1 paper
Yiming Ju, Ziyi Ni, Xingrun Xing +4
Supervised fine-tuning (SFT) is crucial for adapting Large Language Models (LLMs) to specific tasks. In this work, we demonstrate that the order of training data can lead to signif…