Unverified50% confidenceFactExact time
The loss trajectory correlation coefficient usually refers to the Pearson correlation coefficient; 0.99 means the two curves move almost entirely in the same direction, but the correlation coefficient does not reflect absolute differences—even with a fixed offset it can still be close to 1.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe load balancing loss uses the sum of squares of all expert selection frequencies L=ΣFi², and by the Cauchy–Schwarz inequality the loss is minimized when all Fi=1/N.66% similarUnverifiedWith a sample size of 112 questions, the 95% confidence interval near 0.95 is approximately ±4 points, making it impossible to distinguish between scores of 0.94 and 0.92.65% similarUnverifiedFine-tuned DistilBERT achieves F1=0.9993 on in-distribution testing, but loses 12.9 mean average precision points under topic shift, a greater degradation than the linear model's 5.2 points.65% similarUnverifiedTotal variation distance is defined as the maximum difference in probability between two distributions on some event, with a range from 0 to 1; 0 means the distributions are completely identical, and 1 means they do not overlap at all.65% similarUnverifiedIn the fixed-vector self-similarity experiment, when the offset goes from 0 to 1024, the mean of the raw scores always fluctuates within ±0.38, while the mean of the absolute values only drops from 3.17 to 3.02, a change of about 4%.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1032396API
curl https://kongchang.com/api/v1/knowledge/claims/1032396MCP
get_claim(id=1032396)