Code Generation Oversight on APPS
63Safety ScoreFactor(D, U)
Evaluation results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Factor(D, U)Decomposition Complexity (D)=low2025.12 | 63 | 106 | 18 | 13 | 23 | 999 | 2,026 | 338 | 377 | 34 | 19 | |
| Factor(D, U)Decomposition Complexity (D)=medium2025.12 | 56 | 103 | 17 | 13 | 17 | 999 | 2,964 | 243 | 275 | 24 | 9.3 | |
| Factor(D, U)Decomposition Complexity (D)=high2025.12 | 49 | 106 | 17 | 15 | 14 | 999 | 3,538 | 258 | 285 | 26 | 8.1 | |
| Trusted Monitoring2025.12 | 41 | 100 | 24 | 20 | 13 | 999 | 999 | 153 | 153 | 15 | 15 |