Dataset · Hugging Face

giskardai/harmbench-scenarios

Name: harmbench-scenarios
Creator: giskardai
License: mit
Keywords: source_datasets:walledai/HarmBench, language:en, license:mit, size_categories:n<1K, format:json, modality:text, library:datasets, library:pandas, library:polars, library:mlcroissant, arxiv:2402.04249, region:us, safety, red-teaming, llm-evaluation

HarmBench Scenarios Safety-evaluation scenarios derived from the HarmBench behavior dataset, serialized as giskard Scenario objects (one JSON object per line). Each row poses a harmful request the agent should refuse or safely decline, paired with an LLMJudge check (giskard.scan::judges/harmbench_safety.j2) that grades the agent response. Format Every line is a serialized giskard.checks.Scenario: name — "HarmBench #<id>" steps[].interacts[].inputs — the harmful… See the full description on the dataset page: https://huggingface.co/datasets/giskardai/harmbench-scenarios.

View on Hugging FaceReport an issue

Author / organization	giskardai
License (as declared)	mit
Downloads	282 (as reported by Hugging Face)
Likes	0
Created	24 June 2026
Last updated at source	24 June 2026

safetyred-teamingllm-evaluationrefusalharmbench

Embed this entry

<iframe src="https://silytics.si/embed/item/8ed42811-fa52-4744-8ad3-7acae5564a7a" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Free, no tracking. More about feeds and embeds.

Popularity, last 4 days

282 now (+12 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.

More from giskardai

giskardai/do-not-answer-scenariosDataset
giskardai/StereoTalesDataset
giskardai/phareDataset

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 19 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.