1 paper · 1 filter
Manjiang Yu, Hongji Li, Junwei Chen +4
Representation intervention has emerged as a promising paradigm for aligning large language models toward desired behaviors without modifying model weights. Existing methods typica…