paper-with-me

홈 › Papers

Curry-DPO: Enhancing Alignment using Curriculum Learning & Ranked Preferences

2024-03-12 · Pulkit Pattnaik, Rishabh Maheshwary, Kelechi Ogueji, Vikas Yadav, Sathwik Tejaswi Madhusudhan

Direct Preference Optimization (DPO) is an effective technique that leverages pairwise preference data (usually one chosen and rejected response pair per user prompt) to align LLMs to human preferences. In practice, multiple responses can exist for a given prompt with varying quality relative to each other. With availability of such quality ratings for multiple responses, we propose utilizing these responses to create multiple preference pairs for a given prompt. Our work focuses on systematically using the constructed multiple preference pair in DPO training via curriculum learning methodology. In particular, we order these multiple pairs of preference data from easy to hard (emulating curriculum training) according to various criteria. We show detailed comparisons of our proposed approach to the standard single-pair DPO setting. Our method, which we call Curry-DPO consistently shows increased performance gains on MTbench, Vicuna, WizardLM, and the UltraFeedback test set, highlighting its effectiveness. More specifically, Curry-DPO achieves a score of 7.43 on MT-bench with Zephy-7B model outperforming majority of existing LLMs with similar parameter size. Curry-DPO also achieves the highest adjusted win rates on Vicuna, WizardLM, and UltraFeedback test datasets (90.7%, 87.1%, and 87.9% respectively) in our experiments, with notable gains of upto 7.5% when compared to standard DPO technique. We release the preference pairs used in alignment at: https://huggingface.co/datasets/ServiceNow-AI/Curriculum_DPO_preferences

📄 PDF Abstract BibTeX arXiv:2403.07230

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

What and When to Learn: CURriculum Ranking Loss for Large-Scale Speaker Verification

2026-03-25 · Massa Baali, Sarthak Bisht, Rita Singh, Bhiksha Raj arxiv

Speaker verification at large scale remains an open challenge as fixed-margin losses treat all samples equally regardless of quality. We hypothesize that mislabeled or degraded samples introduce noisy gradients that disr…

Speaker Verification

Towards Better Optimization For Listwise Preference in Diffusion Models

2025-10-02 · Jiamu Bai, Xin Yu, Meilong Xu, Weitao Lu 외 arxiv

Reinforcement learning from human feedback (RLHF) has proven effectiveness for aligning text-to-image (T2I) diffusion models with human preferences. Although Direct Preference Optimization (DPO) is widely adopted for its…

Text-to-Image GenerationComputational EfficiencyReinforcement LearningImage Editing

Two Minds Better Than One: Collaborative Reward Modeling for LLM Alignment

2025-05-15 · Jiazheng Zhang, Wenqing Jing, Zizhuo Zhang, Zhiheng Xi 외

Reward models (RMs) play a pivotal role in aligning large language models (LLMs) with human values. However, noisy preferences in human feedback can lead to reward misgeneralization - a phenomenon where reward models lea…

To Go Far, Go Together: Diverse Preferences Induce a Curriculum for Reward Optimization

2026-08-19 · Taehyung Kim, Jongeun Choi arxiv

Learning a reward model from human feedback and optimizing a policy against it is one approach to aligning AI systems with individual users. From a fairness perspective, existing work improves such alignment by developin…

Continuous Control

RAFT: Reward rAnked FineTuning for Generative Foundation Model Alignment

2023-04-13 · Hanze Dong, Wei Xiong, Deepanshu Goyal, Yihan Zhang 외

Generative foundation models are susceptible to implicit biases that can arise from extensive unsupervised training data. Such biases can produce suboptimal samples, skewed outcomes, and unfairness, with potentially seri…

Ethics