Instruction-Tuning - How does instruction fine-tuning work?

By training language models on several tasks simultaneously, Progressive Learning Multitask offers a revolutionary way to instruction fine-tuning. We offer explicit instructions for each task rather than focusing on improving the model for a particular activity. When the model is being fine-tuned, it follows these instructions.