Answer Matching Outperforms Multiple Choice for Language Model Evaluation

TL;DR

This paper demonstrates that answer matching, which evaluates free-form responses against reference answers using language models, outperforms traditional multiple choice benchmarks in aligning with human judgment for language model evaluation.

Contribution

The paper introduces answer matching as a scalable, more accurate alternative to multiple choice evaluation, showing its near-perfect agreement with human grading.

Findings

Answer matching achieves near-perfect agreement with human grading.

Multiple choice evaluation aligns poorly with human judgment.

Rankings of models change significantly when using answer matching.

Abstract

Multiple choice benchmarks have long been the workhorse of language model evaluation because grading multiple choice is objective and easy to automate. However, we show multiple choice questions from popular benchmarks can often be answered without even seeing the question. These shortcuts arise from a fundamental limitation of discriminative evaluation not shared by evaluations of the model's free-form, generative answers. Until recently, there appeared to be no viable, scalable alternative to multiple choice--but, we show that this has changed. We consider generative evaluation via what we call answer matching: Give the candidate model the question without the options, have it generate a free-form response, then use a modern language model with the reference answer to determine if the response matches the reference. To compare the validity of different evaluation strategies, we annotate MMLU-Pro and GPQA-Diamond to obtain human grading data, and measure the agreement of each evaluation approach. We find answer matching using recent models--even small ones--achieves near-perfect agreement, in the range of inter-annotator agreement. In contrast, both multiple choice evaluation and using LLM-as-a-judge without reference answers aligns poorly with human grading. Improving evaluations via answer matching is not merely a conceptual concern: the rankings of several models change significantly when evaluating their free-form responses with answer matching. In light of these findings, we discuss how to move the evaluation ecosystem from multiple choice to answer matching.