Adaptive Testing for LLM Evaluation: A Psychometric Alternative to Static Benchmarks

TL;DR

ATLAS introduces an adaptive testing framework based on Item Response Theory to evaluate large language models efficiently, reducing the number of required items by up to 90% while maintaining measurement accuracy.

Contribution

It presents a novel adaptive testing method for LLM evaluation using IRT, significantly decreasing testing resources needed compared to static benchmarks.

Findings

Reduces item count by up to 90% while maintaining accuracy.

Ability estimates closely match raw accuracy across benchmarks.

Provides finer discrimination among models with similar accuracy.

Abstract

Evaluating large language models (LLMs) typically requires thousands of benchmark items, making the process expensive, slow, and increasingly impractical at scale. Existing evaluation protocols rely on average accuracy over fixed item sets, treating all items as equally informative despite substantial variation in difficulty and discrimination. We introduce ATLAS, an adaptive testing framework based on Item Response Theory (IRT) that estimates model ability using Fisher information-guided item selection. ATLAS reduces the number of required items by up to 90% while maintaining measurement precision. For instance, it matches whole-bank ability estimates using only 41 items (0.157 MAE) on HellaSwag (5,600 items). We further reconstruct accuracy from ATLAS's ability estimates and find that reconstructed accuracies closely match raw accuracies across all five benchmarks, indicating that ability ${\theta}$ preserves the global performance structure. At the same time, ${\theta}$ provides finer discrimination within accuracy-equivalent models: among more than 3,000 evaluated models, 23-31% shift by more than 10 rank positions, and models with identical accuracies receive meaningfully different ability estimates. Code and calibrated item banks are available at https://github.com/Peiyu-Georgia-Li/ATLAS.git.