Factual Recall on Factual (test)
0.756AUROCPost-solution probe
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Post-solution probeProbe Type=Post-solution2026.07 | 0.756 | 0.104 | — | |
| Distilled future confidence predictorProbe Type=Distilled2026.07 | 0.731 | 0.133 | — | |
| Pre-solution probeProbe Type=Pre-solution2026.07 | 0.696 | 0.161 | — |