Unverified80% confidenceFactTime unknown
仅仅改变输入文本的措辞和结构，就能让同一个模型的性能产生数十个百分点的差异
1
Sources
80%
Confidence
Long-term
Relevance
5/31/2026
First Seen
Sources
Related Entities
Related Claims
Unverified同一大语言模型在不同硬件上可能产生不同推理结果，甚至同一硬件上重复运行也可能出现微小差异71% similarUnverifiedWord的Track Changes（修订痕迹）功能每处修改需同时保存原始内容与修订内容并附加作者、时间戳等元数据，使文档结构高度嵌套70% similarUnverified代码的语义更多体现在结构关系而非字面文本，一个函数改动的影响范围必须沿调用图遍历而非仅靠语义相似度捕获70% similarUnverifiedThe output of large language models is fundamentally probabilistic text generation, and a single rephrasing of a prompt can change field names, nesting levels, or enumeration values in the output without triggering compile-time or static analysis warnings.69% similarUnverifiedThe role of an embedding model is to convert text into high-dimensional vectors, enabling machines to measure semantic similarity in a mathematical way.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/14991API
curl https://kongchang.com/api/v1/knowledge/claims/14991MCP
get_claim(id=14991)