Pre-training data contamination detection on CommonsenseQA (test)
51.9AUCProbe w. Synthetic Data
Evaluation Results
| Method | Links | |
|---|---|---|
| Probe w. Synthetic DataCategory=Our Method, Training Data=Synthetic2024.06 | 51.9 | |
| Neighbor AttackCategory=Reference-free2024.06 | 50 | |
| Loss AttackCategory=Reference-free2024.06 | 49.9 | |
| Min-K% ProbCategory=Reference-free2024.06 | 49.6 | |
| Smaller ModelCategory=Reference-based, Reference Model=Contam-Small (124M Params)2024.06 | 49.5 | |
| Lowercased TextCategory=Reference-based2024.06 | 49.2 | |
| Zlib CompressionCategory=Reference-based2024.06 | 48.8 |