strongreject@parity
Safeguards
StrongReject benchmark for evaluating LLM safety and jailbreak resistance. Parity subset with 150 tasks (50 prompts * 3 jailbreaks).
Run this task
CLI:
inspect eval inspect_harbor/strongreject_parity --model openai/gpt-5Python:
from inspect_ai import eval
from inspect_harbor import strongreject_parity
eval(strongreject_parity(), model="openai/gpt-5")Dataset information
| Harbor registry | strongreject@parity |
| Inspect task | strongreject_parity |
| Version | parity |
| Samples | 150 |
| Paper | arxiv |
See Task Parameters for the parameter set shared across all Harbor tasks.