Unverified60% confidenceFactExact time
The MoE architecture faces load balancing challenges, and researchers typically introduce auxiliary loss functions to mitigate the expert collapse problem, while teams such as DeepSeek have explored auxiliary-loss-free balancing strategies.
2
Sources
60%
Confidence
Long-term
Relevance
7/24/2026
First Seen
Sources
阿里开源2.4T Qwen大模型，中国AI席卷全球竞争格局
Related Claims
UnverifiedThe Top-K routing strategy in MoE is prone to causing a small number of experts to be overused (the expert collapse problem), which requires techniques such as auxiliary load-balancing loss functions to address.76% similarUnverified研究者引入辅助负载均衡损失来惩罚专家负载不均匀分布，DeepSeek系列还引入专家亲和度机制73% similarUnverifiedMoE kernels need to simultaneously address both compute regularization and load balancing, as uneven load across experts drags down overall utilization.71% similarUnverifiedMoE架构在长上下文场景下存在能力衰减问题，因路由机制在长序列下决策不稳定，连DeepSeek也未完全解决68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/604723API
curl https://kongchang.com/api/v1/knowledge/claims/604723MCP
get_claim(id=604723)