icaro-lab/ahb
IMPORTANT: Please help protect the integrity of this benchmark by not publicly re-uploading or redistributing the dataset outside the canonical repository. Adversarial Humanities Benchmark π Website | π Paper | π€ Dataset Adversarial Humanities Benchmark (AHB) is a text-only benchmark for measuring whether language models maintain safety behavior under humanities-style adversarial reformulations. AHB applies humanities-style transformations to source prompts from theβ¦ See the full description on the dataset page: https://huggingface.co/datasets/icaro-lab/ahb.
| Author / organization | icaro-lab |
|---|---|
| License (as declared) | mit |
| Downloads | 140 (as reported by Hugging Face) |
| Likes | 0 |
| Created | 19 April 2026 |
| Last updated at source | 21 April 2026 |
Popularity, last 3 days
140 now (-2 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.
Connected in the knowledge graph
Recent changes
- 16 Oct 2026 Description: IMPORTANT: Please help protect the integrity of this benchmark by not publicly re-uploading or redistributing the dataset outside the canonical repository. Adversarial Humanities Benchmark π Website | π Paper | π€ Dataset Adversarial Huma β IMPORTANT: Please help protect the integrity of this benchmark by not publicly re-uploading or redistributing the dataset outside the canonical repository. Adversarial Humanities Benchmark π Website | π Paper | π€ Dataset Adversarial Huma
Changes detected by Silytics between refreshes of the source.
Related papers
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 18 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.