metadata
title: ScopeGuard Benchmark
emoji: 🛡️
colorFrom: orange
colorTo: gray
sdk: static
pinned: false
short_description: Explore every held-out ScopeGuard evaluation result.
ScopeGuard Benchmark Explorer
Interactive static explorer for the complete 100-example held-out evaluation of ScopeGuard Qwen2.5-1.5B LoRA.
This Space does not run inference. It displays committed deterministic evaluation artifacts so reviewers can inspect successes and failures directly.