Downstream Datasets
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Image Classification | Downstream Datasets Average | Average Accuracy85.4 | 68 | |
| Classification | 15 Downstream Datasets | Food-10190.7 | 24 | |
| Image Classification | 12 Downstream Datasets Fine-grained, Natural, Specialized (test) | Accuracy (Fine-grained)72.07 | 21 | |
| Downstream Task Evaluation | Multiple Downstream Datasets (LAMBADA, ARC, WinoGrande, PIQA, HellaSwag, SciQ, RACE) | LAMBADA (OpenAI)45 | 12 | |
| Few-shot Classification | 12 Downstream Datasets (ChestX, CropDisease, Deep Weeds, DTD, EuroSAT, Flowers 102, Kaokore, Omniglot, Resisc45, Sketch, SVHN, ISIC) | ChestX Accuracy (FS)32.57 | 8 | |
| Base-to-New Classification | 11 downstream datasets Highly Imbalanced, τ=0.06 | IN. Score71.58 | 6 | |
| Base-to-New Classification | 11 downstream datasets Balanced, τ=1 | IN Accuracy73.92 | 6 |
Showing 7 of 7 rows