←visualization / Kimi K3

Kimi K3 · Educational Architecture

KDA, Gated MLA, LatentMoE, AttnRes

◆ HYBRID ARCHITECTURE

n_embd

n_head

n_layer

seq_lenT=8

selected layeriKDA 1

macrocyclesi2

comparison focusiBalanced

AttnResiVisible

🔍 Variable Inspector

click any code line

Click a code line to inspect its variables.

About this lesson

Quick answer

Explore a reduced educational Kimi K3 macrocycle combining KDA, Gated MLA, Stable LatentMoE, SiTU, and blockwise AttnRes.