Upload SSL Layer MOS checkpoints (part 15)

results_tencent/hubert_base/layer3/run4/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:631e2d40a5cd386ac92c257983f300df8ed14132d20eb6194a657d9320710394
+size 2749638

results_tencent/hubert_base/layer3/run4/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 8000
+INFO:root:Num val speech clips: 2000
+INFO:root:Num test speech clips: 1563
+INFO:root:Device=cuda
+INFO:root:Average loss=0.585044427037239
+INFO:root:
+[Valid][0][UTT][ MSE = 0.3097 | LCC = 0.9209 | SRCC = 0.9227 ]
+INFO:root:Average loss=0.18621259391307832
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2580 | LCC = 0.9450 | SRCC = 0.9464 ]
+INFO:root:Average loss=0.15076538735628128
+INFO:root:
+[Valid][2][UTT][ MSE = 0.1899 | LCC = 0.9496 | SRCC = 0.9501 ]
+INFO:root:Average loss=0.12869137960672378
+INFO:root:
+[Valid][3][UTT][ MSE = 0.1957 | LCC = 0.9526 | SRCC = 0.9530 ]
+INFO:root:Average loss=0.11905399978160858
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2246 | LCC = 0.9558 | SRCC = 0.9565 ]
+INFO:root:Average loss=0.1064041962325573
+INFO:root:
+[Valid][5][UTT][ MSE = 0.1910 | LCC = 0.9558 | SRCC = 0.9565 ]
+INFO:root:Average loss=0.10023876416683197
+INFO:root:
+[Valid][6][UTT][ MSE = 0.1964 | LCC = 0.9565 | SRCC = 0.9569 ]
+INFO:root:Average loss=0.09155132782459259
+INFO:root:
+[Valid][7][UTT][ MSE = 0.2142 | LCC = 0.9591 | SRCC = 0.9595 ]
+INFO:root:Average loss=0.08545256182551383
+INFO:root:
+[Valid][8][UTT][ MSE = 0.1806 | LCC = 0.9592 | SRCC = 0.9593 ]
+INFO:root:Average loss=0.08068941643834114
+INFO:root:
+[Valid][9][UTT][ MSE = 0.1953 | LCC = 0.9610 | SRCC = 0.9615 ]
+INFO:root:Average loss=0.08015629923343659
+INFO:root:
+[Valid][10][UTT][ MSE = 0.1945 | LCC = 0.9603 | SRCC = 0.9610 ]
+INFO:root:Average loss=0.07107420200109482
+INFO:root:
+[Valid][11][UTT][ MSE = 0.1723 | LCC = 0.9613 | SRCC = 0.9613 ]
+INFO:root:Average loss=0.0715145302414894
+INFO:root:
+[Valid][12][UTT][ MSE = 0.1820 | LCC = 0.9609 | SRCC = 0.9610 ]
+INFO:root:Average loss=0.06508512896299362
+INFO:root:
+[Valid][13][UTT][ MSE = 0.1582 | LCC = 0.9612 | SRCC = 0.9619 ]
+INFO:root:Average loss=0.06264670187234879
+INFO:root:
+[Valid][14][UTT][ MSE = 0.1757 | LCC = 0.9610 | SRCC = 0.9613 ]
+INFO:root:Average loss=0.05488259424269199
+INFO:root:
+[Valid][15][UTT][ MSE = 0.1552 | LCC = 0.9616 | SRCC = 0.9616 ]
+INFO:root:Average loss=0.05504263737797737
+INFO:root:
+[Valid][16][UTT][ MSE = 0.1690 | LCC = 0.9613 | SRCC = 0.9617 ]
+INFO:root:Average loss=0.054654211789369585
+INFO:root:
+[Valid][17][UTT][ MSE = 0.1705 | LCC = 0.9627 | SRCC = 0.9627 ]
+INFO:root:Average loss=0.050548502132296565
+INFO:root:
+[Valid][18][UTT][ MSE = 0.1641 | LCC = 0.9611 | SRCC = 0.9610 ]
+INFO:root:Average loss=0.052234662771224974
+INFO:root:
+[Valid][19][UTT][ MSE = 0.1602 | LCC = 0.9628 | SRCC = 0.9629 ]
+INFO:root:Average loss=0.047292057484388354
+INFO:root:
+[Valid][20][UTT][ MSE = 0.1571 | LCC = 0.9621 | SRCC = 0.9625 ]
+INFO:root:Average loss=0.042859895259141925
+INFO:root:
+[Valid][21][UTT][ MSE = 0.1713 | LCC = 0.9601 | SRCC = 0.9602 ]
+INFO:root:Average loss=0.04676552534103393
+INFO:root:
+[Valid][22][UTT][ MSE = 0.1733 | LCC = 0.9626 | SRCC = 0.9625 ]
+INFO:root:Average loss=0.04063004079461098
+INFO:root:
+[Valid][23][UTT][ MSE = 0.1769 | LCC = 0.9635 | SRCC = 0.9635 ]
+INFO:root:Average loss=0.04098973496258259
+INFO:root:
+[Valid][24][UTT][ MSE = 0.1704 | LCC = 0.9624 | SRCC = 0.9622 ]
+INFO:root:Average loss=0.038265125632286075
+INFO:root:
+[Valid][25][UTT][ MSE = 0.1553 | LCC = 0.9628 | SRCC = 0.9626 ]
+INFO:root:Average loss=0.03658791047334671
+INFO:root:
+[Valid][26][UTT][ MSE = 0.1662 | LCC = 0.9628 | SRCC = 0.9629 ]
+INFO:root:Average loss=0.035070098757743834
+INFO:root:
+[Valid][27][UTT][ MSE = 0.1590 | LCC = 0.9631 | SRCC = 0.9630 ]
+INFO:root:Average loss=0.03913343170285225
+INFO:root:
+[Valid][28][UTT][ MSE = 0.1528 | LCC = 0.9640 | SRCC = 0.9638 ]
+INFO:root:Average loss=0.03600745689868927
+INFO:root:
+[Valid][29][UTT][ MSE = 0.1657 | LCC = 0.9620 | SRCC = 0.9622 ]
+INFO:root:Average loss=0.032543361201882365
+INFO:root:
+[Valid][30][UTT][ MSE = 0.1691 | LCC = 0.9631 | SRCC = 0.9630 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.1539 | LCC = 0.9657 | SRCC = 0.9658 ]

results_tencent/hubert_base/layer4/run0/config.gin

@@ -0,0 +1,31 @@
+# Training config on BVCC dataset.
+
+# ------------------------------
+# Architecture
+# ------------------------------
+ProjectionHead:
+    conv_channels = (32, 32)
+    dense_neurons = (64, 32, 1)
+    use_poolings = (True, True)
+    kernel_size = 5
+    pool_size = 5
+    
+# ------------------------------
+# Dataset
+# ------------------------------
+Tencent:
+    debug = False
+
+get_dataloader:
+    num_workers = 12
+    shuffle = True
+
+# ------------------------------
+# Training loop
+# ------------------------------
+TrainingLoop:
+    num_epochs = 30
+    learning_rate = 1e-4
+    dataset_cls = @Tencent
+    weight_decay = 0.0
+    batch_size_train = 64

results_tencent/hubert_base/layer4/run0/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b5072cbdf0d4aa98d18c580e97f75346cea1701ce367ccdfd9fd3742bddd0cc8
+size 2747979

results_tencent/hubert_base/layer4/run0/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d29521f53773706fa046312d38bdbc0dd52dc65e0b759397171c6cef8db264f
+size 2749638

results_tencent/hubert_base/layer4/run0/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 8000
+INFO:root:Num val speech clips: 2000
+INFO:root:Num test speech clips: 1563
+INFO:root:Device=cuda
+INFO:root:Average loss=0.4132269891500473
+INFO:root:
+[Valid][0][UTT][ MSE = 0.3494 | LCC = 0.9285 | SRCC = 0.9277 ]
+INFO:root:Average loss=0.1983527769446373
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2592 | LCC = 0.9408 | SRCC = 0.9408 ]
+INFO:root:Average loss=0.15936129015684128
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2257 | LCC = 0.9478 | SRCC = 0.9487 ]
+INFO:root:Average loss=0.13534067189693452
+INFO:root:
+[Valid][3][UTT][ MSE = 0.1982 | LCC = 0.9505 | SRCC = 0.9509 ]
+INFO:root:Average loss=0.11947703397274018
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2073 | LCC = 0.9528 | SRCC = 0.9535 ]
+INFO:root:Average loss=0.11286948585510254
+INFO:root:
+[Valid][5][UTT][ MSE = 0.1985 | LCC = 0.9563 | SRCC = 0.9572 ]
+INFO:root:Average loss=0.10170915728807449
+INFO:root:
+[Valid][6][UTT][ MSE = 0.1775 | LCC = 0.9574 | SRCC = 0.9576 ]
+INFO:root:Average loss=0.09279373687505722
+INFO:root:
+[Valid][7][UTT][ MSE = 0.1747 | LCC = 0.9582 | SRCC = 0.9583 ]
+INFO:root:Average loss=0.08608387839794159
+INFO:root:
+[Valid][8][UTT][ MSE = 0.1873 | LCC = 0.9584 | SRCC = 0.9584 ]
+INFO:root:Average loss=0.08231061291694641
+INFO:root:
+[Valid][9][UTT][ MSE = 0.1632 | LCC = 0.9590 | SRCC = 0.9597 ]
+INFO:root:Average loss=0.07583020102977753
+INFO:root:
+[Valid][10][UTT][ MSE = 0.1950 | LCC = 0.9600 | SRCC = 0.9600 ]
+INFO:root:Average loss=0.07406844142079354
+INFO:root:
+[Valid][11][UTT][ MSE = 0.1876 | LCC = 0.9601 | SRCC = 0.9600 ]
+INFO:root:Average loss=0.06619387859106064
+INFO:root:
+[Valid][12][UTT][ MSE = 0.1638 | LCC = 0.9601 | SRCC = 0.9601 ]
+INFO:root:Average loss=0.06396218147873879
+INFO:root:
+[Valid][13][UTT][ MSE = 0.1850 | LCC = 0.9614 | SRCC = 0.9612 ]
+INFO:root:Average loss=0.06071103432774544
+INFO:root:
+[Valid][14][UTT][ MSE = 0.1952 | LCC = 0.9615 | SRCC = 0.9610 ]
+INFO:root:Average loss=0.05676183819770813
+INFO:root:
+[Valid][15][UTT][ MSE = 0.1771 | LCC = 0.9616 | SRCC = 0.9614 ]
+INFO:root:Average loss=0.05342316171526909
+INFO:root:
+[Valid][16][UTT][ MSE = 0.1842 | LCC = 0.9589 | SRCC = 0.9581 ]
+INFO:root:Average loss=0.05234560041129589
+INFO:root:
+[Valid][17][UTT][ MSE = 0.1949 | LCC = 0.9610 | SRCC = 0.9602 ]
+INFO:root:Average loss=0.050308501183986665
+INFO:root:
+[Valid][18][UTT][ MSE = 0.1733 | LCC = 0.9605 | SRCC = 0.9600 ]
+INFO:root:Average loss=0.04784963104128838
+INFO:root:
+[Valid][19][UTT][ MSE = 0.1871 | LCC = 0.9606 | SRCC = 0.9605 ]
+INFO:root:Average loss=0.04462337051331997
+INFO:root:
+[Valid][20][UTT][ MSE = 0.1605 | LCC = 0.9618 | SRCC = 0.9615 ]
+INFO:root:Average loss=0.04357859367132187
+INFO:root:
+[Valid][21][UTT][ MSE = 0.1915 | LCC = 0.9620 | SRCC = 0.9618 ]
+INFO:root:Average loss=0.040906944751739505
+INFO:root:
+[Valid][22][UTT][ MSE = 0.1737 | LCC = 0.9614 | SRCC = 0.9612 ]
+INFO:root:Average loss=0.04050542619824409
+INFO:root:
+[Valid][23][UTT][ MSE = 0.1823 | LCC = 0.9619 | SRCC = 0.9612 ]
+INFO:root:Average loss=0.03871918366849422
+INFO:root:
+[Valid][24][UTT][ MSE = 0.1588 | LCC = 0.9622 | SRCC = 0.9617 ]
+INFO:root:Average loss=0.036375430226325986
+INFO:root:
+[Valid][25][UTT][ MSE = 0.1763 | LCC = 0.9623 | SRCC = 0.9617 ]
+INFO:root:Average loss=0.03660297678411007
+INFO:root:
+[Valid][26][UTT][ MSE = 0.1737 | LCC = 0.9621 | SRCC = 0.9612 ]
+INFO:root:Average loss=0.03497839273512363
+INFO:root:
+[Valid][27][UTT][ MSE = 0.1917 | LCC = 0.9624 | SRCC = 0.9626 ]
+INFO:root:Average loss=0.03361639212071896
+INFO:root:
+[Valid][28][UTT][ MSE = 0.1636 | LCC = 0.9621 | SRCC = 0.9611 ]
+INFO:root:Average loss=0.03166963692009449
+INFO:root:
+[Valid][29][UTT][ MSE = 0.1532 | LCC = 0.9626 | SRCC = 0.9624 ]
+INFO:root:Average loss=0.03618527166545391
+INFO:root:
+[Valid][30][UTT][ MSE = 0.1715 | LCC = 0.9631 | SRCC = 0.9624 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.1782 | LCC = 0.9638 | SRCC = 0.9645 ]

results_tencent/hubert_base/layer4/run1/config.gin

@@ -0,0 +1,31 @@
+# Training config on BVCC dataset.
+
+# ------------------------------
+# Architecture
+# ------------------------------
+ProjectionHead:
+    conv_channels = (32, 32)
+    dense_neurons = (64, 32, 1)
+    use_poolings = (True, True)
+    kernel_size = 5
+    pool_size = 5
+    
+# ------------------------------
+# Dataset
+# ------------------------------
+Tencent:
+    debug = False
+
+get_dataloader:
+    num_workers = 12
+    shuffle = True
+
+# ------------------------------
+# Training loop
+# ------------------------------
+TrainingLoop:
+    num_epochs = 30
+    learning_rate = 1e-4
+    dataset_cls = @Tencent
+    weight_decay = 0.0
+    batch_size_train = 64

results_tencent/hubert_base/layer4/run1/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7774f084d32dabf7c973b897e0a57f680264ad1a7e85eb0133d71abf0cc67af3
+size 2747979

results_tencent/hubert_base/layer4/run1/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cdf0a10b1af29b0ba4bc168d4fceb76de6c3f447d147387ef8f31b11da191d69
+size 2749638

results_tencent/hubert_base/layer4/run1/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 8000
+INFO:root:Num val speech clips: 2000
+INFO:root:Num test speech clips: 1563
+INFO:root:Device=cuda
+INFO:root:Average loss=0.46159341096878054
+INFO:root:
+[Valid][0][UTT][ MSE = 0.2647 | LCC = 0.9246 | SRCC = 0.9232 ]
+INFO:root:Average loss=0.17943217718601226
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2516 | LCC = 0.9424 | SRCC = 0.9412 ]
+INFO:root:Average loss=0.14303919917345048
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2039 | LCC = 0.9527 | SRCC = 0.9529 ]
+INFO:root:Average loss=0.1231202420592308
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2090 | LCC = 0.9524 | SRCC = 0.9523 ]
+INFO:root:Average loss=0.11004208463430405
+INFO:root:
+[Valid][4][UTT][ MSE = 0.1795 | LCC = 0.9580 | SRCC = 0.9575 ]
+INFO:root:Average loss=0.10258146399259567
+INFO:root:
+[Valid][5][UTT][ MSE = 0.1911 | LCC = 0.9556 | SRCC = 0.9560 ]
+INFO:root:Average loss=0.09297062692046165
+INFO:root:
+[Valid][6][UTT][ MSE = 0.1581 | LCC = 0.9597 | SRCC = 0.9599 ]
+INFO:root:Average loss=0.08695806893706322
+INFO:root:
+[Valid][7][UTT][ MSE = 0.1604 | LCC = 0.9600 | SRCC = 0.9601 ]
+INFO:root:Average loss=0.07936018177866935
+INFO:root:
+[Valid][8][UTT][ MSE = 0.1651 | LCC = 0.9610 | SRCC = 0.9611 ]
+INFO:root:Average loss=0.07519185608625412
+INFO:root:
+[Valid][9][UTT][ MSE = 0.1544 | LCC = 0.9620 | SRCC = 0.9618 ]
+INFO:root:Average loss=0.07150062566995621
+INFO:root:
+[Valid][10][UTT][ MSE = 0.1695 | LCC = 0.9620 | SRCC = 0.9619 ]
+INFO:root:Average loss=0.0688721090555191
+INFO:root:
+[Valid][11][UTT][ MSE = 0.1699 | LCC = 0.9620 | SRCC = 0.9619 ]
+INFO:root:Average loss=0.062040556102991105
+INFO:root:
+[Valid][12][UTT][ MSE = 0.1548 | LCC = 0.9621 | SRCC = 0.9617 ]
+INFO:root:Average loss=0.06254649198055268
+INFO:root:
+[Valid][13][UTT][ MSE = 0.1629 | LCC = 0.9624 | SRCC = 0.9619 ]
+INFO:root:Average loss=0.05622623857855797
+INFO:root:
+[Valid][14][UTT][ MSE = 0.1584 | LCC = 0.9627 | SRCC = 0.9622 ]
+INFO:root:Average loss=0.05377035996317864
+INFO:root:
+[Valid][15][UTT][ MSE = 0.1703 | LCC = 0.9625 | SRCC = 0.9621 ]
+INFO:root:Average loss=0.05389455099403858
+INFO:root:
+[Valid][16][UTT][ MSE = 0.1509 | LCC = 0.9626 | SRCC = 0.9623 ]
+INFO:root:Average loss=0.051428915783762935
+INFO:root:
+[Valid][17][UTT][ MSE = 0.1588 | LCC = 0.9629 | SRCC = 0.9629 ]
+INFO:root:Average loss=0.04775519397854805
+INFO:root:
+[Valid][18][UTT][ MSE = 0.1972 | LCC = 0.9626 | SRCC = 0.9625 ]
+INFO:root:Average loss=0.04895164597034454
+INFO:root:
+[Valid][19][UTT][ MSE = 0.1733 | LCC = 0.9615 | SRCC = 0.9608 ]
+INFO:root:Average loss=0.04453029118478298
+INFO:root:
+[Valid][20][UTT][ MSE = 0.1857 | LCC = 0.9628 | SRCC = 0.9622 ]
+INFO:root:Average loss=0.04112897390127182
+INFO:root:
+[Valid][21][UTT][ MSE = 0.1653 | LCC = 0.9629 | SRCC = 0.9629 ]
+INFO:root:Average loss=0.04150425037741661
+INFO:root:
+[Valid][22][UTT][ MSE = 0.1566 | LCC = 0.9629 | SRCC = 0.9627 ]
+INFO:root:Average loss=0.0384238301217556
+INFO:root:
+[Valid][23][UTT][ MSE = 0.1699 | LCC = 0.9625 | SRCC = 0.9622 ]
+INFO:root:Average loss=0.03549708844721317
+INFO:root:
+[Valid][24][UTT][ MSE = 0.1550 | LCC = 0.9628 | SRCC = 0.9627 ]
+INFO:root:Average loss=0.03578303168714046
+INFO:root:
+[Valid][25][UTT][ MSE = 0.1473 | LCC = 0.9626 | SRCC = 0.9621 ]
+INFO:root:Average loss=0.03402412061393261
+INFO:root:
+[Valid][26][UTT][ MSE = 0.1626 | LCC = 0.9614 | SRCC = 0.9611 ]
+INFO:root:Average loss=0.03447216331958771
+INFO:root:
+[Valid][27][UTT][ MSE = 0.1545 | LCC = 0.9626 | SRCC = 0.9624 ]
+INFO:root:Average loss=0.033135580465197566
+INFO:root:
+[Valid][28][UTT][ MSE = 0.1596 | LCC = 0.9632 | SRCC = 0.9625 ]
+INFO:root:Average loss=0.03470871095359326
+INFO:root:
+[Valid][29][UTT][ MSE = 0.1561 | LCC = 0.9624 | SRCC = 0.9621 ]
+INFO:root:Average loss=0.030326837614178658
+INFO:root:
+[Valid][30][UTT][ MSE = 0.1778 | LCC = 0.9636 | SRCC = 0.9631 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.1818 | LCC = 0.9636 | SRCC = 0.9646 ]

results_tencent/hubert_base/layer4/run2/config.gin

@@ -0,0 +1,31 @@
+# Training config on BVCC dataset.
+
+# ------------------------------
+# Architecture
+# ------------------------------
+ProjectionHead:
+    conv_channels = (32, 32)
+    dense_neurons = (64, 32, 1)
+    use_poolings = (True, True)
+    kernel_size = 5
+    pool_size = 5
+    
+# ------------------------------
+# Dataset
+# ------------------------------
+Tencent:
+    debug = False
+
+get_dataloader:
+    num_workers = 12
+    shuffle = True
+
+# ------------------------------
+# Training loop
+# ------------------------------
+TrainingLoop:
+    num_epochs = 30
+    learning_rate = 1e-4
+    dataset_cls = @Tencent
+    weight_decay = 0.0
+    batch_size_train = 64

results_tencent/hubert_base/layer4/run2/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:31920c8a8d46c20a73d25446748cc3081817656848e8e930352e07db191f910f
+size 2747979

results_tencent/hubert_base/layer4/run2/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ba0037fc4ec77b9c450bf9ec8156a248f586e572e0b8aa339723763b8c55cff
+size 2749638

results_tencent/hubert_base/layer4/run2/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 8000
+INFO:root:Num val speech clips: 2000
+INFO:root:Num test speech clips: 1563
+INFO:root:Device=cuda
+INFO:root:Average loss=0.4995166178941727
+INFO:root:
+[Valid][0][UTT][ MSE = 0.2875 | LCC = 0.9220 | SRCC = 0.9156 ]
+INFO:root:Average loss=0.19209776985645294
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2143 | LCC = 0.9409 | SRCC = 0.9387 ]
+INFO:root:Average loss=0.1524425759911537
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2173 | LCC = 0.9470 | SRCC = 0.9460 ]
+INFO:root:Average loss=0.13451571381092073
+INFO:root:
+[Valid][3][UTT][ MSE = 0.1917 | LCC = 0.9519 | SRCC = 0.9509 ]
+INFO:root:Average loss=0.1196074041724205
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2000 | LCC = 0.9542 | SRCC = 0.9536 ]
+INFO:root:Average loss=0.11005720752477646
+INFO:root:
+[Valid][5][UTT][ MSE = 0.1752 | LCC = 0.9562 | SRCC = 0.9562 ]
+INFO:root:Average loss=0.09841415452957153
+INFO:root:
+[Valid][6][UTT][ MSE = 0.1990 | LCC = 0.9575 | SRCC = 0.9573 ]
+INFO:root:Average loss=0.09684673476219177
+INFO:root:
+[Valid][7][UTT][ MSE = 0.1586 | LCC = 0.9583 | SRCC = 0.9583 ]
+INFO:root:Average loss=0.08796339356899262
+INFO:root:
+[Valid][8][UTT][ MSE = 0.1773 | LCC = 0.9587 | SRCC = 0.9585 ]
+INFO:root:Average loss=0.08198543867468834
+INFO:root:
+[Valid][9][UTT][ MSE = 0.1740 | LCC = 0.9586 | SRCC = 0.9584 ]
+INFO:root:Average loss=0.07829059413075447
+INFO:root:
+[Valid][10][UTT][ MSE = 0.1532 | LCC = 0.9606 | SRCC = 0.9605 ]
+INFO:root:Average loss=0.07272988843917846
+INFO:root:
+[Valid][11][UTT][ MSE = 0.1798 | LCC = 0.9604 | SRCC = 0.9603 ]
+INFO:root:Average loss=0.07076172545552253
+INFO:root:
+[Valid][12][UTT][ MSE = 0.1845 | LCC = 0.9605 | SRCC = 0.9601 ]
+INFO:root:Average loss=0.06509678456187248
+INFO:root:
+[Valid][13][UTT][ MSE = 0.1585 | LCC = 0.9612 | SRCC = 0.9609 ]
+INFO:root:Average loss=0.0646875624358654
+INFO:root:
+[Valid][14][UTT][ MSE = 0.1955 | LCC = 0.9604 | SRCC = 0.9599 ]
+INFO:root:Average loss=0.059271256491541864
+INFO:root:
+[Valid][15][UTT][ MSE = 0.1885 | LCC = 0.9617 | SRCC = 0.9613 ]
+INFO:root:Average loss=0.05441617016494274
+INFO:root:
+[Valid][16][UTT][ MSE = 0.1361 | LCC = 0.9611 | SRCC = 0.9606 ]
+INFO:root:Average loss=0.05317151245474815
+INFO:root:
+[Valid][17][UTT][ MSE = 0.1677 | LCC = 0.9618 | SRCC = 0.9616 ]
+INFO:root:Average loss=0.05185954561829567
+INFO:root:
+[Valid][18][UTT][ MSE = 0.1465 | LCC = 0.9613 | SRCC = 0.9611 ]
+INFO:root:Average loss=0.05082720635831356
+INFO:root:
+[Valid][19][UTT][ MSE = 0.1561 | LCC = 0.9617 | SRCC = 0.9611 ]
+INFO:root:Average loss=0.04718521195650101
+INFO:root:
+[Valid][20][UTT][ MSE = 0.1727 | LCC = 0.9616 | SRCC = 0.9613 ]
+INFO:root:Average loss=0.04441496774554252
+INFO:root:
+[Valid][21][UTT][ MSE = 0.1624 | LCC = 0.9611 | SRCC = 0.9609 ]
+INFO:root:Average loss=0.04399840787053108
+INFO:root:
+[Valid][22][UTT][ MSE = 0.1665 | LCC = 0.9621 | SRCC = 0.9617 ]
+INFO:root:Average loss=0.04263019375503063
+INFO:root:
+[Valid][23][UTT][ MSE = 0.1582 | LCC = 0.9622 | SRCC = 0.9616 ]
+INFO:root:Average loss=0.04249238841235638
+INFO:root:
+[Valid][24][UTT][ MSE = 0.1437 | LCC = 0.9624 | SRCC = 0.9617 ]
+INFO:root:Average loss=0.039484340891242024
+INFO:root:
+[Valid][25][UTT][ MSE = 0.1466 | LCC = 0.9624 | SRCC = 0.9618 ]
+INFO:root:Average loss=0.0373617038577795
+INFO:root:
+[Valid][26][UTT][ MSE = 0.1524 | LCC = 0.9629 | SRCC = 0.9623 ]
+INFO:root:Average loss=0.0370854345113039
+INFO:root:
+[Valid][27][UTT][ MSE = 0.1691 | LCC = 0.9615 | SRCC = 0.9606 ]
+INFO:root:Average loss=0.0355366690158844
+INFO:root:
+[Valid][28][UTT][ MSE = 0.1702 | LCC = 0.9618 | SRCC = 0.9609 ]
+INFO:root:Average loss=0.03367051811516285
+INFO:root:
+[Valid][29][UTT][ MSE = 0.1715 | LCC = 0.9621 | SRCC = 0.9617 ]
+INFO:root:Average loss=0.03245631115138531
+INFO:root:
+[Valid][30][UTT][ MSE = 0.1599 | LCC = 0.9619 | SRCC = 0.9612 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.1507 | LCC = 0.9649 | SRCC = 0.9658 ]

results_tencent/hubert_base/layer4/run3/config.gin

@@ -0,0 +1,31 @@
+# Training config on BVCC dataset.
+
+# ------------------------------
+# Architecture
+# ------------------------------
+ProjectionHead:
+    conv_channels = (32, 32)
+    dense_neurons = (64, 32, 1)
+    use_poolings = (True, True)
+    kernel_size = 5
+    pool_size = 5
+    
+# ------------------------------
+# Dataset
+# ------------------------------
+Tencent:
+    debug = False
+
+get_dataloader:
+    num_workers = 12
+    shuffle = True
+
+# ------------------------------
+# Training loop
+# ------------------------------
+TrainingLoop:
+    num_epochs = 30
+    learning_rate = 1e-4
+    dataset_cls = @Tencent
+    weight_decay = 0.0
+    batch_size_train = 64

results_tencent/hubert_base/layer4/run3/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df26764fc7a1bf316af3561f7506ad848a9918d870551939385c58d2eed73420
+size 2747979

results_tencent/hubert_base/layer4/run3/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a704fcf06b002778ad75ab700f7b64e8f3f269e88563cb463f1d2c4c4fbfe336
+size 2749638

results_tencent/hubert_base/layer4/run3/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 8000
+INFO:root:Num val speech clips: 2000
+INFO:root:Num test speech clips: 1563
+INFO:root:Device=cuda
+INFO:root:Average loss=0.44030873358249667
+INFO:root:
+[Valid][0][UTT][ MSE = 0.3020 | LCC = 0.9279 | SRCC = 0.9280 ]
+INFO:root:Average loss=0.18206725883483887
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2405 | LCC = 0.9444 | SRCC = 0.9435 ]
+INFO:root:Average loss=0.1476012277007103
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2160 | LCC = 0.9506 | SRCC = 0.9513 ]
+INFO:root:Average loss=0.129012701690197
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2141 | LCC = 0.9525 | SRCC = 0.9529 ]
+INFO:root:Average loss=0.11525272637605667
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2194 | LCC = 0.9549 | SRCC = 0.9548 ]
+INFO:root:Average loss=0.10601677018404007
+INFO:root:
+[Valid][5][UTT][ MSE = 0.2011 | LCC = 0.9578 | SRCC = 0.9583 ]
+INFO:root:Average loss=0.09903485164046287
+INFO:root:
+[Valid][6][UTT][ MSE = 0.1959 | LCC = 0.9590 | SRCC = 0.9585 ]
+INFO:root:Average loss=0.09287466624379158
+INFO:root:
+[Valid][7][UTT][ MSE = 0.1695 | LCC = 0.9591 | SRCC = 0.9597 ]
+INFO:root:Average loss=0.08375035202503205
+INFO:root:
+[Valid][8][UTT][ MSE = 0.1564 | LCC = 0.9602 | SRCC = 0.9604 ]
+INFO:root:Average loss=0.07879621079564095
+INFO:root:
+[Valid][9][UTT][ MSE = 0.1944 | LCC = 0.9609 | SRCC = 0.9609 ]
+INFO:root:Average loss=0.07444575351476669
+INFO:root:
+[Valid][10][UTT][ MSE = 0.1809 | LCC = 0.9597 | SRCC = 0.9598 ]
+INFO:root:Average loss=0.07514490965008735
+INFO:root:
+[Valid][11][UTT][ MSE = 0.1948 | LCC = 0.9613 | SRCC = 0.9612 ]
+INFO:root:Average loss=0.0638214111328125
+INFO:root:
+[Valid][12][UTT][ MSE = 0.2024 | LCC = 0.9618 | SRCC = 0.9617 ]
+INFO:root:Average loss=0.06424891629815102
+INFO:root:
+[Valid][13][UTT][ MSE = 0.1626 | LCC = 0.9602 | SRCC = 0.9607 ]
+INFO:root:Average loss=0.060301953971385956
+INFO:root:
+[Valid][14][UTT][ MSE = 0.1562 | LCC = 0.9604 | SRCC = 0.9609 ]
+INFO:root:Average loss=0.05698407934606075
+INFO:root:
+[Valid][15][UTT][ MSE = 0.1949 | LCC = 0.9627 | SRCC = 0.9621 ]
+INFO:root:Average loss=0.053353609368205074
+INFO:root:
+[Valid][16][UTT][ MSE = 0.1880 | LCC = 0.9602 | SRCC = 0.9600 ]
+INFO:root:Average loss=0.05258745715022087
+INFO:root:
+[Valid][17][UTT][ MSE = 0.1667 | LCC = 0.9629 | SRCC = 0.9629 ]
+INFO:root:Average loss=0.05299400274455547
+INFO:root:
+[Valid][18][UTT][ MSE = 0.2154 | LCC = 0.9620 | SRCC = 0.9624 ]
+INFO:root:Average loss=0.04674607641994953
+INFO:root:
+[Valid][19][UTT][ MSE = 0.1735 | LCC = 0.9634 | SRCC = 0.9628 ]
+INFO:root:Average loss=0.04557451544702053
+INFO:root:
+[Valid][20][UTT][ MSE = 0.1831 | LCC = 0.9633 | SRCC = 0.9628 ]
+INFO:root:Average loss=0.04382823812961578
+INFO:root:
+[Valid][21][UTT][ MSE = 0.1703 | LCC = 0.9635 | SRCC = 0.9627 ]
+INFO:root:Average loss=0.040797299072146416
+INFO:root:
+[Valid][22][UTT][ MSE = 0.1625 | LCC = 0.9628 | SRCC = 0.9627 ]
+INFO:root:Average loss=0.040290260583162305
+INFO:root:
+[Valid][23][UTT][ MSE = 0.1685 | LCC = 0.9618 | SRCC = 0.9614 ]
+INFO:root:Average loss=0.040101629719138145
+INFO:root:
+[Valid][24][UTT][ MSE = 0.1883 | LCC = 0.9627 | SRCC = 0.9619 ]
+INFO:root:Average loss=0.037949363678693775
+INFO:root:
+[Valid][25][UTT][ MSE = 0.1712 | LCC = 0.9625 | SRCC = 0.9615 ]
+INFO:root:Average loss=0.034947402223944664
+INFO:root:
+[Valid][26][UTT][ MSE = 0.1597 | LCC = 0.9618 | SRCC = 0.9616 ]
+INFO:root:Average loss=0.033867306530475615
+INFO:root:
+[Valid][27][UTT][ MSE = 0.1929 | LCC = 0.9627 | SRCC = 0.9617 ]
+INFO:root:Average loss=0.03201555570960045
+INFO:root:
+[Valid][28][UTT][ MSE = 0.1508 | LCC = 0.9622 | SRCC = 0.9619 ]
+INFO:root:Average loss=0.03369861473143101
+INFO:root:
+[Valid][29][UTT][ MSE = 0.1673 | LCC = 0.9634 | SRCC = 0.9624 ]
+INFO:root:Average loss=0.032580247417092326
+INFO:root:
+[Valid][30][UTT][ MSE = 0.1701 | LCC = 0.9631 | SRCC = 0.9623 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.1754 | LCC = 0.9639 | SRCC = 0.9642 ]

results_tencent/hubert_base/layer4/run4/config.gin

@@ -0,0 +1,31 @@
+# Training config on BVCC dataset.
+
+# ------------------------------
+# Architecture
+# ------------------------------
+ProjectionHead:
+    conv_channels = (32, 32)
+    dense_neurons = (64, 32, 1)
+    use_poolings = (True, True)
+    kernel_size = 5
+    pool_size = 5
+    
+# ------------------------------
+# Dataset
+# ------------------------------
+Tencent:
+    debug = False
+
+get_dataloader:
+    num_workers = 12
+    shuffle = True
+
+# ------------------------------
+# Training loop
+# ------------------------------
+TrainingLoop:
+    num_epochs = 30
+    learning_rate = 1e-4
+    dataset_cls = @Tencent
+    weight_decay = 0.0
+    batch_size_train = 64

results_tencent/hubert_base/layer4/run4/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d750b1338960daf96e5249d2a182270eafbbe2cf2ab6874d63f430eefd1ebc9e
+size 2747979

results_tencent/hubert_base/layer4/run4/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:68a4b8768aa48a21cceb33d42a522e5179cfbdab1a5d75e2ae234858c6a7a0fb
+size 2749638

results_tencent/hubert_base/layer4/run4/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 8000
+INFO:root:Num val speech clips: 2000
+INFO:root:Num test speech clips: 1563
+INFO:root:Device=cuda
+INFO:root:Average loss=0.4231206508874893
+INFO:root:
+[Valid][0][UTT][ MSE = 0.2700 | LCC = 0.9236 | SRCC = 0.9223 ]
+INFO:root:Average loss=0.1873604558110237
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2261 | LCC = 0.9445 | SRCC = 0.9434 ]
+INFO:root:Average loss=0.14630690729618073
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2074 | LCC = 0.9504 | SRCC = 0.9494 ]
+INFO:root:Average loss=0.12512446576356887
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2110 | LCC = 0.9547 | SRCC = 0.9545 ]
+INFO:root:Average loss=0.11086751115322113
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2034 | LCC = 0.9578 | SRCC = 0.9579 ]
+INFO:root:Average loss=0.10226034927368165
+INFO:root:
+[Valid][5][UTT][ MSE = 0.1989 | LCC = 0.9593 | SRCC = 0.9594 ]
+INFO:root:Average loss=0.09708944684267044
+INFO:root:
+[Valid][6][UTT][ MSE = 0.1803 | LCC = 0.9599 | SRCC = 0.9600 ]
+INFO:root:Average loss=0.08576299405097962
+INFO:root:
+[Valid][7][UTT][ MSE = 0.1588 | LCC = 0.9612 | SRCC = 0.9610 ]
+INFO:root:Average loss=0.08215700510144233
+INFO:root:
+[Valid][8][UTT][ MSE = 0.1715 | LCC = 0.9613 | SRCC = 0.9613 ]
+INFO:root:Average loss=0.07669090881943702
+INFO:root:
+[Valid][9][UTT][ MSE = 0.1798 | LCC = 0.9613 | SRCC = 0.9612 ]
+INFO:root:Average loss=0.07130826711654663
+INFO:root:
+[Valid][10][UTT][ MSE = 0.1576 | LCC = 0.9609 | SRCC = 0.9606 ]
+INFO:root:Average loss=0.07080629828572273
+INFO:root:
+[Valid][11][UTT][ MSE = 0.1713 | LCC = 0.9614 | SRCC = 0.9609 ]
+INFO:root:Average loss=0.06348666182160377
+INFO:root:
+[Valid][12][UTT][ MSE = 0.1769 | LCC = 0.9615 | SRCC = 0.9611 ]
+INFO:root:Average loss=0.06148102152347565
+INFO:root:
+[Valid][13][UTT][ MSE = 0.1813 | LCC = 0.9624 | SRCC = 0.9615 ]
+INFO:root:Average loss=0.056300003230571743
+INFO:root:
+[Valid][14][UTT][ MSE = 0.1847 | LCC = 0.9626 | SRCC = 0.9621 ]
+INFO:root:Average loss=0.0544388337135315
+INFO:root:
+[Valid][15][UTT][ MSE = 0.1692 | LCC = 0.9618 | SRCC = 0.9610 ]
+INFO:root:Average loss=0.050749091237783435
+INFO:root:
+[Valid][16][UTT][ MSE = 0.1722 | LCC = 0.9609 | SRCC = 0.9602 ]
+INFO:root:Average loss=0.049141381353139876
+INFO:root:
+[Valid][17][UTT][ MSE = 0.1528 | LCC = 0.9624 | SRCC = 0.9624 ]
+INFO:root:Average loss=0.05031431767344475
+INFO:root:
+[Valid][18][UTT][ MSE = 0.1539 | LCC = 0.9621 | SRCC = 0.9618 ]
+INFO:root:Average loss=0.0469830205142498
+INFO:root:
+[Valid][19][UTT][ MSE = 0.1581 | LCC = 0.9623 | SRCC = 0.9618 ]
+INFO:root:Average loss=0.04456768381595612
+INFO:root:
+[Valid][20][UTT][ MSE = 0.1813 | LCC = 0.9591 | SRCC = 0.9601 ]
+INFO:root:Average loss=0.04225409373641014
+INFO:root:
+[Valid][21][UTT][ MSE = 0.1595 | LCC = 0.9625 | SRCC = 0.9620 ]
+INFO:root:Average loss=0.041059058561921116
+INFO:root:
+[Valid][22][UTT][ MSE = 0.1494 | LCC = 0.9626 | SRCC = 0.9623 ]
+INFO:root:Average loss=0.04062368324398995
+INFO:root:
+[Valid][23][UTT][ MSE = 0.1584 | LCC = 0.9634 | SRCC = 0.9626 ]
+INFO:root:Average loss=0.037790130510926244
+INFO:root:
+[Valid][24][UTT][ MSE = 0.1613 | LCC = 0.9625 | SRCC = 0.9620 ]
+INFO:root:Average loss=0.036218619644641875
+INFO:root:
+[Valid][25][UTT][ MSE = 0.1662 | LCC = 0.9617 | SRCC = 0.9613 ]
+INFO:root:Average loss=0.03467911140620709
+INFO:root:
+[Valid][26][UTT][ MSE = 0.1522 | LCC = 0.9625 | SRCC = 0.9617 ]
+INFO:root:Average loss=0.03379358433187008
+INFO:root:
+[Valid][27][UTT][ MSE = 0.1568 | LCC = 0.9633 | SRCC = 0.9626 ]
+INFO:root:Average loss=0.03456975008547306
+INFO:root:
+[Valid][28][UTT][ MSE = 0.1692 | LCC = 0.9626 | SRCC = 0.9619 ]
+INFO:root:Average loss=0.030540536373853684
+INFO:root:
+[Valid][29][UTT][ MSE = 0.1527 | LCC = 0.9631 | SRCC = 0.9629 ]
+INFO:root:Average loss=0.03388060097396374
+INFO:root:
+[Valid][30][UTT][ MSE = 0.1577 | LCC = 0.9622 | SRCC = 0.9617 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.1579 | LCC = 0.9651 | SRCC = 0.9653 ]

results_tencent/hubert_base/layer5/run0/config.gin

@@ -0,0 +1,31 @@
+# Training config on BVCC dataset.
+
+# ------------------------------
+# Architecture
+# ------------------------------
+ProjectionHead:
+    conv_channels = (32, 32)
+    dense_neurons = (64, 32, 1)
+    use_poolings = (True, True)
+    kernel_size = 5
+    pool_size = 5
+    
+# ------------------------------
+# Dataset
+# ------------------------------
+Tencent:
+    debug = False
+
+get_dataloader:
+    num_workers = 12
+    shuffle = True
+
+# ------------------------------
+# Training loop
+# ------------------------------
+TrainingLoop:
+    num_epochs = 30
+    learning_rate = 1e-4
+    dataset_cls = @Tencent
+    weight_decay = 0.0
+    batch_size_train = 64

results_tencent/hubert_base/layer5/run0/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db67133686c06c5fbad2a8eee70b613c78efff291e00cfca2f58bf9d8a25e298
+size 2747979

results_tencent/hubert_base/layer5/run0/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d032228cee1b180d603c6623adb2dfbd756a037b7b54881a86cb500920628a5c
+size 2749702

results_tencent/hubert_base/layer5/run0/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 8000
+INFO:root:Num val speech clips: 2000
+INFO:root:Num test speech clips: 1563
+INFO:root:Device=cuda
+INFO:root:Average loss=0.509871609210968
+INFO:root:
+[Valid][0][UTT][ MSE = 0.2946 | LCC = 0.9260 | SRCC = 0.9235 ]
+INFO:root:Average loss=0.17839222133159638
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2250 | LCC = 0.9480 | SRCC = 0.9470 ]
+INFO:root:Average loss=0.13951319366693496
+INFO:root:
+[Valid][2][UTT][ MSE = 0.1932 | LCC = 0.9529 | SRCC = 0.9524 ]
+INFO:root:Average loss=0.12330902749300003
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2062 | LCC = 0.9554 | SRCC = 0.9547 ]
+INFO:root:Average loss=0.10927059602737427
+INFO:root:
+[Valid][4][UTT][ MSE = 0.1682 | LCC = 0.9570 | SRCC = 0.9566 ]
+INFO:root:Average loss=0.10175281023979187
+INFO:root:
+[Valid][5][UTT][ MSE = 0.1955 | LCC = 0.9590 | SRCC = 0.9588 ]
+INFO:root:Average loss=0.09398409953713417
+INFO:root:
+[Valid][6][UTT][ MSE = 0.1893 | LCC = 0.9590 | SRCC = 0.9589 ]
+INFO:root:Average loss=0.08592596876621246
+INFO:root:
+[Valid][7][UTT][ MSE = 0.1933 | LCC = 0.9593 | SRCC = 0.9586 ]
+INFO:root:Average loss=0.08179879042506218
+INFO:root:
+[Valid][8][UTT][ MSE = 0.1892 | LCC = 0.9604 | SRCC = 0.9600 ]
+INFO:root:Average loss=0.07720961457490921
+INFO:root:
+[Valid][9][UTT][ MSE = 0.1961 | LCC = 0.9604 | SRCC = 0.9601 ]
+INFO:root:Average loss=0.07389661473035812
+INFO:root:
+[Valid][10][UTT][ MSE = 0.1846 | LCC = 0.9605 | SRCC = 0.9602 ]
+INFO:root:Average loss=0.06821288815140725
+INFO:root:
+[Valid][11][UTT][ MSE = 0.1817 | LCC = 0.9611 | SRCC = 0.9604 ]
+INFO:root:Average loss=0.06425552794337272
+INFO:root:
+[Valid][12][UTT][ MSE = 0.1631 | LCC = 0.9618 | SRCC = 0.9616 ]
+INFO:root:Average loss=0.06127037313580513
+INFO:root:
+[Valid][13][UTT][ MSE = 0.1900 | LCC = 0.9610 | SRCC = 0.9603 ]
+INFO:root:Average loss=0.05727958863973617
+INFO:root:
+[Valid][14][UTT][ MSE = 0.1928 | LCC = 0.9616 | SRCC = 0.9611 ]
+INFO:root:Average loss=0.05375721390545368
+INFO:root:
+[Valid][15][UTT][ MSE = 0.1795 | LCC = 0.9611 | SRCC = 0.9603 ]
+INFO:root:Average loss=0.05360283032059669
+INFO:root:
+[Valid][16][UTT][ MSE = 0.1750 | LCC = 0.9614 | SRCC = 0.9606 ]
+INFO:root:Average loss=0.05083597853779793
+INFO:root:
+[Valid][17][UTT][ MSE = 0.1804 | LCC = 0.9609 | SRCC = 0.9600 ]
+INFO:root:Average loss=0.04944618593156338
+INFO:root:
+[Valid][18][UTT][ MSE = 0.1557 | LCC = 0.9607 | SRCC = 0.9601 ]
+INFO:root:Average loss=0.04685955688357353
+INFO:root:
+[Valid][19][UTT][ MSE = 0.1642 | LCC = 0.9611 | SRCC = 0.9601 ]
+INFO:root:Average loss=0.04339661008119583
+INFO:root:
+[Valid][20][UTT][ MSE = 0.1665 | LCC = 0.9614 | SRCC = 0.9609 ]
+INFO:root:Average loss=0.0437190684825182
+INFO:root:
+[Valid][21][UTT][ MSE = 0.2013 | LCC = 0.9606 | SRCC = 0.9602 ]
+INFO:root:Average loss=0.041316692784428594
+INFO:root:
+[Valid][22][UTT][ MSE = 0.1764 | LCC = 0.9611 | SRCC = 0.9605 ]
+INFO:root:Average loss=0.03744974264502526
+INFO:root:
+[Valid][23][UTT][ MSE = 0.1652 | LCC = 0.9615 | SRCC = 0.9605 ]
+INFO:root:Average loss=0.037872024461627005
+INFO:root:
+[Valid][24][UTT][ MSE = 0.1745 | LCC = 0.9612 | SRCC = 0.9604 ]
+INFO:root:Average loss=0.03666692753136158
+INFO:root:
+[Valid][25][UTT][ MSE = 0.2224 | LCC = 0.9584 | SRCC = 0.9600 ]
+INFO:root:Average loss=0.033550654053688046
+INFO:root:
+[Valid][26][UTT][ MSE = 0.1842 | LCC = 0.9610 | SRCC = 0.9608 ]
+INFO:root:Average loss=0.032786446258425715
+INFO:root:
+[Valid][27][UTT][ MSE = 0.1696 | LCC = 0.9613 | SRCC = 0.9605 ]
+INFO:root:Average loss=0.0310258277207613
+INFO:root:
+[Valid][28][UTT][ MSE = 0.1692 | LCC = 0.9608 | SRCC = 0.9595 ]
+INFO:root:Average loss=0.03211408606171608
+INFO:root:
+[Valid][29][UTT][ MSE = 0.1635 | LCC = 0.9609 | SRCC = 0.9595 ]
+INFO:root:Average loss=0.029530202552676202
+INFO:root:
+[Valid][30][UTT][ MSE = 0.1799 | LCC = 0.9614 | SRCC = 0.9602 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.1713 | LCC = 0.9609 | SRCC = 0.9614 ]

results_tencent/hubert_base/layer5/run1/config.gin

@@ -0,0 +1,31 @@
+# Training config on BVCC dataset.
+
+# ------------------------------
+# Architecture
+# ------------------------------
+ProjectionHead:
+    conv_channels = (32, 32)
+    dense_neurons = (64, 32, 1)
+    use_poolings = (True, True)
+    kernel_size = 5
+    pool_size = 5
+    
+# ------------------------------
+# Dataset
+# ------------------------------
+Tencent:
+    debug = False
+
+get_dataloader:
+    num_workers = 12
+    shuffle = True
+
+# ------------------------------
+# Training loop
+# ------------------------------
+TrainingLoop:
+    num_epochs = 30
+    learning_rate = 1e-4
+    dataset_cls = @Tencent
+    weight_decay = 0.0
+    batch_size_train = 64

results_tencent/hubert_base/layer5/run1/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d0ef3afe630a483b3f2b6e0fa71ae16d937f71c589ccfa4dcdf8e1697450e4b
+size 2747979

results_tencent/hubert_base/layer5/run1/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be3c00bf8079c4660e5f84a864c8a7a6d248c5e711d9031371756f0848572904
+size 2749702

results_tencent/hubert_base/layer5/run1/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 8000
+INFO:root:Num val speech clips: 2000
+INFO:root:Num test speech clips: 1563
+INFO:root:Device=cuda
+INFO:root:Average loss=0.4485912972688675
+INFO:root:
+[Valid][0][UTT][ MSE = 0.2821 | LCC = 0.9280 | SRCC = 0.9262 ]
+INFO:root:Average loss=0.18385643416643144
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2263 | LCC = 0.9450 | SRCC = 0.9430 ]
+INFO:root:Average loss=0.14546715712547303
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2162 | LCC = 0.9490 | SRCC = 0.9477 ]
+INFO:root:Average loss=0.128211097240448
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2271 | LCC = 0.9513 | SRCC = 0.9498 ]
+INFO:root:Average loss=0.11424678936600685
+INFO:root:
+[Valid][4][UTT][ MSE = 0.1871 | LCC = 0.9559 | SRCC = 0.9553 ]
+INFO:root:Average loss=0.1077485965192318
+INFO:root:
+[Valid][5][UTT][ MSE = 0.1991 | LCC = 0.9563 | SRCC = 0.9556 ]
+INFO:root:Average loss=0.0988824163377285
+INFO:root:
+[Valid][6][UTT][ MSE = 0.2076 | LCC = 0.9575 | SRCC = 0.9569 ]
+INFO:root:Average loss=0.08992268896102905
+INFO:root:
+[Valid][7][UTT][ MSE = 0.2025 | LCC = 0.9572 | SRCC = 0.9563 ]
+INFO:root:Average loss=0.08608752313256264
+INFO:root:
+[Valid][8][UTT][ MSE = 0.2205 | LCC = 0.9596 | SRCC = 0.9593 ]
+INFO:root:Average loss=0.07951096522808075
+INFO:root:
+[Valid][9][UTT][ MSE = 0.1917 | LCC = 0.9578 | SRCC = 0.9571 ]
+INFO:root:Average loss=0.07338705691695213
+INFO:root:
+[Valid][10][UTT][ MSE = 0.1596 | LCC = 0.9595 | SRCC = 0.9595 ]
+INFO:root:Average loss=0.06975187465548516
+INFO:root:
+[Valid][11][UTT][ MSE = 0.1824 | LCC = 0.9601 | SRCC = 0.9593 ]
+INFO:root:Average loss=0.06577360579371452
+INFO:root:
+[Valid][12][UTT][ MSE = 0.1613 | LCC = 0.9596 | SRCC = 0.9597 ]
+INFO:root:Average loss=0.06473290187120438
+INFO:root:
+[Valid][13][UTT][ MSE = 0.2088 | LCC = 0.9608 | SRCC = 0.9605 ]
+INFO:root:Average loss=0.06011417979001999
+INFO:root:
+[Valid][14][UTT][ MSE = 0.1673 | LCC = 0.9607 | SRCC = 0.9606 ]
+INFO:root:Average loss=0.05521336568892002
+INFO:root:
+[Valid][15][UTT][ MSE = 0.1820 | LCC = 0.9588 | SRCC = 0.9580 ]
+INFO:root:Average loss=0.05224400761723518
+INFO:root:
+[Valid][16][UTT][ MSE = 0.1741 | LCC = 0.9615 | SRCC = 0.9608 ]
+INFO:root:Average loss=0.04970773628354073
+INFO:root:
+[Valid][17][UTT][ MSE = 0.1725 | LCC = 0.9616 | SRCC = 0.9610 ]
+INFO:root:Average loss=0.04900969077646732
+INFO:root:
+[Valid][18][UTT][ MSE = 0.1653 | LCC = 0.9610 | SRCC = 0.9608 ]
+INFO:root:Average loss=0.04554431425035
+INFO:root:
+[Valid][19][UTT][ MSE = 0.1803 | LCC = 0.9599 | SRCC = 0.9600 ]
+INFO:root:Average loss=0.04634648248553276
+INFO:root:
+[Valid][20][UTT][ MSE = 0.1741 | LCC = 0.9608 | SRCC = 0.9605 ]
+INFO:root:Average loss=0.04277595445513725
+INFO:root:
+[Valid][21][UTT][ MSE = 0.1693 | LCC = 0.9600 | SRCC = 0.9590 ]
+INFO:root:Average loss=0.042401527494192126
+INFO:root:
+[Valid][22][UTT][ MSE = 0.1857 | LCC = 0.9611 | SRCC = 0.9600 ]
+INFO:root:Average loss=0.040588512673974037
+INFO:root:
+[Valid][23][UTT][ MSE = 0.1740 | LCC = 0.9608 | SRCC = 0.9599 ]
+INFO:root:Average loss=0.039168834745883944
+INFO:root:
+[Valid][24][UTT][ MSE = 0.1887 | LCC = 0.9591 | SRCC = 0.9581 ]
+INFO:root:Average loss=0.03717134357988834
+INFO:root:
+[Valid][25][UTT][ MSE = 0.1711 | LCC = 0.9604 | SRCC = 0.9596 ]
+INFO:root:Average loss=0.036077167302370074
+INFO:root:
+[Valid][26][UTT][ MSE = 0.1827 | LCC = 0.9618 | SRCC = 0.9610 ]
+INFO:root:Average loss=0.033301551312208175
+INFO:root:
+[Valid][27][UTT][ MSE = 0.1701 | LCC = 0.9616 | SRCC = 0.9609 ]
+INFO:root:Average loss=0.03407234816253185
+INFO:root:
+[Valid][28][UTT][ MSE = 0.1701 | LCC = 0.9614 | SRCC = 0.9607 ]
+INFO:root:Average loss=0.03406850910186768
+INFO:root:
+[Valid][29][UTT][ MSE = 0.1913 | LCC = 0.9599 | SRCC = 0.9591 ]
+INFO:root:Average loss=0.030076176837086677
+INFO:root:
+[Valid][30][UTT][ MSE = 0.1648 | LCC = 0.9615 | SRCC = 0.9606 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.1903 | LCC = 0.9602 | SRCC = 0.9608 ]

results_tencent/hubert_base/layer5/run2/config.gin

@@ -0,0 +1,31 @@
+# Training config on BVCC dataset.
+
+# ------------------------------
+# Architecture
+# ------------------------------
+ProjectionHead:
+    conv_channels = (32, 32)
+    dense_neurons = (64, 32, 1)
+    use_poolings = (True, True)
+    kernel_size = 5
+    pool_size = 5
+    
+# ------------------------------
+# Dataset
+# ------------------------------
+Tencent:
+    debug = False
+
+get_dataloader:
+    num_workers = 12
+    shuffle = True
+
+# ------------------------------
+# Training loop
+# ------------------------------
+TrainingLoop:
+    num_epochs = 30
+    learning_rate = 1e-4
+    dataset_cls = @Tencent
+    weight_decay = 0.0
+    batch_size_train = 64

results_tencent/hubert_base/layer5/run2/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1262d494f321c5b50f690458620ce480e9832e29253d036b8672d09fc3ce9461
+size 2747979

results_tencent/hubert_base/layer5/run2/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c3eee747a8c205b163eb4452bbd2105d597ee426445187e7a8551ff131ee3eb
+size 2749702

results_tencent/hubert_base/layer5/run2/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 8000
+INFO:root:Num val speech clips: 2000
+INFO:root:Num test speech clips: 1563
+INFO:root:Device=cuda
+INFO:root:Average loss=0.5340962796211243
+INFO:root:
+[Valid][0][UTT][ MSE = 0.3195 | LCC = 0.9308 | SRCC = 0.9277 ]
+INFO:root:Average loss=0.18154503667354585
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2656 | LCC = 0.9422 | SRCC = 0.9392 ]
+INFO:root:Average loss=0.1449539442062378
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2026 | LCC = 0.9515 | SRCC = 0.9513 ]
+INFO:root:Average loss=0.12917825067043304
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2204 | LCC = 0.9543 | SRCC = 0.9538 ]
+INFO:root:Average loss=0.11683966714143754
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2065 | LCC = 0.9574 | SRCC = 0.9565 ]
+INFO:root:Average loss=0.1063873569369316
+INFO:root:
+[Valid][5][UTT][ MSE = 0.2155 | LCC = 0.9578 | SRCC = 0.9578 ]
+INFO:root:Average loss=0.09734632813930512
+INFO:root:
+[Valid][6][UTT][ MSE = 0.2446 | LCC = 0.9568 | SRCC = 0.9578 ]
+INFO:root:Average loss=0.08912296891212464
+INFO:root:
+[Valid][7][UTT][ MSE = 0.2054 | LCC = 0.9591 | SRCC = 0.9592 ]
+INFO:root:Average loss=0.08710124406218529
+INFO:root:
+[Valid][8][UTT][ MSE = 0.2206 | LCC = 0.9593 | SRCC = 0.9596 ]
+INFO:root:Average loss=0.07703453764319419
+INFO:root:
+[Valid][9][UTT][ MSE = 0.1998 | LCC = 0.9600 | SRCC = 0.9595 ]
+INFO:root:Average loss=0.07147296983003616
+INFO:root:
+[Valid][10][UTT][ MSE = 0.2122 | LCC = 0.9605 | SRCC = 0.9608 ]
+INFO:root:Average loss=0.06867362049221992
+INFO:root:
+[Valid][11][UTT][ MSE = 0.1918 | LCC = 0.9606 | SRCC = 0.9603 ]
+INFO:root:Average loss=0.0661231769323349
+INFO:root:
+[Valid][12][UTT][ MSE = 0.2198 | LCC = 0.9596 | SRCC = 0.9602 ]
+INFO:root:Average loss=0.06014017602801323
+INFO:root:
+[Valid][13][UTT][ MSE = 0.2083 | LCC = 0.9608 | SRCC = 0.9616 ]
+INFO:root:Average loss=0.059064438670873645
+INFO:root:
+[Valid][14][UTT][ MSE = 0.1789 | LCC = 0.9606 | SRCC = 0.9597 ]
+INFO:root:Average loss=0.055299815952777866
+INFO:root:
+[Valid][15][UTT][ MSE = 0.1894 | LCC = 0.9615 | SRCC = 0.9613 ]
+INFO:root:Average loss=0.054188394337892534
+INFO:root:
+[Valid][16][UTT][ MSE = 0.1957 | LCC = 0.9619 | SRCC = 0.9617 ]
+INFO:root:Average loss=0.0503035536557436
+INFO:root:
+[Valid][17][UTT][ MSE = 0.1847 | LCC = 0.9625 | SRCC = 0.9621 ]
+INFO:root:Average loss=0.04701666863262653
+INFO:root:
+[Valid][18][UTT][ MSE = 0.1842 | LCC = 0.9621 | SRCC = 0.9612 ]
+INFO:root:Average loss=0.04411008244752884
+INFO:root:
+[Valid][19][UTT][ MSE = 0.1756 | LCC = 0.9607 | SRCC = 0.9602 ]
+INFO:root:Average loss=0.04376040953397751
+INFO:root:
+[Valid][20][UTT][ MSE = 0.2026 | LCC = 0.9607 | SRCC = 0.9599 ]
+INFO:root:Average loss=0.04142147873342037
+INFO:root:
+[Valid][21][UTT][ MSE = 0.1752 | LCC = 0.9610 | SRCC = 0.9603 ]
+INFO:root:Average loss=0.03836024996638298
+INFO:root:
+[Valid][22][UTT][ MSE = 0.1602 | LCC = 0.9611 | SRCC = 0.9602 ]
+INFO:root:Average loss=0.03753511606156826
+INFO:root:
+[Valid][23][UTT][ MSE = 0.1799 | LCC = 0.9627 | SRCC = 0.9619 ]
+INFO:root:Average loss=0.036446436583995816
+INFO:root:
+[Valid][24][UTT][ MSE = 0.1678 | LCC = 0.9616 | SRCC = 0.9606 ]
+INFO:root:Average loss=0.0363024819791317
+INFO:root:
+[Valid][25][UTT][ MSE = 0.1751 | LCC = 0.9617 | SRCC = 0.9606 ]
+INFO:root:Average loss=0.032234612718224524
+INFO:root:
+[Valid][26][UTT][ MSE = 0.1759 | LCC = 0.9622 | SRCC = 0.9611 ]
+INFO:root:Average loss=0.03274307750165462
+INFO:root:
+[Valid][27][UTT][ MSE = 0.1948 | LCC = 0.9622 | SRCC = 0.9615 ]
+INFO:root:Average loss=0.03142600749433041
+INFO:root:
+[Valid][28][UTT][ MSE = 0.1797 | LCC = 0.9614 | SRCC = 0.9604 ]
+INFO:root:Average loss=0.029825103282928467
+INFO:root:
+[Valid][29][UTT][ MSE = 0.1974 | LCC = 0.9622 | SRCC = 0.9615 ]
+INFO:root:Average loss=0.030201896876096725
+INFO:root:
+[Valid][30][UTT][ MSE = 0.1854 | LCC = 0.9620 | SRCC = 0.9610 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.1919 | LCC = 0.9603 | SRCC = 0.9608 ]

results_tencent/hubert_base/layer5/run3/config.gin

@@ -0,0 +1,31 @@
+# Training config on BVCC dataset.
+
+# ------------------------------
+# Architecture
+# ------------------------------
+ProjectionHead:
+    conv_channels = (32, 32)
+    dense_neurons = (64, 32, 1)
+    use_poolings = (True, True)
+    kernel_size = 5
+    pool_size = 5
+    
+# ------------------------------
+# Dataset
+# ------------------------------
+Tencent:
+    debug = False
+
+get_dataloader:
+    num_workers = 12
+    shuffle = True
+
+# ------------------------------
+# Training loop
+# ------------------------------
+TrainingLoop:
+    num_epochs = 30
+    learning_rate = 1e-4
+    dataset_cls = @Tencent
+    weight_decay = 0.0
+    batch_size_train = 64

results_tencent/hubert_base/layer5/run3/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e8c24368de5f1de0d45bd9a53f22e11783c99aa25c64a0a5b35c3e6e597a75f
+size 2747979

results_tencent/hubert_base/layer5/run3/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c6be1a0fa84565f105d7685f765cb57cf73260ac0fe7411d5e2a03ca0c8bf180
+size 2749702

results_tencent/hubert_base/layer5/run3/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 8000
+INFO:root:Num val speech clips: 2000
+INFO:root:Num test speech clips: 1563
+INFO:root:Device=cuda
+INFO:root:Average loss=0.4258647834062576
+INFO:root:
+[Valid][0][UTT][ MSE = 0.2839 | LCC = 0.9219 | SRCC = 0.9187 ]
+INFO:root:Average loss=0.19219473093748093
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2342 | LCC = 0.9411 | SRCC = 0.9403 ]
+INFO:root:Average loss=0.15391053366661073
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2250 | LCC = 0.9477 | SRCC = 0.9482 ]
+INFO:root:Average loss=0.13024138921499254
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2151 | LCC = 0.9524 | SRCC = 0.9524 ]
+INFO:root:Average loss=0.11660204619169236
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2066 | LCC = 0.9552 | SRCC = 0.9548 ]
+INFO:root:Average loss=0.10466550648212433
+INFO:root:
+[Valid][5][UTT][ MSE = 0.1827 | LCC = 0.9533 | SRCC = 0.9534 ]
+INFO:root:Average loss=0.09522219270467758
+INFO:root:
+[Valid][6][UTT][ MSE = 0.1834 | LCC = 0.9553 | SRCC = 0.9545 ]
+INFO:root:Average loss=0.09436587470769882
+INFO:root:
+[Valid][7][UTT][ MSE = 0.1765 | LCC = 0.9509 | SRCC = 0.9521 ]
+INFO:root:Average loss=0.08626363936066628
+INFO:root:
+[Valid][8][UTT][ MSE = 0.1691 | LCC = 0.9563 | SRCC = 0.9558 ]
+INFO:root:Average loss=0.08010093629360199
+INFO:root:
+[Valid][9][UTT][ MSE = 0.1804 | LCC = 0.9543 | SRCC = 0.9555 ]
+INFO:root:Average loss=0.07669223803281784
+INFO:root:
+[Valid][10][UTT][ MSE = 0.1508 | LCC = 0.9567 | SRCC = 0.9578 ]
+INFO:root:Average loss=0.0685383132994175
+INFO:root:
+[Valid][11][UTT][ MSE = 0.1614 | LCC = 0.9586 | SRCC = 0.9588 ]
+INFO:root:Average loss=0.06738250184059143
+INFO:root:
+[Valid][12][UTT][ MSE = 0.1597 | LCC = 0.9585 | SRCC = 0.9589 ]
+INFO:root:Average loss=0.06330720314383506
+INFO:root:
+[Valid][13][UTT][ MSE = 0.1759 | LCC = 0.9575 | SRCC = 0.9577 ]
+INFO:root:Average loss=0.0582011576294899
+INFO:root:
+[Valid][14][UTT][ MSE = 0.1573 | LCC = 0.9570 | SRCC = 0.9578 ]
+INFO:root:Average loss=0.05550464406609535
+INFO:root:
+[Valid][15][UTT][ MSE = 0.1774 | LCC = 0.9600 | SRCC = 0.9598 ]
+INFO:root:Average loss=0.0553211199939251
+INFO:root:
+[Valid][16][UTT][ MSE = 0.1994 | LCC = 0.9580 | SRCC = 0.9577 ]
+INFO:root:Average loss=0.05223307606577873
+INFO:root:
+[Valid][17][UTT][ MSE = 0.1679 | LCC = 0.9572 | SRCC = 0.9576 ]
+INFO:root:Average loss=0.048624135598540306
+INFO:root:
+[Valid][18][UTT][ MSE = 0.1634 | LCC = 0.9603 | SRCC = 0.9598 ]
+INFO:root:Average loss=0.0455592223405838
+INFO:root:
+[Valid][19][UTT][ MSE = 0.1716 | LCC = 0.9608 | SRCC = 0.9604 ]
+INFO:root:Average loss=0.04710148233175278
+INFO:root:
+[Valid][20][UTT][ MSE = 0.1615 | LCC = 0.9578 | SRCC = 0.9592 ]
+INFO:root:Average loss=0.04371048204600811
+INFO:root:
+[Valid][21][UTT][ MSE = 0.1577 | LCC = 0.9589 | SRCC = 0.9586 ]
+INFO:root:Average loss=0.040241401448845866
+INFO:root:
+[Valid][22][UTT][ MSE = 0.1686 | LCC = 0.9596 | SRCC = 0.9587 ]
+INFO:root:Average loss=0.03926335689425468
+INFO:root:
+[Valid][23][UTT][ MSE = 0.1599 | LCC = 0.9571 | SRCC = 0.9572 ]
+INFO:root:Average loss=0.03711808468401432
+INFO:root:
+[Valid][24][UTT][ MSE = 0.1708 | LCC = 0.9601 | SRCC = 0.9597 ]
+INFO:root:Average loss=0.03830269975960255
+INFO:root:
+[Valid][25][UTT][ MSE = 0.1716 | LCC = 0.9596 | SRCC = 0.9585 ]
+INFO:root:Average loss=0.03531160207092762
+INFO:root:
+[Valid][26][UTT][ MSE = 0.1621 | LCC = 0.9597 | SRCC = 0.9594 ]
+INFO:root:Average loss=0.03396486204862595
+INFO:root:
+[Valid][27][UTT][ MSE = 0.1631 | LCC = 0.9576 | SRCC = 0.9579 ]
+INFO:root:Average loss=0.03429593308269978
+INFO:root:
+[Valid][28][UTT][ MSE = 0.1773 | LCC = 0.9593 | SRCC = 0.9593 ]
+INFO:root:Average loss=0.0324753201007843
+INFO:root:
+[Valid][29][UTT][ MSE = 0.1659 | LCC = 0.9572 | SRCC = 0.9580 ]
+INFO:root:Average loss=0.033033864825963974
+INFO:root:
+[Valid][30][UTT][ MSE = 0.1658 | LCC = 0.9604 | SRCC = 0.9597 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.1802 | LCC = 0.9600 | SRCC = 0.9604 ]

results_tencent/hubert_base/layer5/run4/config.gin

@@ -0,0 +1,31 @@
+# Training config on BVCC dataset.
+
+# ------------------------------
+# Architecture
+# ------------------------------
+ProjectionHead:
+    conv_channels = (32, 32)
+    dense_neurons = (64, 32, 1)
+    use_poolings = (True, True)
+    kernel_size = 5
+    pool_size = 5
+    
+# ------------------------------
+# Dataset
+# ------------------------------
+Tencent:
+    debug = False
+
+get_dataloader:
+    num_workers = 12
+    shuffle = True
+
+# ------------------------------
+# Training loop
+# ------------------------------
+TrainingLoop:
+    num_epochs = 30
+    learning_rate = 1e-4
+    dataset_cls = @Tencent
+    weight_decay = 0.0
+    batch_size_train = 64

results_tencent/hubert_base/layer5/run4/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7bb0e742fd88269c090319bfacf25dcc2a6db8c6cfc9e67fac1d9d067313822e
+size 2747979

results_tencent/hubert_base/layer5/run4/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:506a1b4682972103e9ae09ef95a3994c4965217d709f649a10c0f0ea92b779b9
+size 2749702

results_tencent/hubert_base/layer5/run4/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 8000
+INFO:root:Num val speech clips: 2000
+INFO:root:Num test speech clips: 1563
+INFO:root:Device=cuda
+INFO:root:Average loss=0.5370765228271485
+INFO:root:
+[Valid][0][UTT][ MSE = 0.3321 | LCC = 0.9151 | SRCC = 0.9118 ]
+INFO:root:Average loss=0.1923546217083931
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2317 | LCC = 0.9417 | SRCC = 0.9393 ]
+INFO:root:Average loss=0.15295204663276674
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2365 | LCC = 0.9484 | SRCC = 0.9464 ]
+INFO:root:Average loss=0.13356065195798875
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2033 | LCC = 0.9523 | SRCC = 0.9507 ]
+INFO:root:Average loss=0.11815218222141266
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2046 | LCC = 0.9539 | SRCC = 0.9522 ]
+INFO:root:Average loss=0.10718410271406174
+INFO:root:
+[Valid][5][UTT][ MSE = 0.1783 | LCC = 0.9555 | SRCC = 0.9547 ]
+INFO:root:Average loss=0.09637917339801788
+INFO:root:
+[Valid][6][UTT][ MSE = 0.1929 | LCC = 0.9557 | SRCC = 0.9544 ]
+INFO:root:Average loss=0.0902817822098732
+INFO:root:
+[Valid][7][UTT][ MSE = 0.1979 | LCC = 0.9588 | SRCC = 0.9575 ]
+INFO:root:Average loss=0.08579420790076256
+INFO:root:
+[Valid][8][UTT][ MSE = 0.2054 | LCC = 0.9580 | SRCC = 0.9565 ]
+INFO:root:Average loss=0.07984266909956932
+INFO:root:
+[Valid][9][UTT][ MSE = 0.1855 | LCC = 0.9576 | SRCC = 0.9561 ]
+INFO:root:Average loss=0.07675537443161011
+INFO:root:
+[Valid][10][UTT][ MSE = 0.2074 | LCC = 0.9584 | SRCC = 0.9567 ]
+INFO:root:Average loss=0.07353569087386132
+INFO:root:
+[Valid][11][UTT][ MSE = 0.1952 | LCC = 0.9598 | SRCC = 0.9591 ]
+INFO:root:Average loss=0.06693949377536773
+INFO:root:
+[Valid][12][UTT][ MSE = 0.1671 | LCC = 0.9593 | SRCC = 0.9584 ]
+INFO:root:Average loss=0.065351326495409
+INFO:root:
+[Valid][13][UTT][ MSE = 0.1959 | LCC = 0.9600 | SRCC = 0.9589 ]
+INFO:root:Average loss=0.0597770706564188
+INFO:root:
+[Valid][14][UTT][ MSE = 0.1958 | LCC = 0.9599 | SRCC = 0.9586 ]
+INFO:root:Average loss=0.05761149644851685
+INFO:root:
+[Valid][15][UTT][ MSE = 0.1928 | LCC = 0.9602 | SRCC = 0.9595 ]
+INFO:root:Average loss=0.05435423913598061
+INFO:root:
+[Valid][16][UTT][ MSE = 0.2040 | LCC = 0.9602 | SRCC = 0.9596 ]
+INFO:root:Average loss=0.05272797477245331
+INFO:root:
+[Valid][17][UTT][ MSE = 0.1697 | LCC = 0.9606 | SRCC = 0.9594 ]
+INFO:root:Average loss=0.04896588778495788
+INFO:root:
+[Valid][18][UTT][ MSE = 0.1696 | LCC = 0.9613 | SRCC = 0.9601 ]
+INFO:root:Average loss=0.0479572434425354
+INFO:root:
+[Valid][19][UTT][ MSE = 0.1828 | LCC = 0.9610 | SRCC = 0.9596 ]
+INFO:root:Average loss=0.043995730489492416
+INFO:root:
+[Valid][20][UTT][ MSE = 0.1558 | LCC = 0.9603 | SRCC = 0.9589 ]
+INFO:root:Average loss=0.043300458177924155
+INFO:root:
+[Valid][21][UTT][ MSE = 0.1677 | LCC = 0.9610 | SRCC = 0.9598 ]
+INFO:root:Average loss=0.04187612639367581
+INFO:root:
+[Valid][22][UTT][ MSE = 0.2056 | LCC = 0.9598 | SRCC = 0.9594 ]
+INFO:root:Average loss=0.04159242367744446
+INFO:root:
+[Valid][23][UTT][ MSE = 0.1922 | LCC = 0.9604 | SRCC = 0.9593 ]
+INFO:root:Average loss=0.03936800995469093
+INFO:root:
+[Valid][24][UTT][ MSE = 0.1853 | LCC = 0.9607 | SRCC = 0.9596 ]
+INFO:root:Average loss=0.037449844270944596
+INFO:root:
+[Valid][25][UTT][ MSE = 0.1692 | LCC = 0.9607 | SRCC = 0.9596 ]
+INFO:root:Average loss=0.033755977541208265
+INFO:root:
+[Valid][26][UTT][ MSE = 0.1770 | LCC = 0.9603 | SRCC = 0.9592 ]
+INFO:root:Average loss=0.034037769943475724
+INFO:root:
+[Valid][27][UTT][ MSE = 0.1802 | LCC = 0.9605 | SRCC = 0.9588 ]
+INFO:root:Average loss=0.037080147787928584
+INFO:root:
+[Valid][28][UTT][ MSE = 0.2004 | LCC = 0.9605 | SRCC = 0.9600 ]
+INFO:root:Average loss=0.03280720244348049
+INFO:root:
+[Valid][29][UTT][ MSE = 0.1724 | LCC = 0.9601 | SRCC = 0.9588 ]
+INFO:root:Average loss=0.03155486150085926
+INFO:root:
+[Valid][30][UTT][ MSE = 0.1787 | LCC = 0.9607 | SRCC = 0.9593 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.1789 | LCC = 0.9605 | SRCC = 0.9610 ]

results_tencent/hubert_base/layer6/run0/config.gin

@@ -0,0 +1,31 @@
+# Training config on BVCC dataset.
+
+# ------------------------------
+# Architecture
+# ------------------------------
+ProjectionHead:
+    conv_channels = (32, 32)
+    dense_neurons = (64, 32, 1)
+    use_poolings = (True, True)
+    kernel_size = 5
+    pool_size = 5
+    
+# ------------------------------
+# Dataset
+# ------------------------------
+Tencent:
+    debug = False
+
+get_dataloader:
+    num_workers = 12
+    shuffle = True
+
+# ------------------------------
+# Training loop
+# ------------------------------
+TrainingLoop:
+    num_epochs = 30
+    learning_rate = 1e-4
+    dataset_cls = @Tencent
+    weight_decay = 0.0
+    batch_size_train = 64

results_tencent/hubert_base/layer6/run0/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b4793deea6eb25fd1d87c5288f93e8a290cfaa9e6126471dd2dabad6222083e5
+size 2747979

results_tencent/hubert_base/layer6/run0/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae7a0dffc7eea72b8420d150f6159f2366670edd5abacbb0006427b6e11d4bd8
+size 2749702

results_tencent/hubert_base/layer6/run0/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 8000
+INFO:root:Num val speech clips: 2000
+INFO:root:Num test speech clips: 1563
+INFO:root:Device=cuda
+INFO:root:Average loss=0.4860212173461914
+INFO:root:
+[Valid][0][UTT][ MSE = 0.3023 | LCC = 0.9217 | SRCC = 0.9193 ]
+INFO:root:Average loss=0.18159495234489442
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2373 | LCC = 0.9430 | SRCC = 0.9409 ]
+INFO:root:Average loss=0.14760519045591355
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2639 | LCC = 0.9478 | SRCC = 0.9475 ]
+INFO:root:Average loss=0.12826874774694444
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2005 | LCC = 0.9517 | SRCC = 0.9507 ]
+INFO:root:Average loss=0.11358935910463333
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2137 | LCC = 0.9538 | SRCC = 0.9531 ]
+INFO:root:Average loss=0.10559042143821716
+INFO:root:
+[Valid][5][UTT][ MSE = 0.1757 | LCC = 0.9527 | SRCC = 0.9525 ]
+INFO:root:Average loss=0.09736696517467498
+INFO:root:
+[Valid][6][UTT][ MSE = 0.1991 | LCC = 0.9544 | SRCC = 0.9534 ]
+INFO:root:Average loss=0.0931207927763462
+INFO:root:
+[Valid][7][UTT][ MSE = 0.1838 | LCC = 0.9547 | SRCC = 0.9536 ]
+INFO:root:Average loss=0.08458654436469078
+INFO:root:
+[Valid][8][UTT][ MSE = 0.1941 | LCC = 0.9566 | SRCC = 0.9559 ]
+INFO:root:Average loss=0.07769061848521233
+INFO:root:
+[Valid][9][UTT][ MSE = 0.1768 | LCC = 0.9556 | SRCC = 0.9548 ]
+INFO:root:Average loss=0.0725905313193798
+INFO:root:
+[Valid][10][UTT][ MSE = 0.1730 | LCC = 0.9578 | SRCC = 0.9568 ]
+INFO:root:Average loss=0.06779861024022102
+INFO:root:
+[Valid][11][UTT][ MSE = 0.1601 | LCC = 0.9576 | SRCC = 0.9572 ]
+INFO:root:Average loss=0.0643130929172039
+INFO:root:
+[Valid][12][UTT][ MSE = 0.1902 | LCC = 0.9578 | SRCC = 0.9565 ]
+INFO:root:Average loss=0.0600448796749115
+INFO:root:
+[Valid][13][UTT][ MSE = 0.1914 | LCC = 0.9577 | SRCC = 0.9577 ]
+INFO:root:Average loss=0.05863547882437706
+INFO:root:
+[Valid][14][UTT][ MSE = 0.1744 | LCC = 0.9577 | SRCC = 0.9568 ]
+INFO:root:Average loss=0.053800854325294495
+INFO:root:
+[Valid][15][UTT][ MSE = 0.1743 | LCC = 0.9584 | SRCC = 0.9572 ]
+INFO:root:Average loss=0.050546395227313044
+INFO:root:
+[Valid][16][UTT][ MSE = 0.1998 | LCC = 0.9584 | SRCC = 0.9578 ]
+INFO:root:Average loss=0.048676232919096944
+INFO:root:
+[Valid][17][UTT][ MSE = 0.1766 | LCC = 0.9576 | SRCC = 0.9563 ]
+INFO:root:Average loss=0.04625153458118439
+INFO:root:
+[Valid][18][UTT][ MSE = 0.1800 | LCC = 0.9594 | SRCC = 0.9587 ]
+INFO:root:Average loss=0.043752063050866125
+INFO:root:
+[Valid][19][UTT][ MSE = 0.2023 | LCC = 0.9567 | SRCC = 0.9572 ]
+INFO:root:Average loss=0.041726768523454666
+INFO:root:
+[Valid][20][UTT][ MSE = 0.1828 | LCC = 0.9593 | SRCC = 0.9580 ]
+INFO:root:Average loss=0.03992378796637058
+INFO:root:
+[Valid][21][UTT][ MSE = 0.1809 | LCC = 0.9585 | SRCC = 0.9575 ]
+INFO:root:Average loss=0.03818070361018181
+INFO:root:
+[Valid][22][UTT][ MSE = 0.1818 | LCC = 0.9582 | SRCC = 0.9572 ]
+INFO:root:Average loss=0.03598105804622173
+INFO:root:
+[Valid][23][UTT][ MSE = 0.1943 | LCC = 0.9586 | SRCC = 0.9580 ]
+INFO:root:Average loss=0.034207244515419004
+INFO:root:
+[Valid][24][UTT][ MSE = 0.1759 | LCC = 0.9583 | SRCC = 0.9574 ]
+INFO:root:Average loss=0.03371845646202564
+INFO:root:
+[Valid][25][UTT][ MSE = 0.1783 | LCC = 0.9597 | SRCC = 0.9588 ]
+INFO:root:Average loss=0.032926248893141746
+INFO:root:
+[Valid][26][UTT][ MSE = 0.1741 | LCC = 0.9602 | SRCC = 0.9593 ]
+INFO:root:Average loss=0.029713588669896127
+INFO:root:
+[Valid][27][UTT][ MSE = 0.1704 | LCC = 0.9598 | SRCC = 0.9590 ]
+INFO:root:Average loss=0.030475397616624832
+INFO:root:
+[Valid][28][UTT][ MSE = 0.1756 | LCC = 0.9604 | SRCC = 0.9594 ]
+INFO:root:Average loss=0.029691836163401604
+INFO:root:
+[Valid][29][UTT][ MSE = 0.1922 | LCC = 0.9595 | SRCC = 0.9588 ]
+INFO:root:Average loss=0.028685176894068718
+INFO:root:
+[Valid][30][UTT][ MSE = 0.1894 | LCC = 0.9591 | SRCC = 0.9589 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.1852 | LCC = 0.9585 | SRCC = 0.9585 ]

results_tencent/hubert_base/layer6/run1/config.gin

@@ -0,0 +1,31 @@
+# Training config on BVCC dataset.
+
+# ------------------------------
+# Architecture
+# ------------------------------
+ProjectionHead:
+    conv_channels = (32, 32)
+    dense_neurons = (64, 32, 1)
+    use_poolings = (True, True)
+    kernel_size = 5
+    pool_size = 5
+    
+# ------------------------------
+# Dataset
+# ------------------------------
+Tencent:
+    debug = False
+
+get_dataloader:
+    num_workers = 12
+    shuffle = True
+
+# ------------------------------
+# Training loop
+# ------------------------------
+TrainingLoop:
+    num_epochs = 30
+    learning_rate = 1e-4
+    dataset_cls = @Tencent
+    weight_decay = 0.0
+    batch_size_train = 64

results_tencent/hubert_base/layer6/run1/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89403567a273d9a5fd11ecaf843e8f06a3bae3f364a7fb39a3ee5cecf5cc17bf
+size 2747979

results_tencent/hubert_base/layer6/run1/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e11b242c3e5ca0356be5c3eb78477d8ae2b19b7007f48c7e731ca30ae7169215
+size 2749702

results_tencent/hubert_base/layer6/run1/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 8000
+INFO:root:Num val speech clips: 2000
+INFO:root:Num test speech clips: 1563
+INFO:root:Device=cuda
+INFO:root:Average loss=0.45139190936088563
+INFO:root:
+[Valid][0][UTT][ MSE = 0.3029 | LCC = 0.9204 | SRCC = 0.9195 ]
+INFO:root:Average loss=0.18781933081150054
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2384 | LCC = 0.9379 | SRCC = 0.9376 ]
+INFO:root:Average loss=0.15095506459474564
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2074 | LCC = 0.9429 | SRCC = 0.9440 ]
+INFO:root:Average loss=0.13250942754745484
+INFO:root:
+[Valid][3][UTT][ MSE = 0.1992 | LCC = 0.9477 | SRCC = 0.9472 ]
+INFO:root:Average loss=0.12005011171102524
+INFO:root:
+[Valid][4][UTT][ MSE = 0.1759 | LCC = 0.9512 | SRCC = 0.9516 ]
+INFO:root:Average loss=0.10952923047542572
+INFO:root:
+[Valid][5][UTT][ MSE = 0.1848 | LCC = 0.9537 | SRCC = 0.9534 ]
+INFO:root:Average loss=0.09852313885092735
+INFO:root:
+[Valid][6][UTT][ MSE = 0.1853 | LCC = 0.9516 | SRCC = 0.9516 ]
+INFO:root:Average loss=0.09237195628881455
+INFO:root:
+[Valid][7][UTT][ MSE = 0.1794 | LCC = 0.9560 | SRCC = 0.9552 ]
+INFO:root:Average loss=0.08177016216516494
+INFO:root:
+[Valid][8][UTT][ MSE = 0.1775 | LCC = 0.9551 | SRCC = 0.9541 ]
+INFO:root:Average loss=0.08053918594121932
+INFO:root:
+[Valid][9][UTT][ MSE = 0.1877 | LCC = 0.9563 | SRCC = 0.9553 ]
+INFO:root:Average loss=0.07255291226506233
+INFO:root:
+[Valid][10][UTT][ MSE = 0.1627 | LCC = 0.9562 | SRCC = 0.9555 ]
+INFO:root:Average loss=0.07130857136845589
+INFO:root:
+[Valid][11][UTT][ MSE = 0.1682 | LCC = 0.9570 | SRCC = 0.9559 ]
+INFO:root:Average loss=0.06450583267211914
+INFO:root:
+[Valid][12][UTT][ MSE = 0.1548 | LCC = 0.9563 | SRCC = 0.9561 ]
+INFO:root:Average loss=0.06175072214007377
+INFO:root:
+[Valid][13][UTT][ MSE = 0.1790 | LCC = 0.9573 | SRCC = 0.9564 ]
+INFO:root:Average loss=0.05693508513271808
+INFO:root:
+[Valid][14][UTT][ MSE = 0.1912 | LCC = 0.9566 | SRCC = 0.9553 ]
+INFO:root:Average loss=0.05695650634169579
+INFO:root:
+[Valid][15][UTT][ MSE = 0.1694 | LCC = 0.9561 | SRCC = 0.9549 ]
+INFO:root:Average loss=0.04971479913592339
+INFO:root:
+[Valid][16][UTT][ MSE = 0.1613 | LCC = 0.9556 | SRCC = 0.9561 ]
+INFO:root:Average loss=0.04990753784775734
+INFO:root:
+[Valid][17][UTT][ MSE = 0.1783 | LCC = 0.9574 | SRCC = 0.9565 ]
+INFO:root:Average loss=0.04656219618022442
+INFO:root:
+[Valid][18][UTT][ MSE = 0.1590 | LCC = 0.9578 | SRCC = 0.9573 ]
+INFO:root:Average loss=0.04499005725979805
+INFO:root:
+[Valid][19][UTT][ MSE = 0.1799 | LCC = 0.9580 | SRCC = 0.9572 ]
+INFO:root:Average loss=0.04280725106596947
+INFO:root:
+[Valid][20][UTT][ MSE = 0.1967 | LCC = 0.9575 | SRCC = 0.9566 ]
+INFO:root:Average loss=0.04136967197060585
+INFO:root:
+[Valid][21][UTT][ MSE = 0.1813 | LCC = 0.9574 | SRCC = 0.9567 ]
+INFO:root:Average loss=0.04063347715139389
+INFO:root:
+[Valid][22][UTT][ MSE = 0.1614 | LCC = 0.9580 | SRCC = 0.9573 ]
+INFO:root:Average loss=0.037435056164860724
+INFO:root:
+[Valid][23][UTT][ MSE = 0.1836 | LCC = 0.9580 | SRCC = 0.9569 ]
+INFO:root:Average loss=0.036187818929553035
+INFO:root:
+[Valid][24][UTT][ MSE = 0.1881 | LCC = 0.9575 | SRCC = 0.9562 ]
+INFO:root:Average loss=0.03592694228887558
+INFO:root:
+[Valid][25][UTT][ MSE = 0.1838 | LCC = 0.9577 | SRCC = 0.9568 ]
+INFO:root:Average loss=0.034170073181390766
+INFO:root:
+[Valid][26][UTT][ MSE = 0.1940 | LCC = 0.9579 | SRCC = 0.9568 ]
+INFO:root:Average loss=0.03472886946797371
+INFO:root:
+[Valid][27][UTT][ MSE = 0.1771 | LCC = 0.9587 | SRCC = 0.9575 ]
+INFO:root:Average loss=0.033754281431436536
+INFO:root:
+[Valid][28][UTT][ MSE = 0.1748 | LCC = 0.9590 | SRCC = 0.9577 ]
+INFO:root:Average loss=0.030566699728369714
+INFO:root:
+[Valid][29][UTT][ MSE = 0.1775 | LCC = 0.9573 | SRCC = 0.9564 ]
+INFO:root:Average loss=0.02976990607380867
+INFO:root:
+[Valid][30][UTT][ MSE = 0.1704 | LCC = 0.9587 | SRCC = 0.9578 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.1787 | LCC = 0.9582 | SRCC = 0.9580 ]