paper-with-me

홈 › Papers

Direct Alignment of Language Models via Quality-Aware Self-Refinement

2024-05-31 · Runsheng Yu, Yong Wang, Xiaoqi Jiao, Youzhi Zhang, James T. Kwok

Reinforcement Learning from Human Feedback (RLHF) has been commonly used to align the behaviors of Large Language Models (LLMs) with human preferences. Recently, a popular alternative is Direct Policy Optimization (DPO), which replaces an LLM-based reward model with the policy itself, thus obviating the need for extra memory and training time to learn the reward model. However, DPO does not consider the relative qualities of the positive and negative responses, and can lead to sub-optimal training outcomes. To alleviate this problem, we investigate the use of intrinsic knowledge within the on-the-fly fine-tuning LLM to obtain relative qualities and help to refine the loss function. Specifically, we leverage the knowledge of the LLM to design a refinement function to estimate the quality of both the positive and negative responses. We show that the constructed refinement function can help self-refine the loss function under mild assumptions. The refinement function is integrated into DPO and its variant Identity Policy Optimization (IPO). Experiments across various evaluators indicate that they can improve the performance of the fine-tuned models over DPO and IPO.

📄 PDF Abstract BibTeX arXiv:2405.21040

Code (0)

등록된 구현이 없습니다.

Methods 이 논문이 사용한 방법론

DPO 설명 없음
ALIGN In the ALIGN method, visual and language representations are jointly trained from noisy image alt-text data. The image and text encoders are learned via contrastive loss…

Similar Papers 제목 키워드 기반

Cross-Temporal Attention Fusion (CTAF) for Multimodal Physiological Signals in Self-Supervised Learning

2026-02-02 · Arian Khorasani, Théophile Demazure arxiv

We study multimodal affect modeling when EEG and peripheral physiology are asynchronous, which most fusion methods ignore or handle with costly warping. We propose Cross-Temporal Attention Fusion (CTAF), a self-supervise…

Self-Supervised Learning

Learning to Align, Aligning to Learn: A Unified Approach for Self-Optimized Alignment

2025-08-11 · Haowen Wang, Yun Yue, Zhiling Ye, Shuowen Zhang 외 arxiv

Alignment methodologies have emerged as a critical pathway for enhancing language model alignment capabilities. While SFT (supervised fine-tuning) accelerates convergence through direct token-level loss intervention, its…

Reinforcement Learning

Emergently Misaligned Language Models Show Behavioral Self-Awareness That Shifts With Subsequent Realignment

2026-02-16 · Laurène Vaugrante, Anietta Weckauff, Thilo Hagendorff arxiv

Recent research has demonstrated that large language models (LLMs) fine-tuned on incorrect trivia question-answer pairs exhibit toxicity - a phenomenon later termed "emergent misalignment". Moreover, research has shown t…

Self-Pluralising Culture Alignment for Large Language Models

2024-10-16 · Shaoyang Xu, Yongqi Leng, Linhao Yu, Deyi Xiong

As large language models (LLMs) become increasingly accessible in many countries, it is essential to align them to serve pluralistic human values across cultures. However, pluralistic culture alignment in LLMs remain an …

Prompt Engineering

Alignment-Aware Decoding

2025-09-30 · Frédéric Berdoz, Luca A. Lanzendörfer, René Caky, Roger Wattenhofer arxiv

Alignment of large language models remains a central challenge in natural language processing. Preference optimization has emerged as a popular and effective method for improving alignment, typically through training-tim…