Unverified50% confidenceFactExact time
2017年谷歌发表《Outrageously Large Neural Networks》，系统验证了稀疏MoE在Transformer结构中的有效性
1
Sources
50%
Confidence
Long-term
Relevance
7/23/2026
First Seen
Sources
Kimi K3领跑开源模型，AI格局加速重塑
Related Claims
UnverifiedGoogle Brain于2017年在《Outrageously Large Neural Networks》论文中将MoE规模化应用于Transformer框架84% 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)82% similarUnverifiedSince the Transformer architecture was proposed by Google in 2017, its self-attention mechanism has a computational complexity that grows quadratically with sequence length77% similarUnverifiedThe Transformer was released by Google in 2017, and its self-attention mechanism allows the model to dynamically compute the relevance weights between a token and all other tokens in the sequence.73% similarVerifiedThe Transformer architecture was proposed by Google in 2017.73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/598593API
curl https://kongchang.com/api/v1/knowledge/claims/598593MCP
get_claim(id=598593)