Unverified50% confidenceFactExact time
In the Transformer, the Q, K and V matrices are obtained by multiplying the hidden representations of the input sequence by three learnable weight matrices W_Q, W_K and W_V respectively.
1
Sources
50%
Confidence
Long-term
Relevance
10/31/2026
First Seen
Sources
LLM Inference Bottlenecks and KV Cache: Core Mechanisms and Mainstream Optimization Approaches
bilibili先进编译实验室10/20/2026
Related Entities
Related Claims
UnverifiedQ-value updates follow the iterative form of the Bellman equation: Q(s,a) ← Q(s,a) + α[r + γ·max Q(s',a') - Q(s,a)].62% similarUnverifiedMultiplying Q by the transpose of K yields an N × N correlation matrix, where each element represents the strength of the relationship between two words.62% similarUnverified典型的导纳模型可表达为 M·ẍ + B·ẋ + K·x = F_ext，其中M、B、K分别是期望的虚拟惯量、阻尼和刚度矩阵60% similarUnverifiedFor a frozen base matrix W0, LoRA represents the learned update as W' = W0 + (alpha/r)BA, where A and B are trainable compact matrices with rank r much smaller than the dimensions of the original matrix.60% similarUnverifiedAttention 采用 QKV 框架，每个 Token 的 Embedding 经三个线性变换矩阵生成 Query、Key、Value 向量59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1121452API
curl https://kongchang.com/api/v1/knowledge/claims/1121452MCP
get_claim(id=1121452)