Safetyv1.0.0
sh-benchmark-contamination
Benchmark Contamination: inspects eval scores. Verdicts: flag (severity medium). Rules: warn. False-positive story: Canonical public benchmarks overlap most corpora by nature; the override flags the score as overlap-limited in reporting.
Made by QynQue · 14d ago
On QueHub, a guardrail that keeps it honest. Not machine-checked yet — when a check exists, you'll see its verdict here, never before.
The fine print
- Kind
- shield
- Version
- 1.0.0
- Made by
- QynQue
- On the shelf since
- 9/25/2026