paper-with-me

Papers

More RLHF, More Trust? On The Impact of Preference Alignment On Trustworthiness

2024-04-29 · Aaron J. Li, Satyapriya Krishna, Himabindu Lakkaraju

The trustworthiness of Large Language Models (LLMs) refers to the extent to which their outputs are reliable, safe, and ethically aligned, and it has become a crucial consideration alongside their cognitive performance. In practice, Reinforcement Learning From Human Feedback (RLHF) has been widely used to align LLMs with labeled human preferences, but its assumed effect on model trustworthiness hasn't been rigorously evaluated. To bridge this knowledge gap, this study investigates how models aligned with general-purpose preference data perform across five trustworthiness verticals: toxicity, stereotypical bias, machine ethics, truthfulness, and privacy. Our results demonstrate that RLHF on human preferences doesn't automatically guarantee trustworthiness, and reverse effects are often observed. Furthermore, we propose to adapt efficient influence function based data attribution methods to the RLHF setting to better understand the influence of fine-tuning data on individual trustworthiness benchmarks, and show its feasibility by providing our estimated attribution scores. Together, our results underscore the need for more nuanced approaches for model alignment from both the data and framework perspectives, and we hope this research will guide the community towards developing language models that are increasingly capable without sacrificing trustworthiness.

📄 PDF Abstract BibTeX arXiv:2404.18870

Code (1)

ai4life-group/rlhf_trust 공식 구현 pytorch

Tasks

EthicsLanguage Modelling

Methods 이 논문이 사용한 방법론

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…
Focus 설명 없음

Similar Papers 제목 키워드 기반

RLHF-V: Towards Trustworthy MLLMs via Behavior Alignment from Fine-grained Correctional Human Feedback

2023-12-01 · CVPR 2024 1 · Tianyu Yu, Yuan YAO, Haoye Zhang, Taiwen He 외

Multimodal Large Language Models (MLLMs) have recently demonstrated impressive capabilities in multimodal understanding, reasoning, and interaction. However, existing MLLMs prevalently suffer from serious hallucination p…

HallucinationImage CaptioningVisual Question Answering

LLM Misalignment via Adversarial RLHF Platforms

2025-03-04 · Erfan Entezami, Ali Naseh

Reinforcement learning has shown remarkable performance in aligning language models with human preferences, leading to the rise of attention towards developing RLHF platforms. These platforms enable users to fine-tune mo…

Filtered Direct Preference Optimization

2024-04-22 · Tetsuro Morimura, Mitsuki Sakamoto, Yuu Jinnai, Kenshi Abe 외

Reinforcement learning from human feedback (RLHF) plays a crucial role in aligning language models with human preferences. While the significance of dataset quality is generally recognized, explicit investigations into i…

Secrets of RLHF in Large Language Models Part II: Reward Modeling

2024-01-11 · Binghai Wang, Rui Zheng, Lu Chen, Yan Liu 외

Reinforcement Learning from Human Feedback (RLHF) has become a crucial technology for aligning language models with human values and intentions, enabling models to produce more helpful and harmless responses. Reward mode…

Contrastive LearningMeta-Learningreinforcement-learning

RLHS: Mitigating Misalignment in RLHF with Hindsight Simulation

2025-01-15 · Kaiqu Liang, Haimin Hu, Ryan Liu, Thomas L. Griffiths 외

Generative AI systems like foundation models (FMs) must align well with human values to ensure their behavior is helpful and trustworthy. While Reinforcement Learning from Human Feedback (RLHF) has shown promise for opti…