# Zero-shot Image Classification on ImageNet Robustness Variants and ObjectNet Source: [Zero-shot Image Classification on ImageNet Robustness Variants and ObjectNet]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:35:20.384Z Evaluation of zero-shot Image Classification performance across standard ImageNet, its robustness variants (v1, v2, A, R, S), and ObjectNet. 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 and Robustness Variants zero-shot v1, v2, A, R, S, ObjectNet | | Metric | Average Class Accuracy | | 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 - [Average Class Accuracy](): Higher is better; unit: PERCENT. - [ImageNet Accuracy](): Higher is better; unit: PERCENT. - [ImageNet-v2 Accuracy](): Higher is better; unit: PERCENT. - [ObjectNet Accuracy](): Higher is better; unit: PERCENT. - [ImageNet-A Accuracy](): Higher is better; unit: PERCENT. - [ImageNet-R Accuracy](): Higher is better; unit: PERCENT. - [ImageNet-S Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Average Class Accuracy | Result date | Report paper | | --- | --- | --- | --- | | ViT-L/14 Sparse | 75.6 | Not recorded | 697ae972a2b28e761d87cf4a | | ViT-L/14 baseline | 75.1 | Not recorded | 697ae972a2b28e761d87cf4a | | ViT-L/14 Sparse+ | 75.1 | Not recorded | 697ae972a2b28e761d87cf4a | | OpenAI ViT-B/32 | 57.5 | Not recorded | 697ae972a2b28e761d87cf4a | | Prisma ViT-B/32 cls@11 | 56.6 | Not recorded | 697ae972a2b28e761d87cf4a | ### ViT-L/14 Sparse Average Class Accuracy: 75.6. Result ID: 699784855698665ec4d975ac. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Sparsity":"0.66%","Evaluation":"Zero-shot"}. This record does not assert independent reproduction. ### ViT-L/14 baseline Average Class Accuracy: 75.1. Result ID: 699784855698665ec4d975ab. [Reporting paper]() Recorded evaluation configuration: {"Sparsity":"100%","Evaluation":"Zero-shot"}. This record does not assert independent reproduction. ### ViT-L/14 Sparse+ Average Class Accuracy: 75.1. Result ID: 699784855698665ec4d975ad. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Sparsity":"0.47%","Evaluation":"Zero-shot"}. This record does not assert independent reproduction. ### OpenAI ViT-B/32 Average Class Accuracy: 57.5. Result ID: 699784855698665ec4d975a9. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Sparsity":"100%","Evaluation":"Zero-shot"}. This record does not assert independent reproduction. ### Prisma ViT-B/32 cls@11 Average Class Accuracy: 56.6. Result ID: 699784855698665ec4d975aa. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Sparsity":"2.8%","Evaluation":"Zero-shot"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.