Unverified60% confidenceFactExact time
Hy4's activated parameters account for approximately 6.4% of its total parameter count.
2
Sources
60%
Confidence
Long-term
Relevance
8/31/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedHy3 adopts a Mixture of Experts (MoE) architecture with a total parameter count of 295 billion (295B), activating only 21 billion (21B) parameters per inference.68% similarUnverifiedV4-Pro的总参数与活跃参数比例约为32:1的稀疏比68% similarUnverifiedW4A4量化将权重和激活值均压缩至4位整数，理论上相比FP16可将显存占用降低至约1/467% similarUnverifiedMoE模型中'35B'指总参数量，每个token只激活少量Expert（通常2-4个），实际计算量可能仅相当于6-7B的Dense模型65% similarUnverifiedIn speculative decoding, the draft model's parameter count is typically about 1/10 to 1/5 of the target model, and the candidate count k is typically 4-8.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/828813API
curl https://kongchang.com/api/v1/knowledge/claims/828813MCP
get_claim(id=828813)