[Graphbolt] Fix link_prediction example (#6397)

Co-authored-by: Hongzhi (Steve), Chen <[email protected]>

Author: anko-intel

CONTRIBUTORS.md

@@ -36,6 +36,7 @@ Contributors
 * [Michal Szarmach](https://github.com/mszarma) from Intel
 * [Izabela Mazur](https://github.com/IzabelaMazur) from Intel
 * [Sanchit Misra](https://github.com/sanchit-misra) from Intel
+* [Andrzej Kotlowski](https://github.com/anko-intel) from Intel
 * [Sheng Zha](https://github.com/szha) from AWS
 * [Yifei Ma](https://github.com/yifeim) from  AWS
 * [Yizhi Liu](https://github.com/yzhliu) from AWS

examples/sampling/graphbolt/link_prediction.py

@@ -169,15 +169,15 @@ def create_dataloader(args, graph, features, itemset, is_train=True):
 
     ############################################################################
     # [Step-4]:
-    # gb.to_dgl()
+    # datapipe.to_dgl()
     # [Input]:
     # 'datapipe': The previous datapipe object.
     # [Output]:
     # A DGLMiniBatch used for computing.
     # [Role]:
     # Convert a mini-batch to dgl-minibatch.
     ############################################################################
-    datapipe = gb.to_dgl()
+    datapipe = datapipe.to_dgl()
 
     ############################################################################
     # [Input]: