一作ACL 2026Poster· Natural Language Processing
Parameter Importance is Not Static: Evolving Parameter Isolation for Supervised Fine-Tuning
Zekai Lin · Chao Xue · Di Liang · Xingsheng Han · Peiyang Liu · Xianjie Wu · Lei Jiang · Yu Lu · Bob Simons · Shuang Liang · Minlong Peng
摘要
Supervised Fine-Tuning (SFT) of large language models often suffers from task interference and catastrophic forgetting. Recent approaches alleviate this issue by isolating task-critical parameters during training. However, these methods represent a static solution to a dynamic problem, assuming t...