add more modularity

cfg.py

@@ -2,11 +2,11 @@ from pathlib import Path
 import torch
 
 BATCH_SIZE = 128
-NUM_EPOCHS = 1
-NUM_WORKERS = 0
+NUM_EPOCHS = 10
+NUM_WORKERS = 4
 NUM_SEGMENTS = 5
 NUM_SEGMENT_HOPS = 2
 SEEDS = [42, 1337]
 ALL_DATASET_PATHS = list((Path(__file__).parent.absolute() / 'data' / 'mimii').glob('*/'))
 
-DEVICE = 'cuda' if torch.cuda.is_available() else 'cpu'
+DEVICE = 'cuda' if torch.cuda.is_available() else 'cpu'

main.py

@@ -1,60 +1,57 @@
-import numpy as np
-from tqdm import tqdm
-from cfg import *
-from mimii import MIMII
-from models.ae import AE
-import torch.nn as nn
-import torch.optim as optim
-import random
+if __name__ == '__main__':
+    import numpy as np
+    from tqdm import tqdm
+    from cfg import *
+    from mimii import MIMII
+    from models.ae import AE
+    import torch.nn as nn
+    import torch.optim as optim
+    import random
 
-torch.manual_seed(42)
-torch.cuda.manual_seed(42)
-np.random.seed(42)
-random.seed(42)
+    torch.manual_seed(42)
+    torch.cuda.manual_seed(42)
+    np.random.seed(42)
+    random.seed(42)
 
-dataset_path = ALL_DATASET_PATHS[0]
-print(f'Training on {dataset_path.name}')
-mimii = MIMII(dataset_path=dataset_path, machine_id=0)
-mimii.to(DEVICE)
-#mimii.preprocess(n_fft=1024, hop_length=256, n_mels=80, center=False, power=2.0)
+    dataset_path = ALL_DATASET_PATHS[0]
+    print(f'Training on {dataset_path.name}')
+    mimii = MIMII(dataset_path=dataset_path, machine_id=0)
+    mimii.to(DEVICE)
+    #mimii.preprocess(n_fft=1024, hop_length=256, n_mels=80, center=False, power=2.0)
 
-dl = mimii.train_dataloader(
-    segment_len=NUM_SEGMENTS,
-    hop_len=NUM_SEGMENT_HOPS,
-    batch_size=BATCH_SIZE,
-    num_workers=NUM_WORKERS,
-    shuffle=True
-)
+    dl = mimii.train_dataloader(
+        segment_len=NUM_SEGMENTS,
+        hop_len=NUM_SEGMENT_HOPS,
+        batch_size=BATCH_SIZE,
+        num_workers=NUM_WORKERS,
+        shuffle=True
+    )
 
 
-model = AE(400).to(DEVICE)
-model.init_weights()
-criterion = nn.MSELoss()
-optimizer = optim.Adam(model.parameters(), lr=0.001)
+    model = AE(400).to(DEVICE)
+    model.init_weights()
+
+    optimizer = optim.Adam(model.parameters(), lr=0.001)
 
 
-beta_1 = 0.00
-beta_2 = 0.0
+    for epoch in range(NUM_EPOCHS):
+        print(f'EPOCH #{epoch+1}')
+        losses = []
+        for batch in tqdm(dl):
+            data, labels = batch
+            data = data.to(DEVICE)
 
-for epoch in range(NUM_EPOCHS):
-    print(f'EPOCH #{epoch+1}')
-    losses = []
-    for batch in tqdm(dl):
-        data, labels = batch
-        data = data.to(DEVICE)
+            loss = model.train_loss(data)
 
-        y_hat, y = model(data)
-        loss = criterion(y_hat, y)
+            optimizer.zero_grad()
+            loss.backward()
+            optimizer.step()
 
-        optimizer.zero_grad()
-        loss.backward()
-        optimizer.step()
+            losses.append(loss.item())
+        print(f'Loss: {np.mean(losses)}')
 
-        losses.append(loss.item())
-    print(f'Loss: {np.mean(losses)}')
-
-auc = mimii.evaluate_model(model, NUM_SEGMENTS, NUM_SEGMENTS)
-print(f'AUC: {auc}')
+    auc = mimii.evaluate_model(model, NUM_SEGMENTS, NUM_SEGMENTS)
+    print(f'AUC: {auc}')
 
 
 

mimii.py

@@ -92,8 +92,7 @@ class MIMII(object):
                     data, labels = batch
                     data = data.to(self.device)
 
-                    y_hat, y = f(data)
-                    preds = torch.sum((y_hat - y) ** 2, dim=tuple(range(1, y_hat.dim())))
+                    preds = f.test_loss(data)
 
                     file_preds += preds.cpu().data.tolist()
                 y_true.append(labels.max().item())

models/ae.py

@@ -24,6 +24,17 @@ class AE(nn.Module):
         x = data.view(data.shape[0], -1)
         return self.net(x), x
 
+    def train_loss(self, data):
+        criterion = nn.MSELoss()
+        y_hat, y = self.forward(data)
+        loss = criterion(y_hat, y)
+        return loss
+
+    def test_loss(self, data):
+        y_hat, y = self.forward(data)
+        preds = torch.sum((y_hat - y) ** 2, dim=tuple(range(1, y_hat.dim())))
+        return preds
+
     def init_weights(self):
         def _weight_init(m):
             if hasattr(m, 'weight'):