ProSA: Assessing and Understanding the Prompt Sensitivity of LLMs
This work addresses the challenge of prompt sensitivity for users and researchers of LLMs, though it is incremental as it builds on existing research by providing a new evaluation tool.
The authors tackled the problem of prompt sensitivity in large language models (LLMs), which causes performance variability and affects user satisfaction, by introducing ProSA, a framework that evaluates and understands this sensitivity, finding that larger models are more robust and few-shot examples can reduce sensitivity.
Large language models (LLMs) have demonstrated impressive capabilities across various tasks, but their performance is highly sensitive to the prompts utilized. This variability poses challenges for accurate assessment and user satisfaction. Current research frequently overlooks instance-level prompt variations and their implications on subjective evaluations. To address these shortcomings, we introduce ProSA, a framework designed to evaluate and comprehend prompt sensitivity in LLMs. ProSA incorporates a novel sensitivity metric, PromptSensiScore, and leverages decoding confidence to elucidate underlying mechanisms. Our extensive study, spanning multiple tasks, uncovers that prompt sensitivity fluctuates across datasets and models, with larger models exhibiting enhanced robustness. We observe that few-shot examples can alleviate this sensitivity issue, and subjective evaluations are also susceptible to prompt sensitivities, particularly in complex, reasoning-oriented tasks. Furthermore, our findings indicate that higher model confidence correlates with increased prompt robustness. We believe this work will serve as a helpful tool in studying prompt sensitivity of LLMs. The project is released at: https://github.com/open-compass/ProSA .