Unverified50% confidenceBenchmarkExact time
In experiments across 11 language models and 70 relations, the linear readability of the original relation (linear probe accuracy) strongly predicted generalization performance after fine-tuning, with a Pearson correlation coefficient of 0.87 to 0.92 across relations within the same model.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAs the model scales from 8B to 70B, the predictive power of fixed output vector similarity for behavior actually worsens, with a paired mean correlation change of -0.080 (95% confidence interval [-0.127, -0.035]).69% similarUnverifiedRepresentational depth can predict the magnitude of behavioral effects, achieving a Spearman correlation of 0.761 (p < 0.001) across all 17 models.64% similarUnverifiedThe Pearson correlation coefficient r measures the strength of linear correlation between two continuous variables and ranges from -1 to 1: close to 1 indicates a strong positive correlation, and close to 0 indicates almost no linear relationship.63% similarUnverifiedIn a test covering 8 local models and more than 5000 calls, the wording style of the prompt (plain/normal/florid) had almost no effect on accuracy on non-math tasks, with differences of no more than 4 percentage points for models of 8B and above.63% similarUnverifiedIn the first round of experiments, features with verified capabilities averaged 1.43, while features without verified capabilities averaged 1.14, a negligible gap; after fixing the language issues, the same comparison became 2.00 versus 1.07.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1041964API
curl https://kongchang.com/api/v1/knowledge/claims/1041964MCP
get_claim(id=1041964)