LLVM

The 'LLVM' dataset/benchmark contains historical bugs and program features used to evaluate compiler fuzzing techniques and the discovery of defects in compilers.

Papers using LLVM (4)

Discovering 100+ Compiler Defects in 72 Hours via LLM-Driven Semantic Logic Recomposition2026Improving Compiler Bug Isolation by Leveraging Large Language Models2025LaForge: Always-Correct and Fast Incremental Builds from Simple Specifications2021The Unexplored Treasure Trove of Phabricator Code Review2022