Unverified80% confidenceFactTime unknown
MiniMax的模型架构采用了混合专家（MoE）与线性注意力的创新组合
1
Sources
80%
Confidence
Long-term
Relevance
5/31/2026
First Seen
Sources
Related Entities
Related Claims
Unverified近年涌现出混合专家模型（MoE，如 Mixtral）、状态空间模型（SSM，如 Mamba）、线性注意力变体等新兴架构71% similarUnverifiedMeta的LLaMA和微软的Phi系列开源小模型证明精心设计的小模型在特定任务上可接近大模型表现69% similarUnverifiedMost current large models are based on the MoE (Mixture of Experts) architecture.68% similarUnverifiedThe core innovation of the Mixture of Experts (MoE) architecture lies in conditional computation, whereby a gating network dynamically selects a small number of experts to participate in computation rather than activating all parameters.68% similarUnverifiedLoRA fine-tuning is one of the mainstream workflows for model fine-tuning.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/13442API
curl https://kongchang.com/api/v1/knowledge/claims/13442MCP
get_claim(id=13442)