microsoft/CoSApien
CoSApien: A Human-Authored Safety Control Benchmark Overview Paper: Controllable Safety Alignment: Inference-Time Adaptation to Diverse Safety Requirements, published at ICLR 2025. Purpose: Evaluate the controllability of large language models (LLMs) aligned through natural language safety configs, ensuring both helpfulness and adherence to specified safety requirements. Description: CoSApien is a human-authored benchmark comprising real-world scenarios where… See the full description on the dataset page: https://huggingface.co/datasets/microsoft/CoSApien.
| Author / organization | microsoft |
|---|---|
| License (as declared) | cdla-permissive-2.0 |
| Downloads | 384 (as reported by Hugging Face) |
| Likes | 3 |
| Created | 18 April 2025 |
| Last updated at source | 1 August 2025 |
Popularity, last 2 days
384 now. Counts as reported by the source, snapshotted daily by Silytics.
Recent changes
- 17 Oct 2026 Description: CoSApien: A Human-Authored Safety Control Benchmark Overview Paper: Controllable Safety Alignment: Inference-Time Adaptation to Diverse Safety Requirements, published at ICLR 2025. Purpose: Evaluate the controllability of large language mod → CoSApien: A Human-Authored Safety Control Benchmark Overview Paper: Controllable Safety Alignment: Inference-Time Adaptation to Diverse Safety Requirements, published at ICLR 2025. Purpose: Evaluate the controllability of large language mod
Changes detected by Silytics between refreshes of the source.
More from microsoft
- microsoft/ms_marcoDataset
- microsoft/NOTSOFARDataset
- microsoft/orca-math-word-problems-200kDataset
- microsoft/CABRADataset
- microsoft/rStar-CoderDataset
- microsoft/wiki_qaDataset
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.