High-Dimension Human Value Representation in Large Language Models

TL;DR

This paper introduces UniVaR, a high-dimensional, scalable neural representation of human values in large language models, enabling visualization and understanding of value prioritization across diverse languages and cultures.

Contribution

We propose UniVaR, a novel continuous, self-supervised high-dimensional representation of human values in LLMs, independent of architecture and training data.

Findings

UniVaR effectively visualizes value prioritization in 25 languages.

It reveals complex interactions between human values and language modeling.

The approach is evaluated on 15 open-source and commercial LLMs.

Abstract

The widespread application of LLMs across various tasks and fields has necessitated the alignment of these models with human values and preferences. Given various approaches of human value alignment, there is an urgent need to understand the scope and nature of human values injected into these LLMs before their deployment and adoption. We propose UniVaR, a high-dimensional neural representation of symbolic human value distributions in LLMs, orthogonal to model architecture and training data. This is a continuous and scalable representation, self-supervised from the value-relevant output of 8 LLMs and evaluated on 15 open-source and commercial LLMs. Through UniVaR, we visualize and explore how LLMs prioritize different values in 25 languages and cultures, shedding light on complex interplay between human values and language modeling.