Question Answering on BBH n=2395 (test)
100Activation ScoreCRC
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CRCbeta=0.15, T=42026.06 | 100 | 91.6 | 8.4 | |
| Selective Predictionbeta=0.15, T=42026.06 | 99.9 | 91.7 | 8.2 | |
| Calibrated Learnedbeta=0.15, T=42026.06 | 99.7 | 91.4 | 8.6 | |
| Isotonic Confidencebeta=0.15, T=42026.06 | 99.7 | 91.7 | 8.3 | |
| Learned Stopperbeta=0.15, T=42026.06 | 99.4 | 91.7 | 8.3 | |
| Consensusbeta=0.15, T=42026.06 | 98.2 | 91.5 | 8.5 | |
| Oraclebeta=0.15, T=42026.06 | 92.6 | 100 | 0 | |
| Confidence Thresholdbeta=0.15, T=42026.06 | 87.1 | 91.5 | 8.5 | |
| kNN no biasbeta=0.15, T=42026.06 | 79.1 | 97.9 | 1.6 | |
| Adaptive Local k-NN Certification (Ours)beta=0.15, T=42026.06 | 74.9 | 98.5 | 1.1 | |
| Final-roundbeta=0.15, T=42026.06 | 74.7 | 98.3 | 1.3 |