Researchers from IBM and MIT Introduce LAB: A Novel AI Method Designed to Overcome the Scalability Challenges in the Instruction-Tuning Phase of Large Language Model (LLM) Training