Large language models can be empowered with graph understanding and reasoning capabilities through the GraphInstruct benchmark.
