Unverified50% confidenceFactExact time
DP-μP proposes a spectral scaling formula that combines the SNR condition with matrix spectral properties to give concrete scaling rules for different optimizers.
1
Sources
50%
Confidence
Long-term
Relevance
10/19/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe DP-μP framework derives a signal-to-noise ratio (SNR) condition that characterizes the relative strength of signal and noise at each layer under the DP constraint, and based on this determines how learning rates should scale across layers.75% similarUnverifiedThe standard approach to controlling for number width is to add width as a covariate in the regression model, or to analyze separately within subgroups of the same width.67% similarUnverified凸优化理论中的KKT条件是理解支持向量机SVM对偶问题以及RLHF中约束优化设计的必要工具66% similarUnverifiedThe paper introduces a design efficiency factor D* to summarize the way leverage is distributed among observations, enabling the new formula to capture structural differences ignored by traditional methods.64% similarVerifiedCommonly used calibration methods include Platt Scaling (fitting a sigmoid function using logistic regression) and Isotonic Regression (fitting a non-parametric monotonic mapping using isotonic regression).64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1068349API
curl https://kongchang.com/api/v1/knowledge/claims/1068349MCP
get_claim(id=1068349)