Failure prediction stopping on BETOLD
80.8Max Accuracyα-STOP
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| α-STOPPredictor=attention-based failure predictor2026.06 | 80.8 | 0.584 | 0.0005 | |
| ALERT*Predictor=attention-based failure predictor2026.06 | 79.7 | 0.361 | 0.0854 | |
| Plug-in ThresholdPredictor=attention-based failure predictor2026.06 | 79.7 | 0.511 | 0.0327 | |
| LLM Judge (best)Predictor=attention-based failure predictor2026.06 | 71.3 | — | — | |
| FIRMBOUNDPredictor=attention-based failure predictor2026.06 | 69.1 | 0.41 | 0.073 | |
| EndtoEndRLPredictor=attention-based failure predictor2026.06 | 67 | 0.445 | 0.0572 |