Update pipe.py

pipe.py

@@ -35,10 +35,14 @@ class SmilesDiffusionPipe(Pipeline):
 
     def encode(self, char_list):
         smiles_matrix = np.zeros(len(char_list), dtype=np.float32)
+    
+        unk_id = self.voc.vocab.get("[UNK]")
+        if unk_id is None:
+            raise ValueError("[UNK] token not found in vocabulary.")
+    
         for i, char in enumerate(char_list):
-            if char not in self.voc.vocab:
-                raise ValueError(f"Token '{char}' not in vocabulary.")
-            smiles_matrix[i] = self.voc.vocab[char]
+            smiles_matrix[i] = self.voc.vocab.get(char, unk_id)
+    
         return smiles_matrix
         
     def decode(self, matrix):