Unverified50% confidenceFactExact time
Local equivalent weights are constructed by matching the first-order sensitivity of the plug-in post-stratification estimator with the first-order sensitivity of a weighted average fitted response mean, both differentiated with respect to the latent mean scale vector.
1
Sources
50%
Confidence
Long-term
Relevance
10/15/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedCapability evaluations focus on systematically measuring the upper bound of a model's performance in specific dangerous capability domains.75% similarUnverifiedMeasurement Sensitivity refers to whether endpoint metrics change after completely rebuilding the evaluation plus inference configuration while keeping the same model identifier.70% similarUnverifiedGraftOOD adopts a shared velocity-regression objective function, combining two types of supervision signals: group-relative reward correction and sequence-level example supervision.68% similarUnverified'Whether the extraction is faithful to the source text' and 'whether the signal has predictive power' should be separated into two independent evaluation dimensions: the former belongs to a short feedback loop, while the latter requires long-cycle attribution using delayed labels.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1032982API
curl https://kongchang.com/api/v1/knowledge/claims/1032982MCP
get_claim(id=1032982)