Unverified50% confidenceFactExact time
Fin publishes pass^k, Cursor publishes keep rate, and Shopify publishes judge agreement.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/21/2026
First Seen
Valid until: 1/19/2027
Sources
Related Entities
Related Claims
UnverifiedThe GK summary (Greenwald-Khanna) uses counters with error bounds to provide deterministic approximate guarantees for quantiles.58% similarUnverifiedThe retrieval quality metrics reported by Truvec include Hit rate, Recall@k, Precision@k, MRR, nDCG@k, and F1@k, with a default Top-K of 5 and support for 1 to 50.56% similarUnverifiedGPT-6 Sol received a perfect score of 100 on the automatically graded Q3Close financial reconciliation task and 90 points on the Skepticism Audit code review task.55% similarUnverifiedThe component validation gate outputs three metrics: pass@1 (proportion passing validation on the first attempt), pass@repair (proportion mountable after one repair), and reject rate (proportion that should fall back to text).54% similarUnverifiedK折交叉验证的K通常取5或10，这在偏差-方差权衡上被实证为较优选择52% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1080130API
curl https://kongchang.com/api/v1/knowledge/claims/1080130MCP
get_claim(id=1080130)