CLARC: C/C++ Benchmark for Robust Code Search

Kaicheng Wang , Liyan Huang , Weike Fang, Weihang Wang · 2026

Read CLARC: C/C++ Benchmark for Robust Code Search by Kaicheng Wang , Liyan Huang , Weike Fang, Weihang Wang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-04484.txt.

Description

Wang and colleagues benchmark robust C and C++ code search with CLARC. Retrieval aids developers, yet benchmarks focus on Python and rarely stress robustness beyond lexical cues. CLARC draws on real GitHub repos, holding 1,245 evaluation pairs and 5,472 training pairs. Model generated queries pass human scoring and hypothesis testing. The pipeline ensures compilability, sorts by dependency complexity, and stress tests with anonymization plus Assembly and WebAssembly compilation. Six leading models drop sharply, leaning on lexical features over semantic understanding. Code meaning thus matters more than mere words.

Catalog record

CREATORKaicheng Wang , Liyan Huang , Weike Fang, Weihang Wang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE58707
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.04484

Browse the full library · About the Tower of Babel