Unverified50% confidenceFactExact time
Accuracy only checks whether the highest-probability option is correct, while NLL also measures the probability assigned to the correct option, so NLL can worsen while accuracy rises.
1
Sources
50%
Confidence
Long-term
Relevance
10/18/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedCalibration error should be weighted by the consequences of decisions within the uncertainty interval, not simply averaged.72% similarUnverifiedDenser relevance judgments reduce the probability that correct results retrieved by a model are mistakenly judged as irrelevant, thereby improving evaluation reliability.71% similarUnverified精确数值计算需要严格遵循确定性规则，这与基于统计概率的token预测机制存在根本性矛盾70% similarUnverifiedThe accuracy of multi-hop reasoning is highly dependent on the reliability of low-hop (especially single-hop) factual foundations.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1064011API
curl https://kongchang.com/api/v1/knowledge/claims/1064011MCP
get_claim(id=1064011)