Updating information and adding final model.

dmsr.py

@@ -419,5 +419,6 @@ def main(epochs, hidden_dim, lr, weight_decay, dropout):
 
 
 if __name__ == '__main__':
-    # Set of random hyperparameters to train the model.
-    main(400, 32, 0.01, 1e-4, 0)
+    # Set of hyperparameters to train the model.
+     main(2343, 31, 0.0010235455088934942, 0.005144745056173074, 0.5)
+

testRepoDB.py

@@ -163,11 +163,11 @@ if __name__ == '__main__':
     k = 50
     
     # Set of hyperparameters.
-    epochs = 2752
-    hidden_dim = 107
-    lr = 0.0008317
-    weight_decay = 0.006314
-    dropout = 0.8
+    epochs = 2343 
+    hidden_dim = 31
+    lr = 0.0010235455088934942
+    weight_decay = 0.005144745056173074
+    dropout = 0.5
     
     # Train and test k models and obtain their metrics.
     for i in range(k):

topN.py

@@ -28,10 +28,11 @@ def filterPreds(original, pred):
         head = i
         tail = torch.arange(0, len(pred_labels))  # All tails.
         indexH = ((headsO == head).nonzero(as_tuple=True)[0])  # Index of those heads originally present in the graph.
-            
+        #Check
+        print(len(tail))
         for index in indexH:
             tail = tail[tail != index]  # Just get those tails not present in the original graph.
-        
+        print(len(tail))
         new.append([head, tail, pred_labels[tail].cpu().detach().numpy()])  # New predictions are appended.
 
     return new