All should be equal in the eyes of LMs: Counterfactually Aware Fair Text Generation
Fairness in Language Models (LMs) remains a longstanding challenge, given the inherent biases in training data that can be perpetuated …
PDF