Safety
These metrics cover misuse, manipulation, toxicity, harm, bias, and related risks. Each page lists its shortname, fields, and an example payload (and optional metric_args when the metric supports them).
Metrics
The same pages appear under Safety in the docs sidebar:
- Bias (
bias) - Harmfulness (
harmfulness) - Manipulation (
manipulation) - Misuse (
misuse) - Role Violation (
role_viol) - Toxicity (
toxic)