Faithfulness evaluation on MDACE (test)
87Comp ScoreAttInGrad
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AttInGradModel=IGR2024.06 | 87 | 17 | |
| AttInGradModel=BU2024.06 | 86 | 22 | |
| AttInGradModel=BS2024.06 | 86 | 34 | |
| AttInGradModel=TM2024.06 | 86 | 18 | |
| AttInGradModel=PGD2024.06 | 86 | 21 | |
| AttentionModel=IGR2024.06 | 85 | 23 | |
| AttentionModel=BS2024.06 | 84 | 42 | |
| AttentionModel=PGD2024.06 | 84 | 28 | |
| AttentionModel=BU2024.06 | 82 | 29 | |
| a∇aModel=IGR2024.06 | 82 | 23 | |
| AttentionModel=TM2024.06 | 82 | 24 | |
| a∇aModel=PGD2024.06 | 81 | 28 | |
| a∇aModel=BU2024.06 | 77 | 28 | |
| a∇aModel=BS2024.06 | 77 | 41 | |
| InputXGradModel=IGR2024.06 | 77 | 14 | |
| DeepliftModel=IGR2024.06 | 77 | 14 | |
| DeepliftModel=TM2024.06 | 77 | 15 | |
| a∇aModel=TM2024.06 | 77 | 24 | |
| InputXGradModel=PGD2024.06 | 77 | 17 | |
| DeepliftModel=PGD2024.06 | 77 | 17 | |
| DeepliftModel=BU2024.06 | 76 | 18 | |
| InputXGradModel=BS2024.06 | 76 | 26 | |
| DeepliftModel=BS2024.06 | 76 | 26 | |
| InputXGradModel=TM2024.06 | 76 | 15 | |
| InputXGradModel=BU2024.06 | 75 | 17 | |
| IGModel=IGR2024.06 | 75 | 14 | |
| IGModel=PGD2024.06 | 75 | 18 | |
| IGModel=BU2024.06 | 74 | 19 | |
| IGModel=BS2024.06 | 74 | 28 | |
| IGModel=TM2024.06 | 74 | 15 | |
| RolloutModel=PGD2024.06 | 40 | 41 | |
| RolloutModel=BU2024.06 | 39 | 47 | |
| RolloutModel=BS2024.06 | 39 | 47 | |
| RolloutModel=IGR2024.06 | 39 | 41 | |
| RolloutModel=TM2024.06 | 37 | 40 | |
| RandModel=IGR2024.06 | 4 | 95 | |
| RandModel=PGD2024.06 | 4 | 93 | |
| RandModel=BU2024.06 | 3 | 92 | |
| RandModel=BS2024.06 | 2 | 93 | |
| RandModel=TM2024.06 | 2 | 92 |