Question Answering on PubMedQA (AR%, Risk%)
100AR (%)Fixed-Threshold
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Fixed-Thresholdalpha=0.202026.05 | 100 | 23.9 | |
| Always-Actalpha=0.202026.05 | 100 | 23.9 | |
| ACIalpha=0.202026.05 | 84.9 | 21.8 | |
| NEX-Confalpha=0.202026.05 | 81.9 | 18.8 | |
| SAOCPalpha=0.202026.05 | 81.2 | 22.4 | |
| CSAalpha=0.202026.05 | 63.5 | 15.1 | |
| ConfFactalpha=0.202026.05 | 34.8 | 7.6 | |
| Naive-Tuningalpha=0.202026.05 | 1.6 | 24.9 |