# Image Classification on ImageNet16-120 (test) (Robustness Metrics)

Source: [Image Classification on ImageNet16-120 (test) (Robustness Metrics)](<https://www.sota2.com/research/sota/image-classification-on-imagenet16-120-test-robustness-metrics>)

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

Data updated: 2026-02-26T05:39:16.180Z

Evaluation of Image Classification performance on the test split of the ImageNet-16-120 dataset, reporting specific robustness metrics (Clean, CC., FGSM, HRS).

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) | ImageNet16-120 (test) |
| Metric | Clean Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 6 of 6 |
| 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

- [Clean Accuracy](<https://www.sota2.com/research/sota/image-classification-on-imagenet16-120-test-robustness-metrics.md?metric=Clean%20Accuracy>): Higher is better; unit: PERCENT.

- [CC Score](<https://www.sota2.com/research/sota/image-classification-on-imagenet16-120-test-robustness-metrics.md?metric=CC%20Score>): Higher is better; unit: PERCENT.

- [FGSM Accuracy](<https://www.sota2.com/research/sota/image-classification-on-imagenet16-120-test-robustness-metrics.md?metric=FGSM%20Accuracy>): Higher is better; unit: SCALAR.

- [HRS Robustness Score](<https://www.sota2.com/research/sota/image-classification-on-imagenet16-120-test-robustness-metrics.md?metric=HRS%20Robustness%20Score>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Clean Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| PC-DARTS | 50.58 | Not recorded | 68fa70e81e209131cea4de59 |
| DrNAS | 49.63 | Not recorded | 68fa70e81e209131cea4de59 |
| CRoZe | 47.9 | Not recorded | 68fa70e81e209131cea4de59 |
| SynFlow | 43.73 | Not recorded | 68fa70e81e209131cea4de59 |
| GradNorm | 39.13 | Not recorded | 68fa70e81e209131cea4de59 |
| AdvRush | 38.72 | Not recorded | 68fa70e81e209131cea4de59 |

### PC-DARTS

Clean Accuracy: 50.58. Result ID: 699160a03a6c31484907bd04.

[Reporting paper](<https://www.sota2.com/research/paper/68fa70e81e209131cea4de59.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa67511e209131cea1563d.md>)

[Implementation](<https://github.com/yuhuixu1993/PC-DARTS>)

Recorded evaluation configuration: {"Training-free NAS":false,"Params (M)":"5.30","#GPU":"8","Batch Size":"1024"}. This record does not assert independent reproduction.

### DrNAS

Clean Accuracy: 49.63. Result ID: 699160a03a6c31484907bd05.

[Reporting paper](<https://www.sota2.com/research/paper/68fa70e81e209131cea4de59.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69941e209131cea22d24.md>)

[Implementation](<https://github.com/xiangning-chen/DrNAS>)

Recorded evaluation configuration: {"Training-free NAS":false,"Params (M)":"5.70","#GPU":"8","Batch Size":"512"}. This record does not assert independent reproduction.

### CRoZe

Clean Accuracy: 47.9. Result ID: 699160a03a6c31484907bd09.

[Reporting paper](<https://www.sota2.com/research/paper/68fa70e81e209131cea4de59.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70e81e209131cea4de59.md>)

[Implementation](<https://github.com/HyeonjeongHa/CRoZe>)

Recorded evaluation configuration: {"Training-free NAS":true,"Params (M)":"5.87","#GPU":"1","Batch Size":"8"}. This record does not assert independent reproduction.

### SynFlow

Clean Accuracy: 43.73. Result ID: 699160a03a6c31484907bd08.

[Reporting paper](<https://www.sota2.com/research/paper/68fa70e81e209131cea4de59.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69961e209131cea22e24.md>)

Recorded evaluation configuration: {"Training-free NAS":true,"Params (M)":"6.13","#GPU":"1","Batch Size":"8"}. This record does not assert independent reproduction.

### GradNorm

Clean Accuracy: 39.13. Result ID: 699160a03a6c31484907bd07.

[Reporting paper](<https://www.sota2.com/research/paper/68fa70e81e209131cea4de59.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69961e209131cea22e24.md>)

Recorded evaluation configuration: {"Training-free NAS":true,"Params (M)":"5.90","#GPU":"1","Batch Size":"8"}. This record does not assert independent reproduction.

### AdvRush

Clean Accuracy: 38.72. Result ID: 699160a03a6c31484907bd06.

[Reporting paper](<https://www.sota2.com/research/paper/68fa70e81e209131cea4de59.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a931e209131cea2905e.md>)

Recorded evaluation configuration: {"Training-free NAS":false,"Params (M)":"4.20","#GPU":"1","Batch Size":"64"}. This record does not assert independent reproduction.

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