2 papers across 2 sessions
Investigate how language bias originates, why margin mechanisms are effective, and propose a novel Multi-Margin Collaborative Debiasing (MMCD) framework