Excited to share that our paper, “PEEM: Prompt Engineering Evaluation Metrics for Interpretable Joint Evaluation of Prompts and Responses,” has been published in IEEE Access! 🎉

This work introduces PEEM, a novel framework for the joint evaluation of prompts and model responses. Traditional metrics often focus solely on the output quality, neglecting the critical role of prompt engineering. PEEM addresses this gap by providing interpretable metrics that assess how well a prompt guides the model and how consistently the model responds.

Key highlights:

• Joint Evaluation: Simultaneously evaluates prompt quality and response accuracy.
• Interpretability: Provides granular insights into prompt-response dynamics.
• Robustness: Tested across various LLMs and prompt engineering techniques.

You can find the paper at:

• IEEE Xplore: https://ieeexplore.ieee.org/document/11460151
• arXiv: https://arxiv.org/abs/2603.10477