Unverified50% confidenceFactExact time
The output of large models is probabilistic and cannot be simply judged for equality, so they cannot be tested using traditional assertion methods.
1
Sources
50%
Confidence
Long-term
Relevance
9/24/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedBecause the output of large models is non-deterministic, simple assert statements cannot determine whether generated text is correct, which is why evaluations (evals) are needed.79% similarUnverified单次采样无法代表模型对某提示词的真实能力分布，3次以上独立采样才能提供足够的方差估计，是生产级提示词评估的最佳实践76% similarUnverifiedBecause large models themselves cannot strictly distinguish between instructions and data, prompt injection is fundamentally very difficult to completely eliminate.75% similarUnverified贝叶斯最优分类器理论上完美但实践中通常无法直接计算，因真实条件概率分布往往未知72% similarUnverified经典统计学习理论的框架并不能完全解释超大规模模型的行为72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/948504API
curl https://kongchang.com/api/v1/knowledge/claims/948504MCP
get_claim(id=948504)