Benchmark

AttaQ

AttaQ is a unique dataset containing adversarial examples in the form of questions designed to provoke harmful or inappropriate responses from large…

Modality
text
Categories
safety
Openness
unknown
Source
llm_stats
Reported scores
3

Reported scores

llm_stats

ModelOrganizationReported valueReportedEvidence
Granite 3.3 8B BaseIBM0.8852025-04-16evidence
Granite 3.3 8B InstructIBM0.8852025-04-16evidence
IBM Granite 4.0 Tiny PreviewIBM0.8612025-05-02evidence

Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.