Unverified50% confidenceFactExact time
DeepSeek's fine-grained expert splitting innovation increases the number of experts from N to 2N or more, expanding the routing selection combination space from C(8,2)=28 combinations to C(64,6)=over 74.61 million combinations.
1
Sources
50%
Confidence
Long-term
Relevance
9/9/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedDeepSeek's key innovation lies in its fine-grained expert splitting strategy, which breaks down the larger experts in traditional MoE into a greater number of smaller units.77% similarUnverifiedDeepSeek is developing a larger frontier model, aiming to compete at the trillion-parameter scale.69% similarUnverified现代深度网络往往拥有数十亿乃至上万亿参数，远超训练样本数量，按经典统计学习理论应严重过拟合，实际上却展现出惊人的泛化能力65% similarVerifiedDeepSeek-V3 has 671B parameters but activates only 37B, adopting an architecture of 256 routed experts plus 1 shared expert, with each token processed by 8 selected experts.65% similarUnverifiedCompared with the US, DeepSeek lags by more than an order of magnitude in compute.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/885821API
curl https://kongchang.com/api/v1/knowledge/claims/885821MCP
get_claim(id=885821)