作者位次 #2/4NeurIPS 2026Poster
Alignment Needs 'Cognitive Control': On The Role of Regularization in LLM Alignment
Nona Ghazizadeh · Elnaz Rahmati · Morteza Dehghani · Payam Piray
对齐与偏好OpenReview
首次获取该候选人的完整档案及 StartLux 评估共 20 积分；每位用户单独解锁，之后查看免费。主动刷新 20 积分。
实时查找电子邮箱，费用不超过20积分；未找到邮箱不扣费。
Nona Ghazizadeh · Elnaz Rahmati · Morteza Dehghani · Payam Piray
Parsa Hejabi · Elnaz Rahmati · Alireza Salkhordeh Ziabari · Morteza Dehghani
Large Language Models (LLMs) often produce inconsistent answers when faced with different phrasings of the same prompt. In this paper, we propose Flip-Flop Consistency (F ^2 C), an unsupervised training method that improves robustness to such perturbations. F ^2 C is composed of two key component...
Preni Golazizian · Elnaz Rahmati · Jackson Trager · Zhivar Sourati · Nona Ghazizadeh · Georgios Chochlakis · Jose J. Alcocer · Kerby Bennett · Aarya Vijay Devnani · Parsa Hejabi · Harry G. Muttram · Akshay Kiran Padte · Mehrshad Saadatinia · Chenhao Wu · Alireza Salkhordeh Ziabari · Michael Sierra-Arévalo · Nicholas Weller · Shrikanth Narayanan · Benjamin A.T. Graham · Morteza Dehghani
Traffic stops are among the most frequent police–civilian interactions, and body-worn cameras (BWCs) provide a unique record of how these encounters unfold. Respect is a central dimension of these interactions, shaping public trust and perceived legitimacy, yet its interpretation is inherently su...