Hallucination Detection on CoQA (dev)
77.8AUROC (Token Entropy)UA-CLM
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| UA-CLMModel=Gemma-2B2024.12 | 77.8 | 78.37 | 73.58 | 78.71 | |
| CLMModel=Gemma-2B2024.12 | 77.23 | 76.06 | 72.95 | 76.18 | |
| ULTModel=Gemma-2B2024.12 | 70.97 | 69.21 | 65.4 | 71.62 | |
| Pre-trainedModel=Gemma-2B2024.12 | 70.73 | 70.89 | 69.62 | 76.35 | |
| UA-CLMModel=Llama-2-7B2024.12 | 69.55 | 73.98 | 74.13 | 77.41 | |
| UA-CLMModel=Llama-2-13B2024.12 | 67.01 | 72.55 | 73.63 | 76.94 | |
| ULTModel=Llama-2-13B2024.12 | 63.23 | 65.23 | 68.83 | 71.58 | |
| CLMModel=Llama-2-13B2024.12 | 63.02 | 63.48 | 68.15 | 69.1 | |
| CLMModel=Llama-2-7B2024.12 | 62.52 | 63.2 | 66.35 | 68.89 | |
| CTModel=Llama-2-7B2024.12 | 61.75 | 65.71 | 67.06 | 72.92 | |
| Pre-trainedModel=Llama-2-13B2024.12 | 60.27 | 64.04 | 66.79 | 71.11 | |
| Pre-trainedModel=Llama-2-7B2024.12 | 58.13 | 63.24 | 66.86 | 74.67 | |
| ULTModel=Llama-2-7B2024.12 | 57.9 | 59.15 | 67.93 | 64.95 | |
| CTModel=Llama-2-13B2024.12 | 52.99 | 55.99 | 60.72 | 69.58 |