FORTRESS adversarial (static) ORS

FORTRESS adversarial (static) ORS is a benchmark for safe and responsible behaviour. The official description says: "FORTRESS adversarial robustness benchmark, static attack setup, Offense Rate Score (ORS); lower is better". Below we describe in simple words what benchmarks of this type generally examine.

FORTRESS adversarial (static) ORS measures safe and responsible behaviour. In detail, it checks:

• refusing harmful requests while still helping on benign ones
• resisting attempts to bypass safety rules
• avoiding dangerous, inaccurate or policy-violating output
• flagging uncertainty instead of confidently producing unsafe answers

The score is accuracy: the percentage of tasks answered or solved correctly. 100% means every task was solved perfectly; higher is better.

Official methodology note: FORTRESS adversarial robustness benchmark, static attack setup, Offense Rate Score (ORS); lower is better. Reported on the Muse Spark Open model card Safety section.

What is the FORTRESS adversarial (static) ORS benchmark?

FORTRESS adversarial (static) ORS is a benchmark for safe and responsible behaviour. The official description says: 'FORTRESS adversarial robustness benchmark, static attack setup, Offense Rate Score (ORS); lower is better'. Below we describe in simple words what benchmarks of this type generally examine.

What does FORTRESS adversarial (static) ORS test?

FORTRESS adversarial (static) ORS mainly measures safe and responsible behaviour. Concretely, it checks: refusing harmful requests while still helping on benign ones; resisting attempts to bypass safety rules; avoiding dangerous, inaccurate or policy-violating output.

How is the FORTRESS adversarial (static) ORS score calculated?

The score is accuracy: the percentage of tasks answered or solved correctly. 100% means every task was solved perfectly; higher is better. The declared maximum score of this benchmark is 100.0.

Where do I get FORTRESS adversarial (static) ORS data and scores?

The official benchmark page/reference is linked on this page. Every score on Open LLM Radar is also available as open data: download the scores JSON (http://open-llm-radar-api:8000/api/v1/benchmarks/fortress-adversarial-ors/scores) or pull it directly from the Open LLM Radar API with any HTTP client.