| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Image Classification | 26 public benchmarks EVA-CLIP protocol pre-training DataComp-1B (test val) | Accuracy (ImageNet-1k)65.9 | 3 | |
| Image Classification | 26 Public Benchmarks | Average Top-1 Accuracy63 | 3 | |
| Single Image Reflection Removal | Public Benchmarks (test) | Success Rate90.5 | 3 | |
| Open-vocabulary video instance segmentation | Public Benchmarks (test val) | Metric- | 0 |