Upload SSL Layer MOS checkpoints (part 14)

results_nisqa/w2v2_xlsr_2b/layer36/run3/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:937d97ce136bf5023c13f10f8303e89b90dff26f52e0d2140f491d24995362d8
+size 5697099

results_nisqa/w2v2_xlsr_2b/layer36/run3/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33e13706f4263b51110d253420df64526314a1d2ffe29597874169427553bf36
+size 5698822

results_nisqa/w2v2_xlsr_2b/layer36/run3/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 11020
+INFO:root:Num val speech clips: 2700
+INFO:root:Num test speech clips: 232
+INFO:root:Device=cuda
+INFO:root:Average loss=0.34114912875814934
+INFO:root:
+[Valid][0][UTT][ MSE = 0.2653 | LCC = 0.9093 | SRCC = 0.9094 ]
+INFO:root:Average loss=0.21932248125186546
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2829 | LCC = 0.9132 | SRCC = 0.9126 ]
+INFO:root:Average loss=0.19273635822569016
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2599 | LCC = 0.9176 | SRCC = 0.9171 ]
+INFO:root:Average loss=0.17890871923438387
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2395 | LCC = 0.9188 | SRCC = 0.9181 ]
+INFO:root:Average loss=0.17102571469166375
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2625 | LCC = 0.9196 | SRCC = 0.9191 ]
+INFO:root:Average loss=0.15810992004554394
+INFO:root:
+[Valid][5][UTT][ MSE = 0.2633 | LCC = 0.9196 | SRCC = 0.9186 ]
+INFO:root:Average loss=0.15764302389986942
+INFO:root:
+[Valid][6][UTT][ MSE = 0.2954 | LCC = 0.9174 | SRCC = 0.9190 ]
+INFO:root:Average loss=0.1489934727463419
+INFO:root:
+[Valid][7][UTT][ MSE = 0.2519 | LCC = 0.9196 | SRCC = 0.9188 ]
+INFO:root:Average loss=0.1393150646734789
+INFO:root:
+[Valid][8][UTT][ MSE = 0.2957 | LCC = 0.9184 | SRCC = 0.9187 ]
+INFO:root:Average loss=0.1308250139128266
+INFO:root:
+[Valid][9][UTT][ MSE = 0.2629 | LCC = 0.9169 | SRCC = 0.9161 ]
+INFO:root:Average loss=0.13108045879126973
+INFO:root:
+[Valid][10][UTT][ MSE = 0.2528 | LCC = 0.9166 | SRCC = 0.9180 ]
+INFO:root:Average loss=0.11990281271968962
+INFO:root:
+[Valid][11][UTT][ MSE = 0.2211 | LCC = 0.9166 | SRCC = 0.9156 ]
+INFO:root:Average loss=0.11580338277396439
+INFO:root:
+[Valid][12][UTT][ MSE = 0.2243 | LCC = 0.9170 | SRCC = 0.9164 ]
+INFO:root:Average loss=0.10611730172282699
+INFO:root:
+[Valid][13][UTT][ MSE = 0.2446 | LCC = 0.9166 | SRCC = 0.9152 ]
+INFO:root:Average loss=0.10415011019892775
+INFO:root:
+[Valid][14][UTT][ MSE = 0.2438 | LCC = 0.9140 | SRCC = 0.9148 ]
+INFO:root:Average loss=0.09659749560187318
+INFO:root:
+[Valid][15][UTT][ MSE = 0.2338 | LCC = 0.9158 | SRCC = 0.9142 ]
+INFO:root:Average loss=0.08946695960412136
+INFO:root:
+[Valid][16][UTT][ MSE = 0.2415 | LCC = 0.9139 | SRCC = 0.9130 ]
+INFO:root:Average loss=0.08559628326252017
+INFO:root:
+[Valid][17][UTT][ MSE = 0.2441 | LCC = 0.9140 | SRCC = 0.9138 ]
+INFO:root:Average loss=0.0821648394484396
+INFO:root:
+[Valid][18][UTT][ MSE = 0.2407 | LCC = 0.9142 | SRCC = 0.9130 ]
+INFO:root:Average loss=0.07587608707950294
+INFO:root:
+[Valid][19][UTT][ MSE = 0.2540 | LCC = 0.9127 | SRCC = 0.9109 ]
+INFO:root:Average loss=0.07441908726199514
+INFO:root:
+[Valid][20][UTT][ MSE = 0.2468 | LCC = 0.9133 | SRCC = 0.9122 ]
+INFO:root:Average loss=0.06918234888442679
+INFO:root:
+[Valid][21][UTT][ MSE = 0.2532 | LCC = 0.9108 | SRCC = 0.9115 ]
+INFO:root:Average loss=0.06427377828456074
+INFO:root:
+[Valid][22][UTT][ MSE = 0.2436 | LCC = 0.9116 | SRCC = 0.9105 ]
+INFO:root:Average loss=0.06675699493185633
+INFO:root:
+[Valid][23][UTT][ MSE = 0.2592 | LCC = 0.9102 | SRCC = 0.9115 ]
+INFO:root:Average loss=0.06094985429434418
+INFO:root:
+[Valid][24][UTT][ MSE = 0.2579 | LCC = 0.9089 | SRCC = 0.9096 ]
+INFO:root:Average loss=0.05724435702310821
+INFO:root:
+[Valid][25][UTT][ MSE = 0.2395 | LCC = 0.9118 | SRCC = 0.9112 ]
+INFO:root:Average loss=0.05621382077305303
+INFO:root:
+[Valid][26][UTT][ MSE = 0.2652 | LCC = 0.9084 | SRCC = 0.9101 ]
+INFO:root:Average loss=0.05416408233947492
+INFO:root:
+[Valid][27][UTT][ MSE = 0.2413 | LCC = 0.9106 | SRCC = 0.9111 ]
+INFO:root:Average loss=0.051780555827941506
+INFO:root:
+[Valid][28][UTT][ MSE = 0.2481 | LCC = 0.9115 | SRCC = 0.9120 ]
+INFO:root:Average loss=0.04852320665138305
+INFO:root:
+[Valid][29][UTT][ MSE = 0.2354 | LCC = 0.9104 | SRCC = 0.9110 ]
+INFO:root:Average loss=0.0499665065092503
+INFO:root:
+[Valid][30][UTT][ MSE = 0.2509 | LCC = 0.9118 | SRCC = 0.9101 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.2180 | LCC = 0.9139 | SRCC = 0.9094 ]

results_nisqa/w2v2_xlsr_2b/layer36/run4/config.gin

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

results_nisqa/w2v2_xlsr_2b/layer36/run4/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4f278ef0b1a547a26de56afb29d16060fb9b2663c9ded3144470573e4f6da8c
+size 5697099

results_nisqa/w2v2_xlsr_2b/layer36/run4/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cdb1865f72ae4d6425cfa422ad73591deb02cd089ad2b2215e2ce22dbff81c86
+size 5698758

results_nisqa/w2v2_xlsr_2b/layer36/run4/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 11020
+INFO:root:Num val speech clips: 2700
+INFO:root:Num test speech clips: 232
+INFO:root:Device=cuda
+INFO:root:Average loss=0.3633779720161002
+INFO:root:
+[Valid][0][UTT][ MSE = 0.2825 | LCC = 0.9104 | SRCC = 0.9098 ]
+INFO:root:Average loss=0.20649637445549054
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2591 | LCC = 0.9150 | SRCC = 0.9146 ]
+INFO:root:Average loss=0.19608437308686316
+INFO:root:
+[Valid][2][UTT][ MSE = 0.3028 | LCC = 0.9130 | SRCC = 0.9141 ]
+INFO:root:Average loss=0.18323846785836137
+INFO:root:
+[Valid][3][UTT][ MSE = 0.3067 | LCC = 0.9162 | SRCC = 0.9174 ]
+INFO:root:Average loss=0.17238816834254073
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2737 | LCC = 0.9159 | SRCC = 0.9186 ]
+INFO:root:Average loss=0.1656403662152373
+INFO:root:
+[Valid][5][UTT][ MSE = 0.2546 | LCC = 0.9199 | SRCC = 0.9192 ]
+INFO:root:Average loss=0.15183146478812817
+INFO:root:
+[Valid][6][UTT][ MSE = 0.2543 | LCC = 0.9190 | SRCC = 0.9181 ]
+INFO:root:Average loss=0.1443551750799824
+INFO:root:
+[Valid][7][UTT][ MSE = 0.2355 | LCC = 0.9181 | SRCC = 0.9172 ]
+INFO:root:Average loss=0.14240306271293948
+INFO:root:
+[Valid][8][UTT][ MSE = 0.2198 | LCC = 0.9183 | SRCC = 0.9177 ]
+INFO:root:Average loss=0.12912541459922847
+INFO:root:
+[Valid][9][UTT][ MSE = 0.2596 | LCC = 0.9164 | SRCC = 0.9153 ]
+INFO:root:Average loss=0.13196023041113264
+INFO:root:
+[Valid][10][UTT][ MSE = 0.2646 | LCC = 0.9158 | SRCC = 0.9146 ]
+INFO:root:Average loss=0.11997475584893558
+INFO:root:
+[Valid][11][UTT][ MSE = 0.2450 | LCC = 0.9156 | SRCC = 0.9161 ]
+INFO:root:Average loss=0.11280032591840436
+INFO:root:
+[Valid][12][UTT][ MSE = 0.2696 | LCC = 0.9152 | SRCC = 0.9149 ]
+INFO:root:Average loss=0.10825414404359167
+INFO:root:
+[Valid][13][UTT][ MSE = 0.2382 | LCC = 0.9137 | SRCC = 0.9142 ]
+INFO:root:Average loss=0.1020082690553858
+INFO:root:
+[Valid][14][UTT][ MSE = 0.2229 | LCC = 0.9144 | SRCC = 0.9148 ]
+INFO:root:Average loss=0.09829820320010185
+INFO:root:
+[Valid][15][UTT][ MSE = 0.2224 | LCC = 0.9154 | SRCC = 0.9147 ]
+INFO:root:Average loss=0.09135327722898798
+INFO:root:
+[Valid][16][UTT][ MSE = 0.2630 | LCC = 0.9098 | SRCC = 0.9113 ]
+INFO:root:Average loss=0.08702370644993865
+INFO:root:
+[Valid][17][UTT][ MSE = 0.2598 | LCC = 0.9106 | SRCC = 0.9116 ]
+INFO:root:Average loss=0.08361529102387456
+INFO:root:
+[Valid][18][UTT][ MSE = 0.2456 | LCC = 0.9117 | SRCC = 0.9125 ]
+INFO:root:Average loss=0.07584702628368588
+INFO:root:
+[Valid][19][UTT][ MSE = 0.2642 | LCC = 0.9113 | SRCC = 0.9098 ]
+INFO:root:Average loss=0.07364652906625258
+INFO:root:
+[Valid][20][UTT][ MSE = 0.2479 | LCC = 0.9126 | SRCC = 0.9112 ]
+INFO:root:Average loss=0.07163937094842078
+INFO:root:
+[Valid][21][UTT][ MSE = 0.2341 | LCC = 0.9119 | SRCC = 0.9111 ]
+INFO:root:Average loss=0.0705772734144864
+INFO:root:
+[Valid][22][UTT][ MSE = 0.2334 | LCC = 0.9137 | SRCC = 0.9127 ]
+INFO:root:Average loss=0.06145699755046409
+INFO:root:
+[Valid][23][UTT][ MSE = 0.2494 | LCC = 0.9094 | SRCC = 0.9102 ]
+INFO:root:Average loss=0.06300608964496954
+INFO:root:
+[Valid][24][UTT][ MSE = 0.2502 | LCC = 0.9099 | SRCC = 0.9090 ]
+INFO:root:Average loss=0.061450453615085235
+INFO:root:
+[Valid][25][UTT][ MSE = 0.2500 | LCC = 0.9090 | SRCC = 0.9086 ]
+INFO:root:Average loss=0.05682381492301908
+INFO:root:
+[Valid][26][UTT][ MSE = 0.2414 | LCC = 0.9085 | SRCC = 0.9090 ]
+INFO:root:Average loss=0.05406042991604419
+INFO:root:
+[Valid][27][UTT][ MSE = 0.2543 | LCC = 0.9088 | SRCC = 0.9092 ]
+INFO:root:Average loss=0.05193365442787291
+INFO:root:
+[Valid][28][UTT][ MSE = 0.2602 | LCC = 0.9070 | SRCC = 0.9070 ]
+INFO:root:Average loss=0.04985050074325476
+INFO:root:
+[Valid][29][UTT][ MSE = 0.2620 | LCC = 0.9087 | SRCC = 0.9094 ]
+INFO:root:Average loss=0.04771909820766463
+INFO:root:
+[Valid][30][UTT][ MSE = 0.2355 | LCC = 0.9106 | SRCC = 0.9095 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.2199 | LCC = 0.9200 | SRCC = 0.9144 ]

results_nisqa/w2v2_xlsr_2b/layer38/run0/config.gin

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

results_nisqa/w2v2_xlsr_2b/layer38/run0/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4dc67ba1bc5f802e6d0c0b6c8879579d128caf693f5ac6a2d5915fce468e55d
+size 5697099

results_nisqa/w2v2_xlsr_2b/layer38/run0/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9ebe64421afcc60cd6f7b0a3e09258bee7edb9f44458adc0096ebddd5c2b75c
+size 5698758

results_nisqa/w2v2_xlsr_2b/layer38/run0/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 11020
+INFO:root:Num val speech clips: 2700
+INFO:root:Num test speech clips: 232
+INFO:root:Device=cuda
+INFO:root:Average loss=0.3573866956840361
+INFO:root:
+[Valid][0][UTT][ MSE = 0.2639 | LCC = 0.9105 | SRCC = 0.9128 ]
+INFO:root:Average loss=0.2084711740557858
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2858 | LCC = 0.9173 | SRCC = 0.9170 ]
+INFO:root:Average loss=0.18633931784788307
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2582 | LCC = 0.9186 | SRCC = 0.9182 ]
+INFO:root:Average loss=0.17921920874387542
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2471 | LCC = 0.9197 | SRCC = 0.9201 ]
+INFO:root:Average loss=0.16782769272265408
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2443 | LCC = 0.9194 | SRCC = 0.9190 ]
+INFO:root:Average loss=0.16278059392995228
+INFO:root:
+[Valid][5][UTT][ MSE = 0.2649 | LCC = 0.9209 | SRCC = 0.9211 ]
+INFO:root:Average loss=0.1517497328664526
+INFO:root:
+[Valid][6][UTT][ MSE = 0.2339 | LCC = 0.9211 | SRCC = 0.9206 ]
+INFO:root:Average loss=0.14310941147494177
+INFO:root:
+[Valid][7][UTT][ MSE = 0.2440 | LCC = 0.9198 | SRCC = 0.9201 ]
+INFO:root:Average loss=0.13917590152321524
+INFO:root:
+[Valid][8][UTT][ MSE = 0.2677 | LCC = 0.9205 | SRCC = 0.9195 ]
+INFO:root:Average loss=0.1312613497579718
+INFO:root:
+[Valid][9][UTT][ MSE = 0.2443 | LCC = 0.9205 | SRCC = 0.9193 ]
+INFO:root:Average loss=0.12194313384549466
+INFO:root:
+[Valid][10][UTT][ MSE = 0.2701 | LCC = 0.9170 | SRCC = 0.9190 ]
+INFO:root:Average loss=0.12343103615190253
+INFO:root:
+[Valid][11][UTT][ MSE = 0.2479 | LCC = 0.9185 | SRCC = 0.9184 ]
+INFO:root:Average loss=0.11066144248309163
+INFO:root:
+[Valid][12][UTT][ MSE = 0.2355 | LCC = 0.9174 | SRCC = 0.9170 ]
+INFO:root:Average loss=0.10586180815720833
+INFO:root:
+[Valid][13][UTT][ MSE = 0.2499 | LCC = 0.9178 | SRCC = 0.9172 ]
+INFO:root:Average loss=0.1007692071829917
+INFO:root:
+[Valid][14][UTT][ MSE = 0.2370 | LCC = 0.9149 | SRCC = 0.9155 ]
+INFO:root:Average loss=0.09697600533765864
+INFO:root:
+[Valid][15][UTT][ MSE = 0.2840 | LCC = 0.9149 | SRCC = 0.9146 ]
+INFO:root:Average loss=0.09308018886371155
+INFO:root:
+[Valid][16][UTT][ MSE = 0.2536 | LCC = 0.9151 | SRCC = 0.9141 ]
+INFO:root:Average loss=0.09009218631538353
+INFO:root:
+[Valid][17][UTT][ MSE = 0.2392 | LCC = 0.9145 | SRCC = 0.9146 ]
+INFO:root:Average loss=0.0800325024033213
+INFO:root:
+[Valid][18][UTT][ MSE = 0.2505 | LCC = 0.9147 | SRCC = 0.9140 ]
+INFO:root:Average loss=0.07555471017224595
+INFO:root:
+[Valid][19][UTT][ MSE = 0.2402 | LCC = 0.9122 | SRCC = 0.9132 ]
+INFO:root:Average loss=0.07462900818858532
+INFO:root:
+[Valid][20][UTT][ MSE = 0.2598 | LCC = 0.9104 | SRCC = 0.9116 ]
+INFO:root:Average loss=0.07392941957953349
+INFO:root:
+[Valid][21][UTT][ MSE = 0.2592 | LCC = 0.9140 | SRCC = 0.9126 ]
+INFO:root:Average loss=0.06581736142838622
+INFO:root:
+[Valid][22][UTT][ MSE = 0.2722 | LCC = 0.9116 | SRCC = 0.9102 ]
+INFO:root:Average loss=0.06413208495157992
+INFO:root:
+[Valid][23][UTT][ MSE = 0.2470 | LCC = 0.9135 | SRCC = 0.9127 ]
+INFO:root:Average loss=0.06026931763815053
+INFO:root:
+[Valid][24][UTT][ MSE = 0.2497 | LCC = 0.9116 | SRCC = 0.9120 ]
+INFO:root:Average loss=0.056638996617470176
+INFO:root:
+[Valid][25][UTT][ MSE = 0.2321 | LCC = 0.9102 | SRCC = 0.9114 ]
+INFO:root:Average loss=0.053174568318648835
+INFO:root:
+[Valid][26][UTT][ MSE = 0.2611 | LCC = 0.9101 | SRCC = 0.9107 ]
+INFO:root:Average loss=0.05612329578813101
+INFO:root:
+[Valid][27][UTT][ MSE = 0.2582 | LCC = 0.9126 | SRCC = 0.9123 ]
+INFO:root:Average loss=0.05068570681217778
+INFO:root:
+[Valid][28][UTT][ MSE = 0.2516 | LCC = 0.9089 | SRCC = 0.9097 ]
+INFO:root:Average loss=0.04872862295909768
+INFO:root:
+[Valid][29][UTT][ MSE = 0.2409 | LCC = 0.9103 | SRCC = 0.9107 ]
+INFO:root:Average loss=0.04688998436652167
+INFO:root:
+[Valid][30][UTT][ MSE = 0.2576 | LCC = 0.9107 | SRCC = 0.9099 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.1793 | LCC = 0.9263 | SRCC = 0.9217 ]

results_nisqa/w2v2_xlsr_2b/layer38/run1/config.gin

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

results_nisqa/w2v2_xlsr_2b/layer38/run1/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f2c12f1d2a07e54b7a2f7eb1c94102f3cac1b0234dba33f7298efa7bfaaf6625
+size 5697099

results_nisqa/w2v2_xlsr_2b/layer38/run1/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:767570efa2e66a13d14bdb3f19fc1fe8cd39bb92640fef2e1f7c4c60ffd6a76d
+size 5698758

results_nisqa/w2v2_xlsr_2b/layer38/run1/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 11020
+INFO:root:Num val speech clips: 2700
+INFO:root:Num test speech clips: 232
+INFO:root:Device=cuda
+INFO:root:Average loss=0.3701450997866647
+INFO:root:
+[Valid][0][UTT][ MSE = 0.2995 | LCC = 0.9103 | SRCC = 0.9117 ]
+INFO:root:Average loss=0.2066600604811845
+INFO:root:
+[Valid][1][UTT][ MSE = 0.3051 | LCC = 0.9144 | SRCC = 0.9138 ]
+INFO:root:Average loss=0.19030084314546145
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2612 | LCC = 0.9184 | SRCC = 0.9181 ]
+INFO:root:Average loss=0.18301058590756675
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2652 | LCC = 0.9178 | SRCC = 0.9190 ]
+INFO:root:Average loss=0.16856661460475425
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2441 | LCC = 0.9195 | SRCC = 0.9199 ]
+INFO:root:Average loss=0.17055293151064416
+INFO:root:
+[Valid][5][UTT][ MSE = 0.2423 | LCC = 0.9210 | SRCC = 0.9206 ]
+INFO:root:Average loss=0.15293642322968887
+INFO:root:
+[Valid][6][UTT][ MSE = 0.2481 | LCC = 0.9204 | SRCC = 0.9203 ]
+INFO:root:Average loss=0.1536116985485733
+INFO:root:
+[Valid][7][UTT][ MSE = 0.2486 | LCC = 0.9210 | SRCC = 0.9204 ]
+INFO:root:Average loss=0.1417658081805775
+INFO:root:
+[Valid][8][UTT][ MSE = 0.2483 | LCC = 0.9201 | SRCC = 0.9194 ]
+INFO:root:Average loss=0.13323315482780423
+INFO:root:
+[Valid][9][UTT][ MSE = 0.2599 | LCC = 0.9207 | SRCC = 0.9208 ]
+INFO:root:Average loss=0.13293945095959425
+INFO:root:
+[Valid][10][UTT][ MSE = 0.2393 | LCC = 0.9195 | SRCC = 0.9209 ]
+INFO:root:Average loss=0.12451318862465764
+INFO:root:
+[Valid][11][UTT][ MSE = 0.2203 | LCC = 0.9174 | SRCC = 0.9176 ]
+INFO:root:Average loss=0.11940954789260909
+INFO:root:
+[Valid][12][UTT][ MSE = 0.2543 | LCC = 0.9193 | SRCC = 0.9188 ]
+INFO:root:Average loss=0.10964984875280044
+INFO:root:
+[Valid][13][UTT][ MSE = 0.2348 | LCC = 0.9185 | SRCC = 0.9178 ]
+INFO:root:Average loss=0.1041015455919194
+INFO:root:
+[Valid][14][UTT][ MSE = 0.2443 | LCC = 0.9157 | SRCC = 0.9157 ]
+INFO:root:Average loss=0.10195955978652646
+INFO:root:
+[Valid][15][UTT][ MSE = 0.2446 | LCC = 0.9180 | SRCC = 0.9183 ]
+INFO:root:Average loss=0.09697722333531848
+INFO:root:
+[Valid][16][UTT][ MSE = 0.2565 | LCC = 0.9168 | SRCC = 0.9159 ]
+INFO:root:Average loss=0.08986225011462422
+INFO:root:
+[Valid][17][UTT][ MSE = 0.2483 | LCC = 0.9138 | SRCC = 0.9140 ]
+INFO:root:Average loss=0.08316023395843589
+INFO:root:
+[Valid][18][UTT][ MSE = 0.2839 | LCC = 0.9088 | SRCC = 0.9124 ]
+INFO:root:Average loss=0.08225621808471018
+INFO:root:
+[Valid][19][UTT][ MSE = 0.2294 | LCC = 0.9153 | SRCC = 0.9154 ]
+INFO:root:Average loss=0.07777839237985583
+INFO:root:
+[Valid][20][UTT][ MSE = 0.2309 | LCC = 0.9148 | SRCC = 0.9148 ]
+INFO:root:Average loss=0.07256104950012499
+INFO:root:
+[Valid][21][UTT][ MSE = 0.2533 | LCC = 0.9131 | SRCC = 0.9126 ]
+INFO:root:Average loss=0.07099192309586298
+INFO:root:
+[Valid][22][UTT][ MSE = 0.2550 | LCC = 0.9152 | SRCC = 0.9143 ]
+INFO:root:Average loss=0.0644832938651129
+INFO:root:
+[Valid][23][UTT][ MSE = 0.2304 | LCC = 0.9120 | SRCC = 0.9127 ]
+INFO:root:Average loss=0.06363135169265587
+INFO:root:
+[Valid][24][UTT][ MSE = 0.2462 | LCC = 0.9117 | SRCC = 0.9131 ]
+INFO:root:Average loss=0.06210611584041849
+INFO:root:
+[Valid][25][UTT][ MSE = 0.2327 | LCC = 0.9113 | SRCC = 0.9130 ]
+INFO:root:Average loss=0.05778973609898132
+INFO:root:
+[Valid][26][UTT][ MSE = 0.2282 | LCC = 0.9126 | SRCC = 0.9118 ]
+INFO:root:Average loss=0.05881890247581322
+INFO:root:
+[Valid][27][UTT][ MSE = 0.2643 | LCC = 0.9100 | SRCC = 0.9106 ]
+INFO:root:Average loss=0.05335093823643778
+INFO:root:
+[Valid][28][UTT][ MSE = 0.2398 | LCC = 0.9120 | SRCC = 0.9116 ]
+INFO:root:Average loss=0.05170987105007806
+INFO:root:
+[Valid][29][UTT][ MSE = 0.2388 | LCC = 0.9126 | SRCC = 0.9119 ]
+INFO:root:Average loss=0.04872577803241724
+INFO:root:
+[Valid][30][UTT][ MSE = 0.2428 | LCC = 0.9117 | SRCC = 0.9119 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.2048 | LCC = 0.9251 | SRCC = 0.9186 ]

results_nisqa/w2v2_xlsr_2b/layer38/run2/config.gin

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

results_nisqa/w2v2_xlsr_2b/layer38/run2/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec67279a366fddeca9a1c395e1be190c6c7060d1a539dc6e2c6f82f6b5655454
+size 5697099

results_nisqa/w2v2_xlsr_2b/layer38/run2/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8bd4691c8b3c6c54c2383b3ba30444cb8fa8e3eb5681964351a6168c59e0ed8c
+size 5698822

results_nisqa/w2v2_xlsr_2b/layer38/run2/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 11020
+INFO:root:Num val speech clips: 2700
+INFO:root:Num test speech clips: 232
+INFO:root:Device=cuda
+INFO:root:Average loss=0.35254366295358347
+INFO:root:
+[Valid][0][UTT][ MSE = 0.2876 | LCC = 0.9112 | SRCC = 0.9117 ]
+INFO:root:Average loss=0.20144968062122434
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2631 | LCC = 0.9168 | SRCC = 0.9170 ]
+INFO:root:Average loss=0.186714480507236
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2793 | LCC = 0.9167 | SRCC = 0.9177 ]
+INFO:root:Average loss=0.1759697659253385
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2862 | LCC = 0.9196 | SRCC = 0.9204 ]
+INFO:root:Average loss=0.17317569578830908
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2404 | LCC = 0.9210 | SRCC = 0.9203 ]
+INFO:root:Average loss=0.1585929570342764
+INFO:root:
+[Valid][5][UTT][ MSE = 0.2285 | LCC = 0.9202 | SRCC = 0.9199 ]
+INFO:root:Average loss=0.1536314383235281
+INFO:root:
+[Valid][6][UTT][ MSE = 0.2337 | LCC = 0.9202 | SRCC = 0.9206 ]
+INFO:root:Average loss=0.14856330176137086
+INFO:root:
+[Valid][7][UTT][ MSE = 0.2473 | LCC = 0.9210 | SRCC = 0.9204 ]
+INFO:root:Average loss=0.1404270191240862
+INFO:root:
+[Valid][8][UTT][ MSE = 0.2284 | LCC = 0.9222 | SRCC = 0.9219 ]
+INFO:root:Average loss=0.1388624220483565
+INFO:root:
+[Valid][9][UTT][ MSE = 0.2442 | LCC = 0.9204 | SRCC = 0.9202 ]
+INFO:root:Average loss=0.1285193006944105
+INFO:root:
+[Valid][10][UTT][ MSE = 0.2456 | LCC = 0.9197 | SRCC = 0.9199 ]
+INFO:root:Average loss=0.12413563966923366
+INFO:root:
+[Valid][11][UTT][ MSE = 0.2540 | LCC = 0.9194 | SRCC = 0.9191 ]
+INFO:root:Average loss=0.11518122212735214
+INFO:root:
+[Valid][12][UTT][ MSE = 0.2523 | LCC = 0.9189 | SRCC = 0.9191 ]
+INFO:root:Average loss=0.11208878595360441
+INFO:root:
+[Valid][13][UTT][ MSE = 0.2292 | LCC = 0.9183 | SRCC = 0.9171 ]
+INFO:root:Average loss=0.10707279049247675
+INFO:root:
+[Valid][14][UTT][ MSE = 0.2269 | LCC = 0.9190 | SRCC = 0.9178 ]
+INFO:root:Average loss=0.09820114803968827
+INFO:root:
+[Valid][15][UTT][ MSE = 0.2342 | LCC = 0.9173 | SRCC = 0.9162 ]
+INFO:root:Average loss=0.09673374585505855
+INFO:root:
+[Valid][16][UTT][ MSE = 0.2300 | LCC = 0.9181 | SRCC = 0.9166 ]
+INFO:root:Average loss=0.09015973732736758
+INFO:root:
+[Valid][17][UTT][ MSE = 0.2477 | LCC = 0.9157 | SRCC = 0.9148 ]
+INFO:root:Average loss=0.08398223959038713
+INFO:root:
+[Valid][18][UTT][ MSE = 0.2199 | LCC = 0.9168 | SRCC = 0.9159 ]
+INFO:root:Average loss=0.07887843989826351
+INFO:root:
+[Valid][19][UTT][ MSE = 0.2433 | LCC = 0.9145 | SRCC = 0.9144 ]
+INFO:root:Average loss=0.07614357323746461
+INFO:root:
+[Valid][20][UTT][ MSE = 0.2409 | LCC = 0.9153 | SRCC = 0.9148 ]
+INFO:root:Average loss=0.0716572009199272
+INFO:root:
+[Valid][21][UTT][ MSE = 0.2384 | LCC = 0.9140 | SRCC = 0.9131 ]
+INFO:root:Average loss=0.06865802198390052
+INFO:root:
+[Valid][22][UTT][ MSE = 0.2529 | LCC = 0.9135 | SRCC = 0.9136 ]
+INFO:root:Average loss=0.06603247238419993
+INFO:root:
+[Valid][23][UTT][ MSE = 0.2392 | LCC = 0.9132 | SRCC = 0.9128 ]
+INFO:root:Average loss=0.06423431561689157
+INFO:root:
+[Valid][24][UTT][ MSE = 0.2514 | LCC = 0.9132 | SRCC = 0.9124 ]
+INFO:root:Average loss=0.05865720825302118
+INFO:root:
+[Valid][25][UTT][ MSE = 0.2391 | LCC = 0.9134 | SRCC = 0.9124 ]
+INFO:root:Average loss=0.059297107590760796
+INFO:root:
+[Valid][26][UTT][ MSE = 0.2609 | LCC = 0.9126 | SRCC = 0.9119 ]
+INFO:root:Average loss=0.055286865724476775
+INFO:root:
+[Valid][27][UTT][ MSE = 0.2343 | LCC = 0.9150 | SRCC = 0.9138 ]
+INFO:root:Average loss=0.05216811735161467
+INFO:root:
+[Valid][28][UTT][ MSE = 0.2349 | LCC = 0.9134 | SRCC = 0.9123 ]
+INFO:root:Average loss=0.05113409902103719
+INFO:root:
+[Valid][29][UTT][ MSE = 0.2481 | LCC = 0.9126 | SRCC = 0.9125 ]
+INFO:root:Average loss=0.048803837593063455
+INFO:root:
+[Valid][30][UTT][ MSE = 0.2278 | LCC = 0.9137 | SRCC = 0.9122 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.1787 | LCC = 0.9222 | SRCC = 0.9154 ]

results_nisqa/w2v2_xlsr_2b/layer38/run3/config.gin

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

results_nisqa/w2v2_xlsr_2b/layer38/run3/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab6f444a4c1258114adcdaef7732187bcef6e73265dfc4a133dee23a5d7d1173
+size 5697099

results_nisqa/w2v2_xlsr_2b/layer38/run3/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d10b865ae32fbaab69ffe64508600775a935c8f8f36fc4b29b6b0966498db73
+size 5698758

results_nisqa/w2v2_xlsr_2b/layer38/run3/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 11020
+INFO:root:Num val speech clips: 2700
+INFO:root:Num test speech clips: 232
+INFO:root:Device=cuda
+INFO:root:Average loss=0.3337726172684245
+INFO:root:
+[Valid][0][UTT][ MSE = 0.3343 | LCC = 0.9070 | SRCC = 0.9079 ]
+INFO:root:Average loss=0.21050732242578715
+INFO:root:
+[Valid][1][UTT][ MSE = 0.3029 | LCC = 0.9135 | SRCC = 0.9145 ]
+INFO:root:Average loss=0.18591759577824202
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2467 | LCC = 0.9184 | SRCC = 0.9180 ]
+INFO:root:Average loss=0.17901290208101273
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2512 | LCC = 0.9193 | SRCC = 0.9192 ]
+INFO:root:Average loss=0.17746918273798992
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2544 | LCC = 0.9196 | SRCC = 0.9193 ]
+INFO:root:Average loss=0.16187805290511578
+INFO:root:
+[Valid][5][UTT][ MSE = 0.2640 | LCC = 0.9202 | SRCC = 0.9196 ]
+INFO:root:Average loss=0.15155292039661739
+INFO:root:
+[Valid][6][UTT][ MSE = 0.2205 | LCC = 0.9204 | SRCC = 0.9198 ]
+INFO:root:Average loss=0.14731123561114934
+INFO:root:
+[Valid][7][UTT][ MSE = 0.2364 | LCC = 0.9202 | SRCC = 0.9193 ]
+INFO:root:Average loss=0.14066440978146702
+INFO:root:
+[Valid][8][UTT][ MSE = 0.2546 | LCC = 0.9210 | SRCC = 0.9200 ]
+INFO:root:Average loss=0.1329699172234604
+INFO:root:
+[Valid][9][UTT][ MSE = 0.2410 | LCC = 0.9181 | SRCC = 0.9170 ]
+INFO:root:Average loss=0.1269003630459653
+INFO:root:
+[Valid][10][UTT][ MSE = 0.2334 | LCC = 0.9190 | SRCC = 0.9176 ]
+INFO:root:Average loss=0.1231278281852689
+INFO:root:
+[Valid][11][UTT][ MSE = 0.2389 | LCC = 0.9200 | SRCC = 0.9191 ]
+INFO:root:Average loss=0.11786811549022708
+INFO:root:
+[Valid][12][UTT][ MSE = 0.2614 | LCC = 0.9179 | SRCC = 0.9189 ]
+INFO:root:Average loss=0.11110201129162242
+INFO:root:
+[Valid][13][UTT][ MSE = 0.2601 | LCC = 0.9184 | SRCC = 0.9178 ]
+INFO:root:Average loss=0.10376826889080808
+INFO:root:
+[Valid][14][UTT][ MSE = 0.2703 | LCC = 0.9166 | SRCC = 0.9161 ]
+INFO:root:Average loss=0.09984254951163524
+INFO:root:
+[Valid][15][UTT][ MSE = 0.2544 | LCC = 0.9164 | SRCC = 0.9160 ]
+INFO:root:Average loss=0.09393385149581585
+INFO:root:
+[Valid][16][UTT][ MSE = 0.2377 | LCC = 0.9162 | SRCC = 0.9153 ]
+INFO:root:Average loss=0.08805486338073118
+INFO:root:
+[Valid][17][UTT][ MSE = 0.2462 | LCC = 0.9161 | SRCC = 0.9152 ]
+INFO:root:Average loss=0.08314067353575216
+INFO:root:
+[Valid][18][UTT][ MSE = 0.2410 | LCC = 0.9160 | SRCC = 0.9150 ]
+INFO:root:Average loss=0.08018991301600643
+INFO:root:
+[Valid][19][UTT][ MSE = 0.2476 | LCC = 0.9151 | SRCC = 0.9142 ]
+INFO:root:Average loss=0.07220590116746853
+INFO:root:
+[Valid][20][UTT][ MSE = 0.2334 | LCC = 0.9146 | SRCC = 0.9138 ]
+INFO:root:Average loss=0.07081344837828868
+INFO:root:
+[Valid][21][UTT][ MSE = 0.2354 | LCC = 0.9158 | SRCC = 0.9148 ]
+INFO:root:Average loss=0.06739313949997715
+INFO:root:
+[Valid][22][UTT][ MSE = 0.2470 | LCC = 0.9151 | SRCC = 0.9142 ]
+INFO:root:Average loss=0.0653358448964323
+INFO:root:
+[Valid][23][UTT][ MSE = 0.2410 | LCC = 0.9155 | SRCC = 0.9147 ]
+INFO:root:Average loss=0.06552038662609337
+INFO:root:
+[Valid][24][UTT][ MSE = 0.2574 | LCC = 0.9129 | SRCC = 0.9127 ]
+INFO:root:Average loss=0.059740636260867806
+INFO:root:
+[Valid][25][UTT][ MSE = 0.2519 | LCC = 0.9147 | SRCC = 0.9137 ]
+INFO:root:Average loss=0.05701734470149685
+INFO:root:
+[Valid][26][UTT][ MSE = 0.2499 | LCC = 0.9125 | SRCC = 0.9135 ]
+INFO:root:Average loss=0.05758010447455968
+INFO:root:
+[Valid][27][UTT][ MSE = 0.2440 | LCC = 0.9132 | SRCC = 0.9128 ]
+INFO:root:Average loss=0.052428161714635144
+INFO:root:
+[Valid][28][UTT][ MSE = 0.2319 | LCC = 0.9140 | SRCC = 0.9132 ]
+INFO:root:Average loss=0.052885294606575387
+INFO:root:
+[Valid][29][UTT][ MSE = 0.2502 | LCC = 0.9132 | SRCC = 0.9123 ]
+INFO:root:Average loss=0.0485782977516596
+INFO:root:
+[Valid][30][UTT][ MSE = 0.2335 | LCC = 0.9143 | SRCC = 0.9135 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.2183 | LCC = 0.9217 | SRCC = 0.9171 ]

results_nisqa/w2v2_xlsr_2b/layer38/run4/config.gin

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

results_nisqa/w2v2_xlsr_2b/layer38/run4/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:83b3a3bb076c4ed05854c417565eb117023980b77bf7f0f37c78804b6990b2ae
+size 5697099

results_nisqa/w2v2_xlsr_2b/layer38/run4/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8d83c2d92e738c58f55c5a43075abc564d31daf11d094a44a906112322a601c
+size 5698822

results_nisqa/w2v2_xlsr_2b/layer38/run4/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 11020
+INFO:root:Num val speech clips: 2700
+INFO:root:Num test speech clips: 232
+INFO:root:Device=cuda
+INFO:root:Average loss=0.3471773969644756
+INFO:root:
+[Valid][0][UTT][ MSE = 0.2709 | LCC = 0.9113 | SRCC = 0.9130 ]
+INFO:root:Average loss=0.19945772810478432
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2976 | LCC = 0.9190 | SRCC = 0.9187 ]
+INFO:root:Average loss=0.1902710240269672
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2722 | LCC = 0.9184 | SRCC = 0.9199 ]
+INFO:root:Average loss=0.17381934550283962
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2505 | LCC = 0.9190 | SRCC = 0.9193 ]
+INFO:root:Average loss=0.1671163497460371
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2538 | LCC = 0.9198 | SRCC = 0.9197 ]
+INFO:root:Average loss=0.16208018625230458
+INFO:root:
+[Valid][5][UTT][ MSE = 0.2547 | LCC = 0.9213 | SRCC = 0.9205 ]
+INFO:root:Average loss=0.14857746264493535
+INFO:root:
+[Valid][6][UTT][ MSE = 0.2460 | LCC = 0.9218 | SRCC = 0.9210 ]
+INFO:root:Average loss=0.14407421997344563
+INFO:root:
+[Valid][7][UTT][ MSE = 0.2446 | LCC = 0.9210 | SRCC = 0.9203 ]
+INFO:root:Average loss=0.13928423029494424
+INFO:root:
+[Valid][8][UTT][ MSE = 0.2532 | LCC = 0.9212 | SRCC = 0.9207 ]
+INFO:root:Average loss=0.1324793135241277
+INFO:root:
+[Valid][9][UTT][ MSE = 0.2165 | LCC = 0.9201 | SRCC = 0.9197 ]
+INFO:root:Average loss=0.13079995321745128
+INFO:root:
+[Valid][10][UTT][ MSE = 0.2387 | LCC = 0.9201 | SRCC = 0.9198 ]
+INFO:root:Average loss=0.12545014306299948
+INFO:root:
+[Valid][11][UTT][ MSE = 0.2430 | LCC = 0.9182 | SRCC = 0.9172 ]
+INFO:root:Average loss=0.11740366576207166
+INFO:root:
+[Valid][12][UTT][ MSE = 0.2313 | LCC = 0.9179 | SRCC = 0.9182 ]
+INFO:root:Average loss=0.11058789603627486
+INFO:root:
+[Valid][13][UTT][ MSE = 0.2258 | LCC = 0.9172 | SRCC = 0.9171 ]
+INFO:root:Average loss=0.10945285912710807
+INFO:root:
+[Valid][14][UTT][ MSE = 0.2344 | LCC = 0.9180 | SRCC = 0.9174 ]
+INFO:root:Average loss=0.0995788036312671
+INFO:root:
+[Valid][15][UTT][ MSE = 0.2514 | LCC = 0.9167 | SRCC = 0.9159 ]
+INFO:root:Average loss=0.09386388233975868
+INFO:root:
+[Valid][16][UTT][ MSE = 0.2413 | LCC = 0.9149 | SRCC = 0.9157 ]
+INFO:root:Average loss=0.09315454316018634
+INFO:root:
+[Valid][17][UTT][ MSE = 0.2380 | LCC = 0.9153 | SRCC = 0.9151 ]
+INFO:root:Average loss=0.08830646688193944
+INFO:root:
+[Valid][18][UTT][ MSE = 0.2463 | LCC = 0.9154 | SRCC = 0.9147 ]
+INFO:root:Average loss=0.08332038231495488
+INFO:root:
+[Valid][19][UTT][ MSE = 0.2556 | LCC = 0.9114 | SRCC = 0.9121 ]
+INFO:root:Average loss=0.0787379824141891
+INFO:root:
+[Valid][20][UTT][ MSE = 0.2335 | LCC = 0.9148 | SRCC = 0.9147 ]
+INFO:root:Average loss=0.07530239084466345
+INFO:root:
+[Valid][21][UTT][ MSE = 0.2461 | LCC = 0.9105 | SRCC = 0.9116 ]
+INFO:root:Average loss=0.06736693225499522
+INFO:root:
+[Valid][22][UTT][ MSE = 0.2437 | LCC = 0.9125 | SRCC = 0.9124 ]
+INFO:root:Average loss=0.06834407379000174
+INFO:root:
+[Valid][23][UTT][ MSE = 0.2188 | LCC = 0.9132 | SRCC = 0.9131 ]
+INFO:root:Average loss=0.06434519065683977
+INFO:root:
+[Valid][24][UTT][ MSE = 0.2387 | LCC = 0.9139 | SRCC = 0.9139 ]
+INFO:root:Average loss=0.059280676002791854
+INFO:root:
+[Valid][25][UTT][ MSE = 0.2364 | LCC = 0.9121 | SRCC = 0.9114 ]
+INFO:root:Average loss=0.05718778134081405
+INFO:root:
+[Valid][26][UTT][ MSE = 0.2428 | LCC = 0.9121 | SRCC = 0.9117 ]
+INFO:root:Average loss=0.054296088313436235
+INFO:root:
+[Valid][27][UTT][ MSE = 0.2435 | LCC = 0.9119 | SRCC = 0.9108 ]
+INFO:root:Average loss=0.05357172561949388
+INFO:root:
+[Valid][28][UTT][ MSE = 0.2323 | LCC = 0.9109 | SRCC = 0.9113 ]
+INFO:root:Average loss=0.053072828024281246
+INFO:root:
+[Valid][29][UTT][ MSE = 0.2396 | LCC = 0.9109 | SRCC = 0.9108 ]
+INFO:root:Average loss=0.049692496312836
+INFO:root:
+[Valid][30][UTT][ MSE = 0.2392 | LCC = 0.9110 | SRCC = 0.9107 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.2092 | LCC = 0.9168 | SRCC = 0.9116 ]

results_nisqa/w2v2_xlsr_2b/layer4/run0/config.gin

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

results_nisqa/w2v2_xlsr_2b/layer4/run0/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:190b915f02cbdc83cd723b870ffd5accd91966470c0ecfc8c0139918457f2217
+size 5697099

results_nisqa/w2v2_xlsr_2b/layer4/run0/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:832d9b7298866c3109ee091af8256f6af7aaf60473270f92e355ad81e38f534b
+size 5698822

results_nisqa/w2v2_xlsr_2b/layer4/run0/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 11020
+INFO:root:Num val speech clips: 2700
+INFO:root:Num test speech clips: 232
+INFO:root:Device=cuda
+INFO:root:Average loss=0.3581402603256909
+INFO:root:
+[Valid][0][UTT][ MSE = 0.2967 | LCC = 0.8993 | SRCC = 0.9033 ]
+INFO:root:Average loss=0.20647696160167628
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2921 | LCC = 0.9143 | SRCC = 0.9159 ]
+INFO:root:Average loss=0.1859970011376921
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2463 | LCC = 0.9164 | SRCC = 0.9168 ]
+INFO:root:Average loss=0.16731702049688107
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2895 | LCC = 0.9199 | SRCC = 0.9208 ]
+INFO:root:Average loss=0.15894697438602504
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2736 | LCC = 0.9198 | SRCC = 0.9203 ]
+INFO:root:Average loss=0.1516021919043767
+INFO:root:
+[Valid][5][UTT][ MSE = 0.2908 | LCC = 0.9208 | SRCC = 0.9213 ]
+INFO:root:Average loss=0.1437213765403439
+INFO:root:
+[Valid][6][UTT][ MSE = 0.2456 | LCC = 0.9213 | SRCC = 0.9214 ]
+INFO:root:Average loss=0.13218958977329937
+INFO:root:
+[Valid][7][UTT][ MSE = 0.2793 | LCC = 0.9166 | SRCC = 0.9204 ]
+INFO:root:Average loss=0.12767626207343416
+INFO:root:
+[Valid][8][UTT][ MSE = 0.3010 | LCC = 0.9143 | SRCC = 0.9165 ]
+INFO:root:Average loss=0.12329157398787537
+INFO:root:
+[Valid][9][UTT][ MSE = 0.2643 | LCC = 0.9146 | SRCC = 0.9178 ]
+INFO:root:Average loss=0.10957864187442498
+INFO:root:
+[Valid][10][UTT][ MSE = 0.2574 | LCC = 0.9188 | SRCC = 0.9190 ]
+INFO:root:Average loss=0.10425854323571818
+INFO:root:
+[Valid][11][UTT][ MSE = 0.2504 | LCC = 0.9183 | SRCC = 0.9184 ]
+INFO:root:Average loss=0.10047120075522131
+INFO:root:
+[Valid][12][UTT][ MSE = 0.2337 | LCC = 0.9187 | SRCC = 0.9196 ]
+INFO:root:Average loss=0.0938382121395169
+INFO:root:
+[Valid][13][UTT][ MSE = 0.2317 | LCC = 0.9166 | SRCC = 0.9174 ]
+INFO:root:Average loss=0.08862871325687866
+INFO:root:
+[Valid][14][UTT][ MSE = 0.2475 | LCC = 0.9153 | SRCC = 0.9169 ]
+INFO:root:Average loss=0.08157119912908256
+INFO:root:
+[Valid][15][UTT][ MSE = 0.2285 | LCC = 0.9157 | SRCC = 0.9162 ]
+INFO:root:Average loss=0.08001206972437098
+INFO:root:
+[Valid][16][UTT][ MSE = 0.2369 | LCC = 0.9159 | SRCC = 0.9158 ]
+INFO:root:Average loss=0.07347040222903897
+INFO:root:
+[Valid][17][UTT][ MSE = 0.2357 | LCC = 0.9160 | SRCC = 0.9174 ]
+INFO:root:Average loss=0.0695182099679991
+INFO:root:
+[Valid][18][UTT][ MSE = 0.2343 | LCC = 0.9144 | SRCC = 0.9160 ]
+INFO:root:Average loss=0.06634499580529384
+INFO:root:
+[Valid][19][UTT][ MSE = 0.2287 | LCC = 0.9170 | SRCC = 0.9173 ]
+INFO:root:Average loss=0.06400730898011626
+INFO:root:
+[Valid][20][UTT][ MSE = 0.2468 | LCC = 0.9130 | SRCC = 0.9140 ]
+INFO:root:Average loss=0.06144128617093053
+INFO:root:
+[Valid][21][UTT][ MSE = 0.2369 | LCC = 0.9147 | SRCC = 0.9155 ]
+INFO:root:Average loss=0.056704933028345166
+INFO:root:
+[Valid][22][UTT][ MSE = 0.2713 | LCC = 0.9119 | SRCC = 0.9136 ]
+INFO:root:Average loss=0.054233941793269506
+INFO:root:
+[Valid][23][UTT][ MSE = 0.2495 | LCC = 0.9117 | SRCC = 0.9137 ]
+INFO:root:Average loss=0.0514735408407243
+INFO:root:
+[Valid][24][UTT][ MSE = 0.2264 | LCC = 0.9147 | SRCC = 0.9152 ]
+INFO:root:Average loss=0.050327149677397195
+INFO:root:
+[Valid][25][UTT][ MSE = 0.2376 | LCC = 0.9125 | SRCC = 0.9135 ]
+INFO:root:Average loss=0.04875150247546979
+INFO:root:
+[Valid][26][UTT][ MSE = 0.2351 | LCC = 0.9127 | SRCC = 0.9124 ]
+INFO:root:Average loss=0.04597115120446751
+INFO:root:
+[Valid][27][UTT][ MSE = 0.2230 | LCC = 0.9141 | SRCC = 0.9140 ]
+INFO:root:Average loss=0.0458025127821575
+INFO:root:
+[Valid][28][UTT][ MSE = 0.2481 | LCC = 0.9120 | SRCC = 0.9120 ]
+INFO:root:Average loss=0.04500987299992514
+INFO:root:
+[Valid][29][UTT][ MSE = 0.2470 | LCC = 0.9126 | SRCC = 0.9134 ]
+INFO:root:Average loss=0.04042122976025405
+INFO:root:
+[Valid][30][UTT][ MSE = 0.2537 | LCC = 0.9109 | SRCC = 0.9120 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.2771 | LCC = 0.8915 | SRCC = 0.8779 ]

results_nisqa/w2v2_xlsr_2b/layer4/run1/config.gin

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

results_nisqa/w2v2_xlsr_2b/layer4/run1/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f93a9b7088e7422de4c52c8a086c7b78574a4f823cb4d18ec8f0ece18ed99a3
+size 5697099

results_nisqa/w2v2_xlsr_2b/layer4/run1/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f2dba77643515215226ad0517021a964a66d213553c22500414ee48eea157e92
+size 5698822

results_nisqa/w2v2_xlsr_2b/layer4/run1/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 11020
+INFO:root:Num val speech clips: 2700
+INFO:root:Num test speech clips: 232
+INFO:root:Device=cuda
+INFO:root:Average loss=0.3800612460671133
+INFO:root:
+[Valid][0][UTT][ MSE = 0.2681 | LCC = 0.8994 | SRCC = 0.9006 ]
+INFO:root:Average loss=0.21469881196070267
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2920 | LCC = 0.9124 | SRCC = 0.9128 ]
+INFO:root:Average loss=0.1899807188724507
+INFO:root:
+[Valid][2][UTT][ MSE = 0.3209 | LCC = 0.9147 | SRCC = 0.9159 ]
+INFO:root:Average loss=0.17461467904679348
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2784 | LCC = 0.9187 | SRCC = 0.9198 ]
+INFO:root:Average loss=0.16012374173871355
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2807 | LCC = 0.9171 | SRCC = 0.9175 ]
+INFO:root:Average loss=0.15577032899408672
+INFO:root:
+[Valid][5][UTT][ MSE = 0.2508 | LCC = 0.9171 | SRCC = 0.9179 ]
+INFO:root:Average loss=0.14397010995301207
+INFO:root:
+[Valid][6][UTT][ MSE = 0.2559 | LCC = 0.9212 | SRCC = 0.9221 ]
+INFO:root:Average loss=0.1351071561692078
+INFO:root:
+[Valid][7][UTT][ MSE = 0.2573 | LCC = 0.9201 | SRCC = 0.9206 ]
+INFO:root:Average loss=0.12607861181042787
+INFO:root:
+[Valid][8][UTT][ MSE = 0.2340 | LCC = 0.9210 | SRCC = 0.9213 ]
+INFO:root:Average loss=0.12270136005272066
+INFO:root:
+[Valid][9][UTT][ MSE = 0.2370 | LCC = 0.9190 | SRCC = 0.9203 ]
+INFO:root:Average loss=0.11343699837954058
+INFO:root:
+[Valid][10][UTT][ MSE = 0.2692 | LCC = 0.9205 | SRCC = 0.9206 ]
+INFO:root:Average loss=0.10753618421926664
+INFO:root:
+[Valid][11][UTT][ MSE = 0.2337 | LCC = 0.9195 | SRCC = 0.9200 ]
+INFO:root:Average loss=0.10011593698163253
+INFO:root:
+[Valid][12][UTT][ MSE = 0.2501 | LCC = 0.9182 | SRCC = 0.9186 ]
+INFO:root:Average loss=0.09441132955468459
+INFO:root:
+[Valid][13][UTT][ MSE = 0.2355 | LCC = 0.9184 | SRCC = 0.9189 ]
+INFO:root:Average loss=0.08863076533196289
+INFO:root:
+[Valid][14][UTT][ MSE = 0.2468 | LCC = 0.9174 | SRCC = 0.9179 ]
+INFO:root:Average loss=0.08301670726582494
+INFO:root:
+[Valid][15][UTT][ MSE = 0.2873 | LCC = 0.9170 | SRCC = 0.9166 ]
+INFO:root:Average loss=0.08098130139140035
+INFO:root:
+[Valid][16][UTT][ MSE = 0.2799 | LCC = 0.9164 | SRCC = 0.9165 ]
+INFO:root:Average loss=0.07466369183637159
+INFO:root:
+[Valid][17][UTT][ MSE = 0.2796 | LCC = 0.9166 | SRCC = 0.9165 ]
+INFO:root:Average loss=0.07071503779361013
+INFO:root:
+[Valid][18][UTT][ MSE = 0.2708 | LCC = 0.9154 | SRCC = 0.9152 ]
+INFO:root:Average loss=0.06564993430682689
+INFO:root:
+[Valid][19][UTT][ MSE = 0.2685 | LCC = 0.9139 | SRCC = 0.9142 ]
+INFO:root:Average loss=0.06552802648589101
+INFO:root:
+[Valid][20][UTT][ MSE = 0.2694 | LCC = 0.9153 | SRCC = 0.9149 ]
+INFO:root:Average loss=0.059855695727760394
+INFO:root:
+[Valid][21][UTT][ MSE = 0.2608 | LCC = 0.9146 | SRCC = 0.9148 ]
+INFO:root:Average loss=0.055993978752393946
+INFO:root:
+[Valid][22][UTT][ MSE = 0.2834 | LCC = 0.9146 | SRCC = 0.9143 ]
+INFO:root:Average loss=0.053515766289709624
+INFO:root:
+[Valid][23][UTT][ MSE = 0.2754 | LCC = 0.9149 | SRCC = 0.9150 ]
+INFO:root:Average loss=0.051038012619135695
+INFO:root:
+[Valid][24][UTT][ MSE = 0.2653 | LCC = 0.9156 | SRCC = 0.9153 ]
+INFO:root:Average loss=0.04847914419766795
+INFO:root:
+[Valid][25][UTT][ MSE = 0.2802 | LCC = 0.9121 | SRCC = 0.9119 ]
+INFO:root:Average loss=0.04800357624371617
+INFO:root:
+[Valid][26][UTT][ MSE = 0.2844 | LCC = 0.9125 | SRCC = 0.9126 ]
+INFO:root:Average loss=0.045233594759695794
+INFO:root:
+[Valid][27][UTT][ MSE = 0.3064 | LCC = 0.9119 | SRCC = 0.9117 ]
+INFO:root:Average loss=0.042710320571857384
+INFO:root:
+[Valid][28][UTT][ MSE = 0.2614 | LCC = 0.9113 | SRCC = 0.9115 ]
+INFO:root:Average loss=0.04136431798597292
+INFO:root:
+[Valid][29][UTT][ MSE = 0.2912 | LCC = 0.9126 | SRCC = 0.9125 ]
+INFO:root:Average loss=0.039943140764543084
+INFO:root:
+[Valid][30][UTT][ MSE = 0.2700 | LCC = 0.9127 | SRCC = 0.9125 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.3051 | LCC = 0.9018 | SRCC = 0.8889 ]

results_nisqa/w2v2_xlsr_2b/layer4/run2/config.gin

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

results_nisqa/w2v2_xlsr_2b/layer4/run2/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1be8d41f4345cdf1fdb400830e8056ab7919ea921f46dcecebaef0da923c608e
+size 5697099

results_nisqa/w2v2_xlsr_2b/layer4/run2/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e474bfbd1a972dd7a23d99ad04eacd4799c1ea000b9cdc8dc6355ca5e425b5b7
+size 5698822

results_nisqa/w2v2_xlsr_2b/layer4/run2/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 11020
+INFO:root:Num val speech clips: 2700
+INFO:root:Num test speech clips: 232
+INFO:root:Device=cuda
+INFO:root:Average loss=0.3550736455386774
+INFO:root:
+[Valid][0][UTT][ MSE = 0.2647 | LCC = 0.9007 | SRCC = 0.9037 ]
+INFO:root:Average loss=0.20785329973249766
+INFO:root:
+[Valid][1][UTT][ MSE = 0.3193 | LCC = 0.9170 | SRCC = 0.9172 ]
+INFO:root:Average loss=0.18815435469150543
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2907 | LCC = 0.9173 | SRCC = 0.9184 ]
+INFO:root:Average loss=0.17584002569231683
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2860 | LCC = 0.9186 | SRCC = 0.9192 ]
+INFO:root:Average loss=0.16215410061998864
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2577 | LCC = 0.9184 | SRCC = 0.9186 ]
+INFO:root:Average loss=0.15208365707900484
+INFO:root:
+[Valid][5][UTT][ MSE = 0.2737 | LCC = 0.9194 | SRCC = 0.9198 ]
+INFO:root:Average loss=0.1443601214472269
+INFO:root:
+[Valid][6][UTT][ MSE = 0.2543 | LCC = 0.9201 | SRCC = 0.9204 ]
+INFO:root:Average loss=0.1339055060737395
+INFO:root:
+[Valid][7][UTT][ MSE = 0.2501 | LCC = 0.9196 | SRCC = 0.9199 ]
+INFO:root:Average loss=0.1284567914860097
+INFO:root:
+[Valid][8][UTT][ MSE = 0.2449 | LCC = 0.9202 | SRCC = 0.9202 ]
+INFO:root:Average loss=0.12546919673853527
+INFO:root:
+[Valid][9][UTT][ MSE = 0.2559 | LCC = 0.9191 | SRCC = 0.9203 ]
+INFO:root:Average loss=0.11406566934778511
+INFO:root:
+[Valid][10][UTT][ MSE = 0.2313 | LCC = 0.9184 | SRCC = 0.9187 ]
+INFO:root:Average loss=0.10874996487201982
+INFO:root:
+[Valid][11][UTT][ MSE = 0.2802 | LCC = 0.9174 | SRCC = 0.9178 ]
+INFO:root:Average loss=0.09986862152642598
+INFO:root:
+[Valid][12][UTT][ MSE = 0.2409 | LCC = 0.9170 | SRCC = 0.9175 ]
+INFO:root:Average loss=0.09369076191948328
+INFO:root:
+[Valid][13][UTT][ MSE = 0.2481 | LCC = 0.9142 | SRCC = 0.9154 ]
+INFO:root:Average loss=0.08848452570207546
+INFO:root:
+[Valid][14][UTT][ MSE = 0.2386 | LCC = 0.9145 | SRCC = 0.9158 ]
+INFO:root:Average loss=0.08360218103824324
+INFO:root:
+[Valid][15][UTT][ MSE = 0.2357 | LCC = 0.9158 | SRCC = 0.9163 ]
+INFO:root:Average loss=0.07698096836664084
+INFO:root:
+[Valid][16][UTT][ MSE = 0.2716 | LCC = 0.9127 | SRCC = 0.9141 ]
+INFO:root:Average loss=0.07298371266853602
+INFO:root:
+[Valid][17][UTT][ MSE = 0.2445 | LCC = 0.9138 | SRCC = 0.9143 ]
+INFO:root:Average loss=0.06927389382368568
+INFO:root:
+[Valid][18][UTT][ MSE = 0.2325 | LCC = 0.9141 | SRCC = 0.9150 ]
+INFO:root:Average loss=0.06537058311003137
+INFO:root:
+[Valid][19][UTT][ MSE = 0.2775 | LCC = 0.9121 | SRCC = 0.9135 ]
+INFO:root:Average loss=0.058657830979438184
+INFO:root:
+[Valid][20][UTT][ MSE = 0.2511 | LCC = 0.9145 | SRCC = 0.9153 ]
+INFO:root:Average loss=0.05909948156489802
+INFO:root:
+[Valid][21][UTT][ MSE = 0.2494 | LCC = 0.9117 | SRCC = 0.9117 ]
+INFO:root:Average loss=0.05613204884219032
+INFO:root:
+[Valid][22][UTT][ MSE = 0.2381 | LCC = 0.9140 | SRCC = 0.9156 ]
+INFO:root:Average loss=0.05453000409324045
+INFO:root:
+[Valid][23][UTT][ MSE = 0.2665 | LCC = 0.9097 | SRCC = 0.9128 ]
+INFO:root:Average loss=0.049000707826431775
+INFO:root:
+[Valid][24][UTT][ MSE = 0.2526 | LCC = 0.9116 | SRCC = 0.9122 ]
+INFO:root:Average loss=0.047368894638009154
+INFO:root:
+[Valid][25][UTT][ MSE = 0.2772 | LCC = 0.9086 | SRCC = 0.9107 ]
+INFO:root:Average loss=0.047009450701103045
+INFO:root:
+[Valid][26][UTT][ MSE = 0.2379 | LCC = 0.9116 | SRCC = 0.9124 ]
+INFO:root:Average loss=0.046301326849040265
+INFO:root:
+[Valid][27][UTT][ MSE = 0.2412 | LCC = 0.9128 | SRCC = 0.9129 ]
+INFO:root:Average loss=0.04210556603192939
+INFO:root:
+[Valid][28][UTT][ MSE = 0.3331 | LCC = 0.9050 | SRCC = 0.9072 ]
+INFO:root:Average loss=0.042427271981373686
+INFO:root:
+[Valid][29][UTT][ MSE = 0.2596 | LCC = 0.9099 | SRCC = 0.9110 ]
+INFO:root:Average loss=0.04045815938125456
+INFO:root:
+[Valid][30][UTT][ MSE = 0.2936 | LCC = 0.9101 | SRCC = 0.9129 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.2340 | LCC = 0.8916 | SRCC = 0.8766 ]

results_nisqa/w2v2_xlsr_2b/layer4/run3/config.gin

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

results_nisqa/w2v2_xlsr_2b/layer4/run3/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e16b08d15b1fb133f0b43298672f4d10810fc4a322f3da4c4b19f8d2907b990e
+size 5697099

results_nisqa/w2v2_xlsr_2b/layer4/run3/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1fe7a2d31aa6876b0ef815cc34e8f02aec999a99e232ddbe4f3cb63c1f47c18
+size 5698822

results_nisqa/w2v2_xlsr_2b/layer4/run3/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 11020
+INFO:root:Num val speech clips: 2700
+INFO:root:Num test speech clips: 232
+INFO:root:Device=cuda
+INFO:root:Average loss=0.3717152559860593
+INFO:root:
+[Valid][0][UTT][ MSE = 0.2864 | LCC = 0.9041 | SRCC = 0.9048 ]
+INFO:root:Average loss=0.20237747864054806
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2604 | LCC = 0.9146 | SRCC = 0.9151 ]
+INFO:root:Average loss=0.1798612144210435
+INFO:root:
+[Valid][2][UTT][ MSE = 0.2350 | LCC = 0.9172 | SRCC = 0.9182 ]
+INFO:root:Average loss=0.1703812125531924
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2375 | LCC = 0.9149 | SRCC = 0.9155 ]
+INFO:root:Average loss=0.16310373857358976
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2503 | LCC = 0.9143 | SRCC = 0.9152 ]
+INFO:root:Average loss=0.15025387592398362
+INFO:root:
+[Valid][5][UTT][ MSE = 0.2583 | LCC = 0.9171 | SRCC = 0.9174 ]
+INFO:root:Average loss=0.14118046410424862
+INFO:root:
+[Valid][6][UTT][ MSE = 0.2481 | LCC = 0.9180 | SRCC = 0.9182 ]
+INFO:root:Average loss=0.13794153278445923
+INFO:root:
+[Valid][7][UTT][ MSE = 0.2315 | LCC = 0.9196 | SRCC = 0.9202 ]
+INFO:root:Average loss=0.12494209314012802
+INFO:root:
+[Valid][8][UTT][ MSE = 0.2362 | LCC = 0.9185 | SRCC = 0.9191 ]
+INFO:root:Average loss=0.1179865243376335
+INFO:root:
+[Valid][9][UTT][ MSE = 0.2476 | LCC = 0.9189 | SRCC = 0.9188 ]
+INFO:root:Average loss=0.11468784356048342
+INFO:root:
+[Valid][10][UTT][ MSE = 0.2484 | LCC = 0.9175 | SRCC = 0.9186 ]
+INFO:root:Average loss=0.10408213814300608
+INFO:root:
+[Valid][11][UTT][ MSE = 0.2452 | LCC = 0.9170 | SRCC = 0.9172 ]
+INFO:root:Average loss=0.09892154015557614
+INFO:root:
+[Valid][12][UTT][ MSE = 0.2189 | LCC = 0.9157 | SRCC = 0.9166 ]
+INFO:root:Average loss=0.09609674793705775
+INFO:root:
+[Valid][13][UTT][ MSE = 0.2266 | LCC = 0.9165 | SRCC = 0.9165 ]
+INFO:root:Average loss=0.0897136325005851
+INFO:root:
+[Valid][14][UTT][ MSE = 0.2579 | LCC = 0.9140 | SRCC = 0.9146 ]
+INFO:root:Average loss=0.08187602789846458
+INFO:root:
+[Valid][15][UTT][ MSE = 0.2235 | LCC = 0.9153 | SRCC = 0.9155 ]
+INFO:root:Average loss=0.07840798250426446
+INFO:root:
+[Valid][16][UTT][ MSE = 0.2704 | LCC = 0.9131 | SRCC = 0.9133 ]
+INFO:root:Average loss=0.07659680603039747
+INFO:root:
+[Valid][17][UTT][ MSE = 0.2575 | LCC = 0.9135 | SRCC = 0.9134 ]
+INFO:root:Average loss=0.06912071511924611
+INFO:root:
+[Valid][18][UTT][ MSE = 0.2381 | LCC = 0.9119 | SRCC = 0.9116 ]
+INFO:root:Average loss=0.06767546747288952
+INFO:root:
+[Valid][19][UTT][ MSE = 0.2383 | LCC = 0.9141 | SRCC = 0.9136 ]
+INFO:root:Average loss=0.06359595628832117
+INFO:root:
+[Valid][20][UTT][ MSE = 0.2429 | LCC = 0.9118 | SRCC = 0.9115 ]
+INFO:root:Average loss=0.06040944344218756
+INFO:root:
+[Valid][21][UTT][ MSE = 0.2417 | LCC = 0.9087 | SRCC = 0.9093 ]
+INFO:root:Average loss=0.05653111905807016
+INFO:root:
+[Valid][22][UTT][ MSE = 0.2358 | LCC = 0.9124 | SRCC = 0.9121 ]
+INFO:root:Average loss=0.05411457280073896
+INFO:root:
+[Valid][23][UTT][ MSE = 0.2617 | LCC = 0.9107 | SRCC = 0.9106 ]
+INFO:root:Average loss=0.05628861372784383
+INFO:root:
+[Valid][24][UTT][ MSE = 0.2320 | LCC = 0.9101 | SRCC = 0.9111 ]
+INFO:root:Average loss=0.05021226630045499
+INFO:root:
+[Valid][25][UTT][ MSE = 0.2260 | LCC = 0.9114 | SRCC = 0.9117 ]
+INFO:root:Average loss=0.04752957956553194
+INFO:root:
+[Valid][26][UTT][ MSE = 0.2364 | LCC = 0.9096 | SRCC = 0.9101 ]
+INFO:root:Average loss=0.0454946707697273
+INFO:root:
+[Valid][27][UTT][ MSE = 0.2462 | LCC = 0.9073 | SRCC = 0.9077 ]
+INFO:root:Average loss=0.04292764670802334
+INFO:root:
+[Valid][28][UTT][ MSE = 0.2498 | LCC = 0.9123 | SRCC = 0.9121 ]
+INFO:root:Average loss=0.042303518548263294
+INFO:root:
+[Valid][29][UTT][ MSE = 0.2443 | LCC = 0.9118 | SRCC = 0.9114 ]
+INFO:root:Average loss=0.040361330053106895
+INFO:root:
+[Valid][30][UTT][ MSE = 0.2471 | LCC = 0.9087 | SRCC = 0.9084 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.2123 | LCC = 0.8952 | SRCC = 0.8797 ]

results_nisqa/w2v2_xlsr_2b/layer4/run4/config.gin

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

results_nisqa/w2v2_xlsr_2b/layer4/run4/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e5e05a51c8fb575cb6fc9e110ae59477c78f9810c7bf07038376ed23e242b7a1
+size 5697099

results_nisqa/w2v2_xlsr_2b/layer4/run4/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7620d1435578e69651d518dea119642ccf898c7684733b08e5fbedccad12f4dd
+size 5698822

results_nisqa/w2v2_xlsr_2b/layer4/run4/train.log

@@ -0,0 +1,99 @@
+INFO:root:Num train speech clips: 11020
+INFO:root:Num val speech clips: 2700
+INFO:root:Num test speech clips: 232
+INFO:root:Device=cuda
+INFO:root:Average loss=0.3693974719226705
+INFO:root:
+[Valid][0][UTT][ MSE = 0.3756 | LCC = 0.9021 | SRCC = 0.9044 ]
+INFO:root:Average loss=0.2156127299359768
+INFO:root:
+[Valid][1][UTT][ MSE = 0.2905 | LCC = 0.9136 | SRCC = 0.9140 ]
+INFO:root:Average loss=0.19322255240871727
+INFO:root:
+[Valid][2][UTT][ MSE = 0.3187 | LCC = 0.9157 | SRCC = 0.9162 ]
+INFO:root:Average loss=0.1694273598965882
+INFO:root:
+[Valid][3][UTT][ MSE = 0.2477 | LCC = 0.9191 | SRCC = 0.9192 ]
+INFO:root:Average loss=0.16125701772684306
+INFO:root:
+[Valid][4][UTT][ MSE = 0.2406 | LCC = 0.9192 | SRCC = 0.9199 ]
+INFO:root:Average loss=0.14986434398014423
+INFO:root:
+[Valid][5][UTT][ MSE = 0.2805 | LCC = 0.9203 | SRCC = 0.9201 ]
+INFO:root:Average loss=0.1419528916564291
+INFO:root:
+[Valid][6][UTT][ MSE = 0.2594 | LCC = 0.9218 | SRCC = 0.9221 ]
+INFO:root:Average loss=0.13247361482051068
+INFO:root:
+[Valid][7][UTT][ MSE = 0.2607 | LCC = 0.9194 | SRCC = 0.9204 ]
+INFO:root:Average loss=0.13043185973787583
+INFO:root:
+[Valid][8][UTT][ MSE = 0.2538 | LCC = 0.9216 | SRCC = 0.9218 ]
+INFO:root:Average loss=0.11937604805809914
+INFO:root:
+[Valid][9][UTT][ MSE = 0.2393 | LCC = 0.9180 | SRCC = 0.9187 ]
+INFO:root:Average loss=0.1125882656415763
+INFO:root:
+[Valid][10][UTT][ MSE = 0.2555 | LCC = 0.9209 | SRCC = 0.9209 ]
+INFO:root:Average loss=0.10649066967512831
+INFO:root:
+[Valid][11][UTT][ MSE = 0.2299 | LCC = 0.9161 | SRCC = 0.9170 ]
+INFO:root:Average loss=0.10070321650777249
+INFO:root:
+[Valid][12][UTT][ MSE = 0.2303 | LCC = 0.9187 | SRCC = 0.9191 ]
+INFO:root:Average loss=0.09233528821368438
+INFO:root:
+[Valid][13][UTT][ MSE = 0.2250 | LCC = 0.9168 | SRCC = 0.9173 ]
+INFO:root:Average loss=0.08558362957111673
+INFO:root:
+[Valid][14][UTT][ MSE = 0.2418 | LCC = 0.9145 | SRCC = 0.9157 ]
+INFO:root:Average loss=0.08322412621854358
+INFO:root:
+[Valid][15][UTT][ MSE = 0.2448 | LCC = 0.9148 | SRCC = 0.9151 ]
+INFO:root:Average loss=0.07715324885238801
+INFO:root:
+[Valid][16][UTT][ MSE = 0.2634 | LCC = 0.9141 | SRCC = 0.9139 ]
+INFO:root:Average loss=0.0720843670168364
+INFO:root:
+[Valid][17][UTT][ MSE = 0.2540 | LCC = 0.9159 | SRCC = 0.9154 ]
+INFO:root:Average loss=0.06777044702213623
+INFO:root:
+[Valid][18][UTT][ MSE = 0.2496 | LCC = 0.9140 | SRCC = 0.9138 ]
+INFO:root:Average loss=0.06609794551323604
+INFO:root:
+[Valid][19][UTT][ MSE = 0.2480 | LCC = 0.9098 | SRCC = 0.9113 ]
+INFO:root:Average loss=0.06486912502374263
+INFO:root:
+[Valid][20][UTT][ MSE = 0.2487 | LCC = 0.9118 | SRCC = 0.9133 ]
+INFO:root:Average loss=0.058483155013595016
+INFO:root:
+[Valid][21][UTT][ MSE = 0.2734 | LCC = 0.9130 | SRCC = 0.9127 ]
+INFO:root:Average loss=0.05781583009944486
+INFO:root:
+[Valid][22][UTT][ MSE = 0.2718 | LCC = 0.9131 | SRCC = 0.9132 ]
+INFO:root:Average loss=0.052125683732632266
+INFO:root:
+[Valid][23][UTT][ MSE = 0.2519 | LCC = 0.9120 | SRCC = 0.9119 ]
+INFO:root:Average loss=0.04957199217267119
+INFO:root:
+[Valid][24][UTT][ MSE = 0.2410 | LCC = 0.9122 | SRCC = 0.9123 ]
+INFO:root:Average loss=0.04794861538863251
+INFO:root:
+[Valid][25][UTT][ MSE = 0.2484 | LCC = 0.9100 | SRCC = 0.9102 ]
+INFO:root:Average loss=0.046172154292119724
+INFO:root:
+[Valid][26][UTT][ MSE = 0.2455 | LCC = 0.9126 | SRCC = 0.9127 ]
+INFO:root:Average loss=0.042853522744309695
+INFO:root:
+[Valid][27][UTT][ MSE = 0.2371 | LCC = 0.9126 | SRCC = 0.9124 ]
+INFO:root:Average loss=0.041828360839213936
+INFO:root:
+[Valid][28][UTT][ MSE = 0.2550 | LCC = 0.9129 | SRCC = 0.9123 ]
+INFO:root:Average loss=0.040072544149494585
+INFO:root:
+[Valid][29][UTT][ MSE = 0.2439 | LCC = 0.9122 | SRCC = 0.9120 ]
+INFO:root:Average loss=0.03937507874057817
+INFO:root:
+[Valid][30][UTT][ MSE = 0.2451 | LCC = 0.9123 | SRCC = 0.9124 ]
+INFO:root:
+[Test][31][UTT][ MSE = 0.2894 | LCC = 0.8855 | SRCC = 0.8726 ]

results_nisqa/w2v2_xlsr_2b/layer40/run0/config.gin

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

results_nisqa/w2v2_xlsr_2b/layer40/run0/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f76f323dff0c0edfb1a71a1fe13cd83dc9e2558d0d4f0496e5fabe94f85e19e2
+size 5697099

results_nisqa/w2v2_xlsr_2b/layer40/run0/model_best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8087c7be3605c43f5458b69ed57611f48a480076712b00ac74ae130d0603656b
+size 5698822