Failure prediction stopping on P4G
77.5Max Accuracyα-STOP
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| α-STOPPredictor=attention-based failure predictor2026.06 | 77.5 | 0.757 | 0 | |
| LLM Judge (best)Predictor=attention-based failure predictor2026.06 | 68.6 | — | — | |
| Plug-in ThresholdPredictor=attention-based failure predictor2026.06 | 67.6 | 0.675 | 0.0011 | |
| EndtoEndRLPredictor=attention-based failure predictor2026.06 | 66.7 | 0.623 | 0.0068 | |
| FIRMBOUNDPredictor=attention-based failure predictor2026.06 | 62.7 | 0.609 | 0.0083 | |
| ALERT*Predictor=attention-based failure predictor2026.06 | 54.9 | 0.585 | 0.01 |