| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| MS COCO | Harmonica-3rd | Interpretation Error Rate1.83 | 40 | 3mo ago | |
| ImageNet Inception-v3 | EDDP-C | Coverage93 | 12 | 2mo ago | |
| Stanford Cars | AMP | Consistency50.2 | 10 | 2mo ago | |
| CUB-200 2011 | AMP | Consistency Score76.8 | 10 | 2mo ago | |
| Benzene | GCBM/GCBM-E | AUC83.6 | 7 | 3mo ago | |
| Solubility | GCBM/GCBM-E | AUC91.2 | 7 | 3mo ago | |
| Top-activated Texts | CLIP+SAE | Embedding Similarity0.74 | 4 | 1mo ago | |
| Top-activated Images | CLIP+SAE | EmbSim0.17 | 4 | 1mo ago | |
| CUB-200-C 100-sample | ProtoTTA | Focus Relevance4.3 | 4 | 1mo ago | |
| ImageNet | LaViSE | Top-1 Precision46 | 4 | 3mo ago | |
| MAMI (test) | PrismAgent | Faithfulness7.61 | 2 | 28d ago | |
| FHM (test) | PrismAgent | Faithfulness7.56 | 2 | 28d ago | |
| HarM (test) | PrismAgent | Faithfulness7.18 | 2 | 28d ago | |
| BipedalWalker | Interpretability Score3.2 | 2 | 3mo ago | ||
| Hopper | Interpretability Score3.1 | 2 | 3mo ago | ||
| LunarLander | Interpretability Score4 | 2 | 3mo ago | ||
| InvPendSwingup | ESPL | Interpretability Score4.3 | 2 | 3mo ago | |
| InvDoublePend | Interpretability Score5 | 2 | 3mo ago | ||
| Pendulum | ESPL | Interpretability Score4.5 | 2 | 3mo ago | |
| MountainCar | ESPL | Interpretability Score5 | 2 | 3mo ago | |
| CartPole | ESPL | Interpretability Score5 | 2 | 3mo ago | |
| Places365 | LaViSE | Top-1 Precision74 | 2 | 3mo ago |