Unverified50% confidenceSolutionExact time
The core limitations of small language models are insufficient knowledge density, weak generalization capability, and an extremely high degree of dependence on prompts.
1
Sources
50%
Confidence
Long-term
Relevance
9/6/2026
First Seen
Sources
Related Entities
Related Claims
Unverified大语言模型存在三大缺陷：知识陈旧（存在截止日期，无法获取实时数据）、缺乏私有数据、过于被动77% similarVerified大语言模型推理的瓶颈往往不在计算能力，而在于内存带宽75% similarVerified大语言模型在长时间对话中会因上下文窗口限制而丢失早期信息74% similarUnverifiedText generated by large language models tends to have low perplexity, with word choices that are overly consistent with probabilistic expectations, lacking the unexpectedness of human writing.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/864624API
curl https://kongchang.com/api/v1/knowledge/claims/864624MCP
get_claim(id=864624)