multilabel training

models/config/train_local.yaml

@@ -18,10 +18,10 @@ dance_ids: &dance_ids
   - WCS
 
 data_module:
-  batch_size: 64
-  num_workers: 7 # Reduced to avoid over opening files
+  batch_size: 128
+  num_workers: 10
   # data_subset: 0.001
-  test_proportion: 0.2
+  test_proportion: 0.001
 
 datasets:
   preprocessing.dataset.BestBallroomDataset:
@@ -33,7 +33,7 @@ datasets:
     song_data_path: data/songs_cleaned.csv
     song_audio_path: data/samples # data/samples
     class_list: *dance_ids
-    multi_label: False
+    multi_label: True
     min_votes: 1
     audio_window_jitter: 0.7
 

models/residual.py

@@ -9,6 +9,7 @@ import numpy as np
 import torchaudio
 import yaml
 from models.training_environment import TrainingEnvironment
+from models.utils import LabelWeightedBCELoss
 from preprocessing.dataset import DanceDataModule, get_datasets
 from preprocessing.pipelines import (
     SpectrogramTrainingPipeline,
@@ -118,7 +119,9 @@ def train_residual_dancer(config: dict):
     data = DanceDataModule(dataset, **config["data_module"])
     model = ResidualDancer(n_classes=len(TARGET_CLASSES), **config["model"])
     label_weights = data.get_label_weights().to(DEVICE)
-    criterion = nn.CrossEntropyLoss(label_weights)
+    criterion = LabelWeightedBCELoss(
+        label_weights
+    )  # nn.CrossEntropyLoss(label_weights)
 
     train_env = TrainingEnvironment(model, criterion, config)
     callbacks = [
@@ -126,8 +129,10 @@ def train_residual_dancer(config: dict):
         cb.EarlyStopping("val/loss", patience=1),
         cb.StochasticWeightAveraging(1e-2),
         cb.RichProgressBar(),
-        cb.DeviceStatsMonitor(),
     ]
     trainer = pl.Trainer(callbacks=callbacks, **config["trainer"])
     trainer.fit(train_env, datamodule=data)
-    trainer.test(train_env, datamodule=data)
+    trainer.test(
+        train_env,
+        datamodule=data,
+    )

models/{config/production.yaml → weights/ResidualDancer/multiclass/config.yaml}

@@ -1,4 +1,4 @@
-checkpoint: models/weights/ResidualDancer/weights.ckpt
+checkpoint: models/weights/ResidualDancer/multiclass/weights.ckpt
 device: cpu
 seed: 42
 dance_ids: &dance_ids

models/weights/ResidualDancer/multilabel/config.yaml

@@ -0,0 +1,20 @@
+checkpoint: models/weights/ResidualDancer/multilabel/weights.ckpt
+device: cpu
+seed: 42
+dance_ids: &dance_ids
+  - BCH
+  - CHA
+  - JIV
+  - ECS
+  - QST
+  - RMB
+  - SFT
+  - SLS
+  - SMB
+  - SWZ
+  - TGO
+  - VWZ
+  - WCS
+
+model:
+  n_channels: 128

models/weights/ResidualDancer/multilabel/weights.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a3b87ae0e61bc10d720b7e5f35329db3ccb03b5cb47d120821a9ef314f8315c
+size 193642705