Unverified50% confidenceFactExact time
The flagship version of Llama 3 was pretrained on 15.6 trillion tokens, with roughly 50 times the compute of the largest Llama 2.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/17/2026
First Seen
Valid until: 1/15/2027
Sources
Related Entities
Related Claims
UnverifiedA 70-billion-parameter model corresponds to approximately 1.4 trillion training tokens under the Chinchilla optimum69% similarUnverifiedThe vocabulary size of modern LLMs ranges between 50,000 and 200,000.58% similarUnverified训练与美国同等规模模型需要五万张顶级英伟达卡或二十万张华为95055% similarUnverifiedMaia引擎的1亿+训练样本来源于Lichess公开对局数据55% similarUnverifiedMeta used more than 16,000 H100s to train the Llama 3 series, and leading institutions' training clusters generally consist of tens of thousands of GPUs or more.54% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1049789API
curl https://kongchang.com/api/v1/knowledge/claims/1049789MCP
get_claim(id=1049789)