Unverified50% confidenceFactExact time
The vast majority of large model training data comes from human publicly available knowledge such as web text, books, code, and forum discussions.
1
Sources
50%
Confidence
Long-term
Relevance
9/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedTraining data in the instruction tuning phase contains a large volume of creative writing, business email templates, and role-playing dialogue samples, enabling the model to master various textual formats76% similarUnverifiedThe Ziyue large model made extensive use of educational corpora such as textbooks, exercises, and problem-solving processes during the pre-training and fine-tuning stages, giving it stronger professionalism on educational tasks.74% similarUnverifiedThe capabilities and behavior of large language models are deeply dependent on training data.73% similarUnverified对于小说家工具这样的应用型产品，直接基于大模型API构建可能比自己训练模型更明智70% similarUnverifiedCommon Crawl数据集包含了大量Reddit页面，是多数大模型训练的基础语料之一70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/911091API
curl https://kongchang.com/api/v1/knowledge/claims/911091MCP
get_claim(id=911091)