Unverified50% confidenceBenchmarkExact time
DeepMind在2022年发布Chinchilla论文，其中700亿参数的Chinchilla在多项基准上击败了2800亿参数的Gopher
1
Sources
50%
Confidence
Long-term
Relevance
7/11/2026
First Seen
Sources
Related Claims
VerifiedThe Chinchilla law proposed by DeepMind further refined the optimal ratio relationship between data and parameters.69% similarVerifiedDeepSeek-V3 adopts a Mixture of Experts (MoE) architecture with a total parameter count of 671 billion, but only approximately 37 billion parameters are activated per inference.62% similarUnverifiedAt the same parameter scale, DeepSeek Math's mathematical reasoning capability surpasses open-source models such as Llama, Qwen, and WizardMath.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/483255API
curl https://kongchang.com/api/v1/knowledge/claims/483255MCP
get_claim(id=483255)