ShadowBench

What is ShadowBench?

Evaluates autoformalization systems on 178 Lean 4 problems using the SA-Pass metric, which checks compile, forward, and backward semantic alignment via shadows.

Released

2026-08-29

Evaluates

General AI, Language & Knowledge, Safety

Openness

unknown

Importer

Claire Radar

Review status

ai-name-audit-deferred

Reported scores

0

Source provenance

• Original evidence https://arxiv.org/abs/2608.29270

ShadowBench paper

• Paper https://arxiv.org/abs/2608.29270

Which sources cite ShadowBench?

• ShadowBench

Related General AI, Language & Knowledge, Safety benchmarks

• $\tau^\tau$-Bench: An Environment for End-To-End, Realistic Agent Construction
• 3DCodeBench
• 5-Dialects-BN
• A Large-Scale Dataset and Benchmark
• A3S-Bench
• AARRI-Bench
• ABC-Bench
• ActBench