Uncertainty Estimation on WebQA
73.57AUROCSE + MARS
Evaluation Results
| Method | Links | |
|---|---|---|
| SE + MARSBackbone=Falcon-7b2024.02 | 73.57 | |
| SEBackbone=Falcon-7b2024.02 | 73.16 | |
| SE + MARSBackbone=Llama2-7b2024.02 | 69.88 | |
| SE + MARSBackbone=Mistral-7b2024.02 | 69.86 | |
| SEBackbone=Mistral-7b2024.02 | 69.51 | |
| SEBackbone=Llama2-7b2024.02 | 69.44 | |
| Entropy + MARSBackbone=Falcon-7b2024.02 | 68.44 | |
| Confidence + MARSBackbone=Falcon-7b2024.02 | 68.26 | |
| SE + MARSBackbone=Llama2-13b2024.02 | 67.75 | |
| SEBackbone=Llama2-13b2024.02 | 67.31 | |
| SE + MARSBackbone=Llama2-7b-chat2024.02 | 67.27 | |
| Confidence + MARSBackbone=Mistral-7b2024.02 | 67.16 | |
| SEBackbone=Llama2-7b-chat2024.02 | 67.11 | |
| ConfidenceBackbone=Falcon-7b2024.02 | 66.56 | |
| Confidence + MARSBackbone=Llama2-7b2024.02 | 66.04 | |
| EntropyBackbone=Falcon-7b2024.02 | 65.98 | |
| Entropy + MARSBackbone=Llama2-7b2024.02 | 65.83 | |
| Entropy + MARSBackbone=Mistral-7b2024.02 | 65.76 | |
| ConfidenceBackbone=Mistral-7b2024.02 | 65.66 | |
| ConfidenceBackbone=Llama2-7b2024.02 | 64.76 | |
| Entropy + MARSBackbone=Llama2-7b-chat2024.02 | 64.69 | |
| Confidence + MARSBackbone=Llama2-7b-chat2024.02 | 64.48 | |
| Confidence + MARSBackbone=Llama2-13b2024.02 | 64.23 | |
| EntropyBackbone=Mistral-7b2024.02 | 64.15 | |
| ConfidenceBackbone=Llama2-7b-chat2024.02 | 64.06 | |
| EntropyBackbone=Llama2-7b2024.02 | 64.04 | |
| Entropy + MARSBackbone=Llama2-13b2024.02 | 64.02 | |
| EntropyBackbone=Llama2-7b-chat2024.02 | 63.82 | |
| ConfidenceBackbone=Llama2-13b2024.02 | 62.6 | |
| EntropyBackbone=Llama2-13b2024.02 | 62.11 |