Unverified50% confidenceFactExact time
Modern large model training is typically divided into three stages: pre-training, mid-training, and post-training.
1
Sources
50%
Confidence
Long-term
Relevance
9/12/2026
First Seen
Sources
Related Entities
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.72% similarVerifiedIn 2018, BERT introduced the two-stage paradigm of pre-training plus fine-tuning.70% similarUnverified主流大模型的训练数据通常有6-18个月的滞后期69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/911343API
curl https://kongchang.com/api/v1/knowledge/claims/911343MCP
get_claim(id=911343)