Benchmark RadarLCS-Bench
LCS-Bench is a theory-scale benchmark for auto-formalization in logics for computer science. It includes 327 textbook items, over 4,076 Lean declarations, and supports five evaluation tracks with definitional equivalence checkers.
- Released
- 2026-06-25
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
Auto-formalization at theory scale remains challenging. LCS-Bench provides a reusable benchmark to measure consistency, faithfulness, and correctness of formalization systems, enabling progress in scalable verification.
Motivation
Auto-formalization is critical for scalable formal verification, but existing progress largely focuses on isolated statements, while theory-scale auto-formalization, which coherently translates hundreds of interdependent definitions, lemmas, and theorems, remains open due to challenges in consistency, faithfulness, scalability, and correctness.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.