paper-with-me

Papers

PEO: Improving Bi-Factorial Preference Alignment with Post-Training Policy Extrapolation

2025-03-03 · Yuxuan Liu

The alignment of large language models with human values presents a critical challenge, particularly when balancing conflicting objectives like helpfulness and harmlessness. Existing approaches, such as Reinforcement Learning from Human Feedback (RLHF) and Direct Preference Optimization (DPO), face notable limitations: RLHF suffers from instability and inefficiency in multi-objective optimization, while DPO lacks mechanisms for dynamic trade-offs. To address these challenges, we propose Post-Training Extrapolation Optimization (PEO), a novel and efficient framework for bi-factorial alignment. PEO generates a family of Pareto-optimal policies in a single training pass by leveraging a three-phase pipeline: (1) aspect-specific learning, (2) generalist initialization via interpolation, and (3) post-training optimization via extrapolation. PEO enables dynamic adaptation to diverse user preferences at inference time without retraining. Our comprehensive experiments across multiple LLMs demonstrate that PEO achieves superior Pareto fronts compared to baselines, offering improved flexibility and computational efficiency. Theoretical analyses further highlight PEO's capacity to overcome optimization bottlenecks, paving the way for scalable, personalized alignment.

📄 PDF Abstract BibTeX arXiv:2503.01233

Code (0)

등록된 구현이 없습니다.

Tasks

Computational Efficiency

Methods 이 논문이 사용한 방법론

DPO 설명 없음

Similar Papers 제목 키워드 기반

Are LLMs Vulnerable to Preference-Undermining Attacks (PUA)? A Factorial Analysis Methodology for Diagnosing the Trade-off between Preference Alignment and Real-World Validity

2026-01-10 · Hongjun An, Yiliang Song, Jiangan Chen, Jiawei Shao 외 arxiv

Large Language Model (LLM) training often optimizes for preference alignment, rewarding outputs that are perceived as helpful and interaction-friendly. However, this preference-oriented objective can be exploited: manipu…

TakeAD: Preference-based Post-optimization for End-to-end Autonomous Driving with Expert Takeover Data

2025-12-19 · Deqing Liu, Yinfeng Gao, Deheng Qian, Qichao Zhang 외 arxiv

Existing end-to-end autonomous driving methods typically rely on imitation learning (IL) but face a key challenge: the misalignment between open-loop training and closed-loop deployment. This misalignment often triggers …

Autonomous Driving

PITA: Preference-Guided Inference-Time Alignment for LLM Post-Training

2025-07-26 · Sarat Chandra Bobbili, Ujwal Dinesha, Dheeraj Narasimha, Srinivas Shakkottai arxiv

Inference-time alignment enables large language models (LLMs) to generate outputs aligned with end-user preferences without further training. Recent post-training methods achieve this by using small guidance models to mo…

Mathematical Reasoning

AAPA: Adversarially Anchored Preference Alignment for Post-Training of Large Language Models

2025-09-29 · Faqiang Qian, Kang An, Weikun Zhang, Ziliang Wang 외 arxiv

Post-training alignment of large language models often combines supervised fine-tuning (SFT) on expert demonstrations with reinforcement learning (RL) from preference or verifiable feedback. SFT provides a useful behavio…

Reinforcement Learning

Not All Preferences are What You Need for Post-Training: Selective Alignment Strategy for Preference Optimization

2025-07-10 · Zhijin Dong arxiv

Post-training alignment of large language models (LLMs) is a critical challenge, as not all tokens contribute equally to model performance. This paper introduces a selective alignment strategy that prioritizes high-impac…