Fine-tuning LLMs with Cross-Attention-based Weight Decay for Bias Mitigation.

Farsheed Haque, Zhe Fu, Depeng Xu, Shuhan Yuan, Xi Niu

Browse the full EMNLP paper archive.