TreeBoN: Enhancing Inference-Time Alignment with Speculative Tree-Search and Best-of-N Samplingby Jiahao Qiu, Yifu Lu, Yifan…
Understanding and Alleviating Memory Consumption in RLHF for LLMsby Jin Zhou, Hanmei Yang, Steven, Tang,…
On The Global Convergence Of Online RLHF With Neural Parametrizationby Mudit Gaur, Amrit Singh Bedi,…
Mitigating Forgetting in LLM Supervised Fine-Tuning and Preference Learningby Heshan Fernando, Han Shen, Parikshit Ram,…
How to Evaluate Reward Models for RLHFby Evan Frick, Tianle Li, Connor Chen, Wei-Lin Chiang,…
Personalized Adaptation via In-Context Preference Learningby Allison Lau, Younwoo Choi, Vahid Balazadeh, Keertana Chidambaram, Vasilis…
Goal Inference from Open-Ended Dialogby Rachel Ma, Jingyi Qu, Andreea Bobu, Dylan Hadfield-MenellFirst submitted to…
A Common Pitfall of Margin-based Language Model Alignment: Gradient Entanglementby Hui Yuan, Yifan Zeng, Yue…
Generative Reward Modelsby Dakota Mahan, Duy Van Phung, Rafael Rafailov, Chase Blagden, Nathan Lile, Louis…
AdvBDGen: Adversarially Fortified Prompt-Specific Fuzzy Backdoor Generator Against LLM Alignmentby Pankayaraj Pathmanathan, Udari Madhushani Sehwag,…