Truth Forest: Toward Multi-Scale Truthfulness in Large Language Models through Intervention without Tuning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chen, Zhongzhi, Sun, Xingwu, Jiao, Xianfeng, Lian, Fengzong, Kang, Zhanhui, Wang, Di, Xu, Cheng-Zhong
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911760102981632
author Chen, Zhongzhi
Sun, Xingwu
Jiao, Xianfeng
Lian, Fengzong
Kang, Zhanhui
Wang, Di
Xu, Cheng-Zhong
author_facet Chen, Zhongzhi
Sun, Xingwu
Jiao, Xianfeng
Lian, Fengzong
Kang, Zhanhui
Wang, Di
Xu, Cheng-Zhong
contents Despite the great success of large language models (LLMs) in various tasks, they suffer from generating hallucinations. We introduce Truth Forest, a method that enhances truthfulness in LLMs by uncovering hidden truth representations using multi-dimensional orthogonal probes. Specifically, it creates multiple orthogonal bases for modeling truth by incorporating orthogonal constraints into the probes. Moreover, we introduce Random Peek, a systematic technique considering an extended range of positions within the sequence, reducing the gap between discerning and generating truth features in LLMs. By employing this approach, we improved the truthfulness of Llama-2-7B from 40.8\% to 74.5\% on TruthfulQA. Likewise, significant improvements are observed in fine-tuned models. We conducted a thorough analysis of truth features using probes. Our visualization results show that orthogonal probes capture complementary truth-related features, forming well-defined clusters that reveal the inherent structure of the dataset.
format Preprint
id arxiv_https___arxiv_org_abs_2312_17484
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Truth Forest: Toward Multi-Scale Truthfulness in Large Language Models through Intervention without Tuning
Chen, Zhongzhi
Sun, Xingwu
Jiao, Xianfeng
Lian, Fengzong
Kang, Zhanhui
Wang, Di
Xu, Cheng-Zhong
Computation and Language
Artificial Intelligence
Despite the great success of large language models (LLMs) in various tasks, they suffer from generating hallucinations. We introduce Truth Forest, a method that enhances truthfulness in LLMs by uncovering hidden truth representations using multi-dimensional orthogonal probes. Specifically, it creates multiple orthogonal bases for modeling truth by incorporating orthogonal constraints into the probes. Moreover, we introduce Random Peek, a systematic technique considering an extended range of positions within the sequence, reducing the gap between discerning and generating truth features in LLMs. By employing this approach, we improved the truthfulness of Llama-2-7B from 40.8\% to 74.5\% on TruthfulQA. Likewise, significant improvements are observed in fine-tuned models. We conducted a thorough analysis of truth features using probes. Our visualization results show that orthogonal probes capture complementary truth-related features, forming well-defined clusters that reveal the inherent structure of the dataset.
title Truth Forest: Toward Multi-Scale Truthfulness in Large Language Models through Intervention without Tuning
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2312.17484