Misclassification Detection on Cola
81.6ROC-AUCALIEN
Evaluation Results
| Method | Links | |
|---|---|---|
| ALIENBackbone=Qwen2.5 7B, Detection Method=ALIEN2025.05 | 81.6 | |
| ALIENBackbone=Roberta, Detection Method=ALIEN2025.05 | 80.5 | |
| ALIENBackbone=Electra, Detection Method=ALIEN2025.05 | 78.8 | |
| ALIENBackbone=LLaMA2 7B, Detection Method=ALIEN2025.05 | 78 | |
| Probability based (Entropy)Backbone=LLaMA2 7B, Detection Method=Probability based, Method Variant=Entropy2025.05 | 77 | |
| Probability based (SR)Backbone=LLaMA2 7B, Detection Method=Probability based, Method Variant=SR2025.05 | 76.9 | |
| ALIENBackbone=Qwen3 1.7B, Detection Method=ALIEN2025.05 | 73.7 | |
| Linear probing (Last)Backbone=LLaMA2 7B, Detection Method=Linear probing, Layer Selection=Last2025.05 | 72 | |
| Mahalanobis distance (MDR)Backbone=LLaMA2 7B, Detection Method=Mahalanobis distance, Method Variant=MDR2025.05 | 69.9 | |
| RDEBackbone=LLaMA2 7B, Detection Method=RDE2025.05 | 68 | |
| Mahalanobis distance (MD)Backbone=LLaMA2 7B, Detection Method=Mahalanobis distance, Method Variant=MD2025.05 | 62.2 | |
| Mahalanobis distance (MDM)Backbone=LLaMA2 7B, Detection Method=Mahalanobis distance, Method Variant=MDM2025.05 | 62.2 | |
| Attention pooling (Last)Backbone=LLaMA2 7B, Detection Method=Attention pooling, Layer Selection=Last2025.05 | 61.5 | |
| Linear probing (Mid)Backbone=LLaMA2 7B, Detection Method=Linear probing, Layer Selection=Mid2025.05 | 60.6 | |
| Attention pooling (Begin)Backbone=LLaMA2 7B, Detection Method=Attention pooling, Layer Selection=Begin2025.05 | 60.3 | |
| Linear probing (Begin)Backbone=LLaMA2 7B, Detection Method=Linear probing, Layer Selection=Begin2025.05 | 58.1 | |
| Attention pooling (Mid)Backbone=LLaMA2 7B, Detection Method=Attention pooling, Layer Selection=Mid2025.05 | 51.4 | |
| MDBackbone=Qwen2.5 7B2025.05 | 14.5 | |
| Attention PoolingBackbone=LLaMA2 7B, Layer=Last2025.05 | 11.2 | |
| MDMBackbone=LLaMA2 7B2025.05 | 10.4 | |
| MDBackbone=LLaMA2 7B2025.05 | 10.3 | |
| MDRBackbone=LLaMA2 7B2025.05 | 9.3 | |
| ALIENBackbone=Qwen3 1.7B2025.05 | 9 | |
| RDEBackbone=LLaMA2 7B2025.05 | 8.4 | |
| Linear ProbingBackbone=LLaMA2 7B, Layer=Last2025.05 | 7.5 | |
| SRBackbone=LLaMA2 7B2025.05 | 6.8 | |
| EntropyBackbone=LLaMA2 7B2025.05 | 6.5 | |
| ALIENBackbone=LLaMA2 7B2025.05 | 6.2 | |
| ALIENBackbone=Roberta2025.05 | 6 | |
| ALIENBackbone=Qwen2.5 7B2025.05 | 5.3 | |
| ALIENBackbone=Electra2025.05 | 4.9 |