Inductive Bias Extraction and Matching for LLM Prompts

TL;DR

This paper introduces a method to improve prompt engineering for large language models by extracting and matching the model's inductive bias, leading to significant performance improvements in classification and ranking tasks.

Contribution

It proposes a novel inductive bias extraction and matching strategy that enhances prompt effectiveness by aligning with the model's inherent biases.

Findings

Up to 19% improvement in classification Likert ratings

Up to 27% improvement in ranking Likert ratings

Demonstrates the effectiveness of bias matching in prompt engineering

Abstract

The active research topic of prompt engineering makes it evident that LLMs are sensitive to small changes in prompt wording. A portion of this can be ascribed to the inductive bias that is present in the LLM. By using an LLM's output as a portion of its prompt, we can more easily create satisfactory wording for prompts. This has the effect of creating a prompt that matches the inductive bias in model. Empirically, we show that using this Inductive Bias Extraction and Matching strategy improves LLM Likert ratings used for classification by up to 19% and LLM Likert ratings used for ranking by up to 27%.