Does the Order of Fine-tuning Matter and Why?

Qihong Chen, Jiawei Li, Hyunjae Suh +5

To improve the performance on a target task, researchers have fine-tuned language models with an intermediate task before the target task of interest. However, previous works have…