D-STEER - Preference Alignment Techniques Learn to Behave, not to Believe -- Beneath the Surface, DPO as Steering Vector Perturbation in Activation Space

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Raina, Samarth, Aggarwal, Saksham, Chadha, Aman, Jain, Vinija, Das, Amitava
Format:	Preprint
Published:	2025
Subjects:	Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Internet

https://arxiv.org/abs/2512.11838