Auto-Prompt Generation is Not Robust: Prompt Optimization Driven by Pseudo Gradient

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Shi, Zeru, Wang, Zhenting, Su, Yongye, Luo, Weidi, Gao, Hang, Yang, Fan, Tang, Ruixiang, Zhang, Yongfeng
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912660245708800
author Shi, Zeru
Wang, Zhenting
Su, Yongye
Luo, Weidi
Gao, Hang
Yang, Fan
Tang, Ruixiang
Zhang, Yongfeng
author_facet Shi, Zeru
Wang, Zhenting
Su, Yongye
Luo, Weidi
Gao, Hang
Yang, Fan
Tang, Ruixiang
Zhang, Yongfeng
contents While automatic prompt generation methods have recently received significant attention, their robustness remains poorly understood. In this paper, we introduce PertBench, a comprehensive benchmark dataset that includes a wide range of input perturbations, designed to systematically evaluate the robustness of current auto-prompting techniques. Our analysis reveals substantial vulnerabilities in existing prompt generation strategies, where even minor modifications to the prompt can lead to significant differences in model output. To address this issue, we propose PGO, a gradient-free prompt generation framework that leverages perturbation types as pseudo-gradient signals to guide LLMs in producing more robust prompts. In contrast to existing methods that assess prompt quality only on clean, well-structured inputs, our approach explicitly emphasizes robustness under noisy and perturbed conditions. Extensive experiments across diverse tasks and multiple LLMs show PGO consistently outperforms previous methods in maintaining performance under input perturbations.
format Preprint
id arxiv_https___arxiv_org_abs_2412_18196
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Auto-Prompt Generation is Not Robust: Prompt Optimization Driven by Pseudo Gradient
Shi, Zeru
Wang, Zhenting
Su, Yongye
Luo, Weidi
Gao, Hang
Yang, Fan
Tang, Ruixiang
Zhang, Yongfeng
Computation and Language
Machine Learning
While automatic prompt generation methods have recently received significant attention, their robustness remains poorly understood. In this paper, we introduce PertBench, a comprehensive benchmark dataset that includes a wide range of input perturbations, designed to systematically evaluate the robustness of current auto-prompting techniques. Our analysis reveals substantial vulnerabilities in existing prompt generation strategies, where even minor modifications to the prompt can lead to significant differences in model output. To address this issue, we propose PGO, a gradient-free prompt generation framework that leverages perturbation types as pseudo-gradient signals to guide LLMs in producing more robust prompts. In contrast to existing methods that assess prompt quality only on clean, well-structured inputs, our approach explicitly emphasizes robustness under noisy and perturbed conditions. Extensive experiments across diverse tasks and multiple LLMs show PGO consistently outperforms previous methods in maintaining performance under input perturbations.
title Auto-Prompt Generation is Not Robust: Prompt Optimization Driven by Pseudo Gradient
topic Computation and Language
Machine Learning
url https://arxiv.org/abs/2412.18196