Unverified50% confidenceFactExact time
The final trained model has approximately 400 million (0.4B) parameters, targets Bengali as the primary language, and uses a corpus of approximately 2 billion tokens.
1
Sources
50%
Confidence
Long-term
Relevance
9/23/2026
First Seen
Sources
Related Entities
Related Claims
Verified大语言模型通过在数千亿乃至数万亿个词元（Token）的文本数据上进行自监督预训练71% similarUnverifiedQwen的参数规模从70亿到数千亿不等，训练数据涵盖中英文及多语言语料70% similarUnverifiedBefore working with large models, the largest NLP and speech recognition models Lao Xiao's team built had only 4,000 parameters.69% similarUnverifiedThe Qwen team adopted a four-stage training process, with the first stage establishing foundational language capabilities on approximately 36 trillion tokens of multilingual corpus.69% similarUnverifiedAfter training on tens of trillions of tokens, language models exhibit emergent capabilities such as reasoning, creative writing, and programming.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/942815API
curl https://kongchang.com/api/v1/knowledge/claims/942815MCP
get_claim(id=942815)