Misc fudging about

spleen_segmentation_3d.py

@@ -141,7 +141,7 @@
 # In[7]:
 
 
-set_determinism(seed=0)
+set_determinism(seed=42)
 
 
 # ## Setup transforms for training and validation
@@ -253,6 +253,21 @@
 # In[10]:
 
 
+def tensor_desc(T: torch.Tensor, **kwargs) -> torch.Tensor:
+    strAs: str = "meanstd"
+    v1: float = 0.0
+    v2: float = 0.0
+    tensor: torch.Tensor = torch.Tensor([v1, v2])
+    for k, v in kwargs.items():
+        if k.lower() == "as":
+            strAs = v
+    match strAs:
+        case "meanstd":
+            tensor = torch.Tensor([T.mean().item(), T.std().item()])
+    return tensor
+
+
+torch.manual_seed(42)
 train_ds = CacheDataset(
     data=train_files, transform=train_transforms, cache_rate=1.0, num_workers=4
 )
@@ -261,18 +276,29 @@
 # use batch_size=2 to load images and use RandCropByPosNegLabeld
 # to generate 2 x 4 images for network training
 train_loader = DataLoader(train_ds, batch_size=2, shuffle=True, num_workers=4)
+# for _ in range(0, 1):
+#     print("")
+#     for sample in train_loader:
+#         input = sample["image"].to(torch.device("cuda:0"))
+#         print(tensor_desc(input))
 
 val_ds = CacheDataset(
     data=val_files, transform=val_transforms, cache_rate=1.0, num_workers=4
 )
 # val_ds = Dataset(data=val_files, transform=val_transforms)
 val_loader = DataLoader(val_ds, batch_size=1, num_workers=4)
+# for sample in val_loader:
+#     input = sample["image"].to(torch.device("cuda:0"))
+#     print(tensor_desc(input))
 
 
 # ## Create Model, Loss, Optimizer
 
 # In[12]:
-
+print("")
+# for sample in train_loader:
+#     input = sample["image"].to(torch.device("cuda:0"))
+#     print(tensor_desc(input))
 
 # standard PyTorch program style: create UNet, DiceLoss and Adam optimizer
 device = torch.device("cuda:0")
@@ -294,6 +320,10 @@
 
 # In[13]:
 
+print("")
+for sample in train_loader:
+    input = sample["image"].to(torch.device("cuda:0"))
+    print(tensor_desc(input))
 
 max_epochs = 600
 val_interval = 2
@@ -304,6 +334,7 @@
 post_pred = Compose([AsDiscrete(argmax=True, to_onehot=2)])
 post_label = Compose([AsDiscrete(to_onehot=2)])
 
+
 for epoch in range(max_epochs):
     print("-" * 10)
     print(f"epoch {epoch + 1}/{max_epochs}")
@@ -317,7 +348,9 @@
             batch_data["label"].to(device),
         )
         optimizer.zero_grad()
+        print(tensor_desc(inputs))
         outputs = model(inputs)
+        print(tensor_desc(outputs))
         loss = loss_function(outputs, labels)
         loss.backward()
         optimizer.step()