Unverified50% confidenceFactExact time
在NTK框架下训练损失以指数速率e^{-2ηλ_min t}衰减，收敛速度由核矩阵最小特征值λ_min决定
1
Sources
50%
Confidence
Long-term
Relevance
8/22/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedRNN及其变体LSTM和GRU按时间步依次处理序列，存在梯度指数级衰减导致长距离依赖捕捉困难，以及无法并行化训练的缺陷65% similarUnverifiedNeRF训练耗时长、推理速度慢，输出的隐式表示无法直接转化为可编辑的几何网格64% similarUnverifiedNaïve post-training quantization causes individual weight errors to accumulate and amplify across multiple layers—a phenomenon known as compounding error—and at 2-bit or 1-bit precision the model structure collapses63% similarUnverifiedRNN存在两大缺陷：无法并行计算导致速度慢，以及长距离依赖信息衰减的健忘问题62% similarUnverifiedThe core loss functions of contrastive learning (such as InfoNCE and NT-Xent) essentially pull the representations of similar samples closer together while pushing the representations of dissimilar samples further apart.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/787197API
curl https://kongchang.com/api/v1/knowledge/claims/787197MCP
get_claim(id=787197)