Instruction Following
Domain: LLM Concepts
🎯 Motivation
After instruction tuning, the model can follow a wide variety of instructions - summarize, translate, code, explain, compare - without task-specific training.
📋 Concrete Example
‘Summarize this article in 3 bullet points’ - the model follows the instruction format (summary, 3 points). ‘Write a Python function to check if a number is prime’ - it follows the coding instruction.
🔗 Analogy
A well-trained assistant who can handle any instruction: ‘file these’, ‘call them’, ‘order lunch’. No special training needed per task - they just follow the instruction. Instruction-tuned LLMs do this with language tasks.