Mahavir Dabas

1
Papers
2
Total Citations

Papers (1)

Just Enough Shifts: Mitigating Over-Refusal in Aligned Language Models with Targeted Representation Fine-Tuning

ICML 2025
2
citations