AuditEvaluation#
- class scikitplot.agent.AuditEvaluation(runs, summary)[source]#
Per-run summaries and aggregate rates.
- Parameters:
- runstuple of RunSummary
Every run in the trail, in order of first appearance.
- summarydict
Over top-level runs (sub-runs excluded):
n_runs,n_ended,completion_rate(completed / ended),stop_reasons,mean_decisions,mean_tool_calls,denial_rate(denied / steps),tool_failure_rate(failed / (executed + failed)),approval_grant_rate(granted / answered),total_cost,total_tokens; a rate with no denominator isNone.
- Parameters:
runs (tuple[RunSummary, ...])
- runs: tuple[RunSummary, ...][source]#