Search papers, labs, and topics across Lattice.
Affiliation:
1
0
2
4
Activation steering debiasing methods reduce LLM bias not by correcting underlying preferences, but by making the model less confident and more likely to abstain from answering.
