Unverified50% confidenceFactExact time
M-BERT is a multilingual version of BERT jointly trained on Wikipedia corpora in 104 languages.
1
Sources
50%
Confidence
Long-term
Relevance
10/25/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedBERT 在英文维基百科和 BookCorpus(共约 33 亿词)上进行了掩码语言模型(MLM)预训练63% similarUnverifiedmBERT covers 104 languages, and XLM-RoBERTa covers 100 languages.61% similarUnverifiedLaBSE (Language-agnostic BERT Sentence Embedding) was proposed by Google in 2020 and supports 109 languages.59% similarVerifiedElevenLabs v4 supports 90 languages.57% similarUnverifiedWhisper uses a Byte Pair Encoding (BPE) tokenizer with a multilingual vocabulary of approximately 50,000 tokens, supporting 99 languages.57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1097410API
curl https://kongchang.com/api/v1/knowledge/claims/1097410MCP
get_claim(id=1097410)