# Relative Performance Drop on VQA (Visual Question Answering)

Source: [Relative Performance Drop on VQA (Visual Question Answering)](<https://www.sota2.com/research/sota/relative-performance-drop-on-vqa-visual-question-answering>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-21T12:18:06.444Z

Evaluation of model performance on the VQA dataset for Visual Question Answering, focusing on the relative performance drop percentage, typically used for robustness testing.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Visual Question Answering |
| Dataset (including any stated split) | VQA |
| Metric | Relative Performance Drop |
| Unit | PERCENT |
| Direction | Lower is better |
| Results loaded | 7 of 7 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Relative Performance Drop](<https://www.sota2.com/research/sota/relative-performance-drop-on-vqa-visual-question-answering.md?metric=Relative%20Performance%20Drop>): Lower is better; unit: PERCENT.

## Recorded results

| Method | Relative Performance Drop | Result date | Report paper |
| --- | --- | --- | --- |
| RandNeuron | 0.95 | Not recorded | 69e704d47d0172140ec7efea |
| APE | 3.9 | Not recorded | 69e704d47d0172140ec7efea |
| MA | 5.35 | Not recorded | 69e704d47d0172140ec7efea |
| HONES-Gaussian | 7.1 | Not recorded | 69e704d47d0172140ec7efea |
| AP | 12.45 | Not recorded | 69e704d47d0172140ec7efea |
| HONES-RandHead | 13.25 | Not recorded | 69e704d47d0172140ec7efea |
| HONES | 24.15 | Not recorded | 69e704d47d0172140ec7efea |

### RandNeuron

Relative Performance Drop: 0.95. Result ID: 69e76af147f34ce924cef717.

[Reporting paper](<https://www.sota2.com/research/paper/69e704d47d0172140ec7efea.md>)

Recorded evaluation configuration: {"Backbone":"LLaVA-1.5-13B","Masking ratio":"top 1%"}. This record does not assert independent reproduction.

### APE

Relative Performance Drop: 3.9. Result ID: 69e76af147f34ce924cef714.

[Reporting paper](<https://www.sota2.com/research/paper/69e704d47d0172140ec7efea.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78a71e209131cea79a1a.md>)

Recorded evaluation configuration: {"Backbone":"LLaVA-1.5-13B","Masking ratio":"top 1%"}. This record does not assert independent reproduction.

### MA

Relative Performance Drop: 5.35. Result ID: 69e76af147f34ce924cef713.

[Reporting paper](<https://www.sota2.com/research/paper/69e704d47d0172140ec7efea.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa5f861e209131ce9eaa87.md>)

Recorded evaluation configuration: {"Backbone":"LLaVA-1.5-13B","Masking ratio":"top 1%"}. This record does not assert independent reproduction.

### HONES-Gaussian

Relative Performance Drop: 7.1. Result ID: 69e76af147f34ce924cef716.

[Reporting paper](<https://www.sota2.com/research/paper/69e704d47d0172140ec7efea.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69e704d47d0172140ec7efea.md>)

Recorded evaluation configuration: {"Backbone":"LLaVA-1.5-13B","Masking ratio":"top 1%"}. This record does not assert independent reproduction.

### AP

Relative Performance Drop: 12.45. Result ID: 69e76af147f34ce924cef712.

[Reporting paper](<https://www.sota2.com/research/paper/69e704d47d0172140ec7efea.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75161e209131cea65a2c.md>)

Recorded evaluation configuration: {"Backbone":"LLaVA-1.5-13B","Masking ratio":"top 1%"}. This record does not assert independent reproduction.

### HONES-RandHead

Relative Performance Drop: 13.25. Result ID: 69e76af147f34ce924cef715.

[Reporting paper](<https://www.sota2.com/research/paper/69e704d47d0172140ec7efea.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69e704d47d0172140ec7efea.md>)

Recorded evaluation configuration: {"Backbone":"LLaVA-1.5-13B","Masking ratio":"top 1%"}. This record does not assert independent reproduction.

### HONES

Relative Performance Drop: 24.15. Result ID: 69e76af147f34ce924cef718.

[Reporting paper](<https://www.sota2.com/research/paper/69e704d47d0172140ec7efea.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69e704d47d0172140ec7efea.md>)

Recorded evaluation configuration: {"Backbone":"LLaVA-1.5-13B","Masking ratio":"top 1%"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.