Unverified50% confidenceFactExact time
With RAG, when policy documents are updated, only re-ingestion is required—no model retraining is needed.
1
Sources
50%
Confidence
Long-term
Relevance
10/6/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedRAG enables questions about the latest information to be answered simply by updating the external knowledge base, without retraining the model.79% similarUnverifiedThis update no longer resends tool definitions when switching models using the /model command.63% similarUnverifiedLLM在预训练完成后权重被固化，无法通过新数据自动更新，更新必须重新训练或微调61% similarUnverifiedThis version follows the PATCH update strategy in Semantic Versioning, introducing no breaking changes and remaining backward compatible.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/981457API
curl https://kongchang.com/api/v1/knowledge/claims/981457MCP
get_claim(id=981457)