Do prompt positions really matter?

TL;DR

This paper empirically analyzes the impact of prompt position on NLP model performance, revealing its significant influence and highlighting the importance of optimizing prompt placement for better results.

Contribution

It provides the most comprehensive analysis to date on prompt position effects across diverse NLP tasks, emphasizing the need for prompt position optimization.

Findings

Prompt position significantly affects model performance.

Prior prompt positions are often sub-optimal.

Optimizing prompt position can improve NLP model robustness.

Abstract

Prompt-based models have gathered a lot of attention from researchers due to their remarkable advancements in the fields of zero-shot and few-shot learning. Developing an effective prompt template plays a critical role. However, prior studies have mainly focused on prompt vocabulary searching or embedding initialization within a predefined template with the prompt position fixed. In this empirical study, we conduct the most comprehensive analysis to date of prompt position for diverse Natural Language Processing (NLP) tasks. Our findings quantify the substantial impact prompt position has on model performance. We observe that the prompt positions used in prior studies are often sub-optimal, and this observation is consistent even in widely used instruction-tuned models. These findings suggest prompt position optimisation as a valuable research direction to augment prompt engineering methodologies and prompt position-aware instruction tuning as a potential way to build more robust models in the future.