Saved in:
Bibliographic Details
Main Authors: Chen, Xin, Toyer, Sam, Shkurti, Florian
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2407.16025
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909349183488000
author Chen, Xin
Toyer, Sam
Shkurti, Florian
author_facet Chen, Xin
Toyer, Sam
Shkurti, Florian
contents Spurious correlations in a reward model's training data can prevent Reinforcement Learning from Human Feedback (RLHF) from identifying the desired goal and induce unwanted behaviors. This paper shows that offline RLHF is susceptible to reward confusion, especially in the presence of spurious correlations in offline data. We create a benchmark to study this problem and propose a method that can significantly reduce reward confusion by leveraging transitivity of preferences while building a global preference chain with active learning.
format Preprint
id arxiv_https___arxiv_org_abs_2407_16025
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Exploring and Addressing Reward Confusion in Offline Preference Learning
Chen, Xin
Toyer, Sam
Shkurti, Florian
Machine Learning
Artificial Intelligence
Spurious correlations in a reward model's training data can prevent Reinforcement Learning from Human Feedback (RLHF) from identifying the desired goal and induce unwanted behaviors. This paper shows that offline RLHF is susceptible to reward confusion, especially in the presence of spurious correlations in offline data. We create a benchmark to study this problem and propose a method that can significantly reduce reward confusion by leveraging transitivity of preferences while building a global preference chain with active learning.
title Exploring and Addressing Reward Confusion in Offline Preference Learning
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2407.16025