Unverified85% confidenceFactTime unknown
Google翻译团队2006年的论文展示了类似于后来被称为Scaling Law的现象——随着数据量和计算量增加，模型质量呈现规律性提升
1
Sources
85%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
Unverified大语言模型随参数规模增长突然涌现出思维链推理能力，由Google的涌现能力论文系统记录72% similarUnverified谷歌翻译早期的重大突破很大程度上依赖海量平行语料库（双语成对文本）62% similarUnverifiedGoogle's large-scale codebase maintenance practices show that continuous small-scale refactoring is more effective than centralized rewrites.62% similarUnverifiedThe ReAct pattern originates from a 2022 paper by Google/Princeton61% similarUnverified谷歌于2017年通过Shazeer等人的Sparsely-Gated MoE论文将MoE规模化至语言模型61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/22942API
curl https://kongchang.com/api/v1/knowledge/claims/22942MCP
get_claim(id=22942)