Unverified85% confidenceFactTime unknown
MoE架构最早由Google在2017年提出，后被Mistral的Mixtral系列和阿里的Qwen系列等广泛采用
1
Sources
85%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedMoE概念最早由Jacobs等人于1991年提出，谷歌2017年将其规模化至语言模型，Mistral的Mixtral系列将稀疏MoE带入开源社区73% similarUnverifiedMoE架构由 Google Brain 于 2017 年提出，GPT-4、Mixtral 8x7B 等知名模型均采用该架构72% similarVerifiedGoogle systematized the Mixture of Experts architecture design in its 2017 Sparsely-Gated MoE paper.68% similarVerifiedThe MoE architecture was proposed by Shazeer et al. in 2017 in the paper "Outrageously Large Neural Networks" and subsequently validated at scale by Google in Switch Transformer (2021)66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/16882API
curl https://kongchang.com/api/v1/knowledge/claims/16882MCP
get_claim(id=16882)