Safety
SimpleSafetyTests
SimpleSafetyTests checks whether a model responds safely to blunt requests touching severe harms such as self harm, violence and illegal activity, with a classifier marking each reply safe or unsafe. Recorded scores retain intermediate judge ratings; 1 indicates the best rubric score and 0 the worst.
100items
81subjects
unknownlicense
safetydomain
textmodality
item-level responses released
Saturation status: Yes
Response matrix
Responses come from Stanford CRFM's HELM Safety leaderboard (v1.17.0), run without a system prompt; the original evaluation of Vidgen et al. tested each model under two system-prompt conditions.
Loading response matrix…
lowhighUnobserved
Scale: Score in [0, 1]