Deepseek-v3 : Deepseek MOE Architecture — Part 2
Last Updated on October 6, 2026 by Editorial Team
Author(s): Prachi rise
Originally published on Towards AI.
Deepseek-v3 : Deepseek MOE Architecture — Part 2
This is the full series of Deepseek-V3 technical report, where i explain all the technical details in simpler words with code implementation and explanation.
The article continues by motivating why MoE is needed instead of sending every token through the same large FFN, then introduces the core idea of routing tokens to a small subset of expert FFNs. It explains DeepSeek-V3’s specific MOE design with two expert types (shared experts always active, plus routed experts selected by a router using Top‑K), and defines key routing terminology (shared/routed expert counts, activated experts K, gating and affinity scores, centroids, etc.). The piece then walks through how router affinity is computed, how Top‑K experts are selected and normalized, and provides code-style implementation sketches for FFN and an MoE layer. It further addresses the load-balancing problem (uneven token distribution across experts) and describes DeepSeek-V3’s auxiliary-loss-free approach using expert-specific routing bias, followed by a complementary sequence-wise auxiliary loss to avoid extreme imbalance within each sequence. It ends with a conclusion summarizing the two expert types and the two complementary load-balancing mechanisms, plus a combined code outline that incorporates the balancing loss.
Read the full blog for free on Medium.
Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.
Published via Towards AI
Towards AI Academy
We Build Enterprise-Grade AI. We'll Teach You to Master It Too.
15 engineers. 100,000+ students. Towards AI Academy teaches what actually survives production.
Start free — no commitment:
→ 6-Day Agentic AI Engineering Email Guide — one practical lesson per day
→ Agents Architecture Cheatsheet — 3 years of architecture decisions in 6 pages
Our courses:
→ AI Engineering Certification — 90+ lessons from project selection to deployed product. The most comprehensive practical LLM course out there.
→ Agent Engineering Course — Hands on with production agent architectures, memory, routing, and eval frameworks — built from real enterprise engagements.
→ AI for Work — Understand, evaluate, and apply AI for complex work tasks.
Note: Article content contains the views of the contributing authors and not Towards AI.
Related posts
Recent Posts
