Promptsy

Hallucination Severity Rating

U

@

·

Rate hallucination severity using calibrated examples.

22 copies0 forks
Rate hallucination severity in {{model}} responses using {{evaluation_dataset}}.

Severity calibration:
- Critical: {{critical_example}} (completely fabricated facts)
- Moderate: {{moderate_example}} (exaggerated claims)
- Minor: {{minor_example}} (slight inaccuracies)

Rate all hallucinations and calculate severity distribution.

Details

Category

Analysis

Use Cases

Hallucination ratingSeverity assessmentQuality scoring

Works Best With

claude-opus-4.5gpt-5.2gemini-2.0-flash
Created Shared

Create your own prompt vault and start sharing