
106
+ INFO:root:
107
+ [Valid][20][UTT][ MSE = 0.3294 | LCC = 0.8432 | SRCC = 0.8414 ]
108
+ INFO:root:
109
+ [Valid][20][SYS][ MSE = 0.2330 | LCC = 0.9149 | SRCC = 0.9120 ]
110
+ INFO:root:Average loss=0.1957240331058319
111
+ INFO:root:
112
+ [Valid][21][UTT][ MSE = 0.2638 | LCC = 0.8438 | SRCC = 0.8424 ]
113
+ INFO:root:
114
+ [Valid][21][SYS][ MSE = 0.1370 | LCC = 0.9156 | SRCC = 0.9094 ]
115
+ INFO:root:Average loss=0.18055324447460663
116
+ INFO:root:
117
+ [Valid][22][UTT][ MSE = 0.3058 | LCC = 0.8398 | SRCC = 0.8420 ]
118
+ INFO:root:
119
+ [Valid][22][SYS][ MSE = 0.2248 | LCC = 0.9039 | SRCC = 0.9026 ]
120
+ INFO:root:Average loss=0.20639980890047857
121
+ INFO:root:
122
+ [Valid][23][UTT][ MSE = 0.2672 | LCC = 0.8460 | SRCC = 0.8458 ]
123
+ INFO:root:
124
+ [Valid][23][SYS][ MSE = 0.1678 | LCC = 0.9135 | SRCC = 0.9100 ]
125
+ INFO:root:Average loss=0.19168832009801498
126
+ INFO:root:
127
+ [Valid][24][UTT][ MSE = 0.2609 | LCC = 0.8521 | SRCC = 0.8501 ]
128
+ INFO:root:
129
+ [Valid][24][SYS][ MSE = 0.1466 | LCC = 0.9210 | SRCC = 0.9134 ]
130
+ INFO:root:Average loss=0.18861015102802178
131
+ INFO:root:
132
+ [Valid][25][UTT][ MSE = 0.2697 | LCC = 0.8510 | SRCC = 0.8511 ]
133
+ INFO:root:
134
+ [Valid][25][SYS][ MSE = 0.1404 | LCC = 0.9210 | SRCC = 0.9152 ]
135
+ INFO:root:Average loss=0.17494672307601342
136
+ INFO:root:
137
+ [Valid][26][UTT][ MSE = 0.2967 | LCC = 0.8508 | SRCC = 0.8511 ]
138
+ INFO:root:
139
+ [Valid][26][SYS][ MSE = 0.1617 | LCC = 0.9181 | SRCC = 0.9122 ]
140
+ INFO:root:Average loss=0.19273283638251135
141
+ INFO:root:
142
+ [Valid][27][UTT][ MSE = 0.3081 | LCC = 0.8413 | SRCC = 0.8424 ]
143
+ INFO:root:
144
+ [Valid][27][SYS][ MSE = 0.2173 | LCC = 0.9092 | SRCC = 0.9051 ]
145
+ INFO:root:Average loss=0.16662715948545015
146
+ INFO:root:
147
+ [Valid][28][UTT][ MSE = 0.2756 | LCC = 0.8517 | SRCC = 0.8503 ]
148
+ INFO:root:
149
+ [Valid][28][SYS][ MSE = 0.1756 | LCC = 0.9223 | SRCC = 0.9152 ]
150
+ INFO:root:Average loss=0.15800289943432197
151
+ INFO:root:
152
+ [Valid][29][UTT][ MSE = 0.2639 | LCC = 0.8480 | SRCC = 0.8473 ]
153
+ INFO:root:
154
+ [Valid][29][SYS][ MSE = 0.1819 | LCC = 0.9090 | SRCC = 0.9048 ]
155
+ INFO:root:Average loss=0.15854160783764643
156
+ INFO:root:
157
+ [Valid][30][UTT][ MSE = 0.2453 | LCC = 0.8465 | SRCC = 0.8471 ]
158
+ INFO:root:
159
+ [Valid][30][SYS][ MSE = 0.1411 | LCC = 0.9131 | SRCC = 0.9123 ]
160
+ INFO:root:
161
+ [Test][31][UTT][ MSE = 0.2646 | LCC = 0.8492 | SRCC = 0.8482 ]
162
+ INFO:root:
163
+ [Test][31][SYS][ MSE = 0.1480 | LCC = 0.9267 | SRCC = 0.9284 ]
results_bvcc/w2v2_base/layer10/run2/config.gin ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Training config on BVCC dataset.
2
+
3
+ # ------------------------------
4
+ # Architecture
5
+ # ------------------------------
6
+ ProjectionHead:
7
+ conv_channels = (32, 32)
8
+ dense_neurons = (64, 32, 1)
9
+ use_poolings = (True, True)
10
+ kernel_size = 5
11
+ pool_size = 5
12
+
13
+ # ------------------------------
14
+ # Dataset
15
+ # ------------------------------
16
+ Bvcc:
17
+ debug = False
18
+
19
+ get_dataloader:
20
+ num_workers = 12
21
+ shuffle = True
22
+
23
+ # ------------------------------
24
+ # Training loop
25
+ # ------------------------------
26
+ TrainingLoop:
27
+ num_epochs = 30
28
+ learning_rate = 1e-4
29
+ dataset_cls = @Bvcc
30
+ weight_decay = 0.0
31
+ batch_size_train = 64
results_bvcc/w2v2_base/layer10/run2/model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e33ada00c9dfc1300d134ec03d9ddded0705ab909eb5090ab0661e2e5e3aa644
3
+ size 2747979
results_bvcc/w2v2_base/layer10/run2/model_best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1e9d94a3abab797f3aee2f89678a15c9349c120ecbbd8a45c6e6891c822cff0f
3
+ size 2749702
results_bvcc/w2v2_base/layer10/run2/train.log ADDED
@@ -0,0 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ INFO:root:Num train speech clips: 4974
2
+ INFO:root:Num val speech clips: 1066
3
+ INFO:root:Num test speech clips: 1066
4
+ INFO:root:Device=cuda
5
+ INFO:root:Average loss=0.7398962053732995
6
+ INFO:root:
7
+ [Valid][0][UTT][ MSE = 0.5114 | LCC = 0.6916 | SRCC = 0.6911 ]
8
+ INFO:root:
9
+ [Valid][0][SYS][ MSE = 0.4090 | LCC = 0.7632 | SRCC = 0.7634 ]
10
+ INFO:root:Average loss=0.37356871232772487
11
+ INFO:root:
12
+ [Valid][1][UTT][ MSE = 0.3672 | LCC = 0.7884 | SRCC = 0.7919 ]
13
+ INFO:root:
14
+ [Valid][1][SYS][ MSE = 0.2653 | LCC = 0.8618 | SRCC = 0.8618 ]
15
+ INFO:root:Average loss=0.3334188083043465
16
+ INFO:root:
17
+ [Valid][2][UTT][ MSE = 0.3171 | LCC = 0.8014 | SRCC = 0.8030 ]
18
+ INFO:root:
19
+ [Valid][2][SYS][ MSE = 0.2022 | LCC = 0.8692 | SRCC = 0.8707 ]
20
+ INFO:root:Average loss=0.3181392959295175
21
+ INFO:root:
22
+ [Valid][3][UTT][ MSE = 0.3506 | LCC = 0.8125 | SRCC = 0.8092 ]
23
+ INFO:root:
24
+ [Valid][3][SYS][ MSE = 0.2285 | LCC = 0.8834 | SRCC = 0.8795 ]
25
+ INFO:root:Average loss=0.2956486004285323
26
+ INFO:root:
27
+ [Valid][4][UTT][ MSE = 0.3246 | LCC = 0.8153 | SRCC = 0.8148 ]
28
+ INFO:root:
29
+ [Valid][4][SYS][ MSE = 0.2220 | LCC = 0.8831 | SRCC = 0.8801 ]
30
+ INFO:root:Average loss=0.2900301647874025
31
+ INFO:root:
32
+ [Valid][5][UTT][ MSE = 0.3632 | LCC = 0.8221 | SRCC = 0.8200 ]
33
+ INFO:root:
34
+ [Valid][5][SYS][ MSE = 0.2366 | LCC = 0.8955 | SRCC = 0.8903 ]
35
+ INFO:root:Average loss=0.27746068762663084
36
+ INFO:root:
37
+ [Valid][6][UTT][ MSE = 0.3168 | LCC = 0.8305 | SRCC = 0.8286 ]
38
+ INFO:root:
39
+ [Valid][6][SYS][ MSE = 0.2027 | LCC = 0.8982 | SRCC = 0.8937 ]
40
+ INFO:root:Average loss=0.2560170080799323
41
+ INFO:root:
42
+ [Valid][7][UTT][ MSE = 0.2914 | LCC = 0.8352 | SRCC = 0.8338 ]
43
+ INFO:root:
44
+ [Valid][7][SYS][ MSE = 0.1847 | LCC = 0.9009 | SRCC = 0.8981 ]
45
+ INFO:root:Average loss=0.2501956919829051
46
+ INFO:root:
47
+ [Valid][8][UTT][ MSE = 0.4080 | LCC = 0.8264 | SRCC = 0.8285 ]
48
+ INFO:root:
49
+ [Valid][8][SYS][ MSE = 0.3328 | LCC = 0.8892 | SRCC = 0.8856 ]
50
+ INFO:root:Average loss=0.24749419876398185
51
+ INFO:root:
52
+ [Valid][9][UTT][ MSE = 0.4722 | LCC = 0.8236 | SRCC = 0.8263 ]
53
+ INFO:root:
54
+ [Valid][9][SYS][ MSE = 0.4021 | LCC = 0.8900 | SRCC = 0.8886 ]
55
+ INFO:root:Average loss=0.24561113271957788
56
+ INFO:root:
57
+ [Valid][10][UTT][ MSE = 0.3281 | LCC = 0.8385 | SRCC = 0.8402 ]
58
+ INFO:root:
59
+ [Valid][10][SYS][ MSE = 0.2398 | LCC = 0.9049 | SRCC = 0.9064 ]
60
+ INFO:root:Average loss=0.2400172662276488
61
+ INFO:root:
62
+ [Valid][11][UTT][ MSE = 0.3102 | LCC = 0.8428 | SRCC = 0.8429 ]
63
+ INFO:root:
64
+ [Valid][11][SYS][ MSE = 0.2185 | LCC = 0.9077 | SRCC = 0.9060 ]
65
+ INFO:root:Average loss=0.23402776301671296
66
+ INFO:root:
67
+ [Valid][12][UTT][ MSE = 0.3086 | LCC = 0.8419 | SRCC = 0.8401 ]
68
+ INFO:root:
69
+ [Valid][12][SYS][ MSE = 0.2166 | LCC = 0.9024 | SRCC = 0.8960 ]
70
+ INFO:root:Average loss=0.21971334545658186
71
+ INFO:root:
72
+ [Valid][13][UTT][ MSE = 0.3033 | LCC = 0.8444 | SRCC = 0.8451 ]
73
+ INFO:root:
74
+ [Valid][13][SYS][ MSE = 0.2220 | LCC = 0.9051 | SRCC = 0.9021 ]
75
+ INFO:root:Average loss=0.23659904224750322
76
+ INFO:root:
77
+ [Valid][14][UTT][ MSE = 0.3431 | LCC = 0.8426 | SRCC = 0.8446 ]
78
+ INFO:root:
79
+ [Valid][14][SYS][ MSE = 0.2683 | LCC = 0.9018 | SRCC = 0.9008 ]
80
+ INFO:root:Average loss=0.23464051614969206
81
+ INFO:root:
82
+ [Valid][15][UTT][ MSE = 0.2890 | LCC = 0.8491 | SRCC = 0.8482 ]
83
+ INFO:root:
84
+ [Valid][15][SYS][ MSE = 0.1912 | LCC = 0.9092 | SRCC = 0.9054 ]
85
+ INFO:root:Average loss=0.21507561799043265
86
+ INFO:root:
87
+ [Valid][16][UTT][ MSE = 0.2793 | LCC = 0.8515 | SRCC = 0.8499 ]
88
+ INFO:root:
89
+ [Valid][16][SYS][ MSE = 0.1891 | LCC = 0.9127 | SRCC = 0.9065 ]
90
+ INFO:root:Average loss=0.21147131709716258
91
+ INFO:root:
92
+ [Valid][17][UTT][ MSE = 0.2529 | LCC = 0.8499 | SRCC = 0.8496 ]
93
+ INFO:root:
94
+ [Valid][17][SYS][ MSE = 0.1409 | LCC = 0.9103 | SRCC = 0.9051 ]
95
+ INFO:root:Average loss=0.21192121945130518
96
+ INFO:root:
97
+ [Valid][18][UTT][ MSE = 0.2614 | LCC = 0.8458 | SRCC = 0.8458 ]
98
+ INFO:root:
99
+ [Valid][18][SYS][ MSE = 0.1423 | LCC = 0.9126 | SRCC = 0.9092 ]
100
+ INFO:root:Average loss=0.19656659729587725
101
+ INFO:root:
102
+ [Valid][19][UTT][ MSE = 0.2669 | LCC = 0.8525 | SRCC = 0.8516 ]
103
+ INFO:root:
104
+ [Valid][19][SYS][ MSE = 0.1671 | LCC = 0.9199 | SRCC = 0.9171 ]
105
+ INFO:root:Average loss=0.2072015079932335
106
+ INFO:root:
107
+ [Valid][20][UTT][ MSE = 0.2580 | LCC = 0.8494 | SRCC = 0.8492 ]
108
+ INFO:root:
109
+ [Valid][20][SYS][ MSE = 0.1512 | LCC = 0.9145 | SRCC = 0.9120 ]
110
+ INFO:root:Average loss=0.18787010644490904
111
+ INFO:root:
112
+ [Valid][21][UTT][ MSE = 0.2558 | LCC = 0.8537 | SRCC = 0.8530 ]
113
+ INFO:root:
114
+ [Valid][21][SYS][ MSE = 0.1398 | LCC = 0.9229 | SRCC = 0.9196 ]
115
+ INFO:root:Average loss=0.18262528609006834
116
+ INFO:root:
117
+ [Valid][22][UTT][ MSE = 0.3109 | LCC = 0.8497 | SRCC = 0.8504 ]
118
+ INFO:root:
119
+ [Valid][22][SYS][ MSE = 0.2268 | LCC = 0.9188 | SRCC = 0.9193 ]
120
+ INFO:root:Average loss=0.18379567716366205
121
+ INFO:root:
122
+ [Valid][23][UTT][ MSE = 0.2671 | LCC = 0.8502 | SRCC = 0.8511 ]
123
+ INFO:root:
124
+ [Valid][23][SYS][ MSE = 0.1596 | LCC = 0.9184 | SRCC = 0.9178 ]
125
+ INFO:root:Average loss=0.16958561491889831
126
+ INFO:root:
127
+ [Valid][24][UTT][ MSE = 0.3290 | LCC = 0.8524 | SRCC = 0.8527 ]
128
+ INFO:root:
129
+ [Valid][24][SYS][ MSE = 0.2510 | LCC = 0.9205 | SRCC = 0.9204 ]
130
+ INFO:root:Average loss=0.17480431143672037
131
+ INFO:root:
132
+ [Valid][25][UTT][ MSE = 0.2704 | LCC = 0.8578 | SRCC = 0.8573 ]
133
+ INFO:root:
134
+ [Valid][25][SYS][ MSE = 0.1756 | LCC = 0.9267 | SRCC = 0.9243 ]
135
+ INFO:root:Average loss=0.16138773172711715
136
+ INFO:root:
137
+ [Valid][26][UTT][ MSE = 0.2760 | LCC = 0.8530 | SRCC = 0.8534 ]
138
+ INFO:root:
139
+ [Valid][26][SYS][ MSE = 0.1867 | LCC = 0.9202 | SRCC = 0.9194 ]
140
+ INFO:root:Average loss=0.17517884572347006
141
+ INFO:root:
142
+ [Valid][27][UTT][ MSE = 0.2612 | LCC = 0.8536 | SRCC = 0.8533 ]
143
+ INFO:root:
144
+ [Valid][27][SYS][ MSE = 0.1487 | LCC = 0.9164 | SRCC = 0.9122 ]
145
+ INFO:root:Average loss=0.1903034653992225
146
+ INFO:root:
147
+ [Valid][28][UTT][ MSE = 0.2716 | LCC = 0.8538 | SRCC = 0.8544 ]
148
+ INFO:root:
149
+ [Valid][28][SYS][ MSE = 0.1490 | LCC = 0.9221 | SRCC = 0.9188 ]
150
+ INFO:root:Average loss=0.16389165293329802
151
+ INFO:root:
152
+ [Valid][29][UTT][ MSE = 0.3065 | LCC = 0.8502 | SRCC = 0.8510 ]
153
+ INFO:root:
154
+ [Valid][29][SYS][ MSE = 0.2208 | LCC = 0.9189 | SRCC = 0.9184 ]
155
+ INFO:root:Average loss=0.15112435368773264
156
+ INFO:root:
157
+ [Valid][30][UTT][ MSE = 0.2571 | LCC = 0.8564 | SRCC = 0.8551 ]
158
+ INFO:root:
159
+ [Valid][30][SYS][ MSE = 0.1691 | LCC = 0.9202 | SRCC = 0.9159 ]
160
+ INFO:root:
161
+ [Test][31][UTT][ MSE = 0.2634 | LCC = 0.8483 | SRCC = 0.8472 ]
162
+ INFO:root:
163
+ [Test][31][SYS][ MSE = 0.1228 | LCC = 0.9229 | SRCC = 0.9240 ]
results_bvcc/w2v2_base/layer10/run3/config.gin ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Training config on BVCC dataset.
2
+
3
+ # ------------------------------
4
+ # Architecture
5
+ # ------------------------------
6
+ ProjectionHead:
7
+ conv_channels = (32, 32)
8
+ dense_neurons = (64, 32, 1)
9
+ use_poolings = (True, True)
10
+ kernel_size = 5
11
+ pool_size = 5
12
+
13
+ # ------------------------------
14
+ # Dataset
15
+ # ------------------------------
16
+ Bvcc:
17
+ debug = False
18
+
19
+ get_dataloader:
20
+ num_workers = 12
21
+ shuffle = True
22
+
23
+ # ------------------------------
24
+ # Training loop
25
+ # ------------------------------
26
+ TrainingLoop:
27
+ num_epochs = 30
28
+ learning_rate = 1e-4
29
+ dataset_cls = @Bvcc
30
+ weight_decay = 0.0
31
+ batch_size_train = 64
results_bvcc/w2v2_base/layer10/run3/model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e1f35fa31a5b15a56dd2b81dfe4855db0a5288f1d261075e354090f21f607f6
3
+ size 2747979
results_bvcc/w2v2_base/layer10/run3/model_best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:de2c91977e3a45e25c04d693e1efdad18f4a4ebf787fc846577d2a8ffc126cba
3
+ size 2749638
results_bvcc/w2v2_base/layer10/run3/train.log ADDED
@@ -0,0 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ INFO:root:Num train speech clips: 4974
2
+ INFO:root:Num val speech clips: 1066
3
+ INFO:root:Num test speech clips: 1066
4
+ INFO:root:Device=cuda
5
+ INFO:root:Average loss=0.6151072016129127
6
+ INFO:root:
7
+ [Valid][0][UTT][ MSE = 0.4204 | LCC = 0.7370 | SRCC = 0.7415 ]
8
+ INFO:root:
9
+ [Valid][0][SYS][ MSE = 0.2999 | LCC = 0.8171 | SRCC = 0.8177 ]
10
+ INFO:root:Average loss=0.38759309473710185
11
+ INFO:root:
12
+ [Valid][1][UTT][ MSE = 0.3523 | LCC = 0.7804 | SRCC = 0.7795 ]
13
+ INFO:root:
14
+ [Valid][1][SYS][ MSE = 0.2224 | LCC = 0.8444 | SRCC = 0.8431 ]
15
+ INFO:root:Average loss=0.3734013302586017
16
+ INFO:root:
17
+ [Valid][2][UTT][ MSE = 0.3707 | LCC = 0.7873 | SRCC = 0.7909 ]
18
+ INFO:root:
19
+ [Valid][2][SYS][ MSE = 0.2624 | LCC = 0.8608 | SRCC = 0.8615 ]
20
+ INFO:root:Average loss=0.3029930610687305
21
+ INFO:root:
22
+ [Valid][3][UTT][ MSE = 0.3446 | LCC = 0.8111 | SRCC = 0.8094 ]
23
+ INFO:root:
24
+ [Valid][3][SYS][ MSE = 0.2205 | LCC = 0.8816 | SRCC = 0.8775 ]
25
+ INFO:root:Average loss=0.3115337605659778
26
+ INFO:root:
27
+ [Valid][4][UTT][ MSE = 0.4950 | LCC = 0.7943 | SRCC = 0.8059 ]
28
+ INFO:root:
29
+ [Valid][4][SYS][ MSE = 0.4047 | LCC = 0.8674 | SRCC = 0.8729 ]
30
+ INFO:root:Average loss=0.2675825441494966
31
+ INFO:root:
32
+ [Valid][5][UTT][ MSE = 0.2955 | LCC = 0.8212 | SRCC = 0.8203 ]
33
+ INFO:root:
34
+ [Valid][5][SYS][ MSE = 0.1821 | LCC = 0.8854 | SRCC = 0.8803 ]
35
+ INFO:root:Average loss=0.2922194231397066
36
+ INFO:root:
37
+ [Valid][6][UTT][ MSE = 0.3622 | LCC = 0.8166 | SRCC = 0.8248 ]
38
+ INFO:root:
39
+ [Valid][6][SYS][ MSE = 0.2780 | LCC = 0.8809 | SRCC = 0.8855 ]
40
+ INFO:root:Average loss=0.2934482196966807
41
+ INFO:root:
42
+ [Valid][7][UTT][ MSE = 0.4435 | LCC = 0.8111 | SRCC = 0.8185 ]
43
+ INFO:root:
44
+ [Valid][7][SYS][ MSE = 0.3622 | LCC = 0.8790 | SRCC = 0.8799 ]
45
+ INFO:root:Average loss=0.2818548060380496
46
+ INFO:root:
47
+ [Valid][8][UTT][ MSE = 0.3128 | LCC = 0.8260 | SRCC = 0.8290 ]
48
+ INFO:root:
49
+ [Valid][8][SYS][ MSE = 0.2179 | LCC = 0.8915 | SRCC = 0.8921 ]
50
+ INFO:root:Average loss=0.24064893084458816
51
+ INFO:root:
52
+ [Valid][9][UTT][ MSE = 0.2917 | LCC = 0.8292 | SRCC = 0.8278 ]
53
+ INFO:root:
54
+ [Valid][9][SYS][ MSE = 0.1780 | LCC = 0.8979 | SRCC = 0.8924 ]
55
+ INFO:root:Average loss=0.242812563593571
56
+ INFO:root:
57
+ [Valid][10][UTT][ MSE = 0.2794 | LCC = 0.8348 | SRCC = 0.8349 ]
58
+ INFO:root:
59
+ [Valid][10][SYS][ MSE = 0.1536 | LCC = 0.9036 | SRCC = 0.8996 ]
60
+ INFO:root:Average loss=0.23124572883049646
61
+ INFO:root:
62
+ [Valid][11][UTT][ MSE = 0.3585 | LCC = 0.8292 | SRCC = 0.8331 ]
63
+ INFO:root:
64
+ [Valid][11][SYS][ MSE = 0.2724 | LCC = 0.8929 | SRCC = 0.8921 ]
65
+ INFO:root:Average loss=0.23634144434562096
66
+ INFO:root:
67
+ [Valid][12][UTT][ MSE = 0.3049 | LCC = 0.8325 | SRCC = 0.8369 ]
68
+ INFO:root:
69
+ [Valid][12][SYS][ MSE = 0.2208 | LCC = 0.8946 | SRCC = 0.8953 ]
70
+ INFO:root:Average loss=0.25324304974996126
71
+ INFO:root:
72
+ [Valid][13][UTT][ MSE = 0.3333 | LCC = 0.8284 | SRCC = 0.8334 ]
73
+ INFO:root:
74
+ [Valid][13][SYS][ MSE = 0.2473 | LCC = 0.8942 | SRCC = 0.8948 ]
75
+ INFO:root:Average loss=0.22296481732374582
76
+ INFO:root:
77
+ [Valid][14][UTT][ MSE = 0.3288 | LCC = 0.8310 | SRCC = 0.8329 ]
78
+ INFO:root:
79
+ [Valid][14][SYS][ MSE = 0.2406 | LCC = 0.8992 | SRCC = 0.8979 ]
80
+ INFO:root:Average loss=0.2211699033012757
81
+ INFO:root:
82
+ [Valid][15][UTT][ MSE = 0.3366 | LCC = 0.8352 | SRCC = 0.8372 ]
83
+ INFO:root:
84
+ [Valid][15][SYS][ MSE = 0.2561 | LCC = 0.8978 | SRCC = 0.8955 ]
85
+ INFO:root:Average loss=0.22075015917802468
86
+ INFO:root:
87
+ [Valid][16][UTT][ MSE = 0.2678 | LCC = 0.8435 | SRCC = 0.8434 ]
88
+ INFO:root:
89
+ [Valid][16][SYS][ MSE = 0.1686 | LCC = 0.9088 | SRCC = 0.9047 ]
90
+ INFO:root:Average loss=0.20845631261666617
91
+ INFO:root:
92
+ [Valid][17][UTT][ MSE = 0.2795 | LCC = 0.8428 | SRCC = 0.8424 ]
93
+ INFO:root:
94
+ [Valid][17][SYS][ MSE = 0.1774 | LCC = 0.9096 | SRCC = 0.9050 ]
95
+ INFO:root:Average loss=0.21116518544463012
96
+ INFO:root:
97
+ [Valid][18][UTT][ MSE = 0.2620 | LCC = 0.8462 | SRCC = 0.8460 ]
98
+ INFO:root:
99
+ [Valid][18][SYS][ MSE = 0.1636 | LCC = 0.9117 | SRCC = 0.9080 ]
100
+ INFO:root:Average loss=0.19843919689838702
101
+ INFO:root:
102
+ [Valid][19][UTT][ MSE = 0.2834 | LCC = 0.8428 | SRCC = 0.8439 ]
103
+ INFO:root:
104
+ [Valid][19][SYS][ MSE = 0.1841 | LCC = 0.9136 | SRCC = 0.9119 ]
105
+ INFO:root:Average loss=0.19134420423935622
106
+ INFO:root:
107
+ [Valid][20][UTT][ MSE = 0.2904 | LCC = 0.8457 | SRCC = 0.8455 ]
108
+ INFO:root:
109
+ [Valid][20][SYS][ MSE = 0.1989 | LCC = 0.9120 | SRCC = 0.9076 ]
110
+ INFO:root:Average loss=0.20944310189821783
111
+ INFO:root:
112
+ [Valid][21][UTT][ MSE = 0.2646 | LCC = 0.8465 | SRCC = 0.8490 ]
113
+ INFO:root:
114
+ [Valid][21][SYS][ MSE = 0.1666 | LCC = 0.9136 | SRCC = 0.9139 ]
115
+ INFO:root:Average loss=0.19671830009573546
116
+ INFO:root:
117
+ [Valid][22][UTT][ MSE = 0.2879 | LCC = 0.8445 | SRCC = 0.8469 ]
118
+ INFO:root:
119
+ [Valid][22][SYS][ MSE = 0.2027 | LCC = 0.9099 | SRCC = 0.9105 ]
120
+ INFO:root:Average loss=0.1823623925447464
121
+ INFO:root:
122
+ [Valid][23][UTT][ MSE = 0.2827 | LCC = 0.8465 | SRCC = 0.8485 ]
123
+ INFO:root:
124
+ [Valid][23][SYS][ MSE = 0.1454 | LCC = 0.9202 | SRCC = 0.9189 ]
125
+ INFO:root:Average loss=0.19276750651307595
126
+ INFO:root:
127
+ [Valid][24][UTT][ MSE = 0.2823 | LCC = 0.8405 | SRCC = 0.8389 ]
128
+ INFO:root:
129
+ [Valid][24][SYS][ MSE = 0.1512 | LCC = 0.9130 | SRCC = 0.9092 ]
130
+ INFO:root:Average loss=0.17636762196436906
131
+ INFO:root:
132
+ [Valid][25][UTT][ MSE = 0.2556 | LCC = 0.8436 | SRCC = 0.8448 ]
133
+ INFO:root:
134
+ [Valid][25][SYS][ MSE = 0.1602 | LCC = 0.9090 | SRCC = 0.9076 ]
135
+ INFO:root:Average loss=0.20817641866130707
136
+ INFO:root:
137
+ [Valid][26][UTT][ MSE = 0.2682 | LCC = 0.8417 | SRCC = 0.8437 ]
138
+ INFO:root:
139
+ [Valid][26][SYS][ MSE = 0.1759 | LCC = 0.9074 | SRCC = 0.9057 ]
140
+ INFO:root:Average loss=0.16617436869404253
141
+ INFO:root:
142
+ [Valid][27][UTT][ MSE = 0.2511 | LCC = 0.8464 | SRCC = 0.8461 ]
143
+ INFO:root:
144
+ [Valid][27][SYS][ MSE = 0.1193 | LCC = 0.9229 | SRCC = 0.9178 ]
145
+ INFO:root:Average loss=0.16740939957209122
146
+ INFO:root:
147
+ [Valid][28][UTT][ MSE = 0.2690 | LCC = 0.8440 | SRCC = 0.8437 ]
148
+ INFO:root:
149
+ [Valid][28][SYS][ MSE = 0.1364 | LCC = 0.9144 | SRCC = 0.9107 ]
150
+ INFO:root:Average loss=0.16532027176939523
151
+ INFO:root:
152
+ [Valid][29][UTT][ MSE = 0.3082 | LCC = 0.8402 | SRCC = 0.8406 ]
153
+ INFO:root:
154
+ [Valid][29][SYS][ MSE = 0.2247 | LCC = 0.9109 | SRCC = 0.9082 ]
155
+ INFO:root:Average loss=0.15212263835546297
156
+ INFO:root:
157
+ [Valid][30][UTT][ MSE = 0.2527 | LCC = 0.8465 | SRCC = 0.8445 ]
158
+ INFO:root:
159
+ [Valid][30][SYS][ MSE = 0.1419 | LCC = 0.9131 | SRCC = 0.9048 ]
160
+ INFO:root:
161
+ [Test][31][UTT][ MSE = 0.2515 | LCC = 0.8470 | SRCC = 0.8473 ]
162
+ INFO:root:
163
+ [Test][31][SYS][ MSE = 0.1390 | LCC = 0.9138 | SRCC = 0.9160 ]
results_bvcc/w2v2_base/layer10/run4/config.gin ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Training config on BVCC dataset.
2
+
3
+ # ------------------------------
4
+ # Architecture
5
+ # ------------------------------
6
+ ProjectionHead:
7
+ conv_channels = (32, 32)
8
+ dense_neurons = (64, 32, 1)
9
+ use_poolings = (True, True)
10
+ kernel_size = 5
11
+ pool_size = 5
12
+
13
+ # ------------------------------
14
+ # Dataset
15
+ # ------------------------------
16
+ Bvcc:
17
+ debug = False
18
+
19
+ get_dataloader:
20
+ num_workers = 12
21
+ shuffle = True
22
+
23
+ # ------------------------------
24
+ # Training loop
25
+ # ------------------------------
26
+ TrainingLoop:
27
+ num_epochs = 30
28
+ learning_rate = 1e-4
29
+ dataset_cls = @Bvcc
30
+ weight_decay = 0.0
31
+ batch_size_train = 64
results_bvcc/w2v2_base/layer10/run4/model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8233559288f0e7cce5c389b0d98eff5da9230888a799b95da497236000e1a818
3
+ size 2747979
results_bvcc/w2v2_base/layer10/run4/model_best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4568757eef5a6b5eea5b8fcbb5aac40f903c07142d42636a57d741fcc45125b1
3
+ size 2749702