핵심 개념
LPR combines language-generic program reducers with Large Language Models to optimize program reduction across multiple languages.
초록
Program reduction techniques aim to minimize bug-triggering programs efficiently.
Existing techniques are either language-specific or language-generic.
LPR leverages LLMs to perform language-specific program reduction for C, Rust, and JavaScript.
LPR alternates between language-generic reducers and LLMs to optimize program reduction.
LPR outperforms Vulcan in reduction size and efficiency.
LPR's multi-level prompting approach guides LLMs in transformations.
LPR's proposed transformations include Function Inlining, Loop Unrolling, Data Type Elimination, Data Type Simplification, and Variable Elimination.
LPR's effectiveness is demonstrated across three benchmark suites.
통계
LPR는 C, Rust 및 JavaScript에서 프로그램 축소를 위해 LLM을 활용합니다.
LPR은 Vulcan을 능가하여 프로그램 크기와 효율성을 향상시킵니다.
인용구
"This paper proposes LPR, the first LLMs-aided technique leveraging LLMs to perform language-specific program reduction for multiple languages."
"LPR surpasses Vulcan by producing 24.93%, 4.47%, and 11.71% smaller programs on benchmarks in C, Rust, and JavaScript, separately."