作者位次 #3/9ACL 2026Poster· Natural Language Processing
A Data-Efficient Path to Multilingual LLM s: Language Expansion via Post-training PARAM \Delta Integration into Upcycled M o E
Hao Zhou · Tianhao Li · Zhijun Wang · Shuaijie She · Linjuan Wu · Hao-Ran Wei · Baosong Yang · Jiajun Chen · Shujian Huang
摘要
Expanding Large Language Models(LLMs) to new languages is a costly endeavor, demanding extensive Continued Pre-Training(CPT) and data-intensive alignment. While recent data-free merging techniques attempt to bypass alignment by fusing a multilingual CPT-enhanced model with its instruct counterpar...