Data Contamination Detection on AIME 2025
76F1 ScoreSelf-Critique
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Self-CritiqueTarget Model=Qwen2.5-7B-Instruct2025.10 | 76 | 0.72 | |
| Entropy-NoiseTarget Model=DeepSeek-Math-7B-Instruct2025.10 | 72 | 0.69 | |
| Self-CritiqueTarget Model=DeepSeek-Math-7B-Instruct2025.10 | 71 | 0.61 | |
| Entropy-TempTarget Model=DeepSeek-Math-7B-Instruct2025.10 | 70 | 0.54 | |
| CDDTarget Model=Qwen2.5-7B-Instruct2025.10 | 67 | 0.52 | |
| Entropy-NoiseTarget Model=Qwen2.5-7B-Instruct2025.10 | 67 | 0.63 | |
| Min-K%++Target Model=DeepSeek-Math-7B-Instruct2025.10 | 67 | 0.56 | |
| CDDTarget Model=DeepSeek-Math-7B-Instruct2025.10 | 65 | 0.51 | |
| PPLTarget Model=DeepSeek-Math-7B-Instruct2025.10 | 62 | 0.41 | |
| RecallTarget Model=Qwen2.5-7B-Instruct2025.10 | 55 | 0.65 | |
| Min-K%Target Model=Qwen2.5-7B-Instruct2025.10 | 52 | 0.44 | |
| RecallTarget Model=DeepSeek-Math-7B-Instruct2025.10 | 52 | 0.56 | |
| Min-K%++Target Model=Qwen2.5-7B-Instruct2025.10 | 46 | 0.45 | |
| PPLTarget Model=Qwen2.5-7B-Instruct2025.10 | 42 | 0.56 | |
| Entropy-TempTarget Model=Qwen2.5-7B-Instruct2025.10 | 12 | 0.42 | |
| Min-K%Target Model=DeepSeek-Math-7B-Instruct2025.10 | 12 | 0.4 |