Logical Reasoning on Logic (test)
0.722AUROCPost-solution probe
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Post-solution probeProbe Type=Post-solution2026.07 | 0.722 | 0.17 | — | |
| Distilled future confidence predictorProbe Type=Distilled2026.07 | 0.696 | 0.184 | — | |
| Pre-solution probeProbe Type=Pre-solution2026.07 | 0.628 | 0.202 | — |