Unverified50% confidenceFactExact time
Meta's Llama series has made model weights and architecture public but has not disclosed the complete training dataset or curation process, and the license includes a restriction requiring separate authorization for monthly active users exceeding 700 million.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/6/2026
First Seen
Valid until: 12/5/2026
Sources
Related Entities
Related Claims
UnverifiedWhen Meta released the Llama series, it provided weights but did not disclose the complete training dataset and processing pipeline.79% similarUnverifiedMeta开源的Llama系列通过优化训练效率实现了在消费级显卡上运行70亿至700亿参数模型76% similarUnverifiedMeta Llama系列因未完整公开训练数据集，实际上属于开放权重而非真正开源72% similarUnverifiedMeta's Llama Guard series is a specially trained safety classification model used to perform multi-dimensional safety scoring on LLM output content.70% similarUnverifiedMeta在Llama系列最新迭代中开始引入信息密度作为训练目标之一69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/867138API
curl https://kongchang.com/api/v1/knowledge/claims/867138MCP
get_claim(id=867138)