Code Debugging on LiveCodeBench February-May 2025 (evaluation window)
36.6AccuracyMGDebugger
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MGDebuggerModel=GPT-4o2024.10 | 36.6 | 10.8 | |
| ReflexionModel=GPT-4o2024.10 | 33.6 | 5.4 | |
| Self-Debugging (Expl.)Model=GPT-4o2024.10 | 32.8 | 4.3 | |
| No-DebuggingModel=GPT-4o2024.10 | 29 | — |