Optimization before Evaluation: Evaluation with Unoptimised Prompts Can be Misleading
For practitioners evaluating LLMs, this work highlights a critical methodological flaw in current evaluation practices that can lead to incorrect model selection.
The paper shows that prompt optimization significantly changes the ranking of LLMs on benchmarks, meaning evaluations without per-model prompt optimization can be misleading. Results on academic and industry benchmarks demonstrate that rankings shift after optimization.
Current Large Language Model (LLM) evaluation frameworks utilize the same static prompt template across all models under evaluation. This differs from the common industry practice of using prompt optimization (PO) techniques to optimize the prompt for each model to maximize application performance. In this paper, we investigate the effect of PO towards LLM evaluations. Our results on public academic and internal industry benchmarks show that PO greatly affects the final ranking of models. This highlights the importance of practitioners performing PO per model when conducting evaluations to choose the best model for a given task.