# Image Classification on ImageNet (Clean and AutoAttack)

Source: [Image Classification on ImageNet (Clean and AutoAttack)](<https://www.sota2.com/research/sota/image-classification-on-imagenet-clean-and-autoattack>)

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

Data updated: 2026-02-27T11:46:08.831Z

Evaluation of Image Classification performance on the ImageNet dataset, specifically reporting robustness metrics including Clean accuracy and AutoAttack results.

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 | Image Classification |
| Dataset (including any stated split) | ImageNet |
| Metric | Accuracy (Clean) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| 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

- [Accuracy (Clean)](<https://www.sota2.com/research/sota/image-classification-on-imagenet-clean-and-autoattack.md?metric=Accuracy%20(Clean)>): Higher is better; unit: PERCENT.

- [Accuracy (AutoAttack)](<https://www.sota2.com/research/sota/image-classification-on-imagenet-clean-and-autoattack.md?metric=Accuracy%20(AutoAttack)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy (Clean) | Result date | Report paper |
| --- | --- | --- | --- |
| CLIP | 59.15 | Not recorded | 6940e7a8456146936284f897 |
| UCAT | 43.2 | Not recorded | 6940e7a8456146936284f897 |
| ACAT | 28.6 | Not recorded | 6940e7a8456146936284f897 |
| DKL | 27.18 | Not recorded | 6940e7a8456146936284f897 |
| TRADES | 26.31 | Not recorded | 6940e7a8456146936284f897 |

### CLIP

Accuracy (Clean): 59.15. Result ID: 69a183f135dfcabb00e7887f.

[Reporting paper](<https://www.sota2.com/research/paper/6940e7a8456146936284f897.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69581e209131cea21ad0.md>)

[Implementation](<https://github.com/OpenAI/CLIP>)

Recorded evaluation configuration: {"Fine-tuning dataset":"TinyImageNet","Adversarial training":"2-step PGD","Epsilon":"1/255","Evaluation protocol":"Zero-shot"}. This record does not assert independent reproduction.

### UCAT

Accuracy (Clean): 43.2. Result ID: 69a183f135dfcabb00e78883.

[Reporting paper](<https://www.sota2.com/research/paper/6940e7a8456146936284f897.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6940e7a8456146936284f897.md>)

Recorded evaluation configuration: {"Fine-tuning dataset":"TinyImageNet","Adversarial training":"2-step PGD","Epsilon":"1/255","Evaluation protocol":"Zero-shot"}. This record does not assert independent reproduction.

### ACAT

Accuracy (Clean): 28.6. Result ID: 69a183f135dfcabb00e78881.

[Reporting paper](<https://www.sota2.com/research/paper/6940e7a8456146936284f897.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6dc61e209131cea3c0ff.md>)

[Implementation](<https://github.com/val-iisc/DAJAT>)

Recorded evaluation configuration: {"Fine-tuning dataset":"TinyImageNet","Adversarial training":"2-step PGD","Epsilon":"1/255","Evaluation protocol":"Zero-shot"}. This record does not assert independent reproduction.

### DKL

Accuracy (Clean): 27.18. Result ID: 69a183f135dfcabb00e78882.

[Reporting paper](<https://www.sota2.com/research/paper/6940e7a8456146936284f897.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa756d1e209131cea67d56.md>)

[Implementation](<https://github.com/jiequancui/DKL>)

Recorded evaluation configuration: {"Fine-tuning dataset":"TinyImageNet","Adversarial training":"2-step PGD","Epsilon":"1/255","Evaluation protocol":"Zero-shot"}. This record does not assert independent reproduction.

### TRADES

Accuracy (Clean): 26.31. Result ID: 69a183f135dfcabb00e78880.

[Reporting paper](<https://www.sota2.com/research/paper/6940e7a8456146936284f897.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65f71e209131cea0d897.md>)

Recorded evaluation configuration: {"Fine-tuning dataset":"TinyImageNet","Adversarial training":"2-step PGD","Epsilon":"1/255","Evaluation protocol":"Zero-shot"}. This record does not assert independent reproduction.

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