Visual Question Answering on VQA v2 (test-dev) (Robustness)
71.1Accuracy (Noise)Latent Denoising
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Latent DenoisingBackbone=LLaVA+CLIP, Corruption Severity=52026.04 | 71.1 | 69.3 | 73.3 | 73 | |
| BaselineBackbone=LLaVA+CLIP, Corruption Severity=52026.04 | 69.4 | 67.8 | 71.4 | 70.9 |