Audit-relevance Classification on new n=1462 2000 (held-out)
77.3AccuracyPipeline current
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Pipeline currentType=Production pipeline2026.06 | 77.3 | 38 | 0 | |
| Always-approve (degenerate)Type=Degenerate reference2026.06 | 76.6 | 0 | 0 |
| Method | Links | |||
|---|---|---|---|---|
| Pipeline currentType=Production pipeline2026.06 | 77.3 | 38 | 0 | |
| Always-approve (degenerate)Type=Degenerate reference2026.06 | 76.6 | 0 | 0 |