Unverified50% confidenceFactExact time
LLM的行为本质上是统计性的，同一输入在不同上下文中可能产生不同输出，使得传统基于规则的WAF难以有效防护
1
Sources
50%
Confidence
Long-term
Relevance
8/12/2026
First Seen
Sources
Related Claims
UnverifiedLLM输出具有非确定性，同一输入在不同时间可能产生不同输出，因此评估流水线常用LLM-as-Judge方法用另一个强大模型评判目标模型输出质量73% similarUnverifiedAlthough LLMs do not directly output normalized probability vectors, decision confidence can be approximately estimated through various engineering methods.73% similarUnverifiedLLM outputs are probabilistic—the same specification may produce drastically different behaviors across different runs—and therefore declarative specifications for AI Agents more closely resemble constraint declarations rather than state declarations.73% similarVerifiedThe output of LLMs is inherently non-deterministic; even with the same Prompt as input, the model may return results with different formats, content, or even logic at different times.73% similarUnverifiedLLM 的输出非确定性根源于自回归生成机制，即使输入相同，Temperature、Top-p 等采样参数也会导致不同输出72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/736775API
curl https://kongchang.com/api/v1/knowledge/claims/736775MCP
get_claim(id=736775)