Unverified50% confidenceFactExact time
When the student network's convolution kernel size matches the teacher's, only O(D) samples are needed to recover the teacher parameters; if the kernel size is larger, O(D log D) samples are needed (D being the feature dimension).
1
Sources
50%
Confidence
Long-term
Relevance
10/22/2026
First Seen
Sources
Related Entities
Related Claims
Unverified对于有限VC维d的假设类，O((d + log(1/δ))/ε)量级的样本即可实现PAC可学习60% similarUnverifiedSpeculative decoding improves inference speed by using a small draft model to generate candidate tokens that are verified in parallel by the main model; sparse attention reduces computational complexity from O(n²) to near O(n).56% similarUnverifiedDPO uses a mathematical transformation to equivalently rewrite the "reward maximization + KL constraint" objective as a simple classification loss, requiring no reward model and no reinforcement learning loop, and alignment can be completed with supervised learning alone.56% similarUnverifiedThe 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.55% similarUnverifiedGAD-RL weights the forward KL divergence using the student's predicted probability for the teacher's Top-1 token, suppressing local auxiliary updates when the student's support for the teacher's recommended token is low55% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1083911API
curl https://kongchang.com/api/v1/knowledge/claims/1083911MCP
get_claim(id=1083911)