NegBench
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| MCQ-Neg | NegBench | Affirmed Accuracy84.55 | 25 | |
| Negative Retrieval | NegBench | Image-to-Text Recall@164.62 | 25 | |
| Vision-Language Understanding | NegBench VOC2007 | Accuracy95.37 | 11 | |
| Vision-Language Understanding | NegBench COCO | Accuracy93.19 | 11 | |
| Multiple Choice Visual Question Answering | NegBench VOC 2007 (test) | Accuracy95.37 | 11 | |
| Multiple Choice Visual Question Answering | NegBench COCO (test) | Accuracy93.19 | 11 | |
| Multiple Choice Question | NegBench COCO subset | Overall Accuracy32.55 | 4 |
Showing 7 of 7 rows