second_attempt

.gitignore

@@ -1,2 +1,17 @@
 data
-*.zip
+*.zip
+mnist/MNIST/raw/t10k-images-idx3-ubyte
+.gitignore
+mnist/MNIST/raw/t10k-images-idx3-ubyte.gz
+.gitignore
+mnist/MNIST/raw/t10k-labels-idx1-ubyte
+.gitignore
+mnist/MNIST/raw/t10k-labels-idx1-ubyte.gz
+.gitignore
+mnist/MNIST/raw/train-images-idx3-ubyte
+.gitignore
+mnist/MNIST/raw/train-images-idx3-ubyte.gz
+.gitignore
+mnist/MNIST/raw/train-labels-idx1-ubyte
+.gitignore
+mnist/MNIST/raw/train-labels-idx1-ubyte.gz

eval.py

@@ -9,21 +9,20 @@
 from model import Net 
 
 
-def test_epoch(model, data_loader):
-    # write code to test this epoch
+def test_epoch(model, device, data_loader):
     model.eval()
     test_loss = 0
     correct = 0
     with torch.no_grad():
         for data, target in data_loader:
             output = model(data.to(device))
-            test_loss += F.nll_loss(output, target.to(device), reduction='sum').item() # sum up batch loss
-            pred = output.max(1)[1] # get the index of the max log-probability
-            correct += pred.eq(target.to(device)).sum().item()
+            test_loss += F.nll_loss(output, target.to(device), reduction='sum').item()
+            pred = output.argmax(dim=1, keepdim=True)
+            correct += pred.eq(target.to(device).view_as(pred)).sum().item()
 
-    test_loss /= len(data_loader.dataset)      
-    accuracy = 100.0 * correct / len(data_loader.dataset)
-    out = {"Test loss": test_loss, "Accuracy": accuracy}
+    test_loss /= len(data_loader.dataset)
+    accuracy = 100. * correct / len(data_loader.dataset)
+    out = {'Test loss': test_loss, 'Accuracy': accuracy}
     print(out)
     return out
 
@@ -45,7 +44,13 @@ def main():
     )
 
     args, unknown = parser.parse_known_args()
+    use_cuda = torch.cuda.is_available()  
     torch.manual_seed(args.seed)
+
+    device = torch.device("cuda" if use_cuda else "cpu")
+    kwargs = {'batch_size': args.test_batch_size}
+    if use_cuda:
+        kwargs.update({'num_workers': 1, 'pin_memory': True, 'shuffle': True},)
 
     kwargs = {
         "batch_size": args.test_batch_size,
@@ -67,7 +72,7 @@ def main():
     model.load_state_dict(torch.load(checkpoint_path))
 
     # Run the test epoch and collect evaluation results
-    eval_results = test_epoch(model, test_loader)
+    eval_results = test_epoch(model, device, test_loader)
 
     # # Save evaluation results to a JSON file
     # results_path = Path(args.save_dir) / "eval_results.json"

infer.py

@@ -14,14 +14,30 @@ def infer(model, dataset, save_dir, num_samples=5):
     results_dir.mkdir(parents=True, exist_ok=True)
 
     indices = random.sample(range(len(dataset)), num_samples)
-    for idx in indices:
-        image, _ = dataset[idx]
+
+    for idx, i in enumerate(indices):
+        image, _ = dataset[i]
         with torch.no_grad():
             output = model(image.unsqueeze(0))
         pred = output.argmax(dim=1, keepdim=True).item()
 
         img = Image.fromarray(image.squeeze().numpy() * 255).convert("L")
-        img.save(results_dir / f"{pred}.png")
+        # Ensure unique filenames
+        filename = f"{pred}_{idx}.png"
+        img.save(results_dir / filename)
+
+    print(f"Saved {num_samples} inference result images.")
+
+    # for idx in indices:
+    #     image, _ = dataset[idx]
+
+    #     with torch.no_grad():
+    #         output = model(image.unsqueeze(0))
+    #     pred = output.argmax(dim=1, keepdim=True).item()
+
+    #     img = Image.fromarray(image.squeeze().numpy() * 255).convert("L")
+    #     img.save(results_dir / f"{pred}_{i}.png")
+    #     i +=1
 
 
 def main():

train.py

@@ -8,6 +8,7 @@
 from torch.utils.data import DataLoader 
 import torch.multiprocessing as mp
 from model import Net
+import time
 
 def train_epoch(epoch, args, model, train_loader, optimizer):
     model.train()
@@ -48,7 +49,7 @@ def main():
     parser.add_argument("--dry-run", action="store_true", default=False, help="quickly check a single pass")
     parser.add_argument("--save_dir", default="/opt/mount", help="checkpoint will be saved in this directory")
     parser.add_argument('--epochs', type=int, default=1, metavar='N', help='number of epochs to train (default: 1)')
-    parser.add_argument('--save_model', action='store_true', default=False, help='save the trained model to state_dict')
+    parser.add_argument('--save_model', action='store_true', default=True, help='save the trained model to state_dict')
 
     args = parser.parse_args()
     torch.manual_seed(args.seed)
@@ -78,5 +79,8 @@ def main():
         os.makedirs(os.path.join(args.save_dir, "model"), exist_ok=True)
         torch.save(model.state_dict(), os.path.join(args.save_dir, "model", "mnist_cnn.pt"))
 
+    # time.sleep(10000)
+        #/opt/mount/model/mnist.pt
+
 if __name__ == "__main__":
     main()