Phased Instruction Fine-Tuning for Large Language Models

Wei Pang, Chuan Zhou, Xiao-Hua Zhou +1

Instruction Fine-Tuning enhances pre-trained language models from basic next-word prediction to complex instruction-following. However, existing One-off Instruction Fine-Tuning (On…