Unverified50% confidenceSolutionExact time
Combining RAG with fine-tuning and retrieving similar precedents as context can reduce the pressure on the model to generate content from scratch.
1
Sources
50%
Confidence
Long-term
Relevance
9/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedRAG工作流通常涉及两类模型：嵌入模型（将文本转换为向量用于相似度检索）和对话大模型（综合检索结果生成答案）70% similarUnverified嵌入模型的选择对 RAG 系统性能的影响甚至超过生成模型70% similarUnverifiedWhen fine-tuning embedding models, hard negatives (such as positions in the same domain but at different levels) should be used instead of random negatives, in order to learn more fine-grained semantic differences.66% similarUnverified更聪明的模型能在更少的推理步骤内完成任务，减少冗余token输出，这被称为「token压缩效应」，是新一代模型降本的核心来源之一66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/910241API
curl https://kongchang.com/api/v1/knowledge/claims/910241MCP
get_claim(id=910241)