Unverified50% confidenceFactExact time
谷歌发布的mBERT首次展示了一个模型可以通过共享参数在104种语言上同时进行表示学习，并表现出零样本跨语言迁移能力
1
Sources
50%
Confidence
Long-term
Relevance
7/16/2026
First Seen
Sources
Related Claims
Unverified谷歌近期发布了 DiffusionGemma 技术报告，将扩散生成机制引入语言建模65% similarUnverifiedSpeculative decoding was formally proposed by Google in 2023, using a draft model with 1/10 to 1/50 the parameters of the target model to generate multiple candidate tokens at once, which are then verified in parallel by the large model.64% similarUnverifiedSentencePiece由Google于2018年推出，支持无监督训练，无需预先进行空格分词，Google的T5、LLaMA等模型均采用其Unigram语言模型算法64% similarUnverifiedGoogle's Universal Speech Model (USM) supports more than 100 languages.64% similarUnverifiedMUM于2021年由谷歌提出，是BERT的继任者，参数量约为BERT的1000倍，支持75种语言的跨语言理解63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/533737API
curl https://kongchang.com/api/v1/knowledge/claims/533737MCP
get_claim(id=533737)