Hallucination Detection on CoQA (AvgWD, EigenWD, ER, ES, DSE, LNE, LS)
0.742AvgWDAvgWD
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| AvgWDModel=Qwen2-7b2026.03 | 0.742 | — | — | — | — | — | — | |
| AvgWDModel=Qwen2-72b2026.03 | 0.746 | — | — | — | — | — | — | |
| AvgWDModel=Llama-3.1-8b2026.03 | 0.771 | — | — | — | — | — | — | |
| AvgWDModel=Llama-3.2-3b2026.03 | 0.811 | — | — | — | — | — | — | |
| AvgWDModel=Llama-2-7b2026.03 | 0.815 | — | — | — | — | — | — | |
| DSEModel=Llama-3.2-3b2026.03 | — | — | — | — | 0.781 | — | — | |
| DSEModel=Llama-2-7b2026.03 | — | — | — | — | 0.805 | — | — | |
| DSEModel=Llama-3.1-8b2026.03 | — | — | — | — | 0.761 | — | — | |
| DSEModel=Qwen2-7b2026.03 | — | — | — | — | 0.733 | — | — | |
| DSEModel=Qwen2-72b2026.03 | — | — | — | — | 0.715 | — | — | |
| EigenWDModel=Llama-3.2-3b2026.03 | — | 0.816 | — | — | — | — | — | |
| EigenWDModel=Llama-2-7b2026.03 | — | 0.812 | — | — | — | — | — | |
| EigenWDModel=Llama-3.1-8b2026.03 | — | 0.789 | — | — | — | — | — | |
| EigenWDModel=Qwen2-7b2026.03 | — | 0.734 | — | — | — | — | — | |
| EigenWDModel=Qwen2-72b2026.03 | — | 0.727 | — | — | — | — | — | |
| ERModel=Llama-3.2-3b2026.03 | — | — | 0.755 | — | — | — | — | |
| ERModel=Llama-2-7b2026.03 | — | — | 0.794 | — | — | — | — | |
| ERModel=Llama-3.1-8b2026.03 | — | — | 0.729 | — | — | — | — | |
| ERModel=Qwen2-7b2026.03 | — | — | 0.735 | — | — | — | — | |
| ERModel=Qwen2-72b2026.03 | — | — | 0.722 | — | — | — | — | |
| ESModel=Llama-3.2-3b2026.03 | — | — | — | 0.749 | — | — | — | |
| ESModel=Llama-2-7b2026.03 | — | — | — | 0.799 | — | — | — | |
| ESModel=Llama-3.1-8b2026.03 | — | — | — | 0.712 | — | — | — | |
| ESModel=Qwen2-7b2026.03 | — | — | — | 0.733 | — | — | — | |
| ESModel=Qwen2-72b2026.03 | — | — | — | 0.725 | — | — | — | |
| LNEModel=Llama-3.2-3b2026.03 | — | — | — | — | — | 0.602 | — | |
| LNEModel=Llama-2-7b2026.03 | — | — | — | — | — | 0.76 | — | |
| LNEModel=Llama-3.1-8b2026.03 | — | — | — | — | — | 0.618 | — | |
| LNEModel=Qwen2-7b2026.03 | — | — | — | — | — | 0.557 | — | |
| LNEModel=Qwen2-72b2026.03 | — | — | — | — | — | 0.616 | — | |
| LSModel=Llama-3.2-3b2026.03 | — | — | — | — | — | — | 0.773 | |
| LSModel=Llama-2-7b2026.03 | — | — | — | — | — | — | 0.802 | |
| LSModel=Llama-3.1-8b2026.03 | — | — | — | — | — | — | 0.769 | |
| LSModel=Qwen2-7b2026.03 | — | — | — | — | — | — | 0.738 | |
| LSModel=Qwen2-72b2026.03 | — | — | — | — | — | — | 0.758 |