Instructional Fingerprinting of Large Language Models