List your company

Audit-relevance Classification on benchmark_1000 n=993 (train)

79.6AccuracyPipeline current
42.36852.03461.771.366Jun 9, 2026

Evaluation results

MethodLinks
Pipeline currentType=Production pipeline2026.0679.6500
Always-approve (degenerate)Type=Degenerate reference2026.0674.600
Pre-deployment pure LLMMode=Pure LLM2026.0672.7—100
Opus + skills_v8 promptPrompt=skills_v82026.0669.817100
Opus zero-shot (B1b)Mode=Zero-shot, Variant=B1b2026.0643.832100