Verified75% confidenceFactExact time
The macro F1 score calculates the F1 score for each class separately and then takes the simple average, so it is not skewed by the large number of majority-class samples masking performance on minority classes.
3
Sources
75%
Confidence
Long-term
Relevance
9/18/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe F1 score uses the harmonic mean rather than the arithmetic mean because the harmonic mean is more sensitive to smaller values, so F1 is only high when both precision and recall are simultaneously high.72% similarUnverifiedThe compression ratio is calculated as (original context − compressed context) / original context; it is a ratio-type metric that is extremely sensitive to extreme samples, so the average is easily dominated by a few extreme values, and the median and quantile distribution should be examined.57% similarUnverifiedShampoo maintains second-order statistics separately for each dimension of each parameter tensor, avoiding the computation of the full Hessian.56% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/931156API
curl https://kongchang.com/api/v1/knowledge/claims/931156MCP
get_claim(id=931156)