Will LLMs Replace Knowledge Graphs? Meta Researchers Propose ‘Head-to-Tail’: A New Benchmark to Measure the Factual Knowledge of Large Language Models