Unverified50% confidenceDecision RuleExact time
The author gives reference thresholds for comparing two single runs: about 0.15 for faithfulness, about 0.08 for relevancy, about 0.09 for recall, and about 0.10 for precision (at the 18-question scale, and about 40% wider on a 9-question subset); only mean differences exceeding the threshold are worth taking seriously.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/19/2026
First Seen
Valid until: 1/17/2027
Sources
Related Entities
Related Claims
UnverifiedWhen only re-judging the same batch of saved answers, the standard deviation of the 18-question mean ranged from 0.021 (faithfulness) to 0.035 (precision); after regenerating answers, the standard deviation of faithfulness rose to 0.054, and relevancy rose from 0.002 to 0.028.78% similarUnverifiedDifferent papers often use faithfulness threshold values ranging from 70% to 95%, lacking a unified standard.73% similarUnverifiedCohen's Kappa coefficient is suitable for scenarios with two reviewers and categorical variables; its value ranges from -1 to 1, and a value above 0.6 is generally considered moderate agreement, while above 0.8 is considered high agreement.67% similarUnverifiedAccording to data examples from the LlamaIndex blog, setting the confidence threshold to 0.7 corresponds to approximately 95% precision.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1065192API
curl https://kongchang.com/api/v1/knowledge/claims/1065192MCP
get_claim(id=1065192)