# Image Classification on CIFAR-10 (test) (Clean and Robust Accuracy) Source: [Image Classification on CIFAR-10 (test) (Clean and Robust Accuracy)]() Collection index: [llms.txt]() Data updated: 2026-05-12T08:40:40.169Z Evaluation of standard Image Classification accuracy and a general robustness metric on the CIFAR-10 test set. 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) | CIFAR-10 (test) | | Metric | Clean Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 12 of 12 | | 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](): Higher is better; unit: PERCENT. - [Robust Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Clean Accuracy | Result date | Report paper | | --- | --- | --- | --- | | Default | 94.78 | Not recorded | 6a02b429dbc836b87d824739 | | Score-Opt-N | 93.94 | Not recorded | 6a02b429dbc836b87d824739 | | TARO-3-AA | 93.59 | Not recorded | 6a02b429dbc836b87d824739 | | Wang et al. | 93.5 | Not recorded | 6a02b429dbc836b87d824739 | | TARO-2-AA | 92.88 | Not recorded | 6a02b429dbc836b87d824739 | | AAOpt | 92.1 | Not recorded | 6a02b429dbc836b87d824739 | | ADBM | 91.93 | Not recorded | 6a02b429dbc836b87d824739 | | TARO-2 | 90.62 | Not recorded | 6a02b429dbc836b87d824739 | | Score-Opt-O | 90.23 | Not recorded | 6a02b429dbc836b87d824739 | | TARO-3 | 89.4 | Not recorded | 6a02b429dbc836b87d824739 | | Nie et al. | 89.02 | Not recorded | 6a02b429dbc836b87d824739 | | Yoon et al. | 86.14 | Not recorded | 6a02b429dbc836b87d824739 | ### Default Clean Accuracy: 94.78. Result ID: 6a02e61b47f34ce924e5b018. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"WRN-28-10","Attack Type":"BPDA+EOT","Perturbation Type":"l\_infinity","Perturbation Budget (epsilon)":"8/255"}. This record does not assert independent reproduction. ### Score-Opt-N Clean Accuracy: 93.94. Result ID: 6a02e61b47f34ce924e5b01d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"WRN-28-10","Attack Type":"BPDA+EOT","Perturbation Type":"l\_infinity","Perturbation Budget (epsilon)":"8/255"}. This record does not assert independent reproduction. ### TARO-3-AA Clean Accuracy: 93.59. Result ID: 6a02e61b47f34ce924e5b023. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"WRN-28-10","Attack Type":"BPDA+EOT","Perturbation Type":"l\_infinity","Perturbation Budget (epsilon)":"8/255"}. This record does not assert independent reproduction. ### Wang et al. Clean Accuracy: 93.5. Result ID: 6a02e61b47f34ce924e5b019. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"WRN-28-10","Attack Type":"BPDA+EOT","Perturbation Type":"l\_infinity","Perturbation Budget (epsilon)":"8/255"}. This record does not assert independent reproduction. ### TARO-2-AA Clean Accuracy: 92.88. Result ID: 6a02e61b47f34ce924e5b022. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"WRN-28-10","Attack Type":"BPDA+EOT","Perturbation Type":"l\_infinity","Perturbation Budget (epsilon)":"8/255"}. This record does not assert independent reproduction. ### AAOpt Clean Accuracy: 92.1. Result ID: 6a02e61b47f34ce924e5b01f. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"WRN-28-10","Attack Type":"BPDA+EOT","Perturbation Type":"l\_infinity","Perturbation Budget (epsilon)":"8/255"}. This record does not assert independent reproduction. ### ADBM Clean Accuracy: 91.93. Result ID: 6a02e61b47f34ce924e5b01e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"WRN-28-10","Attack Type":"BPDA+EOT","Perturbation Type":"l\_infinity","Perturbation Budget (epsilon)":"8/255"}. This record does not assert independent reproduction. ### TARO-2 Clean Accuracy: 90.62. Result ID: 6a02e61b47f34ce924e5b020. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"WRN-28-10","Attack Type":"BPDA+EOT","Perturbation Type":"l\_infinity","Perturbation Budget (epsilon)":"8/255"}. This record does not assert independent reproduction. ### Score-Opt-O Clean Accuracy: 90.23. Result ID: 6a02e61b47f34ce924e5b01c. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"WRN-28-10","Attack Type":"BPDA+EOT","Perturbation Type":"l\_infinity","Perturbation Budget (epsilon)":"8/255"}. This record does not assert independent reproduction. ### TARO-3 Clean Accuracy: 89.4. Result ID: 6a02e61b47f34ce924e5b021. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"WRN-28-10","Attack Type":"BPDA+EOT","Perturbation Type":"l\_infinity","Perturbation Budget (epsilon)":"8/255"}. This record does not assert independent reproduction. ### Nie et al. Clean Accuracy: 89.02. Result ID: 6a02e61b47f34ce924e5b01b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"WRN-28-10","Attack Type":"BPDA+EOT","Perturbation Type":"l\_infinity","Perturbation Budget (epsilon)":"8/255"}. This record does not assert independent reproduction. ### Yoon et al. Clean Accuracy: 86.14. Result ID: 6a02e61b47f34ce924e5b01a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"WRN-28-10","Attack Type":"BPDA+EOT","Perturbation Type":"l\_infinity","Perturbation Budget (epsilon)":"8/255"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.