diff --git a/tools/distpartitioning/convert_partition.py b/tools/distpartitioning/convert_partition.py
index d2d70b96e5c5..93c696ef3ad1 100644
--- a/tools/distpartitioning/convert_partition.py
+++ b/tools/distpartitioning/convert_partition.py
@@ -541,8 +541,9 @@ def create_graph_object(
     assert np.all(shuffle_global_nids[1:] - shuffle_global_nids[:-1] == 1)
     shuffle_global_nid_range = (shuffle_global_nids[0], shuffle_global_nids[-1])
 
-    # Determine the node ID ranges of different node types.
+# Determine the node ID ranges of different node types.
     prev_last_id = last_ids.get(part_id - 1, 0)
+    max_last_id = prev_last_id  # Initialize max_last_id before the loop
     for ntype_name in global_nid_ranges:
         ntype_id = ntypes_map[ntype_name]
         type_nids = shuffle_global_nids[ntype_ids == ntype_id]
@@ -552,15 +553,17 @@ def create_graph_object(
             node_map_val[ntype_name].append(
                 [int(type_nids[0]), int(type_nids[-1]) + 1]
             )
-            last_id = th.tensor(
-                [max(prev_last_id, int(type_nids[-1]) + 1)], dtype=th.int64
-            )
+            # Update max_last_id with the maximum ID seen in this partition
+            max_last_id = max(max_last_id, int(type_nids[-1]) + 1)
+    
+    # Create the last_id tensor after the loop to ensure it's always initialized
+    last_id = th.tensor([max_last_id], dtype=th.int64)
     id_ntypes = list(global_nid_ranges.keys())
-
+    
     gather_last_ids = [
         th.zeros(1, dtype=th.int64) for _ in range(dist.get_world_size())
     ]
-
+    
     dist.all_gather(gather_last_ids, last_id)
     prev_last_id = _update_node_map(
         node_map_val, gather_last_ids, id_ntypes, prev_last_id