correct

Author: Vladimir Kurmanov

code/train_models.py

@@ -21,7 +21,7 @@
 
 
 ARCHITECTURES_PATH = "/kaggle/input/second-dataset/dataset"
-MAX_EPOCHS = 1
+MAX_EPOCHS = 70
 LEARNING_RATE = 0.025
 BATCH_SIZE = 96
 NUM_MODLES = 2000
@@ -78,19 +78,19 @@ def get_data_loaders(batch_size=512):
     )
     num_samples = len(train_data)
     indices = np.random.permutation(num_samples)
-    split = int(num_samples * 0.75)
+    split = int(num_samples * 0.5)
 
     search_train_loader = DataLoader(
         train_data,
         batch_size=batch_size,
-        num_workers=6,
+        num_workers=10,
         sampler=SubsetRandomSampler(indices[:split]),
     )
 
     search_valid_loader = DataLoader(
             train_data,
         batch_size=batch_size,
-        num_workers=6,
+        num_workers=10,
         sampler=SequentialSampler(indices[split:]),
     )
 
@@ -106,7 +106,7 @@ def train_model(
     fast_dev_run=False
 ):
     with model_context(architecture):
-        model = DartsSpace(width=16, num_cells=10, dataset='cifar')
+        model = DartsSpace(width=16, num_cells=3, dataset='cifar')
 
     device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
     #if torch.cuda.device_count() > 1:
@@ -163,7 +163,7 @@ def evaluate_and_save_results(
 
     with torch.no_grad():
         for images, labels in valid_loader:
-            print(labels)
+            # print(labels)
             images, labels = images.to(device), labels.to(device)
             outputs = model(images)
             outputs = torch.softmax(outputs, dim=1)