Unverified50% confidenceFactExact time
绝大多数模型的预训练语料以数千至数万token的短中等长度文本为主，超过10万token的训练样本极为稀缺
1
Sources
50%
Confidence
Long-term
Relevance
7/16/2026
First Seen
Sources
Related Claims
UnverifiedMid-training (continued pre-training) is a training stage between pre-training and fine-tuning, with data volumes typically reaching hundreds of billions to trillions of tokens.65% similarUnverifiedThe training data volume of current mainstream large models has reached tens of trillions, or even sixty trillion, tokens.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/530706API
curl https://kongchang.com/api/v1/knowledge/claims/530706MCP
get_claim(id=530706)