# Zero-shot Image Classification on Food101 (PGD Robustness)

Source: [Zero-shot Image Classification on Food101 (PGD Robustness)](<https://www.sota2.com/research/sota/zero-shot-image-classification-on-food101-pgd-robustness>)

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

Data updated: 2026-05-19T10:58:26.712Z

Evaluation of Zero-shot Image Classification performance on the Food101 dataset, specifically reporting clean accuracy and robustness against PGD-100 attacks with epsilon 1/255.

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 | Zero-shot Image Classification |
| Dataset (including any stated split) | Food101 |
| Metric | Accuracy (Robust, PGD-100, eps=1/255) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 9 of 9 |
| 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 (Robust, PGD-100, eps=1/255)](<https://www.sota2.com/research/sota/zero-shot-image-classification-on-food101-pgd-robustness.md?metric=Accuracy%20(Robust%2C%20PGD-100%2C%20eps%3D1%2F255)>): Higher is better; unit: PERCENT.

- [Accuracy (Clean)](<https://www.sota2.com/research/sota/zero-shot-image-classification-on-food101-pgd-robustness.md?metric=Accuracy%20(Clean)>): Higher is better; unit: PERCENT.

- [Accuracy (Robust, CW)](<https://www.sota2.com/research/sota/zero-shot-image-classification-on-food101-pgd-robustness.md?metric=Accuracy%20(Robust%2C%20CW)>): Higher is better; unit: PERCENT.

- [Accuracy (DI)](<https://www.sota2.com/research/sota/zero-shot-image-classification-on-food101-pgd-robustness.md?metric=Accuracy%20(DI)>): Higher is better; unit: PERCENT.

- [Accuracy (AA)](<https://www.sota2.com/research/sota/zero-shot-image-classification-on-food101-pgd-robustness.md?metric=Accuracy%20(AA)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy (Robust, PGD-100, eps=1/255) | Result date | Report paper |
| --- | --- | --- | --- |
| TAME | 60.8 | Not recorded | 6a0beee6dbc836b87d82c7fb |
| SAFT-L | 26.19 | Not recorded | 6992a4ad9ebc499c97406d32 |
| FARE | 25.56 | Not recorded | 6992a4ad9ebc499c97406d32 |
| TeCoA | 25.19 | Not recorded | 6992a4ad9ebc499c97406d32 |
| PMG-AFT | 24.44 | Not recorded | 6992a4ad9ebc499c97406d32 |
| TGA-ZSR | 24.41 | Not recorded | 6992a4ad9ebc499c97406d32 |
| SAFT-M | 24.13 | Not recorded | 6992a4ad9ebc499c97406d32 |
| Vanilla | 3.6 | Not recorded | 6a0beee6dbc836b87d82c7fb |
| CLIP | 3.45 | Not recorded | 6992a4ad9ebc499c97406d32 |

### TAME

Accuracy (Robust, PGD-100, eps=1/255): 60.8. Result ID: 6a0c4251fbc373592c032bd5.

[Reporting paper](<https://www.sota2.com/research/paper/6a0beee6dbc836b87d82c7fb.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0beee6dbc836b87d82c7fb.md>)

Recorded evaluation configuration: {"Backbone":"ViT-B/32","Evaluation Protocol":"Zero-shot","Prompt Configuration":"V-L Independent"}. This record does not assert independent reproduction.

### SAFT-L

Accuracy (Robust, PGD-100, eps=1/255): 26.19. Result ID: 69a1992435dfcabb00e7dc61.

[Reporting paper](<https://www.sota2.com/research/paper/6992a4ad9ebc499c97406d32.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6992a4ad9ebc499c97406d32.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### FARE

Accuracy (Robust, PGD-100, eps=1/255): 25.56. Result ID: 69a1992435dfcabb00e7dc5e.

[Reporting paper](<https://www.sota2.com/research/paper/6992a4ad9ebc499c97406d32.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa738c1e209131cea5d24a.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### TeCoA

Accuracy (Robust, PGD-100, eps=1/255): 25.19. Result ID: 69a1992435dfcabb00e7dc5d.

[Reporting paper](<https://www.sota2.com/research/paper/6992a4ad9ebc499c97406d32.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f271e209131cea44177.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### PMG-AFT

Accuracy (Robust, PGD-100, eps=1/255): 24.44. Result ID: 69a1992435dfcabb00e7dc5f.

[Reporting paper](<https://www.sota2.com/research/paper/6992a4ad9ebc499c97406d32.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72df1e209131cea5912e.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### TGA-ZSR

Accuracy (Robust, PGD-100, eps=1/255): 24.41. Result ID: 69a1992435dfcabb00e7dc60.

[Reporting paper](<https://www.sota2.com/research/paper/6992a4ad9ebc499c97406d32.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75771e209131cea68256.md>)

[Implementation](<https://github.com/zhyblue424/TGA-ZSR>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### SAFT-M

Accuracy (Robust, PGD-100, eps=1/255): 24.13. Result ID: 69a1992435dfcabb00e7dc62.

[Reporting paper](<https://www.sota2.com/research/paper/6992a4ad9ebc499c97406d32.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6992a4ad9ebc499c97406d32.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Vanilla

Accuracy (Robust, PGD-100, eps=1/255): 3.6. Result ID: 6a0c4251fbc373592c032bd4.

[Reporting paper](<https://www.sota2.com/research/paper/6a0beee6dbc836b87d82c7fb.md>)

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

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

Recorded evaluation configuration: {"Backbone":"ViT-B/32","Evaluation Protocol":"Zero-shot"}. This record does not assert independent reproduction.

### CLIP

Accuracy (Robust, PGD-100, eps=1/255): 3.45. Result ID: 69a1992435dfcabb00e7dc5c.

[Reporting paper](<https://www.sota2.com/research/paper/6992a4ad9ebc499c97406d32.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa727a1e209131cea56a1e.md>)

[Implementation](<https://github.com/openai/evals>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

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