# ImageNet-A

ImageNet-A is an AI benchmark; this page records what it measures and traces its data, grader, and scores to primary sources.

Tests image classifiers on naturally occurring, unmodified images selected because standard ImageNet models misclassify them.

## Details

- Address: Online, Global
- Website: [github.com/hendrycks/natural-adv-examples](<https://github.com/hendrycks/natural-adv-examples>)
- Tags: Science, Helpful
- Category: Multimodal
- Status: Active
- Data licence: MIT
- Data access: open
- Grader: exact
- Organization: UC Berkeley
- Released: 2019-07
- Paper: [arxiv.org/abs/1907.07174](<https://arxiv.org/abs/1907.07174>)
- Repository: [github.com/hendrycks/natural-adv-examples](<https://github.com/hendrycks/natural-adv-examples>)
- Top-score provenance: unknown
- ABI family: No

## Questions

### Where is ImageNet-A?

ImageNet-A is at Online, Global.

### How do I contact ImageNet-A?

To contact ImageNet-A, visit https://github.com/hendrycks/natural-adv-examples.

## Source

From [Benchmark primary sources](https://aibenchmarkindex.com/about), license CC0-1.0, checked 2026-10-15.

Last updated October 15, 2026.

## Links

- This benchmark on the [web page](<https://aibenchmarkindex.com/global/online/imagenet-a>)
- [Multimodal in Online](<https://aibenchmarkindex.com/global/online/category/multimodal.md>)
- [More benchmarks in Online](<https://aibenchmarkindex.com/global/online.md>)

> **Check before you compare.** Benchmark methods and leaderboards change. Follow the cited primary source before using a result in a decision.