Attention Pruning: Automated Fairness Repair of Language Models via Surrogate Simulated Annealing

Open in new window