Unverified50% confidenceBenchmarkExact time

DeepMind、Meta、Mistral等实验室已在多个基准测试中证明，高质量合成数据可以大幅减少对人工标注的依赖

1

Sources

50%

Confidence

Long-term

Relevance

7/13/2026

First Seen

Sources

Mercor Races Toward a $20B Valuation: The Capital Logic Behind the AI Data Labeling Race

rss7/9/2026

Related Claims

UnverifiedDeepMind的AlphaCode 2和Meta的多智能体代码生成研究验证了用并行化分摊认知负担、缩短挂钟时间的路径有效性70% similar VerifiedDeepSeek V4 Pro采用了MoE（Mixture of Experts，混合专家）架构，在推理时只激活部分参数以降低计算成本68% similar UnverifiedDeepSeek-R1的推理能力在多个基准测试中表现惊艳，训练成本远低于国际竞品66% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/499321

API

curl https://kongchang.com/api/v1/knowledge/claims/499321

MCP

get_claim(id=499321)