Benchmark Inflation: Revealing LLM Performance Gaps Using Retro-Holdouts

Jacob Haimes, Cenny Wenner, Kunvar Thaman, Vassil Tashev, Clement Neo, Esben Kran, Jason Hoelscher-Obermaier

Keywords: Large Language Models, Benchmark, Evaluations, Datasets, TruthfulQA

TL;DR: LLM training data have been contaminated with test data, we quantify the extent to which reported model performance has been exaggerated.

Abstract: Public benchmarks are compromised, as the training data for many Large Language Models (LLMs) is contaminated with test data, suggesting a performance gap between benchmark scores and actual capabilities. Ideally, a private holdout set could be used to accurately verify scores. Unfortunately, such datasets do not exist for most benchmarks, and post-hoc construction of sufficiently similar datasets is non-trivial. To address these issues, we introduce a systematic methodology for (i) retrospectively constructing a holdout dataset for a target dataset, (ii) demonstrating the statistical indistinguishability of this retro-holdout dataset, and (iii) comparing LLMs on the two datasets to quantify the performance gap due to the dataset's public availability. Applying these methods to TruthfulQA, we construct and release Retro-TruthfulQA, on which we evaluate twenty LLMs and find that some have inflated scores by as much as 16 percentage points. Our results demonstrate that public benchmark scores do not always accurately assess model properties, and underscore the importance of improved data practices in the field.

Submission Number: 2029