Unverified50% confidenceFactExact time
SVD将任意m×n矩阵A分解为U·Σ·V^T，图像压缩通过只保留前k个最大奇异值实现最优低秩逼近（Eckart-Young定理）
1
Sources
50%
Confidence
Long-term
Relevance
8/5/2026
First Seen
Sources
Related Claims
UnverifiedSVDQuant将权重矩阵分解为高精度（16-bit）的低秩分支和4-bit量化的残差分支80% similarUnverified低秩分解将m×n权重矩阵近似为A(m×r)与B(r×n)的乘积，存储参数量从m×n降至r×(m+n)69% similarUnverifiedMLA theoretically reduces the GPU memory footprint of KV Cache to 5% to 13% of standard MHA by compressing KV matrices into low-dimensional latent vector caches.67% similarUnverifiedMLA将高维K、V矩阵投影至低秩潜在空间，压缩比可达8-16倍63% similarVerifiedMLA compresses KV pairs into a low-dimensional latent space for caching and recovers the full Keys and Values via an up-projection matrix during inference, compressing the KV cache by several times.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/692484API
curl https://kongchang.com/api/v1/knowledge/claims/692484MCP
get_claim(id=692484)