Preference Optimization with Multi-Sample Comparisons Article Swipe

MAKE IMPACT

Chaoqi Wang , Zhimiao Zhao , Chen Zhu , Karthik Abinav Sankararaman , Michal Vaľko , Xuefei Cao , Zhaorun Chen , Madian Khabsa , Y. Q. Chen , Hao Ma , Sinong Wang · YOU? · · 2024 · Open Access · · DOI: https://doi.org/10.48550/arxiv.2410.12138

Recent advancements in generative models, particularly large language models (LLMs) and diffusion models, have been driven by extensive pretraining on large datasets followed by post-training. However, current post-training methods such as reinforcement learning from human feedback (RLHF) and direct alignment from preference methods (DAP) primarily utilize single-sample comparisons. These approaches often fail to capture critical characteristics such as generative diversity and bias, which are more accurately assessed through multiple samples. To address these limitations, we introduce a novel approach that extends post-training to include multi-sample comparisons. To achieve this, we propose Multi-sample Direct Preference Optimization (mDPO) and Multi-sample Identity Preference Optimization (mIPO). These methods improve traditional DAP methods by focusing on group-wise characteristics. Empirically, we demonstrate that multi-sample comparison is more effective in optimizing collective characteristics~(e.g., diversity and bias) for generative models than single-sample comparison. Additionally, our findings suggest that multi-sample comparisons provide a more robust optimization framework, particularly for dataset with label noise.

Type

preprint

Language

en

Landing Page

http://arxiv.org/abs/2410.12138

PDF

https://arxiv.org/pdf/2410.12138

OA Status

green

Related Works

10

OpenAlex ID

https://openalex.org/W4403577580

Raw OpenAlex JSON

OpenAlex ID

https://openalex.org/W4403577580

Canonical identifier for this work in OpenAlex

DOI

https://doi.org/10.48550/arxiv.2410.12138

Digital Object Identifier

Title

Preference Optimization with Multi-Sample Comparisons

Work title

Type

preprint

OpenAlex work type

Language

en

Primary language

Publication year

2024

Year of publication

Publication date

2024-10-16

Full publication date if available

Authors

Chaoqi Wang, Zhimiao Zhao, Chen Zhu, Karthik Abinav Sankararaman, Michal Vaľko, Xuefei Cao, Zhaorun Chen, Madian Khabsa, Y. Q. Chen, Hao Ma, Sinong Wang

List of authors in order

Landing page

https://arxiv.org/abs/2410.12138

Publisher landing page

PDF URL

https://arxiv.org/pdf/2410.12138

Direct link to full text PDF

Open access

Yes

Whether a free full text is available

OA status

green

Open access status per OpenAlex

OA URL

https://arxiv.org/pdf/2410.12138

Direct OA link when available

Concepts

Preference, Sample (material), Statistics, Mathematics, Computer science, Econometrics, Chemistry, Chromatography

Top concepts (fields/topics) attached by OpenAlex

Cited by

0

Total citation count in OpenAlex

Related works (count)

10

Other works algorithmically related by OpenAlex

Full payload