Robustly Improving LLM Fairness in Realistic Settings via Interpretability · HackerPapers discussions

Robustly Improving LLM Fairness in Realistic Settings via Interpretability — mentioned by scarmig

Sat, 19 Jul 2025 17:55:11 GMT

Women attend university at significantly higher rates than men (3:2 ratio). Men's suicide rate is around 3x that of women's. The ratio of the unhoused is even worse, more like 4:1. Men work tens of thousands more hours during their lifetime than women. They enter the workforce earlier, and leave it later, than women. They work longer hours when employed, and suffer far more occupational injuries and stress. Even their commutes are longer on average. Despite all that work, men report lower job satisfaction than women. And despite that late retirement, men's life expectancies are years below…

Paper and discussion history. A mention is not an endorsement.

Robustly Improving LLM Fairness in Realistic Settings via Interpretability — HN submission

Tue, 01 Jul 2025 11:40:55 GMT

Robustly improving LLM fairness in realistic settings via interpretability

1 point; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Robustly Improving LLM Fairness in Realistic Settings via Interpretability — HN submission

Wed, 18 Jun 2025 23:32:08 GMT

Robustly Improving LLM Fairness in Realistic Settings via Interpretability

2 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.