Prompt Optimization with Human Feedback

TL;DR

This paper introduces APOHF, a new algorithm for optimizing prompts for large language models using only human preference feedback, inspired by dueling bandits, and demonstrates its efficiency across various tasks.

Contribution

The paper proposes APOHF, a theoretically grounded method for prompt optimization with human feedback, addressing the challenge of lacking numeric scores for prompt evaluation.

Findings

APOHF efficiently finds good prompts with few preference feedbacks.

APOHF applies successfully to instruction tuning, text-to-image models, and response refinement.

The method outperforms baseline approaches in prompt optimization tasks.

Abstract

Large language models (LLMs) have demonstrated remarkable performances in various tasks. However, the performance of LLMs heavily depends on the input prompt, which has given rise to a number of recent works on prompt optimization. However, previous works often require the availability of a numeric score to assess the quality of every prompt. Unfortunately, when a human user interacts with a black-box LLM, attaining such a score is often infeasible and unreliable. Instead, it is usually significantly easier and more reliable to obtain preference feedback from a human user, i.e., showing the user the responses generated from a pair of prompts and asking the user which one is preferred. Therefore, in this paper, we study the problem of prompt optimization with human feedback (POHF), in which we aim to optimize the prompt for a black-box LLM using only human preference feedback. Drawing inspiration from dueling bandits, we design a theoretically principled strategy to select a pair of prompts to query for preference feedback in every iteration, and hence introduce our algorithm named automated POHF (APOHF). We apply our APOHF algorithm to various tasks, including optimizing user instructions, prompt optimization for text-to-image generative models, and response optimization with human feedback (i.e., further refining the response using a variant of our APOHF). The results demonstrate that our APOHF can efficiently find a good prompt using a small number of preference feedback instances. Our code can be found at \url{https://github.com/xqlin98/APOHF}.