The Importance of Prompt Tuning for Automated Neuron Explanations

TL;DR

This paper investigates how prompt formatting influences the quality and efficiency of neuron explanations in large language models, demonstrating that natural prompts improve explanations and reduce computational costs.

Contribution

It introduces a prompt reformatting approach that enhances neuron explanation quality and efficiency in large language models.

Findings

Natural prompts significantly improve explanation quality

Reformatted prompts reduce computational costs

Both automated and human evaluations support the improvements

Abstract

Recent advances have greatly increased the capabilities of large language models (LLMs), but our understanding of the models and their safety has not progressed as fast. In this paper we aim to understand LLMs deeper by studying their individual neurons. We build upon previous work showing large language models such as GPT-4 can be useful in explaining what each neuron in a language model does. Specifically, we analyze the effect of the prompt used to generate explanations and show that reformatting the explanation prompt in a more natural way can significantly improve neuron explanation quality and greatly reduce computational cost. We demonstrate the effects of our new prompts in three different ways, incorporating both automated and human evaluations.