EXHIB: A Benchmark for Realistic and Diverse Evaluation of Function Similarity in the Wild

TL;DR

EXHIB is a comprehensive benchmark with five real-world datasets for evaluating binary function similarity detection models, revealing significant generalization gaps and robustness issues.

Contribution

It introduces a diverse, realistic benchmark for BFSD, enabling better evaluation of models across different application scenarios.

Findings

Models show up to 30% performance degradation on firmware and semantic datasets.

Current models lack robustness to high-level semantic differences.

Existing evaluation practices overlook critical generalization challenges.

Abstract

Binary Function Similarity Detection (BFSD) is a core problem in software security, supporting tasks such as vulnerability analysis, malware classification, and patch provenance. In the past few decades, numerous models and tools have been developed for this application; however, due to the lack of a comprehensive universal benchmark in this field, researchers have struggled to compare different models effectively. Existing datasets are limited in scope, often focusing on a narrow set of transformations or types of binaries, and fail to reflect the full diversity of real-world applications. We introduce EXHIB, a benchmark comprising five realistic datasets collected from the wild, each highlighting a distinct aspect of the BFSD problem space. We evaluate 9 representative models spanning multiple BFSD paradigms on EXHIB and observe performance degradations of up to 30% on firmware and semantic datasets compared to standard settings, revealing substantial generalization gaps. Our results show that robustness to low- and mid-level binary variations does not generalize to high-level semantic differences, underscoring a critical blind spot in current BFSD evaluation practices.